Atomic Force Microscopy (AFM) with applications ...

Atomic Force Microscopy (AFM) with applications spanning from nanoscale (bio)catalysis and characterization of molecularly ordered nanostructures to molecular biology Konstantin Balashev, PhD Sofia University “St. Kliment Ohridski”

Thesis for acquiring the degree DSc. in Chemistry (subject Physical Chemistry) 2013

Preface This thesis is presented for acquiring the degree Doctor of Sciences in Chemistry (subject Physical Chemistry) from the University of Sofia. It shears my experience with the Atomic Force Microscopy (AFM) and some other experimental methods utilized in various areas of (Bio)Physical chemistry starting fifteen years back when I joined the Nano Science Center (then the Center for Interdisciplinary Study of Molecular Interactions (CISMI)) at University of Copenhagen as a post doc in Prof. Thomas Bjørnholm group. At this “Danish period” when I was making my first steps in acquainting AFM with excitement and clumsiness I used to bother too many people with awkward questions, experimental failures etc. Although at the end of our mutual scientific journey then all we were able to achieve not so bad results. So I decided to summarize them in this thesis and everybody involved has a fair shear for their publishing elsewhere. Particularly, I want to acknowledge the help of Dr. Lars Kildemark Nielsen (Radiometer A/S) who thought me a lot about AFM as well as my tutor Thomas who gave me the very first instructions on how to set the AFM apparatus in action. Some other Danish folks whom I am especially indebted are Dr. Kristian Kjaer (Risø) and Prof. Torben Rene Jensen (Aarhus University). From them I learned a lot about X- ray scattering methods when we all had to spend hundreds of sleepless hours at HASY lab Synchrotron facilities in Hamburg. Most of the X-ray data summarized in this thesis and used as a basis for particular comparison with AFM data are due to their exceptional professionalism. Here I want to underline that all the results about AFM lipid-lipase investigations and X-ray studies were performed either in Copenhagen or Hamburg. I also decided to include in the thesis results from this period reported in a paper [# 4 from the list of publications involved in the thesis] and the reason is that this was one of the rarest and most valuable of my AFM results which I was ever able to obtain showing that in certain circumstances the atomic resolution can be achieved by AFM. After my habilitation as associate professor at Faculty of chemistry and Pharmacy at the University of Sofia and my assignment as Head of Laboratory for Atomic Force Microscopy which was established together with late Assoc. prof. Ceco Dushkin, Prof. Boryan Radoev and the Dean of Faculty of Chemistry and Pharmacy Prof. Toni Spasov I expanded my research interests in some different areas spanning within polymer chemistry and nanotechnology where my AFM expertise was needed. The summarized results from this period were almost without exception published in the last three years. In relation to that here I have to acknowledge some people hoping not to miss somebody. First comes my PhD student Peter Georgiev together with whom after many experiments followed by a persistent and restless debate with peer reviewers ultimately we were able to convince them of the advantages of AFM when used as a tool for studying kinetics of gold nanoparticle’s growth. I want to acknowledge some people from dept. of Applied Organic Chemistry- namely Prof. George Georgiev, Assoc. prof. Elena Kamenska, Dr. Silvia Simeonova and Dr. Bistra Kostova (Medical University, Sofia) who gave me an opportunity for exploring together polymer nanoparticles and their applications as drug delivery systems. I want also to thank Prof. Peter Kralchevsky, Assoc. prof. Teodor Gurkov and their PhD student Rumyana Stanimirova from the dept. of Chemical Engineering for shearing their knowledge about ubiquitous properties of the protein Hydrophobin. Also many thanks to Assoc. prof. George Miloshev and Dr. Milena Georgieva (IMB-BAS) for our mutual AFM work on the organization of yeast chromatin. I left those last lines just to emphasize my deepest gratitude to most important people to whom I am entirely indebted for my scientific career and personal development- my PhD tutor Prof. Ivan Panaiotov and Assoc. prof. Tzvetanka Ivanova. I want to thank to the other people from the Lab. of Biophysical Chemistry- Prof. Christian Vassilieff, Dr. Kristina Mircheva, Dr. Ivan Minkov and last but not least my office mate Dr. Nickolai Grozev. Also many thanks to all folks from the Dept. of Physical Chemistry for making such a friendly and creative working atmosphere.

1

Contents Introduction 1. Atomic Force Microscopy (AFM)-Methodology, theoretical and experimental considerations

5 7

1.1.

The principle of the Atomic Force Microscope

1.2.

Imaging in liquid environment

12

1.3.

Operating AFM modes

13

8

1.3.1. Contact Mode of Imaging

13

1.3.2. Oscillating (non-contact and intermittent) AFM modes

14

1.3.3. Force spectroscopy mode

16

1.3.4. Electric force microscopy and scanning Kelvin Force Microscopy

18

1.4.

Image interpretation

20

2. Monolayers at Fluid Interfaces and Langmuir- Blodgett filmsStudying the molecular organization (e.g. biological membranes) by monolayers and bilayers as model systems with experimental methods complimentary to the AFM.

23

3. Lipases- Phospholipase A2 (PLA2) and Humicola lanuginose lipase (HLL) - biological importance and experimental approaches of studying lipolytic enzyme reactions at interfaces.

38

3.1.

Lipases- their biological meaning

38

3.2.

Experimental approaches of studying lipolytic enzyme reactions at interfaces

40

3.3.

Lipase hydrolysis of supported lipid bilayers: an Atomic Force Microscopy Approach.

43

4. Phospholipase A2 (PLA2) and Humicola lanuginose lipase (HLL) captured in act by AFM.

50

4.1.

Influence of product phase separation on PLA2 hydrolysis of supported DPPC bilayers studied by AFM

51

4.2.

Kinetics of degradation of DPPC bilayers as a result of vipoxin PLA2 activity

63

2

4.3.

Hydrolysis of 1-mono-oleoyl-rac-glycerol (MOG) by Humicola lanuginose lipase (HLL) at the lipid–water interface observed by AFM

74

4.4.

AFM visualization of lipid bilayer degradation due to simultaneous action of PLA2 and HLL

84

4.5.

Surface sensitive synchrotron X-ray scattering as complementary method of AFM for studying lipids and lipases and their mutual interaction at interfaces

94

5. The protease Savinase captured in act by AFM.

102

5.1.

Savinase action on Bovine Serum Albumin (BSA) monolayers demonstrated with measurements at the air-water interface and liquid AFM imaging

102

5.2.

Savinase proteolysis of insulin Langmuir monolayers studied by surface pressure and surface potential measurements accompanied by atomic force microscopy (AFM) imaging

107

6. AFM applications for studying the lateral organization and structure of Langmuir- Blodgett (LB) Films.

118

6.1.

LB films of amphiphilic Hexa-peri-hexabenzocoronene (HBC): new phase transitions and electronic properties controlled by surface pressure studied by AFM KFM and GIXD.

118

6.2.

AFM imaging of hydrophobin II (HFBII) layers transferred from air/water interface on mica solid supports

127

7. AFM applications of for studying metal and polymer nanoparticles.

131

7.1.

Implementing Atomic Force Microscopy (AFM) for studying kinetics of gold nanoparticle’s growth

131

7.2.

Applications of AFM for characterization of polymer nanoparticles.

144

7.2.1. AFM characterization of novel drug delivery nanoparticles based on poly-zwitterionic copolymers

144

7.2.2. AFM characterization of polyzwitterionic copolymer nanoparticles loaded in situ with Metoprolol Tartrate: Synthesis, morphology and drug release properties

148

8. Applying Atomic Force Microscopy (AFM) for studying the chromatin organization of Saccharomyces cerevisiae.

153

3

Conclusions and summary of the main results

157

APPENDIX A

160

Experimental procedures for studding the influence of product phase separation on PLA2 hydrolysis of supported DPPC bilayers studied by AFM. APPENDIX B

162

Experimental procedures for studying kinetics of degradation of DPPC bilayers as a result of vipoxin PLA2 activity. APPENDIX C

163

Experimental details about hydrolysis of 1-mono-oleoyl-rac-glycerol (MOG) by Humicola lanuginose lipase (HLL) at the lipid–water interface observed by AFM APPENDIX D

164

Experimental details about AFM visualization of lipid bilayer degradation due to simultaneous action of PLA2 and HLL APPENDIX E

165

Experimental details about Savinase action on Bovine Serum Albumin (BSA) monolayers demonstrated with measurements at the air-water interface and liquid AFM imaging APPENDIX F

167

Experimental details about Savinase proteolysis of insulin Langmuir monolayers studied by surface pressure and surface potential measurements accompanied by atomic force microscopy (AFM) imaging APPENDIX G

169

Experimental details about LB films of amphiphilic Hexa-peri-hexabenzocoronene (HBC): new phase transitions and electronic properties controlled by surface pressure studied by AFM KFM and GIXD. APPENDIX H

172

Experimental details about Implementing Atomic Force Microscopy (AFM) for studying kinetics of gold nanoparticle’s growth

Publications which are wholly or partly involved in the thesis

174

REFERENCES

178

4

Introduction This thesis is dedicated to Atomic Force Microscopy (AFM) and has the motivation to present, new experimental approaches within different fields of (bio)physical chemistry where this technique is involved as well as to critically compare the experimental advantages and shortcomings of AFM over other techniques when exploring physical phenomena and molecular events on nanoscale level. The whole material is divided into eight chapters as the first three of them are introductory where in a coherent way are explained the principles of Atomic Force Microscope considered both schematically and experimentally and are presented also other experimental methods complimentary to AFM which are involved in studying of molecular organization and reactions occurring at nanoscale level at monolayer or bilayer interfaces. A substantial part of the thesis is devoted to lipolytic enzyme reactions in which the lipases, Phospholipase A2 (PLA2) and Humicola lanuginose lipase (HLL) play a key role thus in the last chapter of this introductory part shortly are explained the biological importance and various experimental approaches for studying certain lipolytic reactions at interfaces where these two enzymes take part in the catalytic act. Right after the introductory part in chapter 4 is demonstrated how PLA2 and HLL can be captured in act by AFM which offers an access to various physical phenomena on molecular level with nanoscale resolution. As prominent examples in this respect are presented the following AFM studies: the influence of product phase separation on PLA2 hydrolysis of supported DPPC bilayers; the kinetics of degradation of DPPC bilayers as a result of vipoxin PLA2 activity; the hydrolysis of unsaturated lipids by HLL at lipid–water interface; the degradation of lipid bilayers due to simultaneous action of PLA2 and HLL. In the closing part of chapter 4 are summarized AFM and X-ray scattering data of results from biologically relevant investigations demonstrating the synchrotron X-ray scattering methods as complementary of AFM for in situ observation of reactions or interactions at interfaces under near physiological conditions. The chapter 5 is devoted to investigations of other enzyme with biological and industrial importance- the protease Savinase. Firstly, the Savinase action on Bovine Serum Albumin (BSA) monolayers is demonstrated with measurements at air-water interface and by liquid AFM imaging experiments. Secondly, Savinase proteolysis of insulin Langmuir monolayers is studied by surface pressure and surface potential measurements accompanied by AFM imaging. In chapter 6 the focus is moved towards one of the mostly utilized AFM area of applications, namely studying the lateral organization and molecular structure of Langmuir- Blodgett (LB) Films. As a first example the new phase transitions and electronic properties controlled by surface pressure of amphiphilic Hexa-perihexabenzocoronene (HBC) LB films are demonstrated by AFM and its variation 5

technique -Kelvin Force Microscopy (KFM), together with the complementary Xray scattering methods. Next example presents, the AFM images and structural analysis of hydrophobin II (HFBII) LB films transferred from air/water interface on mica solid supports. Chapter 7 is concerned with AFM applications in the field of nanotechnology proving its importance for studying metal and polymer nanoparticles. At the beginning is demonstrated how AFM can be applied as a kinetics tool for studying the gold nanoparticles’ growth and the AFM results are also compared with two other well established experimental methods, TEM and DLS. Next in the chapter are considered the AFM applications for characterization of polymer nanoparticles based on poly-zwitterionic copolymers used as novel drug delivery systems where are demonstrated the capabilities of AFM for reveling the morphology and drug release properties of in situ loaded polymer nanoparticles. In the closing chapter 8 as a short but important example is demonstrated the applicably of AFM for characterization the higher-order chromatin organization in baker's yeast (Saccharomyces cerevisiae) and particularly the role of linker histone. This last chapter explicitly proves why AFM is considered as unique and exceptional for studying biological specimens. At last, for the sake of limpidity of discussed ideas all experimental procedures, used chemicals, details about certain experimental methods etc. were moved in separate sections as appendixes.

6

1. Atomic Force Microscopy (AFM)-Methodology, theoretical and experimental considerations Atomic force microscopy (AFM) is an advanced and broadly utilized technique in the field of nanotechnology which was primarily designed to study the surface structure and topography of varies samples with nanoscale resolution and accuracy. What makes AFM unique though is that it does not only allow to be obtained images which reveal the arrangement and structure of individual molecules but it “can see” also these molecules in action within their natural environment. AFM experiments can be carried out in physiological buffers at temperature 37 °C and the occurred biological reactions to be followed in real time [1,2].The simplified procedures for sample preparation and ability of imaging biological specimens in liquid environment have opened wide perspectives for application of AFM technique in biological and biomedical sciences [3]. As a member of a large family of instruments well known by the name “Scanning probe microscopes (SPMs)” the AFM is often referred in the literature as Scanning Force Microscope (SFM). All SPM techniques share as a common element a very sharp probe, which is part of the sensor element of SPM setup employed for scanning across the sample surface. During the scan as a result of probe/surface interactions is created a high resolution image of sample topography. In principle, the image resolution goes down to sub-nanometer scale level, depending upon the technique and sharpness of the tip. At the beginning, the invention of AFM by Binnig et al [4] as an imaging technique intended to overcome some limitations of Scanning Tunneling Microscope (STM), namely, the imagining of only conductive samples in vacuum. Nowadays AFM has overreached the expectations of its inventors, taking a leading role in the field of surface science by covering two major experimental areas- surface characterization and measuring the interaction forces between surfaces or molecules [5]. The AFM has number of advantages over any microscope technique with its ability to perform measurements either in air or in fluid environment which allows imaging of any biologically important molecule or biological sample in their native state. For this purpose all commercially available AFM instruments are supplied by a sealed liquid chamber (also known as liquid or fluid cell) where the environmental conditions are strictly controlled. Many AFM setups also have mounted an inverted microscope for simultaneous optical imaging of the surface thus allowing the tip to be positioned manually at desired location over the sample. The obtained AFM images show detailed information about surface features with exceptional precision as the scanning area covered by AFM can vary from the atomic and molecular scale up to sizes larger than hundreds of micrometers. For example, AFM can capture very small images of 5 nm in size, thus covering about 100 individual atoms and revealing the crystallographic structure of materials [6,7] and at the same time can be obtained images of 100 micrometers or larger which vividly expose the shape and morphology of living cells [8,9]. The AFM can also scan rough surfaces because its vertical range (so called Z- range) is usually 7

up to 10 ÷ 15 𝜇𝑚. Compared with the scanning electron microscope (SEM) AFM provides higher topographic contrast and direct measurements of the surface features thus providing quantitative information about the sample’s height. Furthermore, because the sample doesn’t have to be electrically conductive, the metallic coating is avoided, therefore no dehydration is required, and samples are imaged in their native state. In comparison to Transmission Electron Microscopes (TEM), the 3D images captured by AFM and then cross-sectioned reveal far more information than 2D images of TEM. The principal element of any AFM instrument is a sharp probe (tip) mounted at the end of flexible cantilever arm. The probe is scanning across the sample surface with simultaneous monitoring of cantilever deflection when tip follows the surface topographic profile. Thus a three-dimensional image of the surface with high resolution is created. Today many different variations of this basic AFM technique are developed. They include static and dynamic techniques depending on whether the probe remains at constant contact with the sample, or it oscillates above the sample. When the cantilever deflection is monitored then the interaction forces between probe and sample can be measured as function of probe/sample distance with piconewton (𝑝𝑁) force sensitivity (i.e., several orders of magnitude better than the weakest chemical bond). It is a distinguishable experimental feature, bacause it makes AFM an ideal tool for probing interactions between various functionalized surfaces, molecules, chemical groups etc. [10] 1.1. The principle of the Atomic Force Microscope The general design of AFM setup as a block diagram is shown at Fig. 1A where the principle of most popular force transducer known as optical lever is also depicted [11]. The principle of AFM very much reminds the gramophone, but it has number of refinements that enable the apparatus to achieve nanoscale resolution. These “new” AFM elements include- very sharp tips, flexible cantilevers, sensitive deflection sensors, and high-resolution electronically controlled tip/sample positioning system, etc. In the principle block scheme at Fig. 1A, the force transducer measures forces between tip and surface while the feedback controller is responsible for keeping the force constant by extending the Z- piezoelectric transducer i.e. it maintains the tip/sample force at a certain value (so called set point) by controlling the tip/sample distance. The X, Y- piezoelectric elements are used to drive the probe across the surface in a raster-like scan pattern. It’s called raster scan because the piezo moves back and forth repeatedly in the X-direction followed by a step in the Y-direction. The distance at which the Z- piezo moves up and down at constant tip/sample force interactions while X, Y-scanning occurs reproduces the sample topography as a height image (i.e. mapping the surface profile) by monitoring the voltage applied to the Z- piezo. 8

Figure 1. Basic AFM set-up. (A) Block diagram of an AFM (B) Principle of the ‘optical lever’ force transducer. The force transducer senses the cantilever movement by a photodetector through the change of the path of a laser beam deflected from the upper side of the cantilever arm. Then this deflection is used to be measured the interaction forces between the probe and the sample. In a feedback control loop the measured force is compared with a set force value and ultimately the control over the piezoscanner movement in z- direction is established feedback and the topography of the surface is reconstructed as a 3D image.

The simple design and high sensitivity to miniscule cantilever deflections are main advantages of the optical lever [12,13]. It is composed by a quadrant photodiode sensor element divided into four parts with a horizontal and vertical dividing lines where each section of the detector is labeled by A, B, C and D respectively (Fig. 1B). The tip is mounted at the apex of a flexible Si or Si3N4 cantilever arm (Fig. 2A). A laser beam reflects from the uppermost side of cantilever onto the photodiode quadrant. Thus any movement of the cantilever produces a change in position of the laser spot on the photodetector allowing the laser beam deflection to be monitored. The deflection signal is calculated from the difference in voltage signals detected by 𝐴 𝑎𝑛𝑑 𝐵 versus 𝐶 𝑎𝑛𝑑 𝐷 quadrants. Comparison of the voltage signals detected by 𝐴 𝑎𝑛𝑑 𝐶 versus 𝐵 𝑎𝑛𝑑 𝐷 allows detection of lateral or torsional 9

bending of the cantilever. When the probe is brought into a contact with the sample surface by means of a stepper motor jointly with the vertical piezoelectric transducer then the scanning across sample is launched. Next, during sample scanning whenever the height change occurs, the cantilever bends causing the reflected laser spot to change its position on the photodiode system causing a generation of photodiode output voltage. This is an error signal which is used as a feedback to the piezo in order to keep the cantilever deflection constant. A typical image is generated by plotting the extension of the Z-piezo in metric units versus (𝑋, 𝑌) position. This basic AFM principle is implemented in two types of AFM systems known as tip-scanning and surfaces-scanning, respectively, depending upon whether the cantilever or the surface itself is mounted on Z-piezoscanner.

Figure 2. (A) SEM image of the AFM cantilever with the tip on its apex. (B) Schematic illustration of the AFM cantilever when the tip probes the surface. (C) Schematic illustration of the action of the capillary forces during the AFM operation. The force is exerted by a thin water layer covering the sample under ambient conditions

The topographic image of the sample is created on the bases of tip/sample interaction at very small tip-sample separations. When the tip is engaged in close proximity to the surface outermost atoms of the tip are repulsed by atoms of the sample (Fig. 2B). This hard core repulsion is extremely distance dependent providing high resolution in AFM imaging allowing in some crystalline samples to be achieved an atomic resolution [14].Other forces which play a significant role in AFM operation are the capillary forces due to thin water layer covering the sample under ambient conditions. Numerical simulations show that for hydrophilic tip 10

substrate-sample interactions, the tip oscillation amplitude increases while in case of hydrophobic materials, there is no dependence of tip oscillation on tip/sample interaction. This dependence has been attributed to the non-conservative character of capillary force (Fig. 2C) [15]. Some other forces also contribute to total tip/sample interactions and have influence on image quality. The nature of dominant forces in AFM and their distance dependence are depicted in Table 1.1 [16]. Table 1.1 Force

Distance dependence

Elastic

𝐹 = −𝑘𝑐 𝑥

(Hook’s Law)

Force nature

Repulsive force From cantilever and any surface deformation

van der Waals

𝐹=−

(Sphere to surface)

𝐴𝑅 6𝐷 2

Attractive (or Repulsive) force Adhesion and friction

Capillary

𝐹=−

4𝜋𝑅𝛾𝐿 𝑐𝑜𝑠𝜃 𝐷 1+ 𝑑

Attractive force Tin layer present on the top of the sample.

Repulsive

𝜎 𝑛 𝐹 = (𝑛 + 1) ( ) 𝑟

Very short range Only effective between the outermost atoms of the tip and the atoms of the sample.

Electrostatic (Sphere to surface)

𝐹=−

4𝜋𝑅𝜎𝑠 𝜎𝑡 −𝐷𝜅 𝑒 𝜅𝜀𝑒 𝜀0

Repulsive (or Attractive) force Long range force.

𝑭- Force, N 𝒌𝒄 - Spring constant, N/m 𝒙- distance the spring is moved away from the position at zero force, m 𝑨- Hamaker constant, J 𝑹- Radius of the tip, m 𝑫- Distance between the tip and the surface, m 𝜸𝑳 - Liquid surface tension, mN/m

𝜽- contact angle between liquid and the tip, deg 𝒅- height of the wetting liquid, m 𝝈- Hard-sphere diameter of an atom, m 𝒏- Determines the steepness of the repulsive potential 𝝈𝑺 - Surface charge density, C/m2 𝝈𝒕 - Tip charge density, C/m2 𝜿- Debye length, m

11

When scanning over soft samples, in order to achieve maximal sensitivity and to avoid destructive effects on sample, the total interaction force should be kept as low as possible. When scanning in air, the attractive force is dominated by a capillary attraction of the tip to the sample. In liquid, thought the capillary force is diminished and the only forces in act are van der Walls’ forces at the nanometer range and long range electrostatic forces caused by the tip and sample surface net charges. The electrostatic interactions are either attractive or repulsive depending on ionic strength of the liquid in use. As it was demonstrated, with a careful adjustment of buffer composition the imaging of 2D crystalline protein arrays with very high resolution is attainable [17]. 1.2. Imaging in liquid environment The AFM experiments can be carried out in liquid environment using so called liquid (fluid) cell shown at Fig. 3.

Figure 3. (A) Schematics of an AFM liquid cell set-up. (B) Photo of Nanoscope V (Bruker) AFM liquid cell. (C). Forming of a drop at the bottom of liquid cell for imaging without use of a sealing O-ring.

The liquid cell is a refined cantilever holder which is made of a quartz crystal glass with a groove around the cantilever for placing of a sealing O-ring. The cantilever is position into an indented pocket and is held by a wire clip (Fig. 3B). The liquid cell has an inlet and an outlet allowing a buffer solution to be flushed trough the cell space where the sample is positioned (Fig. 3A). In a simple liquid setup where the changing of the liquid environment is not required the cell is used without a sealing O-ring. Then by a syringe the liquid is dropped at the cell bottom and when a drop gets into contact with the sample surface the formed meniscus between tip and sample keeps them immersed in liquid (Fig. 3C).

12

1.3. Operating AFM modes The AFM can operate in different imaging modes, each providing a variety of data about sample surfaces. The most common imaging modes according to the acting force regimes are classified as contact, non-contact and intermittent (tapping) modes. All of them are illustrated schematically at Fig.4 by a graph of the force versus tip-sample distance.

Figure 4. Idealized plot of the forces between tip and sample, highlighting where typical imaging modes are operative.

The magnitude of the interaction forces are illustrated on the graph when the probe approaches and contacts the surface. With distance increasing to the right at large tip/sample separations there are no net forces acting between the tip and the surface. As the tip approaches the sample attractive van der Waals interactions begin to pull the cantilever towards the surface and it goes into a noncontact mode until force minimum is reached. Further nearing of the tip towards the sample is followed by a contact between them causing the electron shells of the atoms in opposing surfaces to overlap and net interaction becomes repulsive. At this point the tip is in contact mode of action. At Fig. 4, the repulsive forces are shown as being positive and attractive forces negative. 1.3.1. Contact Mode of Imaging Of all imaging modes the contact mode is simplest and was the first mode developed for AFM. It occurs within the repulsive regime, around the set-point domain of the graph at Fig. 4 where the tip is in a continuous contact with the sample

13

surface. Conceptually the contact-mode pretty much emulates the principle of surface profiler in which AFM captures images with very high-resolution at fastest possible rate in comparison to all of the rest topographic modes. In contact mode the force applied by the tip to the surface (Fig. 2B) is given by Hooke’s law: 𝐹 = −𝑘𝑐 𝑥

(1.1)

where 𝐹 is the force [𝑁], 𝑘𝑐 is the cantilever force constant [𝑁/𝑚] and 𝑥 is the deflection distance [𝑚]. In contact mode AFM operates as the microscope feedback system maintains the cantilever deflection at a certain value i.e. set point voltage. The set-point is one of the most important control parameters of AFM which have to be optimized. The equation (1.1) implies that either a probe with high force constant (i.e. one with a stiff cantilever), or greater deflection (i.e. higher set-point) will cause higher applied force. The vertical deflection usually fluctuates in the course of imaging because the AFM feedback system does not response instantaneously, as depicted by the thick gray line in the area of repulsive force at Fig. 4. The deflection variations depend on the sample topography, flexibility of the cantilever, scanning speed, feedback control parameters etc. and careful consideration of all of these factors determines the quality of AFM images. 1.3.2. Oscillating (non-contact and intermittent) AFM modes In order to avoid the destructive lateral forces present in the contact mode a large number of dynamic AFM modes of operation are developed and all of them are based on the same principle. Usually by means of complimentary piezoelectric element the cantilever is forced to oscillate at its resonant frequency. When the oscillating tip approaches surface and reaches certain separation distance within non-contact regime then the amplitude changes as a result of tip/ sample interactions which causes a damping effect on cantilever oscillations, thus leading to a reduction of frequency and amplitude. The oscillations are monitored by the force transducer (i.e. by the optical lever in most AFMs) and exactly as in contact-mode the z height is adjusted via feedback loop by maintaining the probe at a fixed distance from the sample. The only dissimilarities between the various oscillating modes are related to the magnitude of cantilever oscillation amplitude, as well as the methods used to detect the oscillation changes. The general principle of oscillating and instrumental set-up schematically are shown in Fig. 5A. An oscillating signal is generated and then applied to the cantilever mechanically, in a way that the cantilever is tuned to oscillate close to its resonant frequency. The oscillations are monitored during the engagement of the tip to sample surface. When a change whether via amplitude, phase or frequency is detected, it is then used in a feedback loop as a parameter for maintaining the probe/sample interactions at constant 14

value. The choice of small or large amplitude has a considerable practical effect, as it is illustrated at Fig. 5B. When using small oscillation amplitude (as denoted by the red arrows A), the cantilever is maintained only in attractive regime and then AFM operates in noncontact mode which has the advantages due to low probe tip/sample forces involved in action. When the large oscillation amplitude is applied, then the probe in each oscillation cycle will move from being far from the surface where there is no tip/sample interaction, through attractive regime into repulsive regime, and back (blue arrows B). This scanning mode referred as intermittent contact-mode AFM or also tapping mode involves large tip/sample interaction forces, so it can be more destructive than non-contact mode, but much easier for technical implementation. In tapping mode the feedback control is usually based on the amplitude signal of probe oscillations [18].

Figure 5. (A) Schematic of generalized operation of oscillating AFM modes, showing instrumental set-up. An oscillating input signal is applied to the cantilever to make the probe vibrate up and down. The concrete movement of the probe depends on its interaction with the sample surface. In a feedback control loop the resulting oscillation in the cantilever deflection is measured and compared to the input oscillation to determine the forces acting on the probe. (B) Different operating regimes for oscillating AFM modes. A-mode with a small amplitude of oscillation, the probe can be kept in the attractive regime. B-mode with large oscillation the probe moves through non-interacting, attractive and repulsive regimes, resulting in intermittent contact. (C) Illustration of the effect of intermittent contact on the cantilevers’ oscillation. The free oscillation (solid) is modified when in contact with a surface (dashed) by a reduction in amplitude and a phase shift. 15

In most AFMs, the probe is oscillating in amplitude range of 1 ÷ 100 𝑛𝑚 with the help of an additional piezoelectric element attached to the probe holder. For fluid imaging usually the entire fluid cell holder is exited, rather than the probe directly. In this way the liquid within cell vibrates and acoustically driving the cantilever [19,20]. In some AFM modifications, the cantilever vibrations are existed by other methods, e.g. by an external magnet, with a magnetically coated cantilever [21,22]. The latter is very advantageous for liquid imaging because the fluid vibrations are substantially reduced. Often, in addition to the amplitude signal, the delay in the phase of the probe oscillation is recorded. Oscillation amplitude and phase are illustrated in Fig. 5C. In practice the tip usually is tuned at 5 to 10 % shift from the value of its resonant frequency. In tapping mode the amplitude is a set-point point parameter, hence the decrease of the amplitude (Fig. 5C) is an error signal for the feedback. Similarly to deflection signal in contact mode, the amplitude signal shows where the feedback system has not yet compensated the changes in sample height. That’s why for achieving the best height resolution, the amplitude signal should be minimized as much as possible. The advantages that have made the tapping mode quite popular over the contact mode are the superior vertical and lateral resolutions which are achieved because of mild tip/sample interactions. The great reduction of lateral force distortions allows qualitative imaging of soft samples in liquid environment [23,24,25] thus making this AFM mode of operation mostly preferred for imaging of biological samples. It is also under perpetual improvement, thanks to additional features such as Q-control system [26], implementation of magnetically driven tips, etc. [21, 22]. 1.3.3. Force spectroscopy mode The ability of AFM to measure forces with magnitudes down to few piconewtons attests for the versatility of this technique as a multifaceted nanotool. When the apparatus is operating in so called “force spectroscopy mode” it maintains the X-Y position of the probe fixed, while the piezo is extended and retracted in z-direction, i.e. the tip approaches and retracts from the sample surface. From the instantaneously measurements of cantilever deflection are obtained the force/distance curves (Fig. 6). The raw data are collected and plotted as on x-axis is recorded the displacement of the z-piezo while on y-axis is presented the cantilever deflection as a voltage signal coming from the photodetector. Following the events presented schematically at Fig. 6, at the beginning the cantilever approaches the surface starting from point A where it is a “free lever” with no net forces acting and therefore no deflection at this point is detected. As the probe comes into close proximity with the sample, cantilever begins “to feel” tip/surface long-range force interactions. Repulsive forces then pull the cantilever upwards and away from the surface, whereas attractive forces are pushing it downwards, towards the surface. When 16

the probe reaches point B where the gradient of attractive forces exceeds the cantilever stiffness, then it deflects rapidly downwards allowing the probe to touch the surface in a “snap-in” or “jump-to contact” manner. The small depression in the deflection curve after point B (Fig. 6) demonstrates the existence of attractive surface forces. When the cantilever makes hard contact with surface, it is deflected upwards due to tip/sample repulsion, and a positive force is observed (point C). The cantilever then retracts following the path of approaching trace in the contact region as it is shown with the arrows direction at Fig. 6. On the retrace way back cantilever remains attached to the surface by adhesive forces, thus causing a hysteresis between trace and retrace paths. At point D occurs a “pull-off” because the separation force becomes sufficiently high to overcome the tip- sample adhesion. Eventually, the cantilever snaps back to its initial “free lever” position (point A at Fig. 6).

Figure 6. Force–distance curve. At point A, the probe is far from the surface; at B occurs “snap-in” contact because the attractive forces pull the probe towards the surface. The force becomes repulsive as the probe continuously approaching the sample. At some point C, the travel direction reverses. At point D occurs a “pull-off” as the force applied to the cantilever overcomes tip/sample adhesion.

In force spectroscopy mode AFM directly measures the force between contacting atoms or molecules at the end of the tip and those at the sample surface. Hence using highly flexible cantilevers with higher deflection sensitivity accompanied with refined optical lever-based sensing systems is possible a single-molecule interactions to be detected [27]. In force spectroscopy experiments usually the tip and/or studied surfaces are preliminary manipulated or modified e.g. the probe modification involves imbedding of specific molecules [28] or colloidal probes [29]. Various surfaces have been probed by means of force spectroscopy mode spanning from biological objects i.e. cell membranes, microorganisms or whole living cells 17

[30,31,32] to a wide variety of solid surfaces including polymers, metals and ceramics [33,34,35]. In order to be obtained accurate quantitative measurements of acting forces two prerequisite steps should be performed, namely both cantilever spring constant and cantilever sensitivity must be correctly determined. It allows the voltage data obtained from the photodetector to be transferred into force units. Several different methods for cantilever calibration are developed. A reference spring method involves very simple and straightforward procedure where the cantilever is pressed against another, reference cantilever of known spring constant [36]. Another method of calibration assumes that the cantilever has harmonic behavior. Then, the spring constant is determined by a resonance shift as a function of the added weights [37]. One of the most commonly used methods for quantification of the cantilever spring constants is so-called thermal method. It was introduced by Hutter and Bechhoefer [38] and it is now implemented in many commercial AFMs. The method assumes an idealized spring behavior of the cantilever which is thermally fluctuated under ambient conditions without the presence of a surface. Then, from the area of the fundamental resonant peak the spring constant of the cantilever can be directly calculated. The mean square deflection of the cantilever 〈𝑥 2 〉, due to thermal fluctuations is related to the spring constant by the equation: 𝑘 𝑇

𝑘𝑐 = 〈𝑥𝐵2 〉

(1.2)

where 𝑘𝐵 and 𝑇 are Boltzmann’s constant (𝑘𝐵 = 1.38 × 10−23 𝐽𝐾 −1 ) and absolute temperature, respectively. Once other noise is subtracted from the background signal, the area of fundamental resonance peak will be equal to the mean square displacement. To calibrate the cantilever sensitivity, i.e. to find a relation between the voltage signal generated out from the photodetector in Volts and the bending in nm, the cantilever is pressed against a hard surface. When the cantilever presses the surface all the piezo movement translates directly into cantilever movement, thus giving the required relation Volts per nm. 1.3.4. Electric force microscopy and scanning Kelvin Force Microscopy Electric force microscopy (EFM) is refined AFM mode which measures the electrical fields at the nanoscale level. The technique uses conductive i.e. metalcoated silicon or silicon nitride tips for read/write applications, and more sophisticated electrical modes of AFM operation. The EFM principle is based on the electrostatic interactions between a probe and a surface having different voltage potentials as the forces are determined by the equation: 1

𝑑𝐶

𝐸𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑠𝑡𝑎𝑡𝑖𝑐 = − 2 𝑉 2 𝑑𝑧

(1.3) 18

where the change in resonant frequency is proportional to the changes in capacitance as a function of second derivative of 𝑧 −displacement. In other words, as long as there is a non-zero potential between the probe and surface, the frequency, and thus amplitude and phase of oscillation will be sensitive to capacitance of the surface. EFM is applied for detection of confined charges on surfaces [39] showing a clear contrast on their locations where no visible changes in topography images are visible. Kelvin force microscopy (KFM) is a sophisticated version of EFM modified with an electronics for mechanically vibrating the cantilever for measuring the tip/sample potential [40]. At Fig.7 is illustrated the block diagram of KFM.

Figure 7. Schematic illustration of the set-up of Kelvin Force Microscopy (KFM)

The principle of operation of KFM, uses the fact that when two surfaces have the same potentials, there will be no forces acting between them, so then in Eq. (1.3) the value of 𝑉 should be zero. Technically this is achieve as a DC potential bias (𝑉𝐷𝐶 ) is applied to a conductive probe, which is further modulated by an AC signal (𝑉𝐴𝐶 ), so that: 𝑉𝑏𝑖𝑎𝑠 = 𝑉𝐷𝐶 + 𝑉𝐴𝐶 sin 𝜔𝑡

(1.3)

Hence, the AC voltage is oscillating at the cantilever resonant frequency which means that the electric potential of the tip is varying at frequency 𝜔 [40]. When it is not equal to the sample’s potential, it will cause the electrical signal from the photodetector to modulate at frequency 𝜔𝑚𝑜𝑑 which again means that the appeared error signal 𝜀 = 𝜔 − 𝜔𝑚𝑜𝑑 has to be compensated by the feedback circuit as it happens when it outputs a DC voltage to the sample to minimize the oscillation at 𝜔𝑚𝑜𝑑 . Because the applied potential 𝑉𝐷𝐶 is equivalent to the surface potential 19

𝑉𝑠𝑢𝑟𝑓𝑎𝑐𝑒 it is digitized with the analog to digital (A/D) converter and displayed on PC software generated graph as potential image [40]. 1.4. Image interpretation AFM images are subject to number of artifacts due to dynamic nature of scanning process, forces in act and most importantly from the tip geometry. Usually the vertical AFM resolution is in subnano level about 1 Å but it is about one order of magnitude better than the lateral resolution. The reason for this huge variance is due to shape and geometry of the tip. In the first AFM prototype the probe was made of a tiny diamond crystal glued onto the end of a spring made of thin strip of gold [4]. Nowadays the fabrication technology of commercial tips is similar to that applied in production of integrated circuits where the tip/cantilever assembly typically is made on similar Si or Si3N4 wafers [41]. One of the essential tip parameters is the sharpness of apex, measured by the radius of curvature, and the aspect ratio of the whole tip (Fig. 8).

Figure 8. The essential parameters in a tip are the radius of curvature (𝑹) and the aspect ratio (ratio of 𝒉 to 𝒘 )

It is well established that the apparent lateral dimensions of all imaged objects are broadened caused by the tip geometry which has a finite size in the same order of magnitude, close to the size of a protein molecule, 5 ÷ 10 𝑛𝑚. This artefact is recognized in the literature as a tip-convolution effect. A simplified geometrical model which approximates the tip shape is shown at Fig. 9A [42]. Applying this model the probe geometry artefact on AFM images then can be removed by a process called “deconvolution” using the apparent relation: 𝑑 = 2√𝑟 2 + 2𝑟𝑅

(1.4)

In this approximative geometrical model the protein molecule is considered as a sphere with radius 𝑟 while the probe has a spherical tip with radius R. The illustrated tip trajectory during scanning has bell-like shape which corresponds to the measured image profile. Apparently, the diameter 𝑑 does not coincide with real diameter of the molecule. The real cross-section profile taken from AFM image of a protein cytochrome P450 is shown at Fig. 9B. The “tip-broadening” effect can be 20

calculated if the tip radius 𝑅 is known. The real diameter of molecule 𝐷 can be estimated from formula (1.4) as follows: 𝐷 = 2𝑟 = −2R + 2 (√𝑅 2 +

𝑑2 4

)

(1.5)

Figure 9. (A) Illustration of the tip-broadening effect resulting from the tip geometry. R, tip radius, r, molecule radius; d, apparent molecule diameter. (B) A real cross section profile taken from the AFM image of a protein molecule (cytochrome P450).

In this particular occasion for the measured by AFM size 𝑑 = 17 𝑛𝑚 and typical value for tip radius 𝑅 = 20 𝑛𝑚 after substitution in (1.5) for the real size of protein molecule will be obtained 𝐷 = −2 × 20 + 2 (√202 +

172 4

) ≈ 3.5 𝑛𝑚. The later value is

on the same order as vertically measured size of protein molecule which is ℎ = 2.8 𝑛𝑚. In practice, the tip dimensions are usually not known precisely and differ from tip to tip even on the same wafer, because the method of manufacturing such probes is not perfectly reproducible. Therefore, the proposed geometrical model is only an approximation because the tip is not an ideal sphere of radius R. For this reason lateral dimensions are rarely used in analytical measurements, whereas vertical dimensions are preferable in calculations. Lateral or friction forces also lead to artifacts in height imaging. When the tip scans over the areas of sample surface with different friction properties both lateral and normal force components will be interpreted as normal force component because both voltage signals will appear as a difference between the two upper and two lower photodiode elements of the detector quadrant (Fig. 10). The cantilever bending is induced as a result of a sticking of the tip to high friction areas. This bending will appear as a pseudo-height. The bending of the cantilever which has length 𝐿 and the spring constant 𝑘𝑐 under normal force 𝐹𝑛 can be approximate with the equation [43]: 𝑥 2

𝐹

𝑧 = (𝑘𝑛 ) (𝐿 ) 𝑐

(1.6)

where 𝑥 is the position along the cantilever. Analogously, the lateral force 𝐹𝑡 will lead to a pseudo- height 𝑧𝑓 : 𝐹

ℎ

𝑧𝑓 = (𝑘𝑡 ) (𝐿 ) 𝑐

(1.7) 21

where ℎ is the tip height. This false height can lead to contrast inversion between forward and backward images because the bend induced by friction will be in opposite directions. To circumvent this problem the scanning direction of the tip should be chosen such that to minimize the lateral forces influence which is done usually by setting the scanning angle at 90° (Fig. 10). Due to the tip oscillations, a contrast inversion artifact in tapping mode also may appear as a result of the instabilities in cantilever oscillation that may occur depending on the choice of operating oscillation point and on the forces that dominate the amplitude damping [11]. The tapping mode of operation also causes the broadening of objects by tip convolution. Although, the lack of permanent contact with the sample makes this mode softer at the profile edges and they appear sharper than in contact mode in which the edges get easily compressed under the tip influence. In tapping mode any false height measurement caused by the friction is eliminated because the tip does not drag over the surface but goes in and out of contact.

Figure 10. Illustration of the separation of normal and lateral force component in a quadrant photodiode detector. First the laser spot is adjusted at the center of the quadrant (no bending).When the scanning starts from side to side over sample surface the normal force component comes from the voltage signal (𝑈𝐴 + 𝑈𝐵 ) − (𝑈𝐶 + 𝑈𝐷 ) (vertical bending) whereas the lateral component from (𝑈𝐴 + 𝑈𝐶 ) − (𝑈𝐵 + 𝑈𝐷 ) (horizontal bending). If the cantilever scans back and forth both signals are combined only in the difference (𝑈𝐴 + 𝑈𝐵 ) − (𝑈𝐶 + 𝑈𝐷 ).

In summary: The AFM is a versatile technique which is designed to study with nanoscale resolution and accuracy the surface structure and physical properties of various samples and particularly those of biological importance. Its uniqueness is due to the capability of capturing individual molecules in act when the experiments are carried out in physiological buffers and the occurred biological reactions are followed in real time. AFM can be used not only as a tool for surface characterization but also for measuring the interaction forces between surfaces or molecules. It can operate in variety of scanning modes which allow different physical (e.g. electrical, magnetic, mechanical) properties of studied surfaces to be revealed 22

2. Monolayers at Fluid Interfaces and Langmuir- Blodgett filmsStudying the molecular organization (e.g. biological membranes) by monolayers and bilayers as model systems with experimental methods complimentary to the AFM. The cell membrane is a complex entity and understanding the physical mechanisms which govern its order and structure generally requires simplification. The exclusion of all components except for a single or few lipid species and/ or proteins tremendously facilitates the study of cell ordering mechanisms. Using this approach many studies have directly or indirectly illustrated the organization of lipids and proteins into domains in simple model systems, thereby enlightening the underlying physicochemical mechanisms leading to the observed organization. One of the most convenient membrane model systems are Langmuir monolayers which have many indisputable advantages. The definition of ‘‘monolayer’’ usually denotes a layer of amphiphilic molecules organized at a fluid interface, either a liquid/liquid or a liquid/gas interface. Even when the interface is incompletely covered by the molecules or when it is more than one molecular layer thick is still referred a monolayer. The amphiphilic compounds which contain polar and nonpolar parts of their molecules when contact the water orient at interface so that polar parts are exposed to the aqueous phase while nonpolar to the nonaqueous phase. If the spread amphiphilic molecules at the interface do not dissolve in adjacent phase then the formed monolayers are called insoluble monolayers or as mostly referred in the literature- Langmuir monolayers. The phospholipids with acyl chains containing more than 12 carbon atoms are sufficiently hydrophobic thus forming stable Langmuir monolayers and providing a very convenient experimental platform for modeling the lipid membrane. Moreover, because the biological membrane can be considered as two loosely coupled monolayers hence the phospholipid monolayer is an excellent model system for studying the order in two dimensions where the surface of adjacent liquid phase is an ideally smooth substrate with direct control over thermodynamical parameters where the wide range of monolayer-subphase interactions can be controlled. The molecular orientation, conformational state, packing density, molecular dipole moment, and lateral viscosity are all controlled by means of various intensive (lateral pressure, temperature, pH, ionic strength, etc.), and extensive (area, surface concentration, etc.) thermodynamic variables. There are also many other reasons to use Langmuir monolayers as a model system in membrane biophysics. One particularly is studying in two dimensions of the enzyme reactions which usually occur at the membrane interface. Langmuir monolayers are also required for the fabrication of so called Langmuir-Blodgett (LB) films, which are multilayers transferred layer-by-layer from water surface to a solid support. The Langmuir technique has proved over the years its advantages for studying the two dimensional organization of various amphiphilic molecules at air/water interface providing such control over the interfacial quality of the surface. 23

Figure 11. Typical commercial Langmuir setup (KSV-NIMA): Trough, two barriers, surface pressure sensor and module for Langmuir–Blodgett (LB) deposition. Most troughs are usually made of PTFE (Teflon). The two barriers are driven by a stepper motor mechanism provided with a PC hardware and software control. The tensometer measures surface pressure by Wilhelmy method where a platinum plate or a metal rod, hanging from a piezo-electric crystal, acts as a balance measuring the surface tension.

In principle, the basic experimental setup of a Langmuir trough consists of a reservoir filled with water, one or two movable barriers, and a balance (surface pressure sensor) (Fig. 11). When spread from a volatile organic solvent on water surface in-between the barriers the amphiphilic molecules form an insoluble monolayer. The movement of barriers towards each other compresses the monolayer molecules leading to a decrease in the mean molecular area (MMA) and consequently the increase of the surface pressure. The most common physical observables in Langmuir technique are the lateral pressure (𝜋) and the surface potential (∆𝑉). The surface pressure 𝜋 is defined as: 𝜋 = 𝛾0 − 𝛾

(2.1)

where 𝛾0is the surface tension between air and water and 𝛾is the surface tension when the amphiphilic molecules are present at the air/water interface. The physical interpretation of the surface tension is the work required to be created a unit of interface between the two phases. It is considered as an entropy decrease at the surface due to ordering and increased hydrogen bonding [44], or due to the loss of hydrogen bonds of the fewer neighboring water molecules at the interface [45]. In thermodynamic terms for planar interfaces the surface tension is defined as Gibbs free energy 𝐺 required for creating a surface area 𝐴. Hence it is the derivative of 𝐺 in respect to 𝐴: 24

𝜕𝐺

𝛾=( )

𝜕𝐴 𝑇,𝑝,𝑛 𝑖

(2.2)

where 𝑇 is the temperature, 𝑝 is pressure, and 𝑛𝑖 the amount of molecules in the monolayer [46]. Since 𝛾0 is constant, the lateral pressure can also be expressed as: 𝜕𝐺

𝜋 = − (𝜕𝐴)

𝑇,𝑝,𝑛𝑖

(2.3)

The equation (2.2) is only an approximation because the strict thermodynamic treatment requires the monolayer to be a closed thermodynamic system with no exchange of lipid molecules between surface and aqueous subphase. Moreover, because the ordered monolayer phases are only metastable treating them as being in thermodynamic equilibrium is not entirely correct [52]. Surface potential ∆𝑉 is another important physical quantity of the surface because it provides direct information for the molecular orientation and surface concentration (Fig. 12A) [47].

Figure 12. (A) Schematic representation of the molecules in a Langmuir monolayer having the effective dipole moment 𝜇⊥ ; (B) The Helmholtz model of the surface potential describes the potential difference between the plates of a parallel capacitor separated at distance d while the interfacial film represents flat array of dipoles with area 𝐴 = 𝑘𝑙.

Surface potential ∆𝑉 is defined as a difference between potentials of water subphase and that of a monomolecular layer spread at air/water interface. It is proportional to the effective dipole density 𝜇 for uncharged monolayer (Fig. 12) and in a continuum approximation is given by Helmholtz equation: 𝜇

∆𝑉 = 𝜀𝜀

0

(2.4)

where 𝜀 is the local dielectric constant; 𝜀0 is the dielectric constant of the vacuum and has the value of 𝜀0 = 8.854 × 10−12 , 𝐶 2 𝑁 −1 𝑚−2; 𝜇 = 𝑛𝜇⊥ with 𝑛 corresponding 25

to the number of molecules at the surface, respectively to the surface concentration Γ and 𝜇⊥ is the effective dipole moment perpendicular to the surface[48]. The physical interpretation of the surface potential is not as straightforward as the meaning of surface pressure. Equation (2.4) derived from the Helmholtz model is analogous to the equation describing the potential difference Δ𝑉 between two plates of a parallel capacitor separated at distance 𝑑 with an assumption that the interfacial film represents flat array of dipoles with area 𝐴 = 𝑘𝑙 (Fig. 12B). Two main limitations of Helmholtz model should be considered. The first one arises from the assumption that dielectric constant 𝜀 is unity. The second limitation comes from the elimination of dipoles contribution of the water molecules at the interface [49]. It also have been shown that hydrocarbon and end-group dipolar moments in the monolayer can influence the surface potential [50]. Even given these limitations as an approximation model the surface potential gives a good measure for the long-range repulsion within a monolayer [49]. The insoluble monolayers spread at gas/liquid interface may be transferred to a solid support by so-called Langmuir–Blodgett (LB) technique. It is also referred as vertical transfer procedure because in the multiple movements of a solid plate through the monolayer while the interfacial pressure is kept constant the plate is positioned vertically in respect to interface (Fig. 13).

Figure 13. Formation of a Langmuir–Blodgett (LB) film (Y-type) by monolayer transfer from a liquid/gas interface onto a solid support.

If the surface of the solid support is hydrophilic (e.g. mica, glass, quartz, etc.) the first dip through air/water interface (Fig. 13) will not transfer the monolayer because the hydrophilic parts of amphiphilic surfactant molecules cannot interact favorably with the hydrophilic solid surface. In the subsequent upward stroke though the surfactant molecules from liquid interface will be transferred with their 26

polar head groups oriented down onto solid support and because their nonpolar parts are pointed outward, the solid surface will be hydrophobized. When the second monolayer is deposited on top of the first layer the surface will be again hydrophilic so that in the next upward stroke surfactant molecules will be deposited head-down, and so on. If the surface of solid support is hydrophobic, the monolayer transfer already starts at the first immersion in the polar liquid phase. The (multi)layers obtained by this procedure in which during each passage of the solid plate through the interface a monolayer is deposited with alternating head–head and tail–tail orientations of surfactant molecules are referred as LB-films of Ytype of transfer. One important condition during LB transfer is keeping the surface pressure constant in order to be preserved the monolayer state at the interface. That’s why the LB-film preparation should take place in a Langmuir trough equipped with a feedback controller which maintains the barrier movement during removal of the molecules from the liquid interface. Usually for improvement of the transfer in liquid subphase from which the monolayer is transferred are added different ions and the ionic strength of electrolyte solution is adjusted. It is well established that when ionic surfactant molecules are involved, low-molecularweight ions play an important role in stabilizing the interaction between charged head groups in the Y-type LB film. In particular, divalent ions (Ca2+, Mg2+ etc.) help to stabilize films of monovalent (an)ionic surfactants [51]. Another method for transferring monolayers is Langmuir–Schaefer (LS) transfer also referred as horizontal transfer. In this procedure the solid support is positioned horizontally either above or beneath the monolayer. Further, it is moved downward until it horizontally touches the monolayer at the liquid interface. After a short contact with the surface the solid substrate is withdrawn or pushed down into the liquid subphase. Both LB and LS methods have proven to be very reliable in building molecular architectures constituted of one, two or more layers. It has been shown that by carefully controlling the transfer process it is possible to “snapshot” the state of a monolayer from air/water interface onto a solid plate with only minor distortions and the obtained film to be further imaged and structurally characterized by AFM [51]. Although Langmuir monolayers have been studied for more than a century now, one could easily claim that with the advancement of the novel experimental techniques and the update of traditional methods a remarkable breakthrough has been done in this area of physical chemistry in the last three decades. The employment of scattering techniques like synchrotron x-ray-diffraction in the late 1980s revealed some details of the monolayer structure at the intermolecular level and threw light on some traditional views of Langmuir monolayer phases and phase transitions [52]. The introduction of the monolayer microscopy techniques, such as fluorescence microscopy (also known as epifluorescent) [53,54,55] and Brewsterangle microscopy [56,57] directly visualized the coexisting phases and domain 27

structures that were only guessed from pressure-area isotherms, surface potentials, and other classical monolayer techniques [58] . As an illustration of such kind of experiments the organization in a 1, 2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) phospholipid monolayer is presented in Fig.14. From the image is visible the peculiar shape of dark domains of the gel lipid phase (which depending on terminology sometimes is also referred as liquid condense phase (LC)) which are embedded in the light fluid phase also known as liquid expanded phase (LE). The size of domains is in the range 30-50 m [58].

Figure 14. Fluorescent images showing various phases and domain shapes during the compression of a DPPC monolayer. The domain shapes observed with in the LE (dark)LC (light) phases. The images (A) to (C) are obtained in order of decreasing of mean molecular area. The images are adopted from [58]

The length scales of lateral organization in model membrane systems range typically from sizes of few molecules ( 𝑖. 𝑒. Å - range) up to several micrometers. At the subnanometer scale the scattering techniques threw new light on the lateral lipid packing and alkyl chains orientation when the monolayer thickness of order of alkyl chains and the structure of interface between monolayer and its aqueous surroundings have been determined [59,60]. Details of the membrane structure on a micrometer scale are usually provided by the optical microscopy. Till the late 1990es the length scale of few nanometers up to microns which lays in-between the margins of the length spectrum of membrane structural studies was comparatively less explored. This gap between scattering methods and fluorescent or other optical techniques was substantially filled by the appearance of AFM. Thus the lateral organization of lipid LB films transferred on mica supports was possible for exploring on scales ranging from few nanometers to several microns. It was shown how the lateral structure develops as phase boundaries are crossed leading to coexistence of the LE and LC domains [61,62]. One of the fundamental questions which arises while studying membrane origination using a monolayer or a bilayer as model system is: At what extend phospholipid monolayers and phospholipid bilayers are comparable as model systems? The apparent physical difference between them is the surface tension at their interface. Whereas the surface tension at the air/water interface is controlled 28

by an applied mechanical force from the barriers, the net surface tension in a bilayer is zero balanced by the opposing forces from the contracting hydrophobic effect and expanding effect from steric repulsion between the closely packed lipid molecules [63]. Nevertheless, when comparing the monolayer and bilayer properties both theoretically and experimentally it is concluded that the monolayer/bilayer equivalence lateral pressure lies between 30 ÷ 35 𝑚𝑁/𝑚 at 20 °𝐶 [64,65]. A similar value is found earlier from monolayer isobar measurements when the area change of DPPC molecules at the melting transition is considered [66]. According to this study at surface pressure of 30 𝑚𝑁/𝑚, the melting phase transition lays inbetween molecular areas of 50.5 Å2 and 60 Å2 , although others suggested that the equivalence pressure is 50 𝑚𝑁/𝑚 [67,68]. These discrepancies are examples demonstrating that direct comparison of a monolayer with a bilayer is not trivial. The application of a monolayer as half-a-bilayer should be done cautiously considering the level of complexity of the system and phenomena under investigation. For example, in membrane systems with asymmetric bilayers where coupling between domains across bilayer leaflets have profound effect the direct comparison between a monolayer and a bilayer is not correct.

Figure 15. Lipid domains in native pulmonary surfactant membranes consisting of lipids and proteins. (A) Atomic force microscopy image of a supported bilayer on mica. (B) Fluorescence microscopy image of a giant unilamellar liposome image [58]

One very suitable membrane model system is a bilayer placed on solid support (e.g. mica). Formation of the supported lipid bilayers can be performed either by the use of LB technique or by fusion of lipid vesicles at the solid substrate. Relating the observed structure in supported bilayers to the structure of free bilayers in the form of vesicles is not straightforward. The change of geometry from a sphere to a plane modifies the bilayer curvature leading to substantial consequences in the phase behavior of the lipid bilayer [69]. Moreover the supported bilayer is positioned in very close proximity to the solid support separated only by a thin layer 29

of liquid, which could also affect the phase behavior [70]. As a visual comparison at Fig. 15 are shown an image of a supported bilayer obtained from AFM (Fig. 15A) and fluorescent microscopy image (Fig. 15B) of unilamellar vesicle. The discovery of the synchrotron X-ray sources which provide intense, wellcollimated beams paved the way for implementation of surface-specific X-ray scattering methods for detailed structural investigations of monolayers spread at air/liquid interface. The Grazing-incidence X-ray diffraction (GIXD) and specular X-ray reflectivity (XR) are methods which provide direct information at subnano scale level about the state of Langmuir monolayers by in situ investigations of molecular structures and events at the air-liquid interface. The X-ray scattering experiments are performed with an exceptional sensitivity by means of liquid surface diffractometer equipped with advanced electronic modules for control of thermodynamic parameters such as surface pressure, temperature etc. thus allowing investigation of phase transitions and reactions which occur at air/water interface [71,72].Diffraction and reflection of x-ray from spread at the air/water interface monolayers are special cases of scattering from a surface. Langmuir monolayers which consist of various phases all in practice exhibiting certain angular orientation can be considered as an equivalent of a 3D powder. Hence the essentials of the grazing-incidence of X-rays theory on air/water interface are adopted from classical 3D crystallography for two X-ray scattering geometries (Fig. 16). The one of them is applied for probing the monolayer lateral structure by grazing-incidence X-ray diffraction (GIXD), (Fig. 16A) while the other extracts information about the vertical structure of a monomolecular layer at the air-water interface by measurements of the specular X-ray reflectivity (XR) (Fig. 16B). Grazing-incidence X-ray diffraction (GIXD) is performed with a constant incident glancing angle, 𝛼𝑖 less than the critical angle, 𝛼𝑐 [73], while the diffracted intensity is recorded as function of the horizontal and vertical angles 2𝜃𝑥𝑦 ≠ 0 and 𝛼𝑓 ≥ 0, respectively. The directions of the incident and scattered X-rays are ⃗⃗⃗𝑖 | = |𝑘 ⃗⃗⃗⃗𝑓 | = |𝑘 ⃗ | = 2𝜋. The conveniently given by the wave vectors, ⃗⃗⃗ 𝑘𝑖 and ⃗⃗⃗⃗ 𝑘𝑓 , where |𝑘 𝜆

scattering process is characterized by the scattering vector, 𝑞 = ⃗⃗⃗ 𝑘𝑖 − ⃗⃗⃗⃗ 𝑘𝑓 which can be separated into its horizontal and vertical components, 𝑞𝑥𝑦 and 𝑞𝑧 , respectively, (Fig. 16A) with magnitudes: 2

𝑞𝑥𝑦 = √|𝑞 ⃗⃗⃗⃗𝑥 |2 + |𝑞 ⃗⃗⃗⃗𝑦 | = ⃗ |√(cos 2𝜃𝑥𝑦 cos 𝛼𝑓 − cos 𝛼𝑖 )2 + (sin 2𝜃𝑥𝑦 cos 𝛼𝑓 − 0) = = |𝑘 2𝜋 𝜆

√cos 2 𝛼𝑖 + cos2 𝛼𝑓 − 2cos 𝛼𝑖 cos 𝛼𝑓 cos 2𝜃𝑥𝑦 ≈

4𝜋 𝜆

sin 𝜃𝑥𝑦 (2.5) 30

and ⃗ |(sin 𝛼𝑓 − sin 𝛼𝑖 ) ≈ 2𝜋 sin 𝛼𝑓 𝑞𝑧 = |𝑘 𝜆

(2.6)

Components of internal scattering vector can be obtained from substitution of the extracted from Snell’s law refraction corrected initial (𝛼𝑖′ = √𝛼𝑖 2 − 𝛼𝑐 2 ) and final (𝛼𝑓′ = √𝛼𝑓 2 − 𝛼𝑐 2 ) angles into eqs. (2.5) and (2.6). In mathematical terms the domain of the scattering vector 𝑞 and the wave vectors ⃗⃗⃗ 𝑘𝑖 and ⃗⃗⃗⃗ 𝑘𝑓 is called reciprocal space.

Figure 16. (A) Geometry for grazing-incidence X-ray diffraction from a liquid surface, showing the scattering triangle in reciprocal space. (B) Geometry for specular X-ray reflectivity from a liquid surface, showing the scattering triangle in reciprocal space.

Diffraction experiments on Langmuir monolayers occur under grazing angels to enhance surface sensitivity. For X-rays of wavelength λ~1Å the refractive index n, of condensed media for electromagnetic waves is almost a unity [74]: 𝑛 = 1 − 𝛿 − 𝑖𝛽

(2.7)

31

where 𝛿 =

2𝜋𝜌𝑟0 𝑘2

is the monolayer thickness, 𝑘 =

2𝜋 𝜆

is the X-ray wavenumber, ρ the

electron density 𝑟0 = 2.82 × 10−15 𝑚 is the classical electron radius. Typically δ is of the order 10−6to 10−5 in the condensed matter and only 10−9 in air. The term β has 𝜇 a relation with the linear absorption coefficient μ, with the relation 𝛽 = 2𝑘. When the refractive index is slightly smaller than unity it implies that total external reflection of radiation incident from vacuum (air) on a vacuum/condensed medium interface will occur at incident angles 𝛼𝑖 below a certain critical angle 𝛼𝑐 at which an evanescent wave propagates along the interface with exponentially decreasing amplitude into condensed medium [74]. The vertical decay length 𝛬 of radiation with wave number 𝑘 =

2𝜋 𝜆

then is given by: 2

𝜇 ′2 ′4 𝑖 ) = 𝑘√2 [√𝛼𝑖 + (𝑘 ) − 𝛼𝑖 ]

−1 (𝛼

𝛬

2

(2.8)

For a limit case when 𝛼𝑖 ≪ 𝛼𝑐 the incident grazing angle (𝛼𝑖 ) is imaginary which justifies defining the critical angle as 𝛼𝑐 = √2𝛿. For an air-water interface, at 𝜆 = 1.3 Å the value of the critical angle is 𝛼𝑐 = 0.13°. Thus for the decay length in water after substitution is obtained 𝛬(𝛼) = 𝛬0 = (2𝑘𝛼𝑐 )−1 = 0.0219 Å−1 ≈ 46 Å [74]. Diffraction experiments on Langmuir monolayers use typically grazing incident angles of 𝛼𝑖 = 0.85𝛼𝑐 , which means that the beam illuminates only the upper ~90 Å beneath the water surface, thus enhancing the surface sensitivity and reducing the background scattering from the subphase. According to fundamentals of bulk (3D) crystallography a 3D crystal, built by repetition of identical unit cells on a 3D lattice is defined by the primitive vectors 𝑎, 𝑏⃗ and 𝑐 and because of the interference of scattering from all crystal’s unit cells, Bragg diffraction occurs at points 𝑞 ≡ 𝑞ℎ𝑘𝑙 satisfying the three Laue conditions: 𝑞 ∙ 𝑎 = 2𝜋ℎ,

𝑞 ∙ 𝑏⃗ = 2𝜋𝑘,

𝑞 ∙ 𝑐 = 2𝜋𝑙

(2.9)

where the integers ℎ, 𝑘 and 𝑙 are the Miller indices and the set of points {𝑞ℎ𝑘𝑙 } is the reciprocal lattice defined by the lattice vectors 𝑎 ∗ , 𝑏⃗ ∗ and 𝑐 ∗ as follows: 𝑞 ≡ 𝑞ℎ𝑘𝑙 = 𝑎∗ ℎ + 𝑏⃗ ∗ 𝑘 + 𝑐 ∗ 𝑙,

ℎ, 𝑘, 𝑙 ∈ ℕ

(2.10)

From equations (2.9) and (2.10) for the vector 𝑎 ∗ can be obtained the apparent relations: 𝑎∗ ∙ 𝑎 = 2𝜋, 𝑎∗ ∙ 𝑏⃗ = 0, 𝑎∗ ∙ 𝑐 = 0; and similarly for 𝑏⃗ ∗ and 𝑐 ∗ . The crys2𝜋

tal plane spacing 𝑑 is given by 𝑑 = |𝑞⃗|. 32

Following this theoretical frame the diffraction from Langmuir monolayers which occurs if molecules are packed at air/water interface in 2D- crystalline order with a repeat of a unit cell along only two primitive vectors 𝑎 and 𝑏⃗ within the monolayer plane there are only two Laue conditions: 𝑞 ∙ 𝑎 = 2𝜋ℎ,

𝑞 ∙ 𝑏⃗ = 2𝜋𝑘,

(2.11)

Then, in reciprocal space, the diffracted intensity occurs where the horizontal scattering vector component,⃗⃗𝑞𝑥𝑦 , coincides with a reciprocal lattice vector 𝑞ℎ𝑘 satisfying eq. (2.11), while the 𝑞𝑧 component is not similarly restricted. Thus, instead of reciprocal lattice points (or Bragg points), eq. (2.10), the diffracted intensity is extended along Bragg rods defined by⃗⃗𝑞 = (𝑞𝑥𝑦 ; 𝑞𝑧 ), where the 𝑞𝑧 component is unrestricted and 𝑞𝑥𝑦 ≡ 𝑞ℎ𝑘 = 𝑎∗ ℎ + 𝑏⃗ ∗ 𝑘,

ℎ, 𝑘, ∈ ℕ

(2.12)

Here the two reciprocal lattice vectors 𝑎∗ and 𝑏⃗ ∗ are parallel to the monolayer plane, 2𝜋 𝑎∗ is orthogonal to 𝑎 and 𝑎∗ ∙ 𝑎 = 2𝜋; and similarly for 𝑏⃗ ∗ . Thus, | 𝑎∗ | = 𝑎 sin 𝛾

⃗∗

2𝜋

and | 𝑏 | = 𝑏 sin 𝛾, where 𝛾 is the angle between 𝑎 and 𝑏⃗. Fig. 16A shows the geometry for GIXD in terms of the vertical incidence angle, 𝛼𝑖 ≈ 0 and the exit angle, 𝛼𝑓 of the X-rays and the horizontal scattering angle 2𝜃𝑥𝑦 . The angle 𝛼𝑓 between horizon and the diffracted beam determines the vertical component, 𝑞𝑧 of the scattering vector [75]. Unlikely to 3D crystallography, where often plenty of Bragg reflections are recorded, in monolayers only few Bragg rods can be measured. This is related mainly to the small amount of intercepting matter with the X-ray beam and to various kinds of crystalline disorder. As a result, crystal structure solution using methods directly adapted from conventional 3D crystallography are possible to a limited extent only. Conversely, it must be noted that each Bragg rod pattern contains much more information than just a single Bragg point from a 3D single crystal. Therefore important structural information can be extracted from the Bragg rods and in some cases these data allow the refinement of, e.g., a molecular model defined in terms of a few rigid bodies. It provides complete structural model for the molecular arrangement in the monolayer. The amplitude of diffracted radiation depends on repeated scattering units in Langmuir monolayer and on their internal structure, i.e. the structure factor 𝐹(𝑞 ) of the film. Diffraction intensity is a measurable quantity which results from the absolute square of structure factor i.e. 𝐼(𝑞 )~|𝐹(𝑞 )|2 and is influenced also by other factors. Thus, along the (ℎ, 𝑘) Bragg rod, the diffracted intensity is given by: 𝐼(ℎ, 𝑘, 𝑞𝑧 ) = |𝐹(ℎ, 𝑘, 𝑞𝑧 )|2 ∙ 𝑒 −(𝑞ℎ𝑘

2𝑢 2 ℎ𝑜𝑟 +𝑞𝑧 𝑢𝑧 )

∙ |𝑇(𝛼𝑓 )|

2

(2.13) 33

The structure factor |𝐹(ℎ, 𝑘, 𝑞𝑧 )|, which is Fourier transform of the electron density of molecules in the unit cell, given by: 𝐹(ℎ, 𝑘, 𝑞𝑧 ) = ∫𝑟∈𝑈𝑛𝑖𝑡 𝐶𝑒𝑙𝑙 𝜌 (𝑟)𝑒 𝑖(𝑞ℎ𝑘∙𝑟+𝑞𝑧 𝑧) 𝑑 3 𝑟

(2.14)

The Fourier transform function 𝐹(ℎ, 𝑘, 𝑞𝑧 ) can be also written in terms of the constituent atoms with form factors (𝑓𝑗 ) [75]: 𝐹(ℎ, 𝑘, 𝑞𝑧 ) = ∑𝑗∈𝑈𝑛𝑖𝑡 𝐶𝑒𝑙𝑙 𝑓𝑗 𝑒 𝑖(𝑞ℎ𝑘∙𝑟𝑗+𝑞𝑧 𝑧𝑗) 2

(2.15) 2

The exponential in eq. (2.13), 𝐷𝑊(𝑞) = 𝑒 −(𝑞ℎ𝑘 𝑢ℎ𝑜𝑟 +𝑞𝑧 𝑢𝑧) is Debye-Waller factor, which takes into account the variations in positional order due to thermal disorder or capillary waves excited at the interface. It involves two terms 𝑢ℎ𝑜𝑟 and 𝑢𝑧 , which are the mean-square atomic displacements in the horizontal and vertical 2

directions, respectively. The factor |𝑇(𝛼𝑓 )| describes interference of X-rays diffracted upwards with X-rays diffracted down and subsequently reflected back up 2

by the interface. The factor |𝑇(𝛼𝑓 )| equals unity except near 𝛼𝑓 = 𝛼𝑐 where it peaks sharply (the Yoneda-Vineyard peak) [76,77]. It is convenient for deducing the zero-point of the 𝛼𝑓 scale, which often covers a range of 0 to 10°, but otherwise 2

|𝑇(𝛼𝑓 )| is unimportant for our purposes. So far it has not been possible to prepare a monomolecular layer that was a single 2D-crystal. The films usually consist of a large number of 2D-crystalline domains, each with a different orientation around the surface normal, i.e. a 2D powder. Therefore, the horizontal components, 1⁄

𝑞𝑥 and 𝑞𝑦 , can be measured only in their combination 𝑞𝑥𝑦 = |𝑞𝑥𝑦 | = (𝑞𝑥 2 + 𝑞𝑦 2 ) 2 but not individually, and the measured intensity is a sum over Bragg rods (ℎ, 𝑘) which have the same |𝑞⃗ 𝑥𝑦 |. Thus, denoting this set of reflections {ℎ, 𝑘}, the measured signal is: 𝐼{ℎ𝑘} (𝑞𝑧 )𝑚𝑒𝑎𝑛𝑠 = ∑(ℎ𝑘)∈{ℎ𝑘} 𝐼(ℎ, 𝑘, 𝑞𝑧 )

(2.16)

In terms of the vertical incidence and exit angles, 𝛼𝑖 and 𝛼𝑓 , and 2𝜃𝑥𝑦 , the angle between the horizontal projections of the incident and diffracted beams (Fig. 16A), the vertical and horizontal scattering vector components are given by eqs. (2.5) and (2.6) [75]. Analogous to the plane spacing in 3D, the repeat (line-)spacing 2𝜋

is 𝑑 = 𝑞 . The Bragg peaks can be provided by Miller indexes, ℎ and 𝑘, and then 𝑥𝑦

the 𝑑-spacing (or 𝑞𝑥𝑦 ) can be used to determine the lattice parameters 𝑎, 𝑏, the angle between the vectors 𝑎 and 𝑏⃗, 𝛾 and the area of the unit cell, 𝐴𝐶𝑒𝑙𝑙 = 𝑎𝑏 sin 𝛾, using

34

𝑑=

2𝜋 𝑞𝑥𝑦

ℎ 2

𝑘 2

ℎ𝑘

= [(𝑎) + (𝑏) − 2 (𝑎𝑏) cos 𝛾]

In some cases the quantity 𝑑𝑚𝑖𝑛 =

2𝜋 𝑞𝑚𝑎𝑥

−1⁄2

sin 𝛾

(2.19)

is referred to as the (real space) res-

olution of a diffraction data set. A purely vertical scattering vector is required in order to describe the vertical structure of monomolecular layers or interfaces. For the Specular X-ray reflection (XR) experiments the incident, 𝛼𝑓 and reflected, 𝛼𝑓 angles are the same, and also 𝛼𝑓 (Fig. 16B), and the scattering vector, 𝑞𝑧 = 2𝑘 sin 𝛼 =

4𝜋 𝜆

sin 𝛼. The X-ray re-

flectivity, 𝑅(𝑞𝑧 ) is measured by a detector which is moving on the 𝛼𝑓 arc. Typically, 𝛼𝑓 is in the range 0 < 𝛼𝑓 < 5°. Given the electron density profile, 𝜌(𝑧) across the interface, there are essentially two different methods for calculating the reflectivity 𝑅(𝑞𝑧 ). It can be calculated dynamically by matching the electromagnetic wave fields at each interface, or kinematically using the so-called “master formula for reflectivity”, 𝑅(𝑞𝑧 ) = 𝑅𝐹 (𝑞𝑧 ) |(𝜌𝑆𝑢𝑏 )−1 ∫

𝑑𝜌(𝑧) 𝑑𝑧

′

2

𝑒 𝑖𝑞𝑧 𝑧 𝑑𝑧|

(2.20)

Although the “master formula” from eq. (2.20) is derived by approximation, it is sufficiently accurate when used with Langmuir monolayers [78]. However, it is important to use the refraction corrected of scattering vector 𝑞𝑧 =

4𝜋 𝜆

sin 𝛼𝑖′ ,

where 𝛼𝑖′ = √𝛼𝑖 2 − 𝛼𝑐 2 . The Fresnel reflectivity 𝑅𝐹 (𝑞𝑧 ) is calculated from standard optical principles for a perfectly sharp interface between air and pure subphase by 2

𝛼

Fresnel's law, which in the small-angle limit for 𝛼𝑖 ≫ 𝛼𝑐 becomes 𝑅𝐹 ≈ (2𝛼𝑐 ) . In 𝑖

−3

(2.20) the electron density 𝜌𝑆𝑢𝑏 for water is 𝜌𝑤𝑎𝑡𝑒𝑟 = 0.334 𝑒/Å . The master formula, equation (2.20), shows that only the modulus, not the phases, of complex functions are measured by XR (or GIXD). Nevertheless, the measured normalized reflectivity

𝑅(𝑞𝑧 ) 𝑅𝐹 (𝑞𝑧 )

, usually is inverted to yield the laterally averaged electron den-

sity 𝜌(𝑧) of the structure as a function of the vertical z coordinate. Note that the (dimensionless) reflectivity 𝑅(𝑞𝑧 ) is determined on an absolute scale and that equation (2.20) contains no unknown factors. Therefore the electron density profile can be derived on an absolute scale, as well. The inversion can be performed either by a model independent method or using an explicit molecular or layered model of the interface [75]. It can be difficult to model complex systems such as monolayers of pure proteins or proteins adsorbed to a lipid monomolecular layer at the air-water interface. Although for a monolayer of simple amphiphiles can be applied a strategy for generating a structural model in which the electron density 𝜌(𝑧) is considered as a sum of contributions 𝜌𝑖 (𝑧) 35

from individual atoms at heights 𝑧𝑖 (or of pseudo atoms, say, CH2) modeled by a stack of homogeneous slabs. Then the model parameters can be the slab thicknesses, the number of electrons per molecule in each slab and the area per molecule along with either a common roughness 𝜎 of the slab interfaces, or individual interslab roughness’s, 𝜎𝑖 .

Figure 17. The electron density profile 𝜌(𝑧) (bottom) of a close-packed monolayer of arachidic acid (top). The electron density profile 𝜌(𝑧) can be constructed either as the sum of densities of individual atoms (or pseudo atoms, e.g. 𝐶𝐻2 ) (full lines) or as the sum of two slabs (dotted lines) of constant densities. The electron density due to the slabs is smeared by a roughness 𝜎 of typically ~3 Å (but for display purpose, the figure was constructed using 𝜎 = 1 Å).

Such a slab model is shown by a dotted line at Fig. 17 and the continuous electron density profile 𝜌(𝑧) obtained after applying the roughness 𝜎 is depicted by a dashed line (almost coincident with the solid line). The root-mean-square roughness 𝜎 for a bare water surface is about 3 Å and is due to thermally excited microscopic capillary waves [75]. This method usually is successful for monolayers of known molecules where the model can consist of typically two to five slabs: one section for the “tail” and one or more for the head-group region of the molecule. In a special case of straight tails lipids, comparison of the tail slab thickness 𝐿𝑇 with the calculated maximal length of the carbon chain yields an estimate of the tilt 𝐿

angle t (from cos 𝑡 = 𝐿 𝑇 ). The maximal length of a saturated carbon chain can be 𝐶𝐻

estimated by 𝐿𝐶𝐻 = (𝑛 + 9.8) × 1.265Å where 𝑛 is the number of CH2 groups and 9.8 is the contribution of the terminal CH3. Some simple rules can be derived for preliminary quantitative analysis of reflectivity data. For a two-slab model of, e.g. a monolayer of closely packed fatty acid, an acylglycero- or a phospholipid, consisting of a thin head group slab of 36

higher density in-between nearly equally dense subphase and tail slabs, it can be 𝑅(𝑞𝑧 ) 𝐹 (𝑞𝑧 )

shown that the first minimum in the observed 𝑅

3

data occurs at 𝑞𝑧 ≈ 2 (𝐿

𝜋 𝑇 +0.5𝐿𝐻

)

where the thicknesses of the head and tail regions are denote 𝐿𝐻 and 𝐿𝑇 . The above discussion applies to surfaces covered by homogenous monolayers. The interpretation of reflectivity data from heterogeneous monomolecular layers is more difficult. In either case the reflectivity measurements provide average structural information of the material in the X-ray foot print area. By contrast, a grazing incidence X-ray diffraction experiment will probe only domains with 2Dcrystalline order while the rest of the sample will contribute to the background intensity only. In this respect, and in the nature of the structural information that GIXD and XR deliver, the both methods can be considered as complementary. In summary: On liquid surfaces many amphiphilic molecules (e.g. phospholipids) are practically insoluble and form at the air–water interface Langmuir monolayers. Together with the classical methods starting with the endorsement of the Langmuir trough for measurements of the surface pressure versus surface concentration (i.e. surface area) variety of refinements and new methods are developed. The electrical properties of the monolayers are measured by the surface potential method while the morphological changes within the monolayer can be observed by fluorescence or Brewster angle microscopies. The most recently adapted to Langmuir trough are X-ray reflection and diffraction methods which allow the structural molecular ordering in monolayers to be investigated. From X-ray reflection data are determined the film thickness and electron density distribution normal to water surface while by the X-ray diffraction can be measured positional and oriental ordering within the lateral organization of Langmuir monolayer. Monolayers can be transferred from air/liquid interface onto solid supports by Langmuir– Blodgett (LB) and Langmuir- Schaefer (LS) techniques thus allowing the molecular architectures of mono- or multilayers to be created. The multilayer quality depends on many factors- the nature of solid support and liquid phase, the quality of the first monolayer transfer etc.

37

3. Lipases- Phospholipase A2 (PLA2) and Humicola lanuginose lipase (HLL) - biological importance and experimental approaches of studying lipolytic enzyme reactions at interfaces. 3.1. Lipases- their biological meaning Fatty acids are the metabolic fuel for living cells delivered from three sources: fats from diet, fats deposited in cells as lipid droplets, and fats synthesized in one organ then exported to another. The lipid metabolism involves various biochemical pathways where lipids are molecularly modified in order to meet the needs of a particular cell or tissue type. They are exported to cell organelles where are needed, either as a fuel or as structure builders and signal molecules. Lipids are hydrolyzed to fatty acids and mono-acylglycerols in order to be transported across the membranes. All these processes are catalyzed by specific enzymes and these which play a leading role in lipid metabolism are the lipases [79,80] and the phospholipases [81,82]. They are water-soluble enzymes that facilitate degradation of lipids and fats, by catalyzing the hydrolysis of ester bonds of triglycerides and phospholipids. The process of degradation is usually referred as lipolysis. It is a significant example of heterogeneous catalysis where water-soluble enzymes act at the interface formed by insoluble lipid substrates and in the meantime these reactions occur simultaneously with diverse interfacial phenomena. The activity of lipases is low on monomeric lipid substrates and rapidly increases on aggregated lipid structures. The "interfacial activation" of the lipases is unique property which distinguishes them form the usual esterases which act on water-soluble carboxylic esters [83]. On molecular lever the lipases are group of structurally well-characterized enzymes by means of variety of experimental methods like X-ray crystallography, site-directed mutagenesis, and classical biochemical methods [84]. As an example, Humicola lanuginose lipase (HLL) is part of a subfamily of lipases from microorganisms that all have the same active site defined by a 𝑆𝑒𝑟 − 𝐻𝑖𝑠 − 𝐴𝑠𝑝 triad [85]. It is found that the active site in HLL is accessible only through a hydrophobic pocket which is covered by a ‘‘lid’’, constituted by a two-turn -helix which can roll-over, giving rise to at least two different conformations. The first one is inactive form with closed lid while the second one with opened lid is the active form [86]. It was also discovered that the open form with the lid rolled over is favored in a media of low dielectric strength, e.g. the lipid–water interface [87,88,89]. The phospholipases are the other class of ubiquitous enzymes that selectively break down the phospholipids. These enzymes are widespread in nature where are involved in performing various digestive, remodeling and regulatory functions. The enzyme has molecular weight of around 14000. PLA2 molecule has bean-like geometrical shape with dimensions 22 Å × 30 Å × 42 Å [90].The phospholipases are classified into several subclasses depending on where they cut off a lipid molecule into two or more parts. Phospholipases А1, А2, В, С and D, are named 38

according to the position of hydrolyzed bond in substrate phospholipid molecule (Fig. 18).

Figure 18. Phospholipases classification according to their action of on phospholipid substrate molecule. (a) Phospholipase A1 (PLA1) cuts off a fatty-acid chain of a di-acyl phospholipid in the sn-1 position while (b) Phospholipase A2 (PLA2) attacks the fattyacid chain which is at the sn-2 position. (c) Phospholipase B can cut off both fatty-acid chains of a di-acyl phospholipid. (d) Phospholipase C cuts off the head group of a phospholipid. (e) Phospholipase D can cut off the polar head group of a phospholipid. R1 and R2 are the two fatty-acid chains, and “Polar Head” depicts a variable part of the head group of a phospholipid molecule

Phospholipase A attacks the fatty-acid chains at the glycerol backbone as phospholipase A1 (PLA1) cuts the chain at the sn1-position while the phospholipase A2 (PLA2) does that at sn2-position. As result of the cleavage a fatty acid and a lysolipid molecules occur as products of the enzyme reaction. From the rest of the phospholipases, phospholipase B can cut off both fatty-acid chains, phospholipase C the head group, thereby producing di-acylglycerol and phospholipase D cleaves the polar head group, leading to phosphatidic acid. The PLA2s are also subject of classification according to place where they are synthesized. There are two major groups of PLA depending on whether it happens in the endocrine glands or in the metabolic paths responsible for cell signaling: - secretion PLA2 (extracellular)-these phospholipases are components of the snake’s, bee’s or wasp’s venoms, different animal’s tissues, pancreatic fluid and bacterial secretions. Typically, these PLA2 are calcium dependent, i.e. they require Са2+ ions in order to fulfill their enzymatic action. - cytosol PLA2 (intracellular)- some of these phospholipases are involved in modification processes of the membranes e.g. as in the neural membranes or in forming the permeability barrier of the skin. Others PLA2 take part in the regulatory functions and cell signaling cascades, often in association with membranes, by producing special lipids like di-acylglycerol, phosphatidic acid, and ceramide. 39

Figure 19. Molecular Dynamics (MD) simulation of a Phospholipase A2 (PLA2) molecule bound at the surface of a phospholipid monolayer via domain called i-face which directly interacts with the lipid structure. The active site where the hydrolytic cleavage of the lipid takes place is located at the center of the enzyme. The substrate phospholipid molecule which is prone to fit into the active site is enhanced [91].

It is found from Molecular Dynamics (MD) simulations that distance between the membrane plane and the enzyme’s active site is approximately 1.5 𝑛𝑚 (corresponding to half the length of an extended DPPC molecule) (Fig. 19), thus implying that the molecule of lipid substrate has to protrude significantly in order to be plucked out of the lipid monolayer and to fit into the active site. The applied model assumes that the enzyme partially penetrates the lipid structure (so-called „tight binding‟), stabilizing the protruded lipid configuration by the hydrophobic amino acid residues which are exposed along the sides of the active site cavity [91]. 3.2.

Experimental approaches of studying lipolytic enzyme reactions at interfaces The most common and historically the oldest approach in biochemistry to understanding the mechanisms of enzyme reactions is the enzyme kinetics i.e. finding the rate constants of a certain biochemical reaction and determining how they change in response to variations in experimental parameters. To be fulfilled that aim variety of experimental methods and techniques have been employed over the years for studying of lipid hydrolysis [92]. At the beginning of 1970s with the pioneer work of Verger and de Haas is established the interfacial lipolytic catalysis on monolayers [93]. Their method is systematically developed over the last four decades by many research groups [94,95,96] proving the monolayer technique as one of the most reliable methods for interfacial kinetics studies. The advantage to control the interfacial quality of substrate surface is broadly explored in utilization of the method for studying different enzyme-substrate (monolayer) systems at air40

water interface, for example, Cutinase (enzyme)- polyethylene glycol/poly(D,L-lactide-co-glycolide) polymers (monolayer) [97]; HLL (enzyme)- polycaprolactone (monolayer) [98];Savinase (enzyme)- alpha gliadin (monolayer) [99] etc. In the original work of Verger and de Haas [93] for studying the hydrolysis of short- or medium-chain lipids organized in monolayer at air-water interface the classical Langmuir trough is modified to so-called zero-order trough which has two compartments interconnected with tiny channels (Fig. 20).

Figure 20. Surface barostat balance has two compartments: a reaction compartment where the enzyme solution (𝐸)is injected and a reservoir for supply of substrate molecules (𝑆). The two compartments are interconnected by a narrow surface channels. The surface pressure and the surface substrate concentration are maintained constant during the hydrolysis and solubilization of the reaction product (𝑃).

The first compartment is a place where the enzyme reaction occurs while the second one is a reservoir which supplies the reaction compartment with a constant flow of substrate molecules. Because the reaction generates soluble products which instantaneously desorb in the adjacent water phase it causes the decrease of surface pressure. In order to keep the monolayer in barostatic conditions (i.e. constant surface pressure) the trough barrier should move forward in order to decrease the total surface area occupied by the substrate molecules. Under such barostatic conditions in the time-course of hydrolysis process each desorbed molecule within the reaction compartment is instantaneously replaced by a molecule supplied from the reservoir thus the surface substrate concentration is maintained constant and the decrease of surface area reflects directly the kinetics of product formation. The barostat method is refined over the years, for example by adding a setup for simultaneous measurements of surface potential and it is broadly utilized for studying different enzyme-substrate systems at the air-water interface [100,101,102]. 41

Figure 21. Adaptation of the Michaelis- Menten kinetic scheme describing interfacial catalysis of short- and medium-chain lipids with soluble reaction products.

In order to describe the interfacial mechanism of enzymatic lipolysis on the base of classical Michaelis-Menten scheme are adapted several other kinetics models. The simplest one applied for the catalysis of short- and medium-chain lipids, considers the coupling between chemical Michaelis-Menten step and the process of enzyme penetration into lipid monolayer followed by a process of instantaneous solubilization of reaction products into the water subphase (Fig. 21) [103]. In order to be suited for the lipolysis of natural substrates which are long-chain lipids generating water-insoluble lipolytic products this simplified kinetic approach is modified into more generalized kinetic model where are added steps by taking into account the processes which modify the "interfacial quality" of the lipid structures, namely the interfacial molecular reorganization and the segregation of insoluble lipolytic products [103]. When lipid substrates are organized in micelles, liposomal dispersions or emulsions, the interface/volume ratio of these systems is very large and the magnitude of phenomena which take place (e.g. enzyme activation) at the interface is amplified. For example, the processes which might contribute to the enhancement of enzyme activity are: the possible exchange of enzyme molecules between lipid particles, the lipid exchange and substrate renewal in micelles when individual molecules are hydrolyzed to form water-soluble products, reorganization of the reaction products in the lipid bilayers leading to alteration and destabilization of the spherical liposomal structures, etc. [103]. The proposed kinetic models e.g. "surface dilution model" which describes the catalysis of mixed micelles [104,105] or "scooting and hopping modes" of PLA2 action on liposomal dispersions [106,107] allow some kinetic parameters to be obtained thus giving valuable information about the mechanism of enzyme action. Even given the uniqueness of the monolayer technique for studying enzyme reactions occurring at air/water interface, the method has limited abilities for studying the enzyme/substrate interactions and fine monolayer structure (e.g., monolayer heterogeneity). Fluorescence microscopy has proven to be a well-suited complementary technique in this respect as the first direct visualization of PLA2 action was performed in the late 1980s using wide-field microscopy on a lipid monolayer 42

[108]. Using this method has been shown the existence of LC lipid domains in phospholipid monolayers, which PLA2 can recognize and hydrolyze [109]. Recently, after a refinement of the experimental technique, Gudmand visualized the PLA2 activity on phospholipid monolayers and showed that adsorption and hydrolytic act are coupled with relaxation process of the bean shaped phase domains into circular round shapes (Fig. 22) [58].

Figure 22. Fluorescence image consequence obtained at different time points during the PLA2 induced degradation of LC-domains in a DPPC monolayer. The bright domains are the LE phase, and dark domains the LC-phase. The degradation is induced by present in the monolayer subphase. (A) Typical bean like domain shape immediately after compression is stopped. (B) After 34 minutes the change of the LC-domains from bean like to circular shape is visible. (C) and (D) The domains are degraded as a result of PLA2 action. The figure is adopted from [58]

3.3.

Lipase hydrolysis of supported lipid bilayers: an Atomic Force Microscopy Approach. Nevertheless effectiveness of the fluorescence microscopy as imaging method it only brings space resolution to the micrometer scale level. The method which brought the visualization of molecular events at the nanometer scale level was the Atomic Force Microscope. The very first AFM application for visualization of enzyme lipolysis of supported DPPC bilayers was demonstrated by Gaub’s group and reported by Grandbois et al [110]. They obtained time series of AFM images 43

during the process of degradation of supported DPPC bilayers achieving a spatial resolution close to 10 nm which allowed the enzyme behavior on a gel phase membrane to be analyzed (Fig. 23A-D) and a molecular mechanism of the enzyme action to be proposed (Fig. 23E). The AFM images also showed some details of the lateral changes of the bilayer topography during the lipolytic process, i.e. a formation of individual channels which are result of the single enzyme action. The consequences of AFM images obtained in the time course of the enzyme reaction are software processed and analyzed and data are presented as kinetic curves of hydrolyzed bilayer area versus time. It is also determined that the rate of the enzyme hydrolysis is proportional to the perimeter length of the existing bilayers defects [110].

Figure 23. AFM image consequence obtained at different time points during PLA2 induced degradation of a DPPC bilayer. (A) Topography image of the surface of DPPC bilayer before enzyme injection and (B-D) after enzyme injection into the liquid. The dark areas in the images are defects (holes) in lipid bilayer. (E) Schematic representation of a model of hydrolysis of a supported bilayer. The model shows a lipid organization at the rim of bilayer defects where the enzyme attack takes place. The figure is adopted from [110] 44

Next step in utilization of AFM as a tool for studying the hydrolysis of supported lipid bilayers is made by Nielsen’s et al, who report for the existence of a latency period (also known as “lag-phase”) in PLA2 degradation of supported phospholipid bilayers [111]. Lag-phase is defined as a period of low enzyme activity where no substrate hydrolysis is observed. It is estimated to be in a time range of 8 to 14 min. The duration of lag phase is determined by bilayer properties and its fine molecular structure, i.e. microheterogeneity, product accumulation etc. Following the Gaub’s experimental approach Nielsen et al have combined kinetic and structural study of hydrolysis of supported DPPC bilayers by AFM and revealed the formation of small depressions during the latency period which were indicative of domain formation by the products created in early stages of hydrolysis. Their data showed that the burst in enzyme activity was located around the product domains and was followed by a disruption of the lipid bilayer in the vicinity of these depressions. From the obtained AFM image subsequences are extracted kinetic data which correlate hydrolyzed area and the perimeter length of bilayer defects. From the plot of kinetic curves is further calculated the rate of hydrolysis of supported bilayers and the enzyme turnover. The experimental approach for studying the hydrolysis of supported lipid bilayers by AFM combines the advantages of Langmuir-Blodgett (LB) technique with those of AFM technique allowing a required molecular structure (i.e. lipid bilayer) to be precisely build and then to be visualized with nanoscale precision. A typical experimental AFM setup for such approach is depicted at Fig. 24.

Figure 24. (A) Schematics of an AFM liquid cell set-up accustomed for lipolysis of bilayers experiment. (B) Photo of Nanoscope V AFM optical head with mounted liquid cell connected to a syringe containing the enzyme solution which is flushed trough the sealed liquid chamber where the lipid bilayer sample is placed.

Using the LB-technique the supported lipid bilayers are transferred from air/water interface on freshly cleaved mica at constant surface pressure which corresponds to the liquid condense (LC) phase of the monolayer, i.e. closely packed 45

monomolecular layer (e.g. for DPPC this value is 35 𝑚𝑁/𝑚). The lipid layers usually are transferred vertically onto the mica supports, with a short pause between the first and second transfer. The supported bilayers have to be kept all the time under water and immediately moved to the AFM fluid cell. Another method of obtaining the phospholipid bilayers is a simple procedure of collapsing small unilamellar vesicles (SUVs) on supported substrates [112,113]. It requires obtaining a suspension of multilamellar vesicles (MLVs) where at first the lipid is dissolved in nonpolar solvent (e.g. chloroform) and then the solvent is evaporated by rotary evaporator. As a result a thick lipid film is spread over the wall of the flask. This lipid is further resuspended in buffer solution thus MLVs are formed. Next, from the MLVs are obtained small unilamellar vesicles (SUVs) by sonication of the suspension of MLVs with a titanium probe sonicator till achieving a limpidity of the suspension. In order to be removed the titanium debris emitted by the probe an additional centrifugation of liposome suspension is applied. Next, on freshly cleaved mica sheet a certain amount of newly prepared SUVs solution is deposited and incubated for some time (20 to 30 min) at temperatures above the main phase transition of the constituent lipid (e.g. 55 ℃ for DPPC). The excess of vesicles is then removed by exchanging the solution covering mica surface with a buffer and the sample is installed in the AFM liquid cell sealed with O-ring. The microscope is allowed to thermally equilibrate before imaging. Prior imaging the liquid cell is flushed with buffer without enzyme. Before enzyme injection the bilayers are equilibrated in the fluid cell for at least half an hour to be reduced the cantilever drift. Then the enzyme solution with certain concentration is flushed trough AFM liquid cell and the first available image is captured. All imaging is usually performed in contact mode and the force between tip and the sample is maintained manually at a minimum by adjustment of the set point. This approach of force control assures that the loading force is less than 500 𝑝𝑁. To prevent any mechanical influence of the tip on hydrolysis, every other frame is scanned with the tip away from the surface except at the very beginning of hydrolysis. Under these conditions, the bilayer can be imaged safely preserving its entity [112,113]. A typical time-course set of images from an AFM experiment with collapsed liposomes prior and after injection of PLA2 is shown at Fig. 25. Few bilayer defects (dark areas) are easily distinguishable in the first image (Fig. 25A) obtained prior to the enzyme injection. The whiter islands are also lipid bilayers or multilayers formed on top of mica supported bilayer. In a study of properties of supported bilayers Nielsen has found that the bilayers formed on top of supported bilayer were much softer and very dynamical islands-like structures which shape at the phase transition temperature was always rounded thus indicating the dominance of line tension forces on their shape [16]. The AFM images shown at Fig. 25 reveal the progressive changes of the bilayer surface topology which occur as a result of enzyme action. 46

Figure 25. Action of PLA2 on supported DPPC bilayers. A typical time-course set of images from an experiment in which PLA2 solution is injected into the liquid cell (𝑠𝑐𝑎𝑛 𝑠𝑖𝑧𝑒 = 10 × 10 𝜇𝑚2 ). Image (A) is prior to the enzyme injection. The few bilayers defects (darker areas) are easily distinguishable. Image (B) is captured 5 𝑚𝑖𝑛 after enzyme injection. The enlargement of the existing defects is necked eye detectable. Images (C–F) are taken in the time interval 10 ÷ 40 𝑚𝑖𝑛 after enzyme injection showing occurrence of progressive changes in surface topology of the bilayer as a result of the enzyme action. The darker areas represent structural defects in the lipid bilayer. The whiter islands are also lipid bilayers formed on a top of the bottom bilayer (Images are personal unpublished data).

Every individual image from the obtained sequence is further processed by a software in order to be estimated the geometry (i.e. area and perimeter) and the growth of bilayer defects in time. These data are subsequently employed in different kinetics models for calculation the initial rate of the enzyme reaction, turnover number etc. The reproducible imaging of supported lipid bilayers in gel phase (i.e. LC phase) by contact mode AFM opened many opportunities for utilization of the method in studying the lipase hydrolysis at fluid interfaces. As an illustration, the example at Fig. 26 shows two images of mixed DPPC-DPG bilayers. One is obtained from a liposome suspension (Fig. 26A) the other by LB-deposition (Fig. 26B).

47

Figure 26. AFM images of supported DPPC/DPG bilayers obtained from liposome suspension (A) or from LB film (B). Scanning areas are 𝟑𝟎 × 𝟑𝟎 𝝁𝒎 and 𝟓 × 𝟓 𝝁𝒎 respectively. Bilayer in (A) contains large defects which makes LB technique preferable.

The dark regions represent defects (holes) in the bilayer. The height difference between the lipid film (bright regions) and mica substrate (dark regions) is approximately 6 nm, a value which is in a good agreement with X-ray diffraction [59] and AFM data [114] for the thickness of lipid double layers . The lipid surface has a root mean square (𝑟𝑚𝑠) roughness of about 1 Å. The images show that bilayers obtained by LB deposition typically are better defined than bilayers obtained by collapsing vesicles because the number of membrane defects is much smaller in the LB-film.

Figure 27. (A) Langmuir isotherms of pure DPPC spread on water and recompression isotherm of the same DPPC film after expansion on a HLL containing subphase. (B) Schematic illustration of HLL enzymes trapped in the upper DPPC monolayer; (C) 1 × 1 𝜇𝑚 image in deflection mode of single HLL enzymes evenly dispersed in the DPPC membrane. (D) Zoom (in height mode) on a single HLL enzyme. Inset shows the height profile of enzyme. 48

It was shown that AFM also allows imaging of single enzyme molecules trapped in a lipid bilayer. A typical example is presented at Fig. 27 where the images of single Humicola lanuginose lipase (HLL) molecules are shown trapped in a DPPC bilayer [115].The images are obtained by a transfer of dense DPPC Langmuir film to a mica support followed by a second transfer of a film containing both DPPC and HLL. The latter film is obtained by decompressing of DPPC Langmuir monolayer followed by recompression after HLL injection into the water subphase. The presence of HLL at the surface is detected from an increased surface pressure of the expanded DPPC monolayer as it is shown at the isotherm graph with dotted line at Fig. 27A while Fig. 27B schematically illustrates how HLL molecules are trapped in the upper DPPC monolayer. After imaging in AFM liquid cell the HLL molecules embedded in the LB transferred DPPC bilayer are clearly visible as protrusions in the flat bilayer background (Fig. 27C, D). The HLL molecules seem to have no tendency to cluster in the membrane as they appear uniformly scattered (Fig. 27C). In Summary: Lipases and phospholipases are interfacially activated watersoluble enzymes which take part in fat’s metabolism by facilitating the degradation of lipids and fats in a process referred as lipolysis. For studying the enzyme kinetics i.e. finding the rate constants and determining how they change in response to variations of kinetic parameters variety of experimental methods and techniques are utilized for studying the lipid hydrolysis. With the establishment of interfacial lipolytic catalysis on monolayers by introduction of the “zero order” trough and following refinements of the method over the years, the monolayer technique is recognized as one of most reliable for interfacial kinetics studies. The method though has limited abilities for studying the enzyme/substrate interactions and fine monolayer structure. The involvement of fluorescence and Brewster angle microscopies as a complementary techniques provided direct visualization of lipase action on lipid substrates although as imaging methods the both techniques bring the space resolution only down to the micrometer scale. The method by which the molecular phenomena were monitored for the first time at nanometer scale level was AFM. Equipped with purposely developed liquid cell AFM visualized the degradation of supported bilayers where the obtained in the time course of hydrolysis series of AFM images had spatial resolution close to 10 nm thus some detailed changes in the bilayer lateral organization were observed with nanoscale precision. In the proposed experimental approach from the captured consequences of AFM images in the time course of enzyme reaction are extracted data which are implemented in kinetic curves of hydrolyzed bilayer area versus time thus valuable information for the mechanism of lipase action is obtained.

49

4. Phospholipase A2 (PLA2) and Humicola lanuginose lipase (HLL) captured in act by AFM. Phospholipase A2 (PLA2) is a water-soluble enzyme which acts at the lipid/water interface where both the activity and kinetics of hydrolysis strongly depend on morphology and physicochemical characteristics of the substrate. Nielsen et al on the basis of their AFM data proposed [111] a molecular mechanism of the process of PLA2 hydrolysis of a supported DPPC bilayer which schematically is depicted at Fig. 28. It primarily illustrates what type of data can be acquired from the AFM imaging. Upon the enzyme action the supported bilayer is hydrolyzed and because the reaction products (i.e. lysolipid (LysoPC) and palmitic acid (PA)) are dissolvable into adjacent buffer phase, areas with deep bilayer defects occur. These defects are the primary experimental observable. Assuming that the rate of hydrolysis is proportional to the rate of product desorption, then quantitative data for the enzyme kinetics can be obtained by equating the degree of hydrolysis to the total area of bilayer deep defects. Furthermore, this experimental strategy is followed for studying the influence of product phase separation on PLA2 activity and in study of vipoxin PLA2 hydrolysis of supported phospholipid bilayers.

Figure 28. Schematic illustration how the enzymatic PLA2 hydrolysis is recorded by AFM. The mica supported bilayer is hydrolyzed by PLA2 and as the products dissolve from mica support to the buffer adjacent phase, the occurrence of deep bilayer defects is recorded by AFM. Such defects are defined as a primary observable in the experiment and are used as a measure of the degree of hydrolysis. Regions with an intermediate height comprised of the product molecules can also be detected. The broken line indicates the surface topography recorded by the AFM tip. Figure is adopted from [176].

The same experimental approach is also considered for studying the kinetics of Humicola lanuginose lipase (HLL) action. In this case for the first time is demonstrated how AFM can be applied for studying the HLL action on hybrid lipid/phospholipid (i.e. Mono-Oleoyl-rac-Glycerol (MOG)/DPPC) bilayers as well as the synergetic action of PLA2 and HLL on mixed lipid-phospholipid bilayers is also considered. 50

4.1.

Influence of product phase separation on PLA2 hydrolysis of supported DPPC bilayers studied by AFM The latency period (lag phase) of phospholipids hydrolysis in bilayer cell membranes by PLA2 is characterized by low enzyme activity, which is followed by a burst [116]. It is suggested that the burst is caused by an accumulation of hydrolysis products, free fatty acid (e.g. palmitic acid (PA)) and lysophospholipid (e.g. 1-palmitoyl-2-hydroxy-sn-phosphocholine (lysoPC)) released during the latency period, thus altering the susceptibility of phospholipid membranes to degradation by the phospholipase [117,118]. The course of lipolytic reaction is determined to large extent by the membrane properties. It is demonstrated that the duration of latency period can be modulated by varying the experimentally accessible parameters e.g. Ca2+ concentration, ionic strength, temperature, lipid composition, density fluctuations, vesicle curvature and addition of hydrolysis products [119,120,121,122]. Hence, the changes prompted by addition of products and the importance of products at different stages of hydrolysis have provoke a great scientific interest [123,124,125]. It was reported that in case of large unilamellar DPPC vesicles the product concentration close to 8 mol % abolishes the lag phase [126] which was also confirmed by thermodynamic data indicating that the phase separation occurs at the same product concentration [127]. The individual roles of hydrolysis products were distinguished, namely that ionized fatty acids promote calcium- independent binding of PLA2 to vesicles through electrostatic interactions while lysolipids enhance bilayer susceptibility to the enzyme attack but at the same time attenuate the binding of PLA2 by changing the structure of fatty acid domains within the bilayer [128]. In addition to phase separation, the increasing amounts of hydrolysis products also lead to vast morphological changes in the vesicular systems. The original vesicular suspension changes into a poorly defined mixture of punctured vesicles, disk micelles, and normal micelles. The structural studies by Cryo-Transmission Electron Microscopy (Cryo-TEM) show that these changes take place already during the latency period where only a few percent of the substrate molecules are hydrolyzed. It is suggested that a cascade process occurs where the initial changes in lipid composition alter the morphology of the substrate, which in turn enhances the rate of hydrolysis [129]. The implementation of the high-resolution structural techniques such as AFM and X-ray diffraction brought new insights in characterizing the interactions between lipases and their substrates [130,131,132]). The first visual evidence of the occurrence of a latency period at nanoscale level was presented by Nielsen et al [111]. In this study, the combined kinetic and structural AFM data for hydrolysis of supported DPPC bilayers reveal the existence of small depressions formed during the lag phase which are indicative for domains’ formation by products accumulated within the bilayer in the early stages of hydrolysis (Fig.29).

51

Figure 29. AFM images of a time sequence of PLA2 hydrolysis of DPPC bilayer transferred as LB-film on mica. (A) Image prior to PLA2 injection, (B) (C) and (D) images captured in the time interval 8 ÷ 17 𝑚𝑖𝑛 after PLA2 injection. Bright gray areas are intact bilayer surfaces whereas the dark correspond to the bare mica support. Arrows in B mark small depressions on bilayer surface, which represent product enrich domains. Arrows in C show small narrow channels, 15 ÷ 20 𝑛𝑚 wide, created by a single enzyme molecules. All images are of size 2 × 2 𝜇𝑚. (E) The graph of the hydrolyzed area versus time from the analysis of complete AFM image series. Figure is adapted from [111].

The lag-burst kinetics is clearly illustrated from the sudden change in bilayer morphology occurred between 8 and 14 min after PLA2 injection. This time interval corresponds well to estimated lag-time which is between 8 and 10 min. The data furthermore show that the burst in activity is centered in proximity of the product domains and the lipid bilayer disruption starts within the existing bilayer depressions as it is depicted in the inset of Fig. 29B. 52

One of the important benefits of using supported bilayers as substrates for PLA2 hydrolysis is that they are morphologically well-defined structures which allow the increase of product concentration range with eliminating the shape and curvature changes from the theoretical interpretation. One step further in exploring this advantage is to experiment by varying the initial product concentration within the accessible product concentration range of 0 ÷ 100% with the same initial bilayer morphology for all concentrations. The experimental procedure (Appendix A) involves AFM study of LB films of lipid substrate/ lysoproducts molecular mixtures (e.g. DPPC/PA/LysoPC) with initial product concentrations ranging from 0% to 100%, transferred on mica supports. These mixed lipid substrate/ lysoproducts systems are intended to emulate different degrees of hydrolysis of DPPC bilayer. All these mixtures which form stable monolayers are successfully transferred from air/ water interface to solid supports and then moved into AFM liquid cell for kinetic and structural analysis. It is well known that large-scale domain formation caused by phase separation in mixed molecular systems does not occur in the Langmuir monolayer when the subphase is pure water. Hence, the LB transfer is performed without Ca2+ ions in the subphase to avoid appearance of micronsized domains characteristic for the large-scale phase separation observed in Langmuir monolayers [133]. Without calcium the supported bilayer resembles vesicular systems and thus mimics a bilayer that is partially degraded by the enzyme [111]. Two monolayers which contain an initial product concentration of 25%, 50%, 75%, and 100% are subsequently one by one transferred vertically onto solid mica supports (Fig. 30A). Then the supported lipid bilayers are kept under water and immediately transferred into AFM liquid cell for their structural characterization. Next by an appropriate software is applied AFM image analysis for quantification of the appearance of holes within the bilayer as well as their growth in the presence of PLA2. The height differences measured in bilayer topography are result either of phase separation or incompletely covered by lipid molecules areas of mica support where the latter appear as holes in the bilayer (Fig. 30B). When the enzyme solution is flushed through the AFM liquid cell these bilayer holes start to grow and new ones are created. The hydrolysis of phospholipid molecules leads to the appearance of product molecules which do not remain in the bilayer but desorb from mica support, thus creating holes in the bilayer. Determining the area fraction of holes in the bilayer at a given time serves as a measure of the degree of hydrolysis. Although measured this way, the degree of hydrolysis is subjective because of desorption of unhydrolyzed DPPC molecules from the adjacent to mica bottom layer and it must be taken into account in the kinetics calculations. In the structural appearance of supported bilayers the majority of surface area (between 95% and 100% of all freshly prepared bilayers transferred to the AFM liquid cell) consists of a uniform flat bilayer with a composition which is a replica of the composition of the Langmuir monolayer. In addition to the uniform bilayer structure, two other features are also observed. The first is bilayer deep 53

holes where the mica support is not covered with molecules. These defects are considered as structural defects and they are further used as a measure for the degree of hydrolysis. As it is shown in Table 4.1, the structural defects are present on all bilayer samples before the enzyme injection into the AFM liquid cell. The individual defects have different shapes and sizes and their number varies between samples but in all cases the area fraction covered by this type of defects never exceeds 5% (i.e. areas of less than 0.6 m2). The average measured depth of the defects is 6 nm, a value which is in good agreement with earlier measurements of the thickness of DPPC bilayer [111,134].

Figure 30. (A) Schematic illustration of the transfer of mixed monolayers to a solid support. First a mixed DPPC/Lyso-products monolayers are LB- transferred one by one to the solid support, as the bilayer is formed the down dip of the support. (B) In the AFM liquid cell, the bilayer is hydrolyzed by PLA2 (shown as its crystal structure), creating additional product molecules. When the structure of the bilayer is observed by AFM the height difference between the top of the bilayer and mica substrate and between the top and product regions are measured.

The second type of defects are small depressions with a depth of 3 ÷ 5 Å relative to the top of uniform bilayer. In contrast to the structural defects these depressions appear as a result of the lateral heterogeneity in the bilayer caused by phase separation in lipid monolayer leading to product domain formation. Thus, these defects are considered as compositional defects. Their lateral size also varies but it is generally one order of magnitude smaller than the size of the structural defects and the total bilayer area covered by compositional defects is less than 0.05 m2. Compositional defects are observed in pure DPPC bilayers at the end of the 54

lag phase just before the burst [111] and for bilayers initially containing 75% products. Despite the high product concentration in 25% and 50% samples, no signs of initial compositional defects are found. With 75% products initially present, only very small area fraction (~1%) of the bilayer has the characteristics of the compositional defects. Table 4.1 Summary of the kinetic and structural data obtained from the experiments Sustrate

PLA2

𝑫𝑷𝑷𝑪: 𝑷𝒓𝒐𝒅𝒖𝒄𝒕𝒔

𝑛𝑀

100:0

84

Initial defects observed in AFM

Lag time

Structural

10÷

𝑚𝑖𝑛

𝑽𝒎𝒂𝒙 𝜇𝑚𝑜𝑙 𝐷𝑃𝑃𝐶 𝑚𝑖𝑛 𝑚𝑜𝑙 𝑃𝐿𝐴2

Total hydrolys is %

0.01÷0.02

90÷100%

25 75:25

10

Structural

0

0.02÷0.07

10÷50%

50:50

10

Structural

0

0.02÷0.07

10÷50%

25:75

10

Structural and Compositional

0

0.02÷0.07

10÷50%

0:100

10

Structural

N/A

N/A

N/A

At Fig. 31A are shown AFM images of the initial topography (before PLA2 injection) of a bilayer with 75% products as the arrows mark the two different types of bilayer defects. When the bilayers are composed only of LysoPC and PA products in molar ratio 1:1, a uniform flat bilayer containing only structural defects reappeared (data not shown). These bilayers are fragile and unstable, and the mechanical influence of the tip creates an increasing area fraction of the defects. Product domains are found to be stable only in presence of DPPC where repeated scanning of the same area does not create additional defects. Fig. 32 shows the reaction time course of the hydrolysis of a bilayer initially containing 75% products. The area of the structural defects is quantified by image analysis and it is used as an estimate for the degree of hydrolysis.

55

Figure 31. Images of a bilayer containing 75% product molecules subject to the hydrolysis by PLA2. (A–D) The 5 × 5 𝜇𝑚2 images are selected from larger image of 15 × 15 𝜇𝑚2 (shown as insets in the upper left corner of each image). The white square in the 15 × 15 𝜇𝑚2 image inset in (A) shows where the enlarged region are selected. Arrows mark the two different defect types. The thick black arrows point to structural defects and the thin black arrows to compositional defects. The arrows point to the same defects in each image. (A) Before the injection of the phospholipase; (B) At the 2nd min after PLA2 addition, new structural defects were seen and also many new compositional defects (white arrows); (C) At the 4th min after PLA2 addition, almost all compositional defects had turned into structural ones; (D) At 6 min after PLA2 addition, all the structural defects then grew with time (the kinetic analysis is presented at Fig. 32).

The hydrolysis reaction starts immediately after enzyme injection without any signs of a latency period and proceeds at maximum velocity until it eventually slows down most likely because of the lack of substrate molecules. The reaction kinetics exhibited the same pattern for all bilayers initially containing products. Table 4.1 summarizes the results for PLA2 hydrolysis of bilayers with different initial product concentration. For comparison the AFM data obtained by Nielsen et al. [111] for pure DPPC bilayers with no products initially present are also included. From the data in Table 4.1 is clear that when products are initially present in the bilayer then the hydrolysis starts immediately after enzyme injection. In the Table 4.1 are also compared the growth rates of initial hydrolyzed area, calculated for bilayers which initially contained products, with the maximum hydrolyzed area growth rate obtained for DPPC bilayers without initial products [111]. As data show when products are initially present, the initial area growth rate is slightly higher than the maximum area growth rate after the burst for pure DPPC bilayers. 56

The total hydrolyzed amount quantified as the area fraction of structural defects caused by dissolution of product molecules also changes from 90% to 100% for pure DPPC bilayers to between 10% and 50% for bilayers initially containing products. At Fig. 31A-D, are presented the first four frames from a complete AFM image series analyzed and presented in the graph at Fig. 32A as degree of hydrolysis versus time. These images illustrate the temporal development of the action of PLA2 on a bilayer that initially has both structural and compositional defects (25% DPPC and 75% products). It is evident that upon the addition of PLA2 the number of structural defects increases in time as a sign of a hydrolysis process. However, more significant are the locations at bilayer surface where new structural defects appeared because these defects are generated in the bilayer regions composed by compositional defects which are attacked right away after enzyme addition thus following to solubilization of the molecules from these bilayer areas into the adjacent buffer phase. The initial structural defects are also growing in size, but with a rate of relative area increase which is much slower than the rate for the compositional defects, as illustrated at Fig. 32B. The relative area increase is found by dividing the area difference between two consecutively captured AFM image frames with initial area of the defect, expressed as ∆𝐴𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 =

𝐴𝑡2 −𝐴𝑡1 𝐴𝑡0

where 𝐴𝑡2 and 𝐴𝑡1 are areas of defect determined from

the analysis of two consecutive AFM images captured at time 𝑡2 and 𝑡1 respectively. The area 𝐴𝑡0 is the initial area of defects before addition of PLA2. The rate of relative area increase is 8.6 times larger for regions where the phase separation is distinguishable. Nielsen et al showed that during a fast hydrolysis the area growth rate of the defects is proportional to the edge length around the areas of defects [111]. If this is the case, when normalizing by the area, it is trivial, that the small defects such as the compositional ones will grow faster than the larger structural defects simply because of their larger perimeter-to area ratio. Hence, to illustrate any differences in the growth rate of the two types of defects instead of normalizing by area is more convenient the area increase to be normalized by the defects’ perimeter. This normalization formula can be expressed as ∆𝐴𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 =

𝐴𝑡2 −𝐴𝑡1 𝑝𝑡1

, where

𝐴𝑡2 and 𝐴𝑡1 are the same as above defined and 𝑝𝑡1 is the perimeter of the defect at time 𝑡1 . Fig. 32B shows the results of this calculation. First, the growth rates of compositional defects are larger than these of the structural defects. However, it happens only during the initial enzyme attack on the compositional defects when their growth rate has a maximum value. After that they become holes turning into new structural defects and their growth rates approach that of the other structural defects. The results at Fig. 32B imply that PLA2 has a preference for the compositional rather than structural defects.

57

Figure 32. (A) Kinetic analysis of the hydrolysis of a bilayer which initially contains 75% products. The degree of hydrolysis is determined from image analysis as the area fraction of structural defects equals the degree of hydrolysis. The hydrolysis starts immediately after enzyme injection without any sign of a latency period Comparison between the early time development of the structural and compositional defects. The defects chosen at time zero were identified on later frames such that they were the same defects that had been analyzed. (B) The relative area increase as a function of time. The area determined in each image was normalized by the area at time zero for each defect individually. (C) The normalized growth rate as a function of time. Here the area increase between two consecutive images was divided by the perimeter of the defect in the first image. This illustrates that the initial growth rate was larger for the compositional defects and that once the defects had become holes (i.e., like structural defects) their growth rate fell back to that of the structural defects.

Some important questions arise from the presented results. What is the influence of the high initial product concentration on PLA2 hydrolysis of the supported bilayers and how it changes the lateral bilayer structure? What is the relationship between the non-equilibrium structures created during the initial course of hydrolysis and the bilayer structure with high product content where phase separation occurs? Phase separation has previously been reported once a critical mole 58

fraction (~0.08) has been formed in vesicles [116,135,136,137]. The insertion of hydrolysis products into vesicles besides phase separation has an additional effect, namely that the occurred large morphological changes alone have a pronounced effect on the hydrolysis process [124]. In monolayers, large-scale phase separation between lysoPC and PA has been observed as well as domain formation in the ternary system DPPC: lysoPC: PA [133]. These domains are negatively charged, which indicates that they mainly consist of PA molecules. PLA2 has a preference for these fatty-acid-enriched domains thus the enzyme clusters underneath them. Any domain formation in monolayers, however, cannot be detected in the absence of calcium, independent of product concentration [133]. The AFM images of the bilayers with 75% products, however, show the presence of compositional defects indicative of some lateral phase separation. The domain size is fairly small, and these domains would be undetectable with fluorescence microscopy. The molecules constituting the compositional defects are likely to be highly enriched in PA. This interpretation is on the base of previous results which revealed that during hydrolysis both product molecules are released from the lipid membrane, but the lysolipid is released to a larger extent, thus forming a fatty-acid-rich region in the membrane [138]. In addition, it is well established that PLA2 has a preference for negatively charged bilayers or negatively charged regions of the bilayer caused by its net positive charge [139]. This preference of the enzyme to bind at enriched of lysoproducts bilayer’s domains is clearly illustrated at Fig. 31. The addition of PLA2 causes immediate desorption of the phase-separated regions. At last but not least, the loss of PC head-group makes the PA molecules shorter, thus making the regions in the bilayer enriched in PA to appear at the AFM images as small depressions which is in accordance with the observations. The length decrease of the molecule caused by the loss of the PC head-group is comparable to 3 ÷ 5 Å decrease of the bilayer depth, the same as measured from the AFM images. Compositional defects similar in size and depth to the ones presented in Fig. 31 are also observed and reported in the work of Nielsen et al [111] (See Fig. 29 B). Although, such defects have not been observed in the DPPC bilayer in absence of PLA2 but appeared only after enzyme addition. This corroborates with the interpretation of the chemical nature of the compositional defects. As was previously mentioned with the pioneer work of Grandbois et al [110] for the first time is demonstrated how the AFM can be utilized as a kinetic tool. It has to be underlined, though that the suitability of the AFM method to study enzyme kinetics depends on the experimental conditions. The temporal resolution (for the scanned areas in question) is in order of 1 or 2 min. Generally, scanning smaller areas would improve the temporal resolution but it still remains fairly lower in comparison to monolayer method of Verger and de Haas [93]. When choosing an area on the sample for which the simultaneous kinetic and structural data suppose to be collected, one faces the compromise between high lateral resolution, which would require an area as small as possible, and good kinetic statistics, which 59

would require the imaging of an area as large as possible. The AFM image areas ranging from 5 × 5 𝜇𝑚2 to 15 × 15 𝜇𝑚2, usually give good kinetic data as illustrated at Fig. 31 which shows that the hydrolysis process is clearly taking place within the frame of view and that there are primary attack points that lead to new defect formation and defect growth. The lateral resolution ranges between 10 ÷ 15 𝑛𝑚 , which is enough the existence of small channels that are result of the hydrolysis by a single enzyme to be detected [110,111]. The small temporal resolution of the AFM can be overcome also if in the experiments are used lower enzyme concentrations. It will slow down the reaction rate sufficiently, so the evolution of hydrolysis reaction is possible to be followed. The kinetic analysis provides a good control over the experiments to be carried out under identical conditions and that any influence from the scanning tip is minimized. Indeed, the maximum area growth rate and the total hydrolysis presented in Table 4.1 would change significantly in the repeat experiments if they were a result primarily of the scanning motion i.e. AFM tip effect. The apparent increase in the maximum area growth rate when products are initially present can be rationalized in terms of a primary observable. A bilayer containing products is more susceptible to attack by the enzyme, which in turn leads to desorption of not only newly hydrolyzed lipids but also some of the product molecules already present in the bilayer. Structural defects emerge and the degree of hydrolysis is influenced by desorption of LysoPC and PA initially present in the bilayer. The addition of enzyme thus has a strong perturbing effect on the bilayer. Product molecules that are originally situated within the bilayer and are stable during the AFM scanning suddenly leave the bilayer upon the addition of PLA2. The rate of hydrolysis at the end of experiments predictably is significantly lower for the ternary systems. Again, the estimate of the total hydrolysis rate is based entirely on the area fraction of the structural defects. Any hydrolysis that does not result in holes’ growth or creation of new holes is not included in the measurement. Therefore the significance of these results is associated with the high-resolution structural data obtained from the AFM both as an excellent imaging and as a kinetic tool. Any defects in a phospholipid bilayer will promote the degradation of the bilayer by PLA2, whether the defects are of compositional origin or they are a result of fluctuations or they are holes in the bilayer. The presence of defects and especially products leads to a decrease in the latency period, which is equivalent to an increase in the susceptibility of the bilayer to hydrolysis. As it is shown at Fig. 31A, two distinct types of defects present in the initial substrate are observed. Both defect types have been shown to serve as sites for the enzymatic attack [110,111]. The structural defects are primary attack points probably caused by looser packing of the lipids around the edge of the defects. The compositional defects composed of products promote the enrollment of phospholipase molecules at the surface because of their affinity toward negatively charged fatty acid molecules. This also 60

results in faster break-up of the bilayer because of the elimination of the latency phase. When the two types of defects are simultaneously present they are both primary attack points for the enzyme, judging from the area increase in time for both types of defects. However, the normalized growth rate (Fig. 32C) is much greater for the compositional defects at the beginning of hydrolysis, proving that this type of defect is more effective promoter of increased susceptibility for the enzyme. With time, the compositional defects become holes transforming into structural defects, and their growth rate approaches the rate of other structural defects which is also a proof for the ability of these phase-separated regions to recruit or activate PLA2 molecules. Therefore, these regions play more important role in promotion of the hydrolysis than the structural defects. The ability to compare two different defect types and to quantitatively analyze the evolution of the individual defects are unique features of the performed experiments. The presence of products even in very high concentrations is not enough on its own to destroy the structural integrity of the supported bilayers. A similar statement is valid for vesicles where structural integrity can be attained even after the addition of large mole fractions of products (up to 40%) achieved by co-sonication of the phospholipid and the product molecules [140]. This is in contrast to the situation with vesicles when the enzyme is present where the large-scale morphological changes and loss of structural integrity is observed at very low degrees of hydrolysis (5–10%) during the latency period [124]. Phase-separated product domains are also stable in the AFM experiments as long as the enzyme is not added to the liquid cell. Once present, however, the bilayer immediately begins to lose its structural integrity. Once more, it has to be underlined that after the injection of PLA2 the destabilization of the product domains is caused by the presence of enzyme and its action and it is not provoked by hydrodynamic effects from flushing of AFM liquid cell. It was checked and proved in a control experiment where the liquid cell was flushed with buffer causing a similar flow. The observed compositional defects retained their stability and integrity during and after the buffer injection. Hence, the bilayer is strongly perturbed by the presence of PLA2, which in combination with the hydrolysis products rips the bilayer apart. The generation of new defects and the growth of existing ones are thus the result of combined action of the generated product domains and the bilayer-perturbing effect of the enzyme as it scoots over the bilayer surface [141]. In the current context the bilayer-perturbing effect of the phospholipase could be investigated in greater detail by addition of an inactive PLA2 to see whether this would lead to desorption of product-rich domains. The absence of a lag time and compositional defects for 25% and 50% products suggests a mechanism for activation of PLA2 by the products, which is independent of the presence of compositional defects. This raises few questions about the length scales. How large does a product domain have to be in order to activate PLA2? And how large does it have to be in order to be observed by AFM when kinetic data has to be acquired simultaneously? It is not possible, based on the 61

present experimental approach, to rule out the existence of short lag times for the cases of 25% and 50% products. Similarly, if a compositional defect has to consist of only a small number of molecules before PLA2 can sense it, then it would be undetectable by AFM. New compositional defects are, however, created during the hydrolysis, which puts emphasis on the length scale argument. If areas enriched in products exist then they are too small to be detected. After a short time in the presence of PLA2, these areas become compositional defects as seen in Fig. 31B. The activation of PLA2 is therefore not exclusively dependent on the presence of compositional defects of the sizes shown in Fig. 31. The activation by the product molecules whether present as compositional defects or not, may, however, be the same. The advantage of the compositional defects is that their evolution in time can be followed by AFM imaging, proving that the preference of PLA2 for these defects in comparison with the structural ones. In relation to the structural development of the supported bilayer during the latency period, the comparison with the results of Nielsen et al [111] gives grounds for arguing that he compositional defects are a result of phase separation at high product concentrations and that these defects must be enriched in PA. The similarity between the imaged defects (e.g. from Fig. 31) and those observed toward the end of the latency phase by Nielsen et al (Fig. 29) is striking. Comparison of the defects’ depths gives the same values, supporting the argument that domains created by the enzyme close to the burst have the same composition as the compositional defects. The domains created by PLA2 do not reflect the equilibrium structure of the bilayer, which is especially true for a bilayer in gel phase (LC), whereas the compositional defects created on the Langmuir trough, are closer to equilibrium. Nevertheless, product domains have a pronounced effect on the susceptibility of the bilayer to PLA2 attack observed as a burst in hydrolysis for pure DPPC bilayers and as a high hydrolysis growth rate in the systems containing products [111]. In summary: The investigation of hydrolysis of supported product containing bilayers by PLA2 using AFM showed that no latency period is observed for the product concentrations in question. Bilayers that initially contained 75% products show two distinct types of defects, structural and compositional. Using image analysis, it is possible to quantify the relative growth rates of these defect types undergoing enzyme action. The results show an initial preference of PLA2 for the compositional defects over the structural ones. The previously reported domain formation during the slow hydrolysis in the latency period shows remarkable similarity to the compositional defects shown in the reported data, supporting the hypothesis that products are important determinants of bilayer susceptibility and consequently PLA2 activity.

62

4.2.

Kinetics of degradation of DPPC bilayers as a result of vipoxin PLA2 activity One of the natural sources of secreted phospholipases A2 (sPLA2) is the venom extracted from different snakes. In most cases the toxicity of the venom complex is associated with the pharmacological functions of enzyme action which involves variety of effects like neurotoxic, myotoxic, edematogenic, hypotensive, cardiotoxic, platelet aggregating and anticoagulant activities [142]. The PLA2 is also utilized as a biomarker of pathophysiological processes [143]. The molecular structure of PLA2s isolated from snake venoms represent either single-chain proteins (PLA2 isolated from cobra venoms, ammodytoxins) or non-covalent ionic type complex, where the enzyme is associated with an acidic and non-toxic component (crotoxin, vipoxin) [142]. The vipoxin’s PLA2 isolated from the venom of Bulgarian long-nose viper (Vipera ammodytes meridionalis) is a heterodimer ionic complex composed of two protein subunits- basic and strongly toxic His48 phospholipase A2 and an acidic, enzymatically inactive and nontoxic component. Both subunits have the same polypeptide length (122 amino acid residues) and are closely linked, sharing 62% sequence identity. The isolated sPLA2 from vipoxin is one of the most toxic phospholipases, although when dissociated, the sPLA2 subunit loses its toxicity and enzyme activity in about 10 to 14 days. [144,145] Following the already established experimental AFM strategy for visualizing the degradation process of lipid bilayers the action of vipoxin PLA2 on DPPC bilayers is considered [146]. The obtained time-series of AFM images in the course of bilayer hydrolysis are employed to identify the occurrence of bilayer defects and subsequently to evaluate their growth rate. The special resolution of the AFM images also allows the area and perimeter length of these defects to be measured and on that grounds some conclusions about the kinetics of enzyme reaction to be drawn. It is also observed an appearance and growth of three-dimensional (3D) crystal-like structures within the formed defects of degraded bilayers which imply of some unique characteristics of the vipoxin’s PLA2 action on organized lipid substrates. These characteristics are distinguishable from earlier reported cases for PLA2 lipolysis of DPPC bilayers imaged by AFM [110,111]. In effort to explain the nature of the observed topological features within the lipid bilayers so called bearing image analysis is applied. This is an image processing method which gives the approximate volume of the observed 3D crystal like structures. It is found that their growth rate follows a similar kinetic pattern as the rate of degradation of supported bilayer. The experimental part (Appendix B) involves obtaining bilayers by a procedure of collapsing small unilamellar vesicles (SUVs) on supported mica substrate [112,113]. Following this experimental protocol in the obtained AFM images are observed two morphological types of bilayer areas which will be considered separately. The first type bilayers are with relatively large number of defected areas defined as bilayer holes (Fig. 33A). The second type bilayers are with smaller in 63

size defects that appeared in the AFM images as surface depressions (Fig. 33B). These concavities at the surface are result from an imaging artifact because their dimensions are small enough to be comparable with the size of AFM tip (Fig. 33D), i.e. the tip apex is too “blunt” to reach the bottom of the defect as it compared at Fig. 33C and Fig. 33D. Following the established experimental procedure for the AFM characterization of the topography of supported DPPC bilayers in liquid as a primarily experimental task the alterations in the bilayer’s structure before and after injection of vipoxin PLA2 into the AFM liquid cell are visualized. The obtained AFM images showing the enlargement of the appeared within the scanned bilayer area defects (holes) are analyzed and quantified on the basis of the present height differences in the bilayers images which are domains of an incomplete coverage of the supported mica (Fig. 33A). When the enzyme is injected into the AFM liquid cell, these bilayer defects start to grow and at some stage of hydrolysis new holes also appear. Determining the area fraction of holes in the bilayer at a given time is a measure of the degree of hydrolysis. The structural appearance of supported bilayers once they are transferred into the AFM liquid cell could be characterized as follows. The total imaged area of all freshly prepared bilayers transferred to the AFM liquid cell is predominantly composed of uniformly flat bilayer. Only limited areas (between 0.5 and 5% of the total area) are covered with bilayer defects. At the images of Fig. 33 two types of bilayer defects are easily distinguishable. The first type defects are relatively large holes as it shown in Fig. 33A, while the second type are smaller, depression-like defects (Fig. 33B). The concentration of both types vary from sample to sample as expected from the experimental procedure [112,113]. Nevertheless, for all samples, the defected area fraction never exceeds 5% of the total imaged area. The average depth of the first type of defects is measured to be about 5.0 ± 0.5 nm which is in a good agreement with earlier reported measurements of the thickness of a DPPC bilayer in the gel phase [110, 111]. The imaging of the individual defects shows that they have different shapes and sizes and cover areas of less than 0.2 m2. The second type of defects are characterized as small surface concavities about 1 nm in depth relative to the top of the uniform bilayer. These defects appear as depressions because their lateral size is comparable with the size of the scanning probe. The lateral size of these defects also varies and is generally a factor of 10 smaller than the size of the hole like defects. The total area covered by depression like defects is determined to be less than 0.05 m2. It is repeatedly proven that supported phospholipid bilayers in gel phase (LC-phase) remain stable and mechanically intact upon imaging in contact mode with AFM equipped with a liquid cell. The performed control experiments for checking their stability and tip-sample interaction effects show no observable changes in the state of imaged DPPC bilayers even during an extensive scanning for period of 2 h prior to the enzyme injection. In all test experiments the imaged 64

bilayers remain stable and undamaged. Only after flushing the AFM liquid cell with vipoxin PLA2 solution the changes of lipid bilayer structure are observed. These structural alterations are due to enzyme-induced bilayer degradation and are followed over time by image capturing.

Figure 33. Two types of imaged bilayer areas. (A) First type is a bilayer with relatively large defected area (holes). The height differences present in the bilayer are a result of the incomplete coverage of the supported mica with phospholipid molecules. The inset shows the profile along the drawn line proving the depth of these holes (about 5.0 ± 0.5 𝑛𝑚); (B) Second type of bilayer defects that appeared in the AFM images as small surface depressions. The inset is a profile along the drawn line showing height differences between darker and brighter areas of the bilayer of about 1.0 ± 0.5 𝑛𝑚; (C) A model for the organization of the phospholipid molecules at the edge of the bilayer larger defects; (D) A model for the organization of the phospholipid molecules at the edge of the second type of bilayer defects. These type of defects have dimensions comparable (one order of magnitude) with the size of AFM tip which prevents the tip to reach the defect’s bottom.

At Fig. 34 is presented a typical time-course set of images from an experiment prior and after injection of vipoxin PLA2. At the first image (Fig. 34A) obtained just prior to enzyme injection few bilayer defects are easily distinguishable. The subsequent images (Fig. 34B to 34P) reveal progressive changes in the bilayer surface topology as a result of the enzyme action. 65

Figure 34. A typical time-course set of images from an experiment in which 100 𝑛𝑀 of vipoxin's PLA2 solution is injected into the AFM liquid cell. The scan size of all images is 5 × 5 𝜇𝑚2 . (A) Image is prior to the enzyme injection. The few bilayers defects (darker areas) are easily distinguishable. (B) Image captured 185 s after enzyme injection. The enlargement of the existing defects is necked eye detectable. (C) to (P) Images taken in the time interval 315 ÷ 6340 𝑠 after enzyme injection showing occurrence of progressive changes in surface topology of the bilayer as a result of the enzyme action. The darker areas represent structural defects in the lipid bilayer. After enzyme injection, the initial structural defect expands as the top lipid layer is hydrolyzed and the bottom layer spontaneously desorbs.

66

An appropriate software package is utilized for analysis of the obtained AFM images and for the estimation of the defect’s growth rate. With the PLA2 present, the holes in bilayer start to enlarge in size as a result of hydrolysis of the lipid molecules from upper to layer the adjacent liquid phase followed by a desorption of products of the enzyme reaction (the molecules of LysoPC and palmitic acid). Simultaneously DPPC molecules from the bottom layer adjacent to the mica support also desorb because as soon as the adjoining upper layer “disappears” their hydrophobic tails happened to be exposed unfavorably towards the liquid phase.

Figure 35. AFM images captured after the vipoxin injection showing the appearance of small depression areas in the bilayer (A) 30 min after the vipoxin injection. The inserts show a software zoom of the boxed area with one of several small depression area observed with a typical depth of 0.3– 0.5 𝑛𝑚 (the z scale has been extended to clearly show the depression); also shown is a section through the image; (B) 37min after the vipoxin injection. Arrows indicate newly appeared defects are presumably created by a single enzyme molecule. Image size is 3 × 3 𝜇𝑚.

Two already accustomed views concerning the mechanisms of binding and activation of PLA2 should be noted. First, PLA2 predominantly hydrolyzes bilayer heterogeneous areas of co-existing lipid/lysolipid domains where product accumulation has created phase segregation and/or the border areas of the existing structural defects because of the loose phospholipid packing at the holes’ rims [110]. 67

Second, PLA2 hydrolysis is characterized by the existence of a lag time period, i.e. a period of low enzyme activity, followed by a burst [111]. The reaction rate is determined primarily by the bilayer’s properties, i.e. the morphology of the surface, number of defects etc. The very first image (Fig. 34B), obtained about 3 min after the addition of PLA2 into AFM liquid cell, clearly shows an increase of the size of already existed defects. Here must be underlined the fact that primary targets of the enzyme action are only the residing holes within the bilayer. In all experiments no other but only the areas of existed membrane structural defects and small bilayer depressions are observed to be affected by the enzyme action. Ultimately, the growing hydrolyzed areas merge into one common defect. In the course of hydrolysis at later phases is also observed the appearance of new holes (Fig. 34P). Fig. 34H shows, that first afresh hydrolyzed bilayer area has appeared about 45 min after PLA2 injection. It is experimentally impossible to determine whether these areas of secondary enzyme attack are phase segregated bilayer domains because the resolution of the captured AFM images is fairly low when the scanned area is range 9 ÷ 25 𝜇𝑚2 . Nevertheless, in one experimental occasion about 30 min after the vipoxin injection it was possible to observe the appearance of small depression areas in the bilayer (Fig 35). The measured depth of these depressions is approximately 0.3 ÷ 0.5 𝑛𝑚 (Fig 35A). Latter, as the imaging revealed the emergence of new bilayer defects as a result of the enzyme action was predominantly within these depression areas depicted with the arrows at Fig. 35B. These observations support the notion that compositional domains formed in the bilayer during lag phase are always observed very close to the burst and thus play leading role in triggering PLA2 activity [111]. In order to quantify the change of desorbed area and perimeter length of existed defects as a function of time the AFM images obtained in time intervals of about 150 min after PLA2 injection are analyzed. A typical result of the kinetics analysis of growing bilayer holes is presented at Fig. 36. The data points extracted from the time series images (Fig. 34A to Fig. 34P) are presented in a format- desorbed area as a primary ordinate versus time and perimeter length as a secondary Y-axis. At Fig. 34A the locations of bilayer structural defects prior to enzyme addition are easily detectable. After injection of PLA2 solution into the AFM liquid cell the growth of these holes is initiated (Fig. 34B) and it continues until they merge together to form one big bilayer defect (Fig. 34F). As was mentioned previously as a result of the low enzyme activity the appearance of new holes is not observed for some time after the PLA2 injection into the AFM liquid cell. Once the enzyme is added and the holes are initially formed their number remains the same for relatively long period before the appearance of new bilayer defects. For different samples this period varied between 40 and 60 minutes. It might be assume that, if ever the enzyme binds to the bilayer, it is activated and prefers to “scoot” trough the lipid bilayer. This assumption favors the “scooting mode” as most realistic molecular model of PLA2 action [106]. 68

Figure 36. The change of hydrolyzed bilayer area and the perimeter length of existed defects as a function of time. Data points are obtained from the analysis of time series images from Fig 2A÷P and presented in a format- desorbed area as a primary ordinate versus time (solid curve with filled squares) and perimeter length (dash curve with open squares) as a secondary Y-axis. The lag phase of this experiment is between 2 to 2.5 times longer than reported in [111]. Nevertheless the two kinetic curves prove the correlation between the hydrolysis rate and the length of the edge of the bilayer defect. From the edge length of defects depends the number of potential spots for the enzyme attack.

The kinetic curve of the hydrolyzed area versus time shows two distinct slopes which correspond to two regimes of enzyme hydrolysis. The first one, prior to appearance of new bilayer defects is a slow hydrolysis rate while the second one is a burst of the enzyme activity. This kinetic behavior supports the hypothesis for the existence of a lag phase characterized by low enzyme activity, although the lag period measured in the present experiments is between 1.5 and 2.5 times longer than the one reported in [111]. The first stage of bilayer hydrolysis occurs preferably at the rims of existing structural defects and the kinetic curves at Fig. 36 prove the correlation between the hydrolysis rate and the edge length of bilayer defects. The appearance of new holes at the end of lag phase indicates that some other domains of DPPC bilayer were also targets of the enzyme attack. As was pointed previously, one can only speculate if these areas are bilayer phase separated domains because when the scanning area is 25 𝜇𝑚2 the AFM resolution is moderately low to show nanoscale details. In order to further investigate the influence of initial size of bilayer defects on the rate of hydrolysis at Fig. 37 are presented two typical data curves corresponding to the two morphological types of bilayer areas. 69

Figure 37. The change of hydrolyzed area versus time for the two morphological types of bilayer areas. Two regimes of hydrolysis -lag phase and burst- are easily distinguishable. In the time interval 0- 2000s (lag phase) the initial slop of curve 1 is bigger than the slope of curve 2. If the initial rate is hydrolyzed area per time, then the estimation for the initial enzyme activity is 5 × 10−4 𝜇𝑚2 ⁄𝑠 for the bilayer with bigger number of defects (curve 1) and respectively 10−4 𝜇𝑚2 ⁄𝑠 for the bilayer with smaller defective area (curve 2). The hydrolysis rate in the burst regime rises up to value of 20 × 10−4 𝜇𝑚2 ⁄𝑠 for both morphological types of bilayers.

First is a bilayer with relatively big number of holes (between 0.05% and 0.5% of the total scanned area) (Fig. 33A and Fig. 34) and the second type are smaller number of defects that appear in the AFM images as small surface depressions (the first image from a typical sequence is shown at Fig. 33B). The defected area in this case is less than 0.05% of the total scanned area. From the kinetic curves at Fig. 37 can be distinguished two regimes of hydrolysis- lag phase and burst. The curves’ initial slops in the time interval 0 ÷ 2000 𝑠 show that initial hydrolysis rate during lag phase is bigger when the bilayer has larger defected area (Fig. 37 curve 1). If the initial rate is defined as hydrolyzed area per time, then the estimation for the initial enzyme activity is 5 × 10−4 𝜇𝑚2 ⁄𝑠, for the bilayer with bigger number of defects (Fig. 37, curve 1) and respectively five times smaller 10−4 𝜇𝑚2 ⁄𝑠 for the bilayer with smaller defective area (Fig. 37, curve 2). These values are coherent with the assumption that when bilayer defects are bigger they accommodate at their rim larger number of enzymes thus the expected hydrolysis rate is bigger. The initial rates can be further translated by means of the number of DPPC molecules hydrolyzed per second, knowing that a mean area per lipid molecule is 0.45 nm2 [110]. As an estimate for the values of the initial enzyme 70

activity are obtained 1111 DPPC molecules hydrolyzed per second, corresponding to the bilayer with bigger number of defects and 222 DPPC molecules hydrolyzed per second, for the bilayer with smaller defective area, respectively. Further, assuming that the initial hydrolysis rate is as reported in [110] i.e. 88±30 DPPC molecules hydrolyzed/ second per enzyme, it means that 15±5 (curve 1) and 3±1 (curve 2) number of enzyme molecules are involved in the initial attack. This estimate can be further extended by using the available perimeter length for the initial defects2.6 m and 0.226 m, respectively measured from the AFM images. As it is reported in [147] the diameter of the vipoxin molecule is about 5.8 𝑛𝑚 which means that the perimeter of the defected area is available for 448 and 40 enzyme molecules respectively seated along the age of the defects in shoulder by shoulder manner. If the last two numbers are compared with the experimentally obtained 15 ± 5 and 3 ± 1 one can conclude that only between 3.3 ÷ 7.5 𝑝𝑒𝑟𝑐𝑒𝑛𝑡 of the perimeter of the existed bilayer defects are occupied by the enzyme molecules. This result is in a very good agreement with the reported in [110]. Once the enzyme reaction steps into the regime where the activity bursts out, the hydrolysis rate has almost the same value for both morphological types of bilayers rising up to 20 × 10−4 𝜇𝑚2 ⁄𝑠 (4444 DPPC molecules hydrolyzed per second). It can be seen from Fig. 36 that after 𝑡 = 2000 𝑠 the slopes of both kinetic curves (1 and 2) are almost the same. It is also observed that when the enzyme reaction steps into the burst stage the growing defects had a finger-like shape with a more profoundly developed channel-like pattern, similar to one previously observed [110]. A new phenomenon observed during the process of bilayer degradation is further considered, namely the occurrence and growth of three-dimensional crystal-like structures in the course of the enzyme reaction. The first thing that has to be emphasized is that these structures appeared in all the occasions only within the frame area of some but not all of the existed defects. From the image subsequence (Fig. 34) as a typical example at Fig. 34J is shown such three-dimensional (3D) structure. It appears at about 1 hour and 10 min after PLA2 injection and in the succeeding images (Fig. 2K÷P) its growth is easily detectable. Explanation of the nature of this phenomenon provokes some curious questions. The enzyme hydrolysis leads to product formation and their release from upper layer to the adjacent liquid phase together with desorption of unfavorably attached to mica support phospholipid molecules from the bottom layer. All these molecules could form micelles, lipid aggregates etc. which further to self-assemble into 3D structures within the appeared structural defects. Although, contradictorily the concentration of desorbed LysoPC, palmitic acid and DPPC molecules is estimated to be far below the CMC [148,149]. One explanation of this ambiguity could be hiding in the nature of the vipoxin’s PLA2 which is not very well studied and could play an important part in the molecular aggregation and the growth of these structures.

71

Figure 38. Images of 3D crystals formed within the lipid bilayer defect during the enzyme hydrolysis obtained approximately (A-C) 4000s and (D-F) 6000s after the enzyme injection into the AFM liquid cell. (A) and (D) are the images of the crystals accompanied with the graphs of the cross sections represented at (C) and (F) respectively, (B) and (E) are 3D image views. 3D images and the section analysis show a well-defined sharpcornered structure with number of terraces and a flat plateau forming the top of the structure. The heights of the formed terraces: 6nm, 14nm and 21nm coincide with one (5.5 ± 0.5nm), three (16.5 ± 1.5nm) and five (22 ± 2nm) double layer thicknesses, respectively. (G) Comparison between the shapes of the curves- crystals volume versus time (dashed line with triangles) and change hydrolyzed bilayer area (secondary Y-axis) as a function of time (solid line with squares).

72

The other reason, could be that the products of the enzyme reaction- LysoPC and palmitic acid, in addition with desorbed DPPC molecules from the bottom layer which are not affected by the enzyme attack create an organic matrix for the heterogeneous nucleation of the calcium salts [150], although the low calcium level in the buffer does not favor this hypothesis. Hence any further speculations about the intimated nature of these 3D structures should be avoided before performing new investigations with employment of additional experimental methods. Using the advantages of AFM for obtaining the high-resolution structural at nanoscale level the morphology of the 3D structures are investigated in some close proximity. As an example Fig. 38 reveals details of the emergence and evolution of a typical crystalline structure. From the presented 3D images and section analysis one could identify a well-defined sharp-cornered structure which has number of terraces and a flat plateau forming the top of the structure. Another interesting observation is related to the measured heights of the formed terraces which are- 6 nm, 14 nm and 21 nm, respectively (Fig. 38C). These dimensions coincide with one (5.5±0.5 nm), three (16.5±1.5 nm) and five (22±2 nm) double layer thicknesses, respectively. These sizes support a suggestion that 3D crystalline structures which occur within the decomposed by vipoxin’s PLA2 bilayer area are self-organized multilayer structures composed from the lysoproducts of the enzyme reaction and desorbed unhydrolyzed DPPC molecules. Moreover, from the obtained time-course set of images the growth rate of the 3D structures is estimated by means of their volume size by applying the bearing function analysis. The graphical results are presented at Fig. 38G as volume (primary axe) and hydrolyzed area (secondary axe) change versus time for both morphological types of bilayer area defects. The graph shows that in the time interval 0 ÷ 3000 𝑠 which coincide with the lag phase the hydrolyzed area slightly changes because of the low enzyme activity and no existence of 3D structures is observed. As soon as the enzyme activity bursts out in the time interval after 3000s, the hydrolyzed area and the volume of the appeared 3D structures start to change following similar kinetic pattern. Once the first traces of 3D structures appeared they start to grow steadily with the volume rate of about 5.5 × 10−3 𝜇𝑚3 /𝑠. In summary: A time course sets of images of supported DPPC bilayers hydrolyzed by vipoxin’s PLA2 are obtained proving the capability of AFM as analytical instrument for studying enzyme reactions at nanoscale level. The captured topographic AFM images are studied for the occurrence and evolution of two types of bilayer defects where the initial area and perimeter length of these defects are first measured and subsequently their growth is followed. Experimentally for the first time the appearance and growth of three-dimensional (3D) crystal-like structures are observed and studied by means of their size and shape as well as the kinetics of their growth. 73

4.3.

Hydrolysis of 1-mono-oleoyl-rac-glycerol (MOG) by Humicola lanuginose lipase (HLL) at the lipid–water interface observed by AFM Lipases (Triglyceride hydrolases) are group of structurally well-characterized interfacially activated enzymes [151,152], which have virtually zero activity on monomeric substrate molecules and high activity on organized substrate molecules constituting a lipid/water interface [84]. Humicola lanuginose lipase (HLL) has more complex molecular structure in comparison to PLA2. The active site in HLL is covered by a ‘‘lid’’ and the enzyme has two alternative conformations, an inactive form with the lid closed and an active form with the lid open [85]. Triglyceride lipases have always received considerable attention as they have diverse physiological function in food and fats degradation [153]. However, the study of the enzyme reaction kinetics in organized substrate media still faces difficulties due to the very weak amphiphilic nature of the triglyceride substrate molecules. These lipids are virtually insoluble in water and either absorbs on the available surfaces or forms droplets of uncontrolled dispersity [86]. The problem with preparing organized and well-defined lipid substrates with known surface areas makes it extremely difficult obtaining the kinetic constants for the lipase reactions. One complete set of kinetics constant for HLL hydrolysis is determined using p-nitrophenylbutarate (PNPB) as the substrate is embedded in anionic 1- palmitoyl-2-oleoylglycero-sn-3-phosphoglycerol (POPG) vesicles. There is indirect evidence that HLL activation is induced by a curved anionic interface [86]. The hydrolysis of 1-mono-oleoyl-rac-glycerol (MOG) by HLL is also performed using the ‘‘zero-order trough’’ technique [154]. A promising approach in avoiding the experimental obstacles concerning organized lipid substrates is by combing the advantages of the LB technique and AFM technique. The first method is utilized to create supported bilayer immersed in water as an appropriate and stable molecular architecture which is further visualized and analyzed by liquid cell AFM in the time-course of the degradation process undergoing the HLL action. Moreover, it will be demonstrated how the AFM is used as tool for studying the kinetics of lipolytic degradation of supported hybrid DPPC/MOG bilayer proving again that such experimental approach is capable of revealing crucial information regarding the mechanism of lipolytic events at the molecular level as well as to be obtained an estimate of the specific activity of HLL. The experimental procedure (Appendix C) involves obtaining of LB- films constituted of hybrid DPPC/MOG bilayers transferred on supported mica sheets. According to experimental protocol, first the DPPC layer is transferred on the supported mica and then on the top of this layer is transferred the MOG layer which is the actual substrate for the HLL action. As it was previously underlined, a substantial obstacle in lipase kinetics studies is the preparation of well-defined substrates due to the very weak am74

phiphilic nature of simple lipids. It is demonstrated that the problem can be overcome by using a strongly amphiphilic phospholipid as a supporting first layer transferred to the supported mica surface [148]. The DPPC-modified surface exhibits strongly hydrophobic qualities upon which a second monolayer of MOG can be transferred, resulting in a well-defined planar MOG surface where the headgroup of the lipid is exposed towards the aqueous solution containing the lipase. The prepared bilayers are all imaged using an AFM liquid cell. In a control experiment for proving the bilayer stability and tip–sample interaction effects, one of the MOG/DPPC bilayers is imaged for more than 2 hours. Then, after the required thermal equilibration of the AFM liquid cell, the enzyme is injected and the changes of the lipid bilayer topology are followed over time. Three sets of experiments are conducted with three different enzyme concentrations: 2.5 𝑛𝑀, 15 𝑛𝑀 and 45 𝑛𝑀, respectively. By varying the enzyme concentration prior to injection into the AFM liquid cell and analyzing the resulting sequence of real-time AFM images showing the changes in desorbed area of DPPC/MOG bilayer, the overall rate of desorption of lipid due to lipase action is measured, i.e. the growth of holes (bilayer defects) as a function of time after the enzyme injection. As expected all prepared bilayers have initial structural defects (Fig. 39A) which are result of the experimental procedure. The heights’ differences between the DPPC/MOG film (bright domains) and the mica (dark domains) in the experiments are predominantly measured about 3.5 ± 0.5 𝑛𝑚 (inset in Fig. 39A). This value is in good agreement with the X-ray data for the thickness of a MOG monolayer which is approximately 1.3 𝑛𝑚 [155] and the thickness of the DPPC monolayer which is approximately 2.4 𝑛𝑚 [156]. In occasional transfers the height difference between the DPPC/MOG film and the mica is found to be slightly smaller 2.5 ± 0.5 𝑛𝑚. This lower value indicates that the lipid film has relaxed after the LB-transfer, prior to imaging, and the surface density of the imaged film is expected to be slightly below the surface density of the MOG film in the Langmuir trough where the surface pressure is kept at 25 𝑚𝑁/𝑚 during the transfer process. The average area of single structural defects measured in all of the performed experiments is estimated between 0.4 ÷ 1.0 𝜇𝑚2 . At Fig. 39 is presented typical series of AFM images from HLL/MOG experiment with enzyme concentration 15 𝑛𝑀. The first image (Fig. 39A) is taken prior to the enzyme injection while the images at Fig. 39B to 39H show consecutive changes due to enzyme action, with a clear increase of desorbed area observed just a few minutes (~3 min) after the enzyme injection into the liquid cell (Fig. 39B). As it was previously shown PLA2 preferably hydrolyses areas in the bilayer where the product incubation has ripened the phospholipid structure [111]. For HLL the edge of existed holes are also the most susceptible sites for the enzyme attack. This is probably due to both the high curvature and loose molecular organization at the edge of bilayer defect thus giving ideal conditions for enzyme penetration and activation. 75

Figure 39. Time sequence of AFM images from experiment where 15 𝑛𝑀 HLL solution is injected into the liquid cell (scan size 15 × 15 𝜇𝑚2 ). (A) Image prior to enzyme injection; (B) Image ~3 𝑚𝑖𝑛 after, and (C) to (H) are selected images 5 ÷ 180 𝑚𝑖𝑛 after enzyme injection. The dark areas are structural defects in the lipid bilayer. After enzyme injection, the initial structural defects expand as the top layer MOG is hydrolyzed and the supporting DPPC layer spontaneously desorb. Inset (A) shows the section analysis (white line) of the double layer height which is measured about 𝟑. 𝟓 𝒏𝒎.

In a quantitative analysis, all holes in the AFM images from all three concentrations of HLL (2.5 𝑛𝑀, 15 𝑛𝑀 and 45 𝑛𝑀) spanning the first 30 𝑚𝑖𝑛 are analyzed with respect to the change in desorbed area (∆𝐴), the initial perimeter length (𝑃0 ), and the change in time (∆𝑡). In the following, hydrolysis and

∆𝐴 𝑃0 ∆𝑡

∆𝐴 ∆𝑡

is referred as the rate of

as the normalized initial rate of hydrolysis.

A plot of the desorbed area from each image series versus time (Fig. 40A) shows that almost equal amounts of lipid is desorbed in the first minutes of 15 𝑛𝑀 and 45 𝑛𝑀 experiments, i.e. the initial rates of hydrolysis are nearly identical. The kinetic curves do not follow as expected the Michaelis–Menten pattern, which postulates the proportionality to the enzyme concentration [157]. This inconsistency can be solved if the perimeter length of the individual structural defects is taken into account in the rate equation as a measure of the effective substrate concentration. The physical meaning of such act is supported by the qualitative analysis of the degradation mode which clearly shows how the edge of the structural defects constitutes the most potent enzyme activation sites.

76

Figure 40. Desorbed lipid area versus time after enzyme injection. (A) Curves for the three performed experiments using 2.5 𝑛𝑀 (♦), 15 𝑛𝑀 (■) and 45 𝑛𝑀 (▲) concentration of enzyme, respectively. All three curves show the same characteristics where an initial burst in hydrolysis is followed by a lower (virtually zero) steady rate. The initial rate ( = slope) of 15 𝑛𝑀 and 45 𝑛𝑀 experiments is nearly superimposable and not as expected proportional to enzyme concentration. (B) The normalized initial rates.

The relatively large initial defect in the 15 nM series, compared to the 45 nM series, hence accounts for the relatively large initial rates in the 15 nM series. Therefore the initial rate is proposed to obey the relation; 𝑣𝑖𝑛𝑖𝑡 = 𝑘𝑃0 𝑐𝐻𝐿𝐿 where 𝑣𝑖𝑛𝑖𝑡 is the initial rate (𝑣𝑖𝑛𝑖𝑡 =

∆𝐴0 ∆𝑡0

(4.1)

), 𝑘 is an empirical constant, 𝑐𝐻𝐿𝐿 is the bulk

concentration of enzyme and 𝑃0 is the initial perimeter length of the structural defects. To check the validity of Eq. (4.1), the initial change in desorbed area (∆𝐴0 ) per initial perimeter length (𝑃0 ) versus time (∆𝑡) is plotted as it is shown in Fig. 40B. In this plot, the expected proportionality between initial rate per perimeter and enzyme concentration is observed. The plots at Fig. 40 furthermore show two well distinguishable phases in the rate of hydrolyses; an initial burst followed by a practically zero rate. Consequently, none of the bilayers are completely desorbed after 3 h of enzyme action. In fact, in all three experiments, the rate of hydrolysis decreases dramatically after less than 30 min once the enzyme is injected. This incomplete hydrolysis suggests that the hydrolysis products and the desorbed DPPC molecules or possibly both inhibit the enzyme. Most likely, the inhibition is caused by the extremely low critical micelles concentration of DPPC (5 × 10−10 𝑀) [158], which cause all desorbed DPPC molecules to form aggregates of micelles, vesicles, etc. In the first scan about 3 min after the enzyme injection in the 2.5 𝑛𝑀 experiment, the desorbed DPPC area is approximately 0.71 m2. The number of DPPC molecules corresponding to this area is 𝑛𝐷𝑃𝑃𝐶 = 1.6 × 106 calculated from the 77

relation 𝑛𝐷𝑃𝑃𝐶 =

𝐷𝑒𝑠𝑜𝑟𝑏𝑒𝑑 𝑎𝑟𝑒𝑎 𝑀𝑀𝐴𝐷𝑃𝑃𝐶

(where 𝑀𝑀𝐴𝐷𝑃𝑃𝐶 ~45Å2 is the mean molecular area of

DPPC molecule). The same calculation for the two other experiments (15 𝑛𝑀 and 45 𝑛𝑀), yield the number of desorbed DPPC molecules after the first scan to be 2.9 × 107 and 6.4 × 107 , respectively. Introducing the term scan area volume as the volume of water over the scan area, i.e. 𝑠𝑐𝑎𝑛 𝑎𝑟𝑒𝑎 × ℎ𝑒𝑖𝑔ℎ 𝑜𝑓 𝑙𝑖𝑞𝑢𝑖𝑑 𝑐𝑒𝑙𝑙 = 102 (𝜇𝑚2 ) × 0.5(𝑚𝑚) = 5 × 10−11 (𝑑𝑚3 ) the formal concentration of DPPC after the 3rd min is estimated to be 5 × 10−8 𝑀 in the 2.5 𝑛𝑀 experiment, 1 × 10−6 𝑀 in the 15 𝑛𝑀 experiment, and 2 × 10−6 𝑀 in the 45 𝑛𝑀 experiment, respectively. In all cases, the CMC of DPPC 𝐶𝑀𝐶𝐷𝑃𝑃𝐶 = 5 × 10−10 𝑀 is exceeded several hundred times. The aggregates formed as a consequence of exceeding the CMC are ideal for binding free enzyme [159], and hydrolysis is most likely halted as a result of HLL binding to the DPPC aggregates. Another explanation of the incomplete hydrolysis is based on the formation of free fatty acids. Some of the fatty acids produced by hydrolysis of MOG may stay in the lipid layer. It is shown that formation of more than 5 to 10 % free fatty acids in monolayers inhibit HLL’s activity in monolayers [160]. Possible changes in the dielectric constant () of the exposed monolayer surroundings could cause a decrease in number of functional enzymes on the lipid surface [89]. The mechanism of HLL adsorption and activation at the lipid–water interfaces on vesicles is also proposed [161]. The visualization by AFM of the molecular events occurring at nanoscale level proves that lipid hydrolysis takes place at the edge of structural defects, and the structural defects that are present before the enzyme injection expand after being exposed to the HLL action. At the same time, large areas without structural defects are surprisingly stable during the duration of the experiment. These directly observed phenomena in the AFM experiment can be described by the following molecular mechanism summarized at Fig. 43. First, the enzymes adsorb evenly to the lipid film [159] and they are preferentially activated at the edge of a structural defect (Fig. 43B). In a hypothesized rate-limiting step, HLL hydrolyses MOG to glycerol and oleic acid products (Fig. 43C). The hydrolysis of MOG molecules leads to the exposure of DPPC’s hydrophobic tail to the aqueous phase. In a kinetically fast step, these unfavorable oriented DPPC molecules spontaneously desorb and most likely form lipid aggregates (vesicles, micelles, etc.), which are either dissolved in the aqueous phase (Fig. 43C) or remain as adsorbed aggregates inside the structural defects. In the end, the hydrolysis of the lipid substrate is inhibited due to binding of all free enzymes to the formed DPPC aggregates of either soluble vesicles (Fig. 43D) or double layer structures formed at the solid– liquid interface (Fig. 43E). Alternatively, free fatty acids which remain at the interface may change the physicochemical nature of the substrate as discussed above (Fig. 43F). In order to eliminate these inhibition effects from the image analysis, the focus is pointed on the initial rates. 78

Figure 41. Schematic illustration of the kinetic steps involved in the lipid bilayer desorption observed by AFM. (A) The enzyme adsorbs to the lipid film. (B) The enzyme is activated at the edge of structural defects. (C) Hydrolysis occurs in the hypothesized rate-limiting step, and spontaneously desorbed DPPC molecules form aggregates. Incomplete hydrolysis is likely to be caused by one or more of the following events: (D) Binding of HLL to DPPC aggregates either to soluble aggregates. (E) Binding to aggregates formed at the solid surface. (F) Changes in the physicochemical properties of the interface due to fatty acid hydrolysis products remaining in the lipid film.

The initial rate (𝑣𝑖𝑛𝑖𝑡 =

∆𝐴0 ∆𝑡0

) is defined as the change in internal area per time

of a given hole between the two first scans (e.g. Fig. 39A and 39B). The number of individual structural defects is determined for any particular experiments e.g. 22 are in the experiment with 𝑐𝐻𝐿𝐿 = 2.5 𝑛𝑀, 6 in the experiment with 𝑐𝐻𝐿𝐿 = 15 𝑛𝑀 and 6 in the experiment with 𝑐𝐻𝐿𝐿 = 45 𝑛𝑀. Only the evolution of the structural defects within the scan area are analyzed, e.g. structural defects, expanding beyond the AFM scan area are excluded from the analysis. Then the initial rate of the change in desorbed area of each structural is deduced from the six images shown in Fig. 44A to 44F. As stated previously, the rate of hydrolysis is found to be proportional to the perimeter length of the initial structural defects. Consequently, the initial rate is normalized with respect to the initial perimeter length and this is referred to as the normalized initial rate in units of 𝑚𝑖𝑐𝑟𝑜𝑚𝑒𝑡𝑒𝑟𝑠/ 𝑚𝑖𝑛𝑢𝑡𝑒. 79

Figure 42. The six images used to deduce the initial rate of hydrolysis. (A) The last image prior to injection and (B) the first image after injection of 2.5 𝑛𝑀 enzyme solution. (C,D) Two similar images injecting 15 𝑛𝑀 enzyme solution. (E,F) Two similar images injecting 45 𝑛𝑀 enzyme solutions. (G) The change in desorbed area of the individual structural defects depicted in (A-F) plotted versus the initial perimeter length; 2.5 𝑛𝑀 (♦), 15 𝑛𝑀 (▲) and 45 𝑛𝑀 (■). In general, the initial rate of the single structural defects is proportional to its initial perimeter length. The inset of (G) is zoom in showing the data points from the 2.5 𝑛𝑀 experiment in close proximity. (H) The average initial rates for the three experiments (equal to the slopes of the lines in (G) and the control experiment (𝑒𝑛𝑧𝑦𝑚𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 = 0, 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑟𝑎𝑡𝑒 = 0) versus the concentration of enzyme.

As discussed previously, due to the temporal resolution of the AFM, which is in the order of one measurement per 3 min, it is apparent that AFM method will give a lower estimate for the normalized initial rates, since inhibitory effects are likely to set in during the first scan. In order to test Eq. (4.1), at Fig. 44G the individual initial rates are plotted versus the initial perimeter length of each hole. The fair linear correlation corroborates the assumption that the rate is proportional to the perimeter length for all values of 𝑃0 as for the 2.5 𝑛𝑀 experiment the values are more scattered (the inset of Fig. 44G). At the Fig. 44H are plotted the average normalized initial rates

80

(𝑠𝑙𝑜𝑝𝑒 =

𝑣𝑖𝑛𝑖𝑡 𝑃0

) for each experimental series versus enzyme concentration. The val-

ues show a strong linear correlation (𝑅 2 = 0.998) with the bulk enzyme concentration range in question. Together, the data presented in Figs. 44G and 44H prove the validity of Eq. (4.1), and this in turn implies that the edges of the structural defects are not saturated with enzyme. In the case of saturation, the initial rate would be independent of the bulk enzyme concentration. The value for the 2.5 𝑛𝑀 series is slightly below the fitted line. One possible explanation of the small deviation is that it is caused by non-specific binding of enzyme molecules (i) onto the sides of the container in which the solution was prepared, (ii) the syringe used to inject the enzyme into the liquid cell (iii) and the liquid cell itself. At higher enzyme concentrations (i.e. one order of magnitude) all these effects are negligible and take place only in much diluted enzyme solutions. There are several reasons for putting special emphasis on the 2.5 𝑛𝑀 series. The series has many small initial defects present before enzyme injection (22 of these are analyzed) and they present the unique opportunity of having many ‘‘separate experiments’’ under exactly the same conditions. Secondly, this is the only series where holes are observed to emerge from sites of no visible initial structural defect in the bilayer. As previously discussed the most plausible reason is that the new holes start to grow from structural defects that are too small to be imaged with the permitted AFM resolution. This corroborates by the fact that some of the new holes start where dim shades are visible, which are likely holes that are too small to be penetrated by the AFM tip. Some new holes emerge at sites which seem defect-free, but the relaxation processes which occur within the lipid substrate during the hydrolysis make the bilayer less dense and apparently facilitate the enzyme access. The third reason for giving 2.5 𝑛𝑀 experimental series special emphasis is that a very large data spread of the measured initial rates (the inset of Fig. 44G) is observed compared to the other experiments when using higher enzyme concentration. The large data spread most likely can be attributed to a stochastic distribution of single enzymes in the various structural defects due to the very low bulk concentration of the enzyme. A nominal enzyme concentration of 2.5 𝑛𝑀 corresponds to 7.5 × 103 enzyme molecules in the scanning area volume with an average distance around 0.9 m between each enzyme 𝑑3𝐷 = (

1 𝑁𝐴 𝑐𝐻𝐿𝐿

1/3

)

. With all this

in mind, it seems likely that the new holes that emerge in the bilayer are result of a single enzyme action. The average initial rates of these emerging holes 𝜇𝑚2

are ~10−2 𝑚𝑖𝑛 . Given that the mean molecular area of MOG at 25 𝑚𝑁/𝑚 is 40 Å2 (2.5 × 106 𝑀𝑂𝐺 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠/Å2 ), the estimated specific activity of HLL-MOG hydrol𝑙𝑖𝑝𝑖𝑑𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠

ysis is ~2.5 × 104 𝑚𝑖𝑛×𝑒𝑛𝑧𝑦𝑚𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒.

81

Table 4.2 Summary of the experiments ConcentraNumber of acTotal initial pea tion (𝒏𝑴) tive rimeter length b Enzymes (𝜇𝑚)c 𝟐. 𝟓 𝟏𝟓 𝟒𝟓

30 4.1 × 102 5.9 × 102

45 29 15

Maximum number of accessible substrate sites d 9.0 × 103 5.8 × 103 3.0 × 103

Occupancy of perimeter (%)e

0.3 7.0 20.0

a Bulk

concentration of enzyme solution injected into the liquid cell. of active enzymes in the scan area during the first scan (e.g. in the 2.5 nM experiment: ∆𝐴0 ⁄∆𝑡0 × by theinitial rate of hydrolysis per enzyme molecule at the surface = 0.7 𝜇𝑚2 ⁄2.3 𝑚𝑖𝑛 × 0.01 𝜇𝑚2 ⁄𝑚𝑖𝑛 × 𝑒𝑛𝑧𝑦𝑚𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 ) c The total measured perimeter length in the scan area. d Maximum number of potential activation sites along the perimeter (total perimeter length /diameter of HLL enzyme molecule 45 × 10−6 [𝑚]⁄5 × 10−9 [𝑚] = 9 × 103). e Percentage of the potential activation sites occupied by enzyme molecules = 30⁄9 × 103 [× 100 %] = 0.3% b Number

Using this estimate, one can calculate the percentage of active enzymes in all of the performed experiments. With 2.5 𝑛𝑀 enzyme concentration, a total degradation of 0.7 𝜇𝑚2 of MOG is found in the first 3 minutes. Using an initial rate of 10−2

𝜇𝑚2 𝑚𝑖𝑛

𝑝𝑒𝑟 𝑒𝑛𝑧𝑦𝑚𝑒, the total number of active enzymes within the scan area is

then estimated to be 30. Since the injected amount of enzyme is 2.5 𝑛𝑀 (i.e. ~7.5 × 103 enzymes in the scanning volume) it means that only 0.04% of the enzymes are active. The ‘‘occupancy’’ of active enzymes at the edge of structural defects can also be estimated as follows. In the 2.5 𝑛𝑀 experiment the total edge length of structural defects in the scan area is 45 𝜇𝑚, therefore if the enzymes are sitting at the rim of the defects in a shoulder-to-shoulder fashion, then each enzyme will occupy approximately 5 𝑛𝑚 of the edge. Then, dividing the edge length by the HLL diameter it gives

45×10−6 [𝑚] 5×10−9 [𝑚]

= 9 × 103 for the total number of potential activation sites.

Only 30 of these enzymes are used in the enzymatic act hence yielding to 0.3% occupancy. The estimated values for the occupancy of the perimeter for all three experiments are given in Table 4.2. For comparison, the data for the specific activity of MOG hydrolysis by HLL measured by Verger et al. [154] using zero-order trough technique are given in Table 4.3. The measured rates are in surprisingly good agreement with the measured by AFM values, considering that there are some fundamental differences between the two experiments: (i) Verger et al. [153] assume that 1% of the enzymes are active at the lipid–water interface, (ii) the zero-order trough experiments are carried out in TRIS buffer close to HLL’s pH optimum, (iii) -cyclodextrin ( -CD) is added to the subphase to help remove products from the interface by binding them inside the water-soluble  -CD molecules (in fact no hydrolysis is observed without  -CD in the subphase), (iv) the substrate is under constant pressure keeping it defect-free, and (v) hydrolysis of MOG in the zero-order trough creates unfavorable air–water interface. Conversely, in the AFM experiment, it is assumed that 82

most new holes in the lipid bilayer are caused by a single active enzyme in each of these holes. Also, the experiment is carried out in MilliQ water below the pH activity optimum of HLL because Tris buffer is reported to destabilize the bilayers by inducing the so called ripple phase in phospholipid bilayers [162], while the  CD is dismissed because preliminary AFM experiments showed that the  -CD solution flushed into the AFM liquid cell also destabilizes in the DPPC/MOG bilayer. This latter effect is also observed and compensated in the zero-order trough experiment, where both substrate molecules (MOG) and product molecules (oleic acid) are removed from the interface by  -CD. Table 4.3 Comparison of experimental techniques Experimental Conditions Initial rate of MOG hydrolysis 𝑴𝑶𝑮 technique ( ) 𝒎𝒊𝒏×𝑵𝑯𝑳𝑳,𝒂𝒄𝒕𝒊𝒗𝒆

AFM

𝑀𝑖𝑙𝑙𝑖𝑄 𝑤𝑎𝑡𝑒𝑟, 𝑝𝐻 6.5, 𝐶𝐻𝐿𝐿,𝐵𝑢𝑙𝑘 = 2.5 𝑛𝑀

2.5 × 104

Zero-order trough

𝑇𝑟𝑖𝑠 𝑏𝑢𝑓𝑓𝑒𝑟, 𝑝𝐻 8.0 𝐶𝐻𝐿𝐿,𝐵𝑢𝑙𝑘 = 0.16 𝑛𝑀 𝐶𝛽𝐶𝐷 = 0.8 𝑚𝑔/𝑚𝑙

5.8 × 104

The experimental conditions and the initial rates of hydrolysis measured in the presented AFM experiment and calculated from the zero-order trough experiment reported in Ref. [154].

Finally hydrolysis of MOG and desorption of DPPC in the AFM liquid cell creates mica–water interface, which is energetically comparable to the initial lipid/water interface. The estimated specific activities are summarized in Table 4.3 𝑀𝑂𝐺 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠

showing that the zero-order trough value is in the order of 6 × 104 𝑚𝑖𝑛×𝑒𝑛𝑧𝑦𝑚𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 and the value obtained from the AFM experiment is in the same order 2.5 × 𝑀𝑂𝐺 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠

104 𝑚𝑖𝑛×𝑒𝑛𝑧𝑦𝑚𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒. The zero-order trough value is calculated from Fig. 5 in Ref. [154] by subtracting the slope of ‘‘decrease of surface area per minute’’ with only  -CD in the subphase and the slope of ‘‘decrease of surface area per minute’’ with both  -CD and HLL in the subphase of the zero-order trough, i.e. the increased rate of ‘‘decrease of surface area’’ after enzyme injection and assuming 1% of the bulk concentration of enzyme is active at the lipid– water interface. Combining the observed phenomena of the AFM and zero-order trough experiments, it seems possible that the hydrolysis in the zero-order trough is directly related to formation of phase segregated domain areas in the monolayer that are induced by the  -CD. Since the observation with the AFM show that the intact lipid areas are resistant to enzyme attack one may speculate that  -CD induces phase segregated domains in the bilayer. Although these domains are compensated for in the zero-order trough by compression of the barriers, the enzyme should have sufficient time to activate and start hydrolysis in these  -CD induced areas. It has been reported [163] that the rate of desorption of the hydrolysis product (oleic acid) 83

from the lipid–water interface becomes constant and independent of the  -CD concentration above 0.5 𝑚𝑀  -CD. However, if  -CD induces phase segregation domains which are the areas of the enzyme activation, then the concentration of  CD could play a significant role. From the AFM experiment is estimated that approximately 0.05% of the enzymes in the bulk are active at the perimeter of the structural defects within the scan area. This is 20 times less than the estimate of 1% proposed by Verger et al [154]. This indicates that if the estimated values from AFM and the zero-order trough experiments are both correct, then it is likely that there is 20 times less enzyme activating perimeter in the lipid monolayer in the zero-order trough than in the bilayer in the AFM liquid cell. In summary, using the LB-technique is demonstrated a method for preparing substrates for the HLL action- hybrid lipid/phospholipid (MOG/DPPC) bilayer with the lipid layer exposed to the aqueous phase. This substrate has a well-defined structure, and proved to be very suitable for both qualitative and quantitative analysis of interfacial lipolytic action of HLL. The substrate degradation by HLL is imaged in liquid cell AFM and the obtained images give information of the preferred sites of enzymatic action. In two of three experimental series only initial structural defects observed at the bilayer surface are attacked by the enzyme, confirming earlier experimental findings that HLL is activated by substrates with high curvature. Although in some experimental series, not only existing structural defects were attacked but new randomly emerged holes in the MOG layer were observed in various areas during the first 30 min. The calculations based on an assumption that the new holes which emerge from initially planer sites on the bilayer are a result of single enzyme action giving an initial rate of hydrolysis 2.5 × 𝑀𝑂𝐺 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠

104 𝑚𝑖𝑛×𝑒𝑛𝑧𝑦𝑚𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒, which agrees well with estimates based on zero-order trough experiments. The findings herein are also consistent with the earlier discussed results regarding preferred sites of activation of PLA2, which have been found to discriminate between structural and compositional defects. For three enzyme concentrations it is found that the rate of hydrolysis is proportional to the perimeter length of the initial structural defects in the bilayer and to the bulk enzyme concentration. This observation for all used enzyme concentrations shows that the edges of the structural defects are not saturated with enzyme. 4.4.

AFM visualization of lipid bilayer degradation due to simultaneous action of PLA2 and HLL Another interesting issue regarding hydrolysis of a bilayer is the synergetic (simultaneous) action of different lipases and phospholipases on mixed lipid/phospholipid substrates. As reported by Verger, et al. [164] when a mixed glyceride/phospholipid emulsion (Intralipid TM) is used as a substrate, typically PLA2 or gastric lipase alone cannot degrade the glyceride molecules. However, after partial 84

lipolysis of the monomolecular film by gastric lipase, the catalytic action of pancreatic lipase can commence immediately. It is concluded that the delay of hydrolysis is a result of lower penetration ability of pancreatic lipase. In the frame view of these findings the AFM is employed for obtaining timeseries images showing the degradation of supported mixed 1,2-Dipalmitoyl-snglycero-3-phosphocholine (DPPC) / 1,2-Dipalmitoyl-sn-glycerol (DPG) (70 % / 30 %, mol/mol) bilayers due to synergetic action of HLL and PLA2. The inability of lipids to generate well-defined bilayer structures is overcome by choosing the phospholipid / lipid mixtures with molar ratio 70% / 30 % as optimum for obtaining stable bilayers and reliable imaging with AFM. Because the two bilayer components, DPPC and DPG are substrates for PLA2 and HLL, respectively, these bilayers are very suitable as a substrates for simultaneous action of HLL and PLA2 thus allowing the synergetic action of the two enzymes to be investigated [149]. In order to distinguish the enzyme action either on DPPC or on DPG molecules, three sets of experiments are carried out. The first set involves only PLA2 present in the bulk phase of the AFM liquid cell. In this case only DPPC molecules embedded in the bilayer are targets for the enzyme attack. In the second set when HLL is introduced into the AFM liquid cell only DPG molecules in the bilayer are expected to be hydrolyzed by the enzyme. In the third experimental set, both enzymes are present in the solution. In the experiments with PLA2 acting on mixed DPPC/DPG bilayers at the beginning the transferred by LB technique supported phospholipid/lipid bilayers are checked for stability and tip–sample interaction effects upon imaging. In a control experiment, one of DPPC/DPG bilayers is imaged extensively prior to enzyme injection in AFM liquid cell proving the bilayer stability. Next, following the established experimental procedure after thermal equilibration of AFM liquid cell, the enzyme is injected into the cell and any structural changes of the lipid bilayer due to enzyme-induced degradation are followed over time (Appendix D). The AFM images show that the majority of bilayer area (95-100%) of all freshly prepared by LB transfer onto mica supports mixed DPPC/DPG bilayers are constituted of a uniform, flat bilayers with a composition corresponding to the structure of monolayers deposited from the Langmuir trough. In addition to the regular bilayer structure are observed holes in the bilayers where the mica is not covered with lipid molecules. These defects are single bilayer deep and are used as an indicator for the degree of hydrolysis upon enzyme injection. The height differences measured from the topographic AFM images (Fig. 45) between the top of the DPPC/DPG bilayer (brighter areas) and the mica support (dark domains) are 4.8 ± 0.5 𝑛𝑚 (inset in Fig. 45B), a value which is in a good agreement with X-ray data for the thickness of the DPPC/DPG monolayer which is reported to be ~2.4 nm [156]. For the scanning areas of 20 × 20 𝜇𝑚2 as the example at Fig. 45A shows, the structural bilayer defects are present in almost all of the studied samples before enzyme injection. The number of defects vary between samples as a result of the transfer 85

process and sample handling but for all the studied samples, the area fraction covered by this type of defect is below 5 %. At Fig. 45 is presented a typical series of AFM images from a PLA2 experiment. The first image (Fig. 45A) is obtained prior to PLA2 injection. Images at Fig. 45B to 45L reveal consecutive changes due to enzyme action, with a clear increase of the defect size observed just ~4 min after enzyme addition to the AFM liquid cell (Fig. 45B). The growth of the existing holes is calculated by section analysis of the captured AFM images. With enzyme present, the bilayer holes grow as a result of hydrolysis of the phospholipid molecules. The product molecules do not remain in the bilayer but desorb from the support, thus enlarging the existing holes. Determining the area fraction of the bilayer holes as a function of time gives a measure of the degree of hydrolysis at various stages of the enzyme reaction as two important issues about PLA2 action should be kept in mind. First, PLA2 preferably hydrolyses areas in the bilayer where product accumulation has created phase segregation in the membrane or at the border of the membrane, where the phospholipids are not well packed. It occurs as a result of both the high curvature and loose molecular organization at the edge of a bilayer defect. Such molecular organization gives the best conditions for enzyme penetration, activation and catalytic activity. Secondly, the hydrolysis of phospholipids by PLA2 is characterized by a latency period, i.e., a period of low activity, followed by a burst. The course of reaction is determined primarily by the properties of bilayer structure which the PLA2 interacts with. Analysis of the AFM images from Fig. 45 shows that hydrolysis of mixed DPPC/DPG bilayer is initiated at certain bilayer locations where either traces of structural defects (shown by black arrows in Fig. 45A) are identified or some small depressions in the bilayer have been observed (the zoom in image in the inset in Fig. 45A ). This observation supports the assertion that the most susceptible places for enzyme interactions are either at the rim of the structural defects in the bilayer or at the boundaries of coexisting phases. It is evident that hydrolysis of the bilayer starts immediately, i.e., in the first minutes, after injection of PLA2 without any lag phase activity (Fig. 45B). One possible explanation is the existence of phase separation between DPPC and DPG but the resolution of AFM images does not allow strictly confirming the presence of separate discrete domains or phases although the detected bilayer depressions could be considered as a compositional defects i.e. phase separation domains (Fig. 45A). It is consistent with the previous discussion about mixed DPPC/Palmitic Acid (PA) bilayer where the phase segregation was proved to exist and because PA has a similar amphiphilic structure as DPG, the assumption of phase segregation within the mixed DPPC/DPG bilayer is highly plausible. Phase segregation would accommodate the adsorption and activation of PLA2 because of the enzyme preference to cluster on the fatty-acid-enriched domains. 86

Figure 43. Time sequence of AFM images with scan size 20 × 20 µ𝑚2 from experiment in which 100 𝑛𝑀 𝑃𝐿𝐴2 solution is injected into the liquid cell. (A) Image prior to enzyme injection; the black arrows show existing initial defects (depressions) in the bilayer, inset shows a zoom in and close-ups of two of these depressions. (B) Image captured 4 min after enzyme injection; the growing defects are numbered, and the inset is a section plot made across the marked with the line zone showing the height of the bilayer which is 3.5 𝑛𝑚. (C) to (L) Images taken in the time interval 8 ÷ 120 𝑚𝑖𝑛 after the enzyme injection. The darker areas represent structural defects in the lipid bilayer. After enzyme injection, the initial structural defect expands as the top lipid layer is hydrolyzed and the bottom layer spontaneously desorbs 87

A quantitative analysis of all holes in the AFM images is performed spanning the time interval of 120 min after enzyme injection to determine the change of the desorbed area (∆𝐴) as a function of time (∆𝑡). Fig. 46A represents a typical result of the kinetics analysis of the growing bilayer holes.

Figure 44. (A) Desorbed lipid areas versus time after injection of PLA2 with concentration 100 𝑛𝑀. Curves show the desorbed area of the structural defects numbered at Fig. 43B with numbers 1 to 10. The inset shows the initial slope of the curves. (B) Closer view of two bilayer defects (#1 and #2) showing the numerous small channels starting at the edge of defects present in the supported bilayer. The figure is adopted from [110].

In all experiments, the number of holes created after PLA2 addition remains the same. This allows to conclude that, once the enzyme binds to the bilayer, it prefers “to scoot” trough the bilayer. From the graphs in Fig. 46A it can be seen that the hydrolyzed area of all structural defects grows linearly in the first 20-25 min and then levels off to an equilibrium. The different initial slopes (as the inset in Fig. 46A shows) indicate that different numbers of enzymes are acting in the scope of one defect. A challenging problem is to determine the number of active enzymes in order to establish the surface enzyme kinetics scheme. As an attempt to determine this, Verger group [165] adapted fluorescence resonance energy transfer technique (FRET) to study the process in which lipase is adsorbed to the monomolecular lipid films spread at the air/water interface. In their earlier studies [166] is suggested that only 1% of the enzyme molecules in the bulk phase are activated and take place in the catalytic act. Though, using AFM where the nanoscale lateral resolution is achieved the patterns of tiny channels in the bilayer are imaged assuming to be tunneled by single enzymes [110,111]. If the channel formation is indeed a result of single enzyme action then the AFM data reported in [110] can be further analyzed knowing that a cross-sectional diameter of PLA2 is 5 nm [93]. The one of the defects (#1 in Fig. 46B) is measured to have an initial perimeter length of 1.45 𝜇𝑚 and could accommodate about 290 enzymes if the PLA2 molecules are sitting at the rim of bilayer defect in a shoulder-to shoulder fashion. 88

Instead, only 13 channels are counted to appear. Assuming that one enzyme occupies 5 nm, then 13 enzymes will occupy 0.065 m and therefore the occupancy of the perimeter will be about 4.8%. The defect #2 in Fig. 46B is estimated to have an initial perimeter length 0.8 𝜇𝑚 and could accommodate about 160 enzymes if the PLA2 molecules were sitting at the edge one after the other. Instead, only 7 channels are counted to appear. Again, assuming that one enzyme occupies 5 nm, 7 enzymes will occupy 0.035 𝜇𝑚, or the occupancy of the perimeter will be 4.4%. Furthermore, as was previously demonstrated for the HLL action on hybrid MOG/DPPC layers the determined occupancy of the perimeter of the existing structural defects by lipase molecules varied depending on the bulk enzyme concentration i.e. the values of 0.3% for very dilute HLL concentrations of 2.5nM, and between 7-20% for enzyme concentrations of 15-45nM respectfully were obtained. To analyze the initial rate of hydrolysis, in Fig. 45B, all the locations in the DPPC/DPG bilayer where structural defects occurred after the injection of PLA2 are numbered. The hydrolyzed area versus time for each observed defect is shown in Fig. 46A. Because these defects appear immediately i.e. without lag phase in areas without detectable structural defects and they are growing at different rates, one can conclude that a different number of enzymes per hole are responsible for the bilayer degradation. Assuming that only ~5% of the edge of the structural defects in the bilayer is covered by the active enzyme molecules, the number of enzymes working per hole can be estimated and, on that basis the hydrolysis rate. Thus, 84 ± 14 DPPC molecules /sec /enzyme is obtained in a good agreement with the reported 88 ± 30 DPPC molecules /sec /enzyme [110]. Another observation regarding PLA2 action is important- the incomplete degradation of a supported DPPC/DPG bilayer. In all experiments, is observed that hydrolysis takes place at the edge of structural defects. Therefore, those structural defects which exist in the bilayers before the enzyme injection further expand after being exposed to PLA2 action. At the same time for the duration of the experiment, large areas without structural defects remain stable and intact. The curves in Fig. 46A show two easily distinguished kinetics regimes. For small times there is a linear growth, i.e. a constant hydrolysis rate, followed by a reduction in hydrolysis rate. In fact, the rate begins decreasing ~25 min after enzyme addition. Consequently, none of the bilayers are completely desorbed after 2.5 hours. It is proposed that the molecular mechanism which stays behind the directly observed phenomena in the AFM experiment is similar to the one depicted in Fig. 43 thus concluding that the hydrolysis is most likely halted as a result of PLA2 binding to the DPPC and/or DPG aggregates which are the reaction products. Let now consider the experiments with HLL on mixed DPPC/DPG bilayers. In contrast to the previous experiments, the injection of HLL into the AFM liquid cell does not result in any detectable changes in the bilayer structure. A typical example of AFM images before and after HLL injection into the liquid cell is given at Fig. 45. 89

Figure 45. AFM images from experiment where a 200𝑛𝑀 𝐻𝐿𝐿 solution is injected into the liquid cell (scan size 20 × 20 𝜇𝑚2 ). Image captured (A) prior to enzyme injection and (B) 210 𝑚𝑖𝑛 after enzyme injection. No visible changes in the bilayer structure are detected. Initially some spots pointed by arrows at (A) with sizes between 0.1 ÷ 0.5 𝜇𝑚 before enzyme injection are noticed then after some time they gradually disappeared.

Initially, some small aggregates with sizes between 0.1 and 0.5 m (indicated with arrows in Fig. 47A) are noticed before the enzyme injection. After 3.5 h these aggregates completely disappeared. Although their molecular nature is unknown it could be assumed that they are either some excess lipid aggregates or some salt crystals that have been eventually dissolved in the bulk phase. If these were lipid aggregates, it is not clear if they were dissolved as a result of the lipase action. In all of the performed experiments, no visible changes in the bilayer are observed. The bilayers remained intact and stable, indicating lack of HLL activity. Such observation is not unexpected, though. Phospholipids are considered as lipase inhibitors [166]. Moreover, Brockman [167] has shown, that for pancreatic lipases below 0.4 mol DPG fraction and the absence of co-lipase, less than 10% hydrolysis occurs. This result is explained as due to poor accessibility of the substrate molecules and when the mol fraction increases between 0.4 and 0.6 then the hydrolysis rate also increases substantially. Almost identical observation is reported for pancreatic carboxylester lipase and an appropriate model for the initiation of the lipolysis at the phosphatidylcholine-rich surface is suggested [168]. Accordingly, small lipid fractions correspond to small number of sites at the interface where the enzyme can be adsorbed. Additionally, even if attached, the enzyme has limited amount of substrate molecules available for hydrolysis. Although clear evidence is not presented, as possible explanations could be considered the restricted lateral diffusion of the substrate molecules or the inactive (open) conformation of the enzyme at the interface [168]. In the frame view of these findings it is reasonable to assume that the lack of HLL activity on mixed DPPC/DPG (0.7 mol /0.3 mol) is a result of the low DPG fraction. The experiments with higher than 0.3 mol fractions of DPG is not possible to be performed because the obtained bilayers are unstable and easily destructible after their transfer on mica support. 90

Figure 46. Time sequence of AFM images from an experiment where both enzymes are present. First HLL is injected and incubated in the liquid cell for 2h. Then PLA2 solution is introduced into the AFM liquid cell. Image (A) is prior to PLA2 injection and the arrow shows existing of an initial structural defect in the bilayer, (B) is 4 𝑚𝑖𝑛 after PLA2 injection. The insets are the magnification of the growing defect and its depth respectively, (C) to (L) are selected images taken in the time interval 8 ÷ 270 𝑚𝑖𝑛 after the enzyme injection. The darker areas represent bilayer structural defects. After enzyme injection, the initial structural defect expands as the top lipid layer is hydrolyzed and the bottom layer spontaneously desorbs. Inset in (C) shows the magnification of the structural defect and numbers correspond to possible number of enzymes hydrolyzed the darkened area. Scan size of all images is 20 × 20 µ𝑚2

91

Ultimately are considered the experiments of simultaneously action of PLA2 and HLL on mixed DPPC-DPG bilayers. In these experiments, after incubating the mixed phospholipid-lipid bilayer with HLL for about 2 h and observing no structural changes, the liquid cell is flushed with PLA2 solution. The injected PLA2 is not expected to flush away the HLL molecules that are firmly adsorbed on DPPC/DPG bilayer, but only these that are in the bulk phase of the liquid phase and those that are loosely attached to the bilayer surface. At Fig. 48 is presented a series of AFM images from a typical experiment. The first image (Fig. 48A) is just prior to PLA2 injection and with HLL present in the AFM liquid cell. Only one structural defect in the double layer is observed. The AFM images at Fig. 48B-L show the consecutive changes which are result of the enzyme action of both enzymes. It is apparent that the desorbed area increased gradually in time. Again, only locations of the surface with the imperfections of the bilayer structure (indicated by arrow in Fig. 48A) are targets of the enzyme attack. The growing defects have finger-like shapes with more profoundly developed channel-like pattern. That pattern is similar to the one observed in the action of PLA2 on DPPC bilayers [110,111]. In the inset of Fig. 48C, the darkened areas of the hole depict the areas degraded by single enzymes. Assuming that about 5% of the edge of the structural defect can accommodate enzyme molecules and, taking into account the fact that the perimeter length of the initial defect is 0.69 m, then the expected number of enzymes is ~9 and the number of finger-like protrusions (artificially darkened regions in Fig. 48C) are similar. At Fig. 49, the kinetics from the PLA2-only experiment (curve B) is compared with the experiment where both PLA2 and HLL are present (curve A). The curve A at Fig. 49 shows the enzyme kinetics obtained from the image sequence at Fig. 48 where only one defect is present. It is apparent that the lag-burst phenomenon is absent and after injecting PLA2 into the liquid cell an immediate increase of hydrolyzed area occurs. To plot curve B, at first the total hydrolyzed area is normalized by the number of defects, which is 10 (Fig. 45). As was previously underlined it is more accurately to use the normalized instead of total area because of the assumptions that only the edge of the defects will accommodate enzymes and the number of these enzymes remains constant. It is apparent that when both PLA2 and HLL are acting the kinetics curve has an “S” shape and exhibits three welldistinguished phases (Fig. 49, curve A). It is reasonable to assume that the first phase of the bilayer hydrolysis is a result only of PLA2 action. The same initial rates for both experiments and the overlap of curves A and B (as the inset in Fig. 49 shows) in the first 25 min supports that assumption. The second phase (indicated by arrow 1 in Fig. 49, curve A) is a burst of enzyme activity. It may be assumed that this second burst is triggered by the HLL activity. As the reaction of PLA2-catalysed DPPC hydrolysis occurs, a percentage of DPPC molecules decreases transferring to Palmitic acid and LysoPC product molecules. When this percentage is sufficiently low (below 0.6 mol fraction) relative to DPG the HLL 92

begins to act as suggested in [168]. Another molecular-level process that favors HLL activation is PLA2-induced compositional change in the upper monolayer (top layer of the bilayer). It is also possible that small DPG enriched domains move by lateral diffusion and merge to form larger DPG areas in the original bilayer structure. Eventually, after 1 h, the hydrolysis of the supported bilayer slows down sufficiently and the hydrolyzed surface area again reaches saturation as a result of the enzyme inhibition described by the molecular mechanism at Fig. 43 of the previous section. To explain all molecular events occurring at the bilayer surface, one can apply the same molecular scheme, adding also the lateral diffusion and merging of DPG-domains in the bilayer structure and HLL activation and action on the bilayer.

Figure 47. Normalized by the number of defects area versus time after PLA2 injection with or without a presence in the liquid cell of HLL. Curve A (PLA2 and HLL) and curve B (PLA2) show the (desorbed area/number of defects) calculated from image series at Fig. 48 and Fig. 45, respectively. The inset shows the initial slopes of the two curves. Arrows show the start of a new phase in the enzyme kinetics

In summary, using LB technique, a mixed DPPC/DPG bilayers are prepared as a substrate for phospholipase (PLA2) and lipase (HLL) action. The bilayers with such composition exhibited a well-defined structures, and proved to be very suitable for both qualitative and quantitative analysis of interfacial lipolytic action. The degradation process is visualized using AFM liquid-cell and in three separate sets of experiments the bilayers are exposed to the action of PLA2, HLL, or both PLA2 and HLL. In the first and the third experimental series (PLA2 and PLA2+HLL), is 93

observed clearly a degradation of supported bilayers as only the initial structural defects are attacked by the enzyme, confirming earlier experimental findings that PLA2 and HLL are activated by the substrates with high curvature. In the second experimental series, the injection of HLL into the AFM liquid cell does not lead to hydrolyses of supported bilayers. This lack of enzyme activity is explained as a result of the low mol fraction of DPG. It is also found that higher than 0.3 mol fractions of DPG make the DPPC/DPG bilayers unstable and easily destructible after their transfer to mica solid support. In the third series of experiments (PLA2+HLL) the enzyme activity is triggered when the PLA2 partially hydrolyzes the supported bilayer. After analyzing the kinetics it is determined that when the two enzymes are present in the AFM liquid cell. PLA2 partially hydrolyzes the bilayer and then the burst of activity occurs as result of HLL involvement in the catalytic act. The obtained AFM images also allow to determine the preferred sites of enzymatic action. Observation and analysis of the pattern of the hydrolyzed bilayers show that the edges of the bilayer structural defects are not saturated with enzyme molecules in shoulder to shoulder fashion but only about 5 % of the available perimeter length of these defects is covered by PLA2 molecules. Finally, the initial rate of hydrolysis is determined to be around 88 DPPC molecules/ s/ enzyme which agrees well with previously published results using a different experimental approach. 4.5.

Surface sensitive synchrotron X-ray scattering as complementary method of AFM for studying lipids and lipases and their mutual interaction at interfaces Since the lipases act on aggregated lipids, monolayer technique is modified by the implementation of barostat set up with the “zero order trough” and properly utilized for lipolysis study [93]. In principle, such experiments relate the macroscopic movement of a barrier on a Langmuir trough to the molecular kinetic event through a micro kinetic model. As an illustration Fig. 50 compares in a schematic way this classical method with the AFM. Some advantages of the AFM for investigation of enzyme kinetics and biological membranes on nanoscale level are apparent. Firstly, AFM “looks” from the cytosol or extracellular space towards the head groups of the membrane lipids. Secondly, AFM is capable of imaging single objects, defects and phase phenomena in a bilayer. As indicated in figure 50B the AFM capture images of the enzyme substrate from the side of aqueous phase, allowing kinetic information to be obtained when the enzyme is added into to the AFM liquid cell.

94

Figure 48 Complementary methods for studying enzyme kinetics of aggregated substrates. (A) The barostat method (i.e. “zero order trough” method). (B) Atomic force microscope (C) Surface sensitive X-ray scattering

It can provide consecutive images of the bilayer change while the enzyme is active. A refinement of the classical monolayer technique by the X-ray scattering method (figure 50C) may be considered as a “competitive” or complementary of the AFM technique on nanoscale level because it provides a way of gaining structural information about the lipids, lipases and lipid-lipase interaction at the air-water interface. Thus, some results on biologically relevant model systems obtained from surface sensitive X-ray scattering (XR and GIXD) are considered. Surface sensitive X-ray scattering has been successfully applied during the past decade for exploring a wide variety of biochemical systems and has provided structural information on a molecular scale on, e.g., bacterial surface-layer proteins, trans-membrane proteins such as bacteriorhodopsin or lipases [169,170,171]. 95

Surface sensitive X-ray scattering is measured in two different geometries: grazing-incidence X-ray diffraction (GIXD) and specular X-ray reflectivity (XR). X-ray reflectivity studies are analogous to the observation of the colors of the rainbow in the reflected visible light from an oil film on a wet road. The phenomenon is due to the interference of light scattered from the air/oil and oil/water interfaces, the thickness of the oil films being of the same order of magnitude as the wavelength of visible light (𝜆 ≈ 5000 Å). This suggests that X-rays (𝜆 ≈ 1 Å) can be reflected from a monomolecular layer to provide information about the molecular packing. GIXD utilizes that a laterally crystalline organization of the molecules gives rise to X-ray diffraction in the same way that visible light is scattered by, e.g., the 2D grating found on a compact disk. Intuitively, very intense X-ray radiation is needed for studying the minute amounts of matter present in a monomolecular layer on a liquid surface. Synchrotron sources dedicated to the production of X-rays have become more available during the past decades with important applications within a broad range of research fields, such as biology, medicine, chemistry and physics [172]. Here is presented how the X-ray radiation from a synchrotron source is applied for investigation of lipid packing and lipid-protein interactions in a Langmuir monolayer of lipids as a model system for biological membranes. One of the advantages of the method is that it allows very small amounts of material to be probed: The illuminated sample area of 2 × 50 𝑚𝑚2 is equivalent to as little as about 1015 lipid molecules spread at the air/water interface. Grazing-incidence X-ray diffraction (CIXD) also demonstrates to be a promising method for obtaining structural information about protein monolayers studied under near-physiological conditions: a 50 Å thick monolayer of from purple membrane constituents (mixture of lipids and the protein Bacteriorhodopsin) was found to form hexagonal 2D crystals at the air/water interface with a lattice constant of 61.3 Å [170] as the diffraction data had a resolution of about 9 Å. In an investigation of bacterial surface proteins the combined information from GIXD and specular reflectivity provided an understanding of the molecular nature of the protein-lipid coupling in the purple membranes. When the protein is recrystallized under the monomolecular lipid film and it is possible to detect the rearrangement of the lipid head groups. The electron density profile of the 90 Å thick recrystallized S-protein layer reveals water-filled cavities near its center where four S-layer protein monomers are located within the unit cell of a square lattice with a lattice constant of about 131 Å. The presented results on the interaction of Humicola lanuginose lipase (HLL) with the lipid 1-mono-oleoyl-rac-glycerol (MOG) and on the pure lipids MOG and 1,2- Dipalmitoyl- sn-glycerol (DPG) provide the steps towards developing of 2D protein crystallography at the air/water interface by means of grazing-incidence X-ray diffraction and specular reflectivity. 96

The theory of the XRD was explained in some details. The X-ray scattering experiments are performed at the synchrotron radiation facility HASYLAB at DESY, Hamburg, Germany, using the liquid surface diffractometer installed at the beam line BW1. A wavelength of 𝜆 = 1.304 Å is used. The supplied by Novo Nordisk A/S (Copenhagen, Denmark) inactive mutant of microbial HLL is a protein with one folded chain of 269 amino acids with a molecular weight of 2.93 × 104 𝑔/𝑚𝑜𝑙 and with an isoelectric point of 4.5. The lipase is dissolved in water giving a clear stock solution of 1.1 𝑚𝑔/𝑚𝐿. Following the experimental procedure for monolayer preparation the lipids (DPG and MOG) were dissolved in chloroform forming clear solutions of 1 𝑚𝑔/𝑚𝐿. The liquid subphase is ultrapure water and the atmosphere is helium gas saturated with water vapor. The lipid is spread on the surface and then compressed slowly with 1 Å2 /𝑚𝑖𝑛 to a moderate surface pressure corresponding to a closely packed monomolecular layer. Prior to lipase injection with a syringe into the subphase, the water level is increased about 0.5 mm underneath the X-ray illuminated spot in order to assist diffusion process. The amount of the injected lipase corresponds to a subphase concentration of about 57 𝑛𝑀. The sample is then equilibrated for approximately one hour before re-lowering the subphase level and performing the X-ray scattering experiments. Further experimental details are given in Table 4.4. Table 4.4 Grazing-incidence X-ray diffraction. Bragg peak data for pure DPG and MOG. The 2D lattice parameter aHex and unit cell area Дсец are calculated assuming hexagonal packing of the hydrocarbon chains. Substance

Temperature Surface Pressure °𝑪 𝒎𝑵/𝒎 Grazing-incidence X-ray diffraction (GIXD) DPG 20.4 29 MOG 10 35 Specular X-ray Reflectivity (XR) DPG 20.3 16 MOG 20.3 15 MOG + Lipase 20.3 21

Area Å𝟐 /𝒎𝒐𝒍 45 37 45 50 47

At first, the obtained results from the Grazing-incidence X-ray diffraction (GIXD) experiments will be summarized. Monomolecular layers of the pure lipids DPG and MOG are investigated and the results are shown at Fig. 51. MOG diffracts only weakly and the Bragg peak can be described as a broad hump above the background. The background is similar to the signal scattered from a pure water surface (dashed line in Fig. 51B). By contrast, DPG gives an intense and sharp single diffraction peak from which can be assumed that the alkyl chains of DPG are hexagonally packed with a lattice constant 𝛼𝐻𝑒𝑥 = 4.80 Å corresponding to an

97

2

area per hydrocarbon chain of 20 Å . The Bragg peak data, given in Table 4.5, reveal a large difference between the estimated sizes, 𝐿𝑥𝑦 , of the crystalline patches of the MOG and DPG monolayers. Table 4.5. Experimental conditions for the samples investigated. All samples had a subphase of neutral pH. The area per lipid molecule, A, is calculated as the area of the Langmuir trough (as limited by the moveable barrier) divided by the number of molecules spread. An XR data set was recorded ca. 2.5 hours after injection of lipase (HLL) under the MOG monolayer. Lipid

𝟐𝜽𝒉𝒌 𝒙𝒚

𝑭𝑾𝑯𝑴

DPG MOG

18.03 15.8

0.164 3.1

𝜶𝑯𝒆𝒙 Å 4.80 5.48

𝑨𝑪𝒆𝒍𝒍 Å𝟐 19.99 26.04

𝑳𝒙𝒚 Å 406 21

The data obtained from Specular X-ray Reflectivity (XR) for monolayers of the pure lipids DPG and MOG are shown in Fig. 51C. Layer models are refined to the measured data (points) and the resulting calculated reflectivity curves are shown as lines in Fig. 51C. The parameters of the layer models are given in Table 4.6 and the plots of the corresponding electron densities 𝜌(𝑧) are shown in Fig. 51D. Table 4.6. Specular X-ray Reflectivity data for pure DPG and MOG analysed by layer models. Lipid: Molecular area, 𝑨 Smearing, 𝝈 Head: Layer thickness,𝑳𝑯 Number of electrons in a layer Electron density, 𝝆𝑯 Tail: Layer thickness,𝑳𝑻 Number of electrons in a layer Electron density, 𝝆𝑻 Subphase: Electron density, 𝝆𝒘𝒂𝒕𝒆𝒓

DPG 2

Å Å Å

Å−3 Å Å−3

41.7 ± 0.2 3.05 ±0.03

MOG 42 ±1 3.50 ±0.06

C5H6O5

C4H7O4

3.64 ±0.05 76 0.50

3.7 ±0.1 63 0.40

C30H62

C17H33

18.05 ±0.06 242 0.32

9.3 ±0.2 135 0.34

H2O −3

Å

0.33

From the obtained electron density profiles is obvious that the hydrocarbon chains of MOG and DPG have similar electron densities in the two models though due to the smaller thickness of the MOG tail layer, the distinct head and tail regions are not so apparent in 𝜌(𝑧) for MOG, as it is for DPG. After the reflectivity data for a pure MOG monolayer are recorded (red triangles in Fig. 51C), lipase is injected in the liquid phase beneath the MOG film. After approximately two hours of incubation the reflectivity is recorded again and the two data sets (with or without enzyme) are compared in Fig. 52. 98

Figure 49 Bragg peaks of (A) DPG and (B) MOG measured by grazing-incidence X-ray diffraction. The X-ray scattering from a pure water surface is shown (dashed curve) for comparison. (C) Specular X-ray reflectivity normalized by the Fresnel reflectivity 𝑅⁄𝑅𝐹 shown as a function of the vertical scattering vector 𝑞𝑧 . Points: Measured data; lines: refined layer models. Black quadrats / full line: DPG; Red triangles/ dashed line: MOG (offset vertically for clarity) (D) Electron density profiles 𝜌(𝑧) normalized by the electron density of water as a function of the depth, z. Solid full line: DPG; dashed line: MOG

The reflectivity is significantly increased at low 𝑞𝑧 , indicating a modification of the interface, presumably by adsorption of the enzyme to the monolayer lipid. From the positions of the pronounced interferences (indicated by dashed lines), a 2𝜋

total film thickness can be estimated as follows 𝐿𝑧 ~ ∆𝑞 = 80 Å. Since the diameter 𝑧

of the enzyme is ~35 Å, this indicates that there may be more than one monolayer of lipase under the MOG monolayer. A biological membrane is a mixture of a large number of different lipids usually comprising such fatty acid derivatives as the lipids investigated here. MOG diffracts only weakly because the (cis) unsaturated bond in the hydrocarbon chain disrupts ordering of the alkyl chains. In contrast, DPG is found to form a hexagonal packing of the hydrocarbon chains with long range crystalline order. This is somewhat analogous to the behavior in 3D: The melting point of DPG is higher than that of MOG. The physical and chemical differences between DPG and MOG alkyl chain packing (Fig. 51A, B and Table 2) are important for the regulation of the fluidity of some biological membranes. 99

The Grazing-incidence X-ray diffraction can be applied not only for investigation of pure lipid monolayers but it also can be utilized for studying the monolayers of proteins adsorbed at air/water interface or of investigation of protein-lipid interactions. Changes in the packing of the hydro carbon chains, i.e. effective length, tilt angle or 2D-crystal system, can be probed using GIXD during adsorption of proteins as a function of e.g. time and/or surface area during expansion or compression of the monolayer. The obtained experimental data for the specular reflectivity show that DPC exhibit more pronounced interferences than MOG (Fig. 51C). This is due to the appearance of more distinct layers of hydrocarbon tails and glycerol head groups for DPG as discussed above. Adsorption of protein under a monolayer of lipids can introduce dramatic changes in the reflectivity as illustrated in Fig. 52 where additional narrow interference fringes are visible indicating an increase of the thickness of the interface. Even through each reflectivity experiment takes approximately 45 min., the adsorption of lipase under a lipid monolayer is possible to be followed as a function of time.

Figure 50 Normalized specular X-ray reflectivity R/RF, for a monomolecular layer of MOG before (black triangles) and 2.5 ℎ𝑜𝑢𝑟𝑠 after injection of lipase under the MOG monolayer (red diamonds). Performed at 𝑇 = 20 °𝐶 and with a surface pressure of 15 𝑚𝑁/𝑚 and 22 𝑚𝑁/𝑚 for pure MOG and MOG/HLL, respectively

In summary: Living cells contain a variety of macromolecular structures in an aqueous phase and their interactions with lipid membranes and other macromolecules play a vital role in many complex biological processes. Consequently, developing model systems for investigation of various phenomena on micro- and nanoscopic scale, are of considerable importance. For example, mono- or bilayers of lipids or mixtures of lipids and macromolecules adsorbed at an interface, as are 100

convenient molecularly structured systems which can be characterized on nanoscale level, by advanced surface sensitive techniques AFM and X-ray scattering (GIXD and XR). These methods allow in situ observation of reactions or interactions at interfaces under near physiological conditions and during the past decade have yielded a multitude of important results, providing good prospects for the future. The presented AFM and X-ray scattering data summarize some of the results from biologically relevant investigations demonstrating the considerable impact of both technique on our present understanding of the molecular structure and phase diagrams of lipids as well as of other surface active molecules, including enzymes and some other biologically important proteins.

101

5. The protease Savinase captured in act by AFM. 5.1.

Savinase action on Bovine Serum Albumin (BSA) monolayers demonstrated with measurements at the air-water interface and liquid AFM imaging At first, the monolayer technique is utilized for hydrolysis of protein monolayers composed of Bovine Serum Albumin (BSA) undergoing proteolytic action of Savinase (EC 3.4.21.14). It is well experimentally established that proteins and particularly BSA can be assembled as monomolecular layers at the air/water interface [48]. Once formed at the interface protein monolayers represent a well define organized substrates for the enzyme action of certain proteases. Savinase is a protease secreted by the alkalophilic bacterium Bacillus lentus and belongs to the subgroup of subtilisin-like enzymes. The crystal structure of the native form of Savinase has been refined from X-ray diffraction data [173] and the enzyme is reported as widely used industrial applicant in detergents [174].The presented experimental data from the BSA monolayer study demonstrate changes of the surface parameters-surface pressure and surface area of spread BSA monolayers at the air/water interface before and after injection of Savinase into the monolayer liquid subphase. On the basis of the experimental data presented as surface area versus time is estimated the turnover number (TON) for the enzyme reaction at the interface. Secondly, the successful application of AFM for visualization of BSA films adsorbed on mica surfaces and subsequently exposed to Savinase proteolytic action is demonstrated. For this purpose the AFM liquid cell is used where the AFM imaging of the BSA monolayers adsorbed on a mica solid supports is performed before and after flushing the cell with enzyme solution. The obtained real time images at different time intervals show a direct visual evidence for the Savinase action (Appendix E). Following the established experimental procedure at first the evolution of surface pressure (𝜋) during spreading of BSA and immediately after injection of Savinase solution in the monolayer subphase are measured. A typical curve  versus t is presented at Fig.53 A. After spreading of 4 𝜇𝐿 of BSA solution on TRIS buffer the surface pressure gradually increases and reaches equilibrium value of 11.5 ± 0.2 𝑚𝑁/𝑚. This equilibrium value is previously reported by Krause et al. [175], although according to dynamic isotherms of BSA, surface pressure versus surface area reported by Boury et al. [176] the inflection point in the isotherm which corresponds to closely packed amino acid segments was at 13 𝑚𝑁/𝑚 giving molecular area of 15 Å2 ⁄𝑎𝑚𝑖𝑛𝑜 𝑎𝑐𝑖𝑑 𝑟𝑒𝑠𝑖𝑑𝑢𝑒. The smaller equilibrium value could be explained for example by the different grade of BSA used. At Fig. 53 A at time 𝑡~ 600 𝑠 the solution of Savinase is injected in the buffer subphase. Initially for about 100 s only a small drop of the surface pressure (about0.5 𝑚𝑁/𝑚) is observed. This drop is further followed by profound decrease of the surface pressure as a 102

result of the hydrolytic action of Savinase. The enzyme cuts portions of the BSA molecules and subsequently, the chopped amino acid segments dissolve in the liquid subphase. As a result the surface pressure decreases.

Figure 51. (A) Typical curve of the surface pressure change during spreading of BSA at the air-water interface. After spreading the BSA solution at 𝑡 = 0, the surface pressure started to increase and reaches equilibrium value of about 11.5 𝑚𝑁⁄𝑚. Than at 𝑡 = 510 𝑠 the solution of the enzyme Savinase is injected (the arrow) into the subphase. (B) Kinetic curves of the decrease of the surface pressure  versus time recorded after the injection of Savinase solution into the monolayer subphase. The kinetics curves correspond to the enzyme concentrations (𝐸0 ) 2.5 𝑛𝑀, 5 𝑛𝑀,17 𝑛𝑀 and 35 𝑛𝑀, respectively. (C) Kinetic curves of the decrease of the surface area A versus time recorded after the injection of Savinase solution into the monolayer subphase. The kinetics curve correspond to the enzyme concentrations (𝐸0 ) 2.5 𝑛𝑀, 5 𝑛𝑀,17 𝑛𝑀 and 35 𝑛𝑀, respectively.

The dependence of the surface pressure decrease ∆𝜋(𝑡) from the enzyme concentration (𝐸0 ) in the subphase is measured at constant surface area corresponding to a closely packed monolayer of BSA molecules. At Fig. 53B is presented the decrease of the surface pressure versus time recorded in the time-course of interfacial hydrolysis after injection of Savinase solution into the subphase. The graph contains four kinetic curves obtained for enzyme concentrations (𝐸0 ) 2.5 𝑛𝑀, 5 𝑛𝑀,17 𝑛𝑀 and 35 𝑛𝑀, respectively. The observed effects of the surface pressure decrease correlate with the increase of the enzyme concentration. It is 103

difficult, though, to use the surface pressure data for theoretical interpretation of the enzyme kinetics because it is hard to establish the equation of state for the BSA monolayer which would relate the surface pressure with surface area. Therefore in the second set of experiments, the surface pressure is kept constant in the time-course of enzyme reaction and the change of surface area (∆𝐴) occupied by BSA molecules at the air-water interface is measured versus time. The value of A is reciprocals to the concentration of BSA at the air water interface. The dependence from the enzyme concentration (E0) of the kinetic curves ∆𝐴(𝑡) is measured at constant surface pressure ∆𝜋 = 11.5 𝑚𝑁/𝑚 (Fig. 53C). Again, the observed effects for the surface area decrease correlates with the increase of the enzyme concentration. The enzyme molecular activity is also called the turn-over number (𝑇𝑂𝑁 ≡ 𝑘𝑐𝑎𝑡 ) and the international units for 𝑇𝑂𝑁 are −1 ⁄ 𝜇 𝑚𝑜𝑙[𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑒] × 𝑠 𝜇𝑚𝑜𝑙[𝑒𝑛𝑧𝑦𝑚𝑒]i.e. number of substrate molecules hydrolyzed per second per number of enzyme molecules. Therefore, from the initial slop of the kinetic curves one could try to estimate the turnover number (TON) of the enzyme reaction for BSA molecules hydrolyzed per second per number of enzyme molecules assuming: (𝑖) the BSA molecules are closely packed at the air-water interface; (𝑖𝑖)that when the whole BSA molecule is hydrolyzed it is instantaneously dissolved into the liquid subphase; (𝑖𝑖𝑖) all enzymes in the subphase are involved in the enzymatic act. Following these assumptions the 𝑇𝑂𝑁 can be calculated from the slopes of the kinetic curves presented at Fig. 53C. The data for 𝑇𝑂𝑁 are presented at Table 5.1. For the calculation are used the initial surface concentration of the BSA in

𝜇𝑚𝑜𝑙 𝑐𝑚2

units, the BSA molecular

mass 𝑀𝑊𝐵𝑆𝐴 = 69000, and the initial enzyme concentration of the Savinase in 𝜇𝑚𝑜𝑙 units. Table 5.1. 𝑺𝒍𝒐𝒑𝒆 𝒄𝒎𝟐 [ ] 𝒔

𝟎. 𝟎𝟎𝟗 𝟎. 𝟎𝟎𝟕𝟐 𝟎. 𝟎𝟎𝟐𝟖 𝟎. 𝟎𝟎𝟎𝟖

𝑹𝒂𝒕𝒆 𝜇𝑚𝑜𝑙 [ ] 𝑠

−8

𝑻𝑶𝑵 1 [ ] 𝑠

𝑪𝑺𝒂𝒗𝒊𝒏𝒂𝒛𝒆 [ 𝜇𝑚𝑜𝑙 ] −7

1.8 × 10 5.3 × 10 1.5 × 10−8 1.5 × 10−5 5.8 × 10−9 2.1 × 10−5 1.7 × 10−9 1.2 × 10−5 𝑨𝒗𝒆𝒓𝒂𝒈𝒆 𝑻𝑶𝑵 = 𝟏. 𝟓 × 𝟏𝟎−𝟓

1.9 × 10−3 9.6 × 10−4 2.8 × 10−4 1.3 × 10−4

1

The obtained mean value for 𝑇𝑂𝑁 = 1.5 × 10−5 , 𝑠 is three orders of magnitude smaller than the lowest value reported in the literature for the action of subtilisin on urea-denatured hemoglobin [177]. The possible explanations of the discrepancy between these numbers could be:

104

(1) The enzyme reaction occurs at the air- water interface which means that the initial enzyme action is hampered by BSA organization at the interface because not all the amino acids are easily available for the enzyme attack. Savinase hydrolyses randomly the peptide bonds within the BSA molecules and as assumed until certain number of amino acids are cut off the BSA molecule does not dissolve into the subphase and remains at the surface, while it is more likely that not the whole BSA molecules but just parts of them are hydrolyzed and subsequently solubilized in the subphase. (2) Not all the enzymes in the subphase are involved in the enzymatic act. As it was previously mentioned according to Verger [93] no more than 1% of the adsorbed human pancreatic lipase (HPL) injected in the monolayer subphase, takes part in the enzyme reaction at the air-water interface. Then, if (1) and (2) are taken into account the TON will substantially increase.

Figure 52. Images 5 × 5 𝜇𝑚2 showing topography changes of the BSA monolayer in the time course of the enzyme hydrolysis: (A) Before Savinase injection into the liquid cell. A defect with a rectangular shape with size 1 × 1 𝜇𝑚2 formed by scratching the surface using the AFM tip; (B) 10 𝑚𝑖𝑛 and (C) 25 𝑚𝑖𝑛 after the enzyme injection. (D) Bigger scan 8 × 8 𝜇𝑚2 made 50 𝑚𝑖𝑛 after enzyme injection as a check for mechanical changes inserted by the tip on the sample. (E) Cross sections of the three images A-C (black lines through the defect) presented at the same graph. (F) Dependence of the average depth of the defect versus time obtained from the whole image sequence (six images in the time interval 0 ÷ 40 𝑚𝑖𝑛 during the enzyme hydrolysis.

105

In a second experimental approach the degradation of the adsorbed BSA monolayers on solid supported mica are visualized using AFM equipped with liquid cell. Fig. 54A to Fig. 54C are images which show the topography changes of the BSA monolayer in the time-course of the hydrolysis. Before enzyme injection into the liquid cell by applying a substantial force on the AFM tip i.e. using the tip as an indenting tool, a defect with a rectangular shape and size 1 × 1 𝜇𝑚2 in the BSA monolayer is formed (Fig. 54A). This defect is created, firstly to serve as a marker during the enzyme reaction and secondly as a check for the depth of the adsorbed on mica surface BSA monolayer. Indeed, as the image cross section shows the measured depth of the defect is 1.6 ± 0.2 𝑛𝑚 which corresponds to the reported thickness of 1.5 ± 0.5 𝑛𝑚 for the BSA monolayer adsorbed on a flat oxidized silicon wafer surface [178]. After injection of Savinase solution into the liquid cell the substantial changes in morphology of BSA surface are observed at the 10th min (Fig. 45B) and 25th min (Fig. 45C). As a result of the enzyme action which has chopped over the BSA surface the depth of the defect is considerably decreased down to depth of 0.3 𝑛𝑚. The cross sections of the three images are compared at the same graph at Fig. 54E. The image at Fig. 54D is a bigger scan made as a check for topographical traces on the BSA monolayer which might have been induced by the tapping of the AFM tip on the sample. Since no detectable changes are present it proves no mechanical influence from the AFM tip on the BSA monolayer. Finally the dependence of the average depth of the defect versus time (are plotted at Fig. 54F. The data are extracted from the whole AFM image sequence obtained during the enzyme hydrolysis. Because, after the enzyme injection the first available for the analysis image appeared after about 8 ÷ 10 𝑚𝑖𝑛 thus it is hard to follow the exact path of the enzyme kinetics in the first minutes of hydrolysis. Nevertheless, again some estimates for the TON can be made by calculating the hydrolyzed volume as multiplying the decrease of the defect height (ordinate of graph at Fig. 54F) by the difference between scanned area and defect area: (𝐴𝑠𝑐𝑎𝑛 ) and the area of defect (𝐴𝑑𝑒𝑓𝑒𝑐𝑡 ), or (𝐴𝑠𝑐𝑎𝑛 − 𝐴𝑑𝑒𝑓𝑒𝑐𝑡 ) = (25 − 1) , 𝜇𝑚2 = 24 × 106 , 𝑛𝑚2 , is multiply by the slope of the curve, 𝑛 presented at Fig. 54F, which is 𝑛 =

1.6 14

= 0.11 𝑛𝑚/𝑚𝑖𝑛 = 1.8 × 10−3 𝑛𝑚/𝑠. Now, if this value is multiplied by the

area difference (𝐴𝑠𝑐𝑎𝑛 − 𝐴𝑑𝑒𝑓𝑒𝑐𝑡 ), the rate of desorbed into the bulk phase BSA, presented in units 𝑛𝑚3 /𝑠, or 𝑛′ = 24 × 106 × 1.8 × 10−3 = 4.3 × 104 𝑛𝑚3 /𝑠. Therefore if assumed that one AAR as a sphere with radius ~2.5 Å, it will have a volume of 4

𝑉𝐴𝐴𝑅 = 3 𝜋(2.5 × 10−1 )3 = 6.5 × 10−2 𝑛𝑚3 /𝐴𝐴𝑅, which means that because BSA con𝑛′

tains 565 AAR then the rate of hydrolysis will be 𝑣ℎ𝑦𝑑 = 565×𝑉 3

𝐴𝐴𝑅

4.3×104

= 565×6.5×10−2 =

1.2 × 10 , BSA molecules/s. From the other hand the injected Savinase has concentration of 10 𝑛𝑀, and the volume of the liquid cell is about 20𝜇𝐿, which means that the number of enzyme molecules in the cell is 𝑛𝑆𝑎𝑣𝑖𝑛𝑎𝑠𝑒 = 𝐶𝑆𝑎𝑣𝑖𝑛𝑎𝑠𝑒 × 𝑉𝐿𝑖𝑞𝑢𝑖𝑑 𝑐𝑒𝑙𝑙 × 1.2×103

𝑁𝐴 = 1.2 × 1011 , [Savinase molecules], hence the calculated TON is 𝑇𝑂𝑁 = 1.2×1011 = 106

10−8 𝑠 −1. This value is sufficiently smaller that the values obtained from the monolayer technique. Therefore, rather to calculate TON to the total number of enzymes in the liquid cell is more correctly to take into account only these enzyme molecules which are active within the area “under surveillance” by AFM. It means that the scanning area of 25 × 106 𝑛𝑚2 , could accommodate the maximal number 𝐴

of Savinase molecules calculated by the formula: 𝑛𝑆𝑎𝑣𝑖𝑛𝑎𝑠𝑒𝑚𝑎𝑥 = 𝐴 𝑠𝑐𝑎𝑛𝑖𝑛𝑔 = 25×106 𝜋×𝑟𝑆𝑎𝑣𝑖𝑛𝑎𝑠𝑒 2

=

25×106 𝜋×32

𝑆𝑎𝑣𝑖𝑛𝑎𝑠𝑒

4

= 8.8 × 10 , [Savinase molecules]. Then the calculated TON will

1.2×103

be 𝑇𝑂𝑁 = 8.8×104 = 1.4 × 10−2 𝑠 −1 which value is almost two orders of magnitude bigger than the one obtained by the monolayer method. In conclusion, one can safely claim the TON spans in-between 10−8 𝑠 −1 < 𝑇𝑂𝑁 < 1.4 × 10−2 𝑠 −1 . In summary, it is demonstrates the enzyme action of Savinase on BSA monolayers either spread at the air/water interface or adsorbed at solid mica supports. In the first occasion the experimentally is studied the degradation of the protein monolayer by measuring the surface pressure and surface area decrease versus time. In a second experimental step the AFM imaging of the supported BSA monolayers adsorbed on mica supports is applied and the information for the enzyme action is extracted by analyzing the obtained AFM images in the time-course of hydrolysis. In both experimental approaches estimates for the turnover of enzyme reaction are obtained as the TONs are comparable although three order smaller than the smallest TON reported for this reaction in the literature. The explanation of this inconsistency could be twofold- (i) when the substrate is organized in monolayer the reaction is retarded (ii) not the whole amount but a certain percent of enzyme is involved in the enzymatic act. In order to be resolved these issues a more detailed kinetic model has to be developed and other appropriate experimental techniques employed. 5.2.

Savinase proteolysis of insulin Langmuir monolayers studied by surface pressure and surface potential measurements accompanied by atomic force microscopy (AFM) imaging The enzyme reactions which occur at the interfaces strongly depend on the molecular organization of substrate molecules. Therefore, one prerequisite step for studying the enzyme kinetics in organized molecular media is to find a suitable molecular architecture of the substrate molecules, either by controlled molecular assembly (i.e. Langmuir- Blodgett films) [179] or by self-assembly procedures [180,181]. The enzyme hydrolysis of insulin monolayers at the air- water interface is considered where the results obtained from classical measurements are accompanied with AFM imaging of the same monolayers transferred on solid supports either by Langmuir- Blodgett or Langmuir- Schaeffer techniques. In principle as it

107

was shown, the polymer monolayers at air–water interface and in particular protein monolayers are substrates suitably organized at a molecular level for the action of proteases [100,101,102]. Once formed at the interface, the monolayer represents structurally well-defined and interfacially controlled substrate for the proteolytic act. In order to find the optimal conditions for the enzyme reaction at the interface, usually as a first step, the state and molecular structure of the monolayers are studied by means of isotherm measurements and, if available, other advanced experimental techniques- fluorescence microscopy (FM), Brewster angle microscopy (BAM), atomic force microscopy (AFM), grazing incidence X-ray diffraction (GIXD) etc. Once the optimal conditions for the enzyme reaction (such as surface pressure, temperature, pH of the subphase etc.) are established, the next experimental step involves the measurements of the evolution of surface parameters i.e. pressure and potential during the enzymatic act. In a case of protein hydrolysis, the surface potential and pressure will change because the hydrolytic scission of the peptide bonds is coupled with the variety of interfacial processes of mass and charge transfer between the surface and the bulk phase as a result of the release of reaction products, solubilization of small segments, etc. In the proposed experimental approach (Appendix F) the next experimental strategy is followed; firstly, the state of the insulin monolayers is studied by surface pressure and surface potential isotherm measurements and AFM imaging. In the second set of experiments the insulin monolayers are spread on Savinase aqueous solution and then simultaneously are measured the decrease of the surface area and the evolution of the surface potential under constant surface pressure. In order to analyze the mechanisms of degradation of insulin molecules, a previously developed kinetic model for -gliadin/Savinase system [100,101,102] is adopted to fit the experimental data. The measured decrease of monolayer area in time is a direct result from the random scission of peptide bonds leading to a progressive fragmentation of the polypeptide chain and subsequent solubilization of the amino acid residues in adjacent aqueous phase. The interpretation of surface potential data is based on the contribution of dipole moments of the intact and broken peptide groups which remain at interface during the proteolysis. Combing the mechanical and electrical data with appropriate kinetic model eventually allows the global kinetic constant for proteolytic action of Savinase to be obtained. The AFM is used to visualize the state of insulin monolayers before and after the proteolytic act. The insulin monolayers are transferred by LB technique onto mica solid supports and the films’ surface topology is studied by means of section and roughness analysis in order to be compared the changes in films’ structure and morphology as a result of the Savinase action. The insulin molecular structure is well established since the pioneer works by Sanger et al published in 1955 [182,183,184]. The molecular weight is determined to be around 6000. The insulin molecule has two polypeptide chains A and B (Fig. 55) which are interconnected by two disulfide bridges, and one intrachain 108

bridge within the 𝑐ℎ𝑎𝑖𝑛 𝐴. The chain A consists of 21 amino acid residues. It is hydrosoluble and negatively charged at pH 7 while 𝑐ℎ𝑎𝑖𝑛 𝐵 is hydrophobic, has 30 amino acid residues, and becomes negatively charged only at higher pH values. For the area occupied by insulin molecules at the air/water interface few models are fairly reliable. For example, using data for the radius of gyration 𝑟 = 10.2965 Å according to the structure 3INC.pdb, a value for the area per a closely compacted insulin molecule 𝐴 ≈ 102 × 𝜋 = 333 Å2 is calculated. However, it seems more reasonable to expect the orientation of the amphiphilic insulin molecule at the airwater interface to be as the chain A is spread at the air water interface while the chain B is oriented towards the adjacent gas phase (the inset of Fig. 55)[185]. Since the polypeptide chain A has 21 residues and the area per amino acid at the airwater interface is reported to be around 20 Å2 /𝑟𝑒𝑠𝑖𝑑𝑢𝑒 [186] therefore in a closely packed monolayer of insulin molecules the expected area is ~20 × 20 = 420 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒.

Figure 53. The insulin molecular structure. The inset- Orientation of the two insulin chains at the air- water interface.

As a first experimental step the insulin is spread from 0.06𝑀 𝐻𝐶𝑙 aqueous solution at pure water surface and simultaneously the change of the surface pressure (𝜋) and the surface potential (∆𝑉) are recorded during the compression of the insulin monolayer. The two isotherms are plotted at the same graph and presented at Fig. 54. Curves 1 and 2 are typical compression isotherms of protein monolayer. The curve 1 represents the surface pressure change (𝜋) vs. surface area (𝐴) and virtually reproduces previously reported results [187,188]. At low monolayer density are measured the values of the surface pressure (𝜋 less than 0.1

𝑚𝑁 𝑚

. When

sufficiently small amount of insulin molecules are spread at the interface is formed 109

a monolayer in molecular state where individual molecules are separated at distances from each other in a thermodynamic state which is considered as 2D gas.

Figure 54. Surface pressure (𝜋) versus surface area (A) (curve 1), and surface potential (𝜋) versus surface area (𝐴) (curve 2) obtained for insulin monolayers spread at air-water interface. The area has scaled to the area required by one amino acid residue. The insets are AFM images showing the surface of LB films picked up when the monolayer has been compressed to areas of ~750 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 (image 1) and at ~425 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 (image 2), respectively.

Further compression of the monolayer leads to the appearance of a gas-condensed (G-C) phase transition at about 800 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 (point A1 at curve 1) with two coexisting phases (G and C). Looking at the curve 2 one could see that the change of the surface potential starts from value ∆𝑉 = 190 𝑚𝑉 corresponding to the area of ~1100 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 and gradually raises up to the value of ∆𝑉 = 270 𝑚𝑉 at the area ~800

Å2 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒

. The surface potential indicates also a G-C

phase transition during the monolayer compression from the area of ~800 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 (point A2 at curve 2) to the area 2⁄ of ~700 Å 𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒. The minimal value of the surface compressibility (point B1 at curve 1) and the saturation of the surface potential (point B2 at curve 2) observed about ~450 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 correspond to a closely packed monolayer of insulin molecules. This finding seems to confirm the idea of the orientation of insulin molecules at the air-water interface with the A-chain, spread at the air/water interface and B-chain facing the adjacent gas phase. In attempt to reveal the fine molecular structure of the insulin monolayer the curves 1 and 2 are considered more closely in a viewpoint of the captured AFM images of LB and LS 110

insulin films transferred onto mica solid supports. It is established that if the monolayers at air/water interface exist in a two phase regime then using different experimental techniques can be observed discrete domains of one phase surrounded by a second coexisting phase. Moreover, such phase domains have been consistently observed in wide variety of shapes, ranging from simple circular disks to complex fractal patterns [189,190]. Buzin et al, experimentally demonstrated by Brewster-Angle Microscopy and AFM that monolayers of Cyclolinear Poly(organosiloxane)s undergo this kind of phase transitions [191].The authors obtained BAM images which clearly showed the coexisting phases in the form of islands of a condensed phase separated by the regions of the other coexisting phase. Thus looking at the isotherms at Fig.54, it could be assumed that the change of the surface potential in the area interval 700 ÷ 800 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 is a result of a similar phase transition process. Nielsen et al [192] has proved that the lateral structure of the monolayer can be successfully preserved by transferring the monolayer from air-water interface to a solid mica support using LB or LS techniques. Hence, following the experimental protocol established in [192] in attempt to explain the molecular structure of the insulin monolayer the LB films transferred at two different molecular areas- at ~750 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 and at ~450 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒, respectively are obtained. The first area correspond to the state of the insulin monolayer in the phase coexistence region while the second is corresponding to closely packed monolayer in state of one condensed phase. The captured AFM images shown at Fig. 55 together with the inserted images in Fig. 56 illustrate the substantial differences in morphology of the LB films’ surface. The first image at Fig. 57A shows the topography of the surface of LB film picked up when the monolayer has been compressed to area of ~750 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒. The coexistence of two phases in the monolayer is evident at Fig. 55A where the islands of one phase are distinguishable and are determined to have an average high of about 0.75 𝑛𝑚 (Fig. 55B). The comparison of the AFM phase image at Fig. 55C with the image at Fig. 55D shows more clearly the coexisting phase domains in the monolayer. The AFM images of LB films obtained at ~425 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 are flat with only one phase present. Some structures with the size of individual insulin molecules are also detectable on the surface though their existence is more likely to be an artifact from the LB transfer. As a next experimental step the proteolytic action of Savinase on monolayer of closely packed at the air-water interface insulin molecules is studied by “zero order” trough. In order to follow the enzymatic kinetics simultaneously under barostatic conditions (∆𝜋 = 𝑐𝑜𝑛𝑠𝑡) are measured the decrease in the surface area (∆𝐴), as well as the evolution of the surface potential (∆𝑉).

111

Figure 55 (A) AFM image of the surface of the LB film picked up when the monolayer has been compressed to area of ~750 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒, and showing the coexistence of the two monolayer phases. The “islands” of one phase (brighter areas) are distinguishable. (B) Section analysis shows an average high of these phase domains about 0.75 nm (D) AFM image of the insulin LB films obtained at ~425 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 shows a flat film surface with only one phase present (C, F) The phase images which show more clearly the coexisting phase domains within the monolayer.

Fig. 56A illustrates the decrease of the surface area in the time-course of proteolysis of insulin monolayer in a condensed phase state, corresponding of ~450 Å2 ⁄𝑖𝑛𝑠𝑢𝑙𝑖𝑛 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒 and at a constant surface pressure 𝜋 = 15 𝑚𝑁/𝑚 corresponding to an almost closely compacted monolayer and a maximum of the enzyme activity. Simultaneously, the changes of surface potential are also recorded and are presented at Fig. 58B.

112

Figure 56. (A) Decrease in the surface area (∆𝐴) and (B) evolution of the surface potential (∆𝑉) versus time (t) during the proteolysis of insulin monolayer. The monolayer is spread at the air-water interface and compressed up to surface pressure 𝜋 = 15 𝑚𝑁/𝑚 then the barostat is turned on to maintain the surface pressure constant and the Savinase solution is injected in the reaction compartment. The final enzyme concentration in the reaction compartment is 10 𝑛𝑀

The initial point at 𝑡 = 0 corresponds to the injecting time of enzyme solution into reaction compartment with final enzyme concentration of 𝐶𝐸 = 10 𝑛𝑀. For time interval 0 ÷ 60 𝑚𝑖𝑛 after the enzyme injection from the presented data at Fig. 58A can be determined the value of the decrease surface area, ∆𝐴 ≈ 12 𝑐𝑚2 , which corresponds to about 4.28% from the initial available surface area 𝐴 = 280 𝑐𝑚2. In the same time is also observed the decrease of surface potential from the initial value of ∆𝑉 = 164 𝑚𝑉 at 𝑡 = 0 to a saturation value of ∆𝑉 = 164 𝑚𝑉 at 𝑡 = 60 𝑚𝑖𝑛. This potential drop of 20 𝑚𝑉 implies for the occurrence of additional charges at the air/water interface as a result of the scission of the peptide bonds by the Savinase. Complementary information about the structure of insulin monolayer is obtained from the analysis of LB films morphology which are transferred before and after Savinase action. The comparison of topography of both kinds of films is presented at Fig 57. The LB film transferred before the enzyme action appears on AFM image flat as the roughness analysis gives for the mean roughness value 𝑅𝑎 = 0.15 ± 0.05 𝑛𝑚 and for the standard deviation of Z (𝑅𝑞 ) the value of 0.22 ± 0.05 𝑛𝑚 (Fig. 59A B). The incidental scratch by the AFM tip has produced a defect on film surface and as it seen from the section analysis at Fig. 59B the depth of this defect is about 0.8 nm which corresponds to the thickness of an insulin monolayer. Sixty minutes after the enzyme injection the transferred LB films show completely different morphology. The observed 3D structures with heights 2.5 ± 0.5 𝑛𝑚 occupy about 6% of the scanned area and they are believed to appear from the collapsed B chains. Also the measured from AFM images values of 𝑅𝑎 and 𝑅𝑞 are almost doubled having values up to 0.26 ± 0.04 𝑛𝑚 and 0.38 ± 0.07 𝑛𝑚, respectively. So, if refer back to the molecular picture of the orientation of insulin molecules at air/water interface (Fig. 53) where the A-chain is facing the water phase and the B-chain is 113

oriented towards the adjacent gas phase then the random scission of the A-chain peptide bonds which are only accessible to the enzyme action will occur and will be followed by a progressive fragmentation of the A-chain. The solubilization of the amino acid residues will eventually lead to formation of 3D collapsed phase from the destabilized B-chains.

Figure 57. The AFM images of the insulin LB films transferred: (A) before the enzyme action and, (C) after the enzyme action, at the end of proteolysis. (B) and (D) are the section analysis curves, made for the images (A) and (C), respectively.

For the interpretation of the experimental data is applied theoretical approach thoroughly developed by Panaiotov et al [100,101,102] and some details of this theoretical framework will be outlined. The decrease of the monolayer area is considered as direct result of the solubilization from the interface of continuously appearing soluble peptide residues of the fragmented 𝐴 − 𝑐ℎ𝑎𝑖𝑛. The equation of mass conservation which takes into account the progressive cleavage of peptides which are initially present in the reaction compartment (with area 𝐴𝑅 ) as well as the contribution of the 𝑖 − 𝑝𝑜𝑟𝑡𝑖𝑜𝑛 (𝐴𝑡=𝑡𝑖 ) continuously supplied from the reservoir into the reaction compartment is [102]: ∆𝐴(𝑡) = 𝐴𝑅 𝛽(𝑡) + ∑ ∆𝐴(𝑡𝑖 ) 𝛽(𝑡 − 𝑡𝑖 )

(5.1)

where 𝛽(𝑡) is normalized decrease per unit area which can be obtained from the experimental data ∆𝐴(𝑡). The meaning of 𝛽(𝑡) obtained from Eq. (5.1) is the fraction of solubilized peptide residues which belong to A-chain per one peptide molecule present at unit area. 114

The degree of completion 𝛼(𝑡) of the proteolysis (the fraction of the broken peptide bonds) is given by the following relation: 𝛼(𝑡) = 1 −

𝑆(𝑡) 𝑆0

(5.2)

where 𝑆0 and 𝑆(𝑡) are the numbers of intact peptide bonds belonging to the chain A at time 𝑡 = 0 and 𝑡, respectively. The mechanism of random fragmentation for a linear polymer assumes that the reaction activities of all peptide groups belonging to the A-chains are equivalent thus the cleavage of the peptide bonds is a random scission process in which the probability of creating peptide molecule with 𝑥 residues starting from one peptide chain with initial number (𝑆0 + 1) residues will be expressed as 𝜔𝑥 (𝛼) = 𝛼(1 − 𝛼)𝑥−1 [2 + (𝑆0 − 𝑥)𝛼] [100]. Then, in case of one solubilized amino acid residue i.e. 𝑥 = 1 after substituting follows: 𝜔1 (𝛼) = 𝛼[2 + (𝑆0 − 1)𝛼]

(5.3)

And therefore the fraction of solubilized residues can be expressed as: 𝛽(𝛼) =

𝜔1 (𝛼) 𝑆0 +1

(5.4)

By fitting the experimental data giving decrease of surface area (∆𝐴) in time (𝑡) (Fig. 58A) with the eqs. (5.1)÷ (5.4), the fractions of reduced area 𝛽(𝑡) and the degree of completion 𝛼(𝑡) in the time course of enzyme reaction may be obtained as it shown in the graph at Fig. 60A.

Figure 58. (A) Fraction per unit area of solubilized broken peptide group 𝛽 and degree of completion of the hydrolysis 𝛼 as a function of time (𝑡); (B) The change of the interfacial concentration of intact (Γ𝑆 (𝑡) ) and broken (Γ𝑃 (𝑡) ) peptide groups with time (𝑡), during the proteolysis, obtained from the experimental data ∆𝐴(𝑡) and Eq. (5.6a) and (5.6b) (𝐶𝐸 = 10 𝑛𝑀; 𝜋𝑐𝑜𝑛𝑠𝑡 = 15 𝑚𝑁⁄𝑚). 115

In the interpretation of experimental data for ∆𝑉(𝑡) (Fig. 58B) is applied the Helmholtz formula [49] as it is taken on account the contribution of dipole moments of intact and broken peptide groups which remain at the interface during proteolysis: ∆𝑉(𝑡) = 4𝜋𝜇⊥𝑆 Γ𝑆 (𝑡) + 4𝜋𝜇⊥𝑃 Γ𝑃 (𝑡)

(5.5)

𝜇

where 𝜇⊥𝑆 and 𝜇⊥𝑃 , (𝜇⊥ = 𝜀 ) are the sum of the vertical components of dipole moments of intact S and broken P peptide groups respectively; 𝜀 is the dielectric constant of water; Γ𝑆 (𝑡) and Γ𝑃 (𝑡) are the surface densities of intact and broken peptide group, respectively. The value for 𝜇⊥𝑆 ≈ 200 𝑚𝐷 is determined from the saturation in the corresponding ∆𝑉(𝐴) isotherm (curve 2, Fig. 56) and Γ𝑆 (𝑡) = 5.8 × 1014 𝑟𝑒𝑠𝑒𝑑𝑢𝑒𝑠/𝑐𝑚2 . At Fig.58B are compared the experimental data ∆𝑉(𝑡) (full line) with the theoretical prediction of ∆𝑉(𝑡) (triangles) where for 𝜇⊥𝑆 and 𝜇⊥𝑃 are plugged the values 𝜇⊥𝑆 = 100 𝑚𝐷 and 𝜇⊥𝑃 = 94 𝑚𝐷 respectively. In a simplified kinetic scheme, adapting the classical Michaelis– Menten model to the interfacial hydrolysis of peptide groups (Fig.61) are added two kinetics steps coupled with the process of the enzymatic binding [102]. The first one takes into account the desorption of broken peptide groups 𝑃 due to the solubilization of segments cut to one single residue while the second step considers the flux of intact peptide groups streaming from the reservoir into the reaction compartment when the barrier moves to compensate the product solubilization and to maintain constant surface pressure.

Figure 59. Schematic model of peptidase action at air/water interface: 𝐸, enzyme in the bulk; 𝐸 ∗ , enzyme fixed at the surface, 𝑆 and 𝑃 denote the intact and broken peptide groups, respectively; ires, the flux of intact peptide groups from the reservoir to the reaction compartment; 𝑘1 , 𝑘−1, 𝑘𝑐𝑎𝑡 are the rate constants of formation and decomposition of the enzymatic complex 𝐸 ∗ 𝑆 ; 𝑘𝑝 and 𝑘𝑑 are the rate constants of penetration and desorption of the enzyme, respectively

Because the enzyme catalytic act occurs at the interface, the enzyme concentration 𝐸 in the reaction compartment must be expressed in units as bulk concentration 𝐶𝐸 while those of 𝐸 ∗ , 𝑆, 𝐸 ∗ 𝑆 and 𝑃 in units of surface concentrations, Γ𝐸∗ , Γ𝑆 , Γ𝐸∗𝑆 and Γ𝑃 , respectively. The system of differential equations which correspond to the proposed kinetic model are solved for the steady state and the final solutions for the surface concentrations Γ𝑆 and Γ𝑃 are [100,101,102]: 116

Γ𝑆 (𝑡) =

𝑡

1 𝑡 𝑑𝛽 ∫ (𝑄𝑚 𝑐𝐸 + 𝑑𝑡 )𝑑𝑡 𝑒0

Γ𝑃 (𝑡) = where 𝑄𝑚 =

𝑘𝑐𝑎𝑡 𝑘𝑑 ∗ ( )𝐾𝑚 𝑘𝑝

[Γ𝑆0 + ∫0 1

𝑡 𝑑𝛽

∫ ( )𝑑𝑡 𝑒 0 𝑑𝑡

𝑡

1

𝑑∆𝐴

𝛼𝑆 𝐴𝑅 𝑑𝑡

𝑡

𝑑𝛽

𝑒 ∫0 (𝑄𝑚𝑐𝐸 + 𝑑𝑡 )𝑑𝑡 𝑑𝑡]

∫0 𝑄𝑚 𝑐𝐸 Γ𝑆 (𝑡)𝑒

𝑡 𝑑𝛽

∫0 ( 𝑑𝑡 )𝑑𝑡

𝑑𝑡

(5.6a)

(5.6b)

, 𝑐𝑚3 × 𝑚𝑖𝑛−1 × 𝑚𝑜𝑙 −1 is a global kinetic constant called ‘‘inter-

∗ facial quality” and 𝐾𝑚 =

𝑘𝑐𝑎𝑡 +𝑘−1 𝑘1

, 𝑚𝑜𝑙 × 𝑐𝑚−2 is the interfacial Michaelis – Menten

constant. From the integrals in Eq. (5.6 a, b) the values of Γ𝑆 (𝑡) and Γ𝑃 (𝑡) respectively can be generated numerically using the experimental data for ∆𝐴(𝑡) and 𝛽(𝑡). Moreover, from the experimental values of ∆𝑉(𝑡) used by means of Eq. (5.5) is determined the time-change of interfacial concentrations of intact Γ𝑆 (𝑡) and broken Γ𝑃 (𝑡) peptide groups. Thus, the two sets of measurements- ∆𝐴(𝑡) and ∆𝑉(𝑡) allow to estimate the values of global constant 𝑄𝑚 by applying an appropriate fitting procedure. The typical result showing evolution of Γ𝑆 (𝑡) and Γ𝑃 (𝑡) is presented at Fig. 60B and the estimated valued of global kinetic constant of hydrolysis is 𝑄𝑚 = 4 × 10−15

𝑐𝑚3

𝑚𝑜𝑙×𝑚𝑖𝑛

. It should be pointed out that 𝑄𝑚 does not depend on enzyme concen-

tration at given surface pressure (𝜋). In comparison with 𝑄𝑚 for the hydrolysis of the protein -gliadin by Savinase [102] obtained value of 𝑄𝑚 is approximately two times smaller but in the same time is two–three orders of magnitude higher than the values of the global lipolytic kinetic constants for small short-chain lipid molecules [92]. In summary: The action of Savinase on insulin monolayer is studied combining two experimental approaches: firstly- by measurements of the decrease of surface area and surface potential in time at constant surface pressure; and secondly, by AFM imaging. The change of surface area gives information about the transformation of substrate into reaction products which are further solubilized in the bulk phase, while the evolution of the surface potential shows the presence of dipole moments of reaction products at the interface. The applied kinetic model, of the interfacial proteolysis, allows to be determined the global kinetic constant 𝑄𝑚 without any fitting parameter. Its value in comparison with the values of 𝑄𝑚 for polymer molecules shows that they are comparable, while the values for the lipolysis kinetics of short-chain lipid substantially differ. The AFM is utilized for study the states of the insulin monolayer at different surface pressures. The captured AFM images of insulin LB films transferred before and after the Savinase action show that the LB film transferred before the enzyme action are flat while at the end of hydrolysis their roughness is increased almost twofold and the appearance of 3D structures from the collapsed insulin 𝐵 − 𝑐ℎ𝑎𝑖𝑛𝑠 is also observed. 117

6. AFM applications for studying the lateral organization and structure of Langmuir- Blodgett (LB) Films. 6.1.

LB films of amphiphilic Hexa-peri-hexabenzocoronene (HBC): new phase transitions and electronic properties controlled by surface pressure studied by AFM KFM and GIXD. It is well established that some polycyclic aromatic hydrocarbons spontaneously organize into mesogenic columnar molecular architectures as a result of the strong 𝜋 − 𝜋 interactions. In the last three decades the materials having such molecular structure attracted considerable attention owing to their fascinating physicochemical behavior as well as to the various potential applications as vectorial transport layers in new organic devices utilized in photography, molecular electronics, solar cells, etc. [193,194,195,196,197]. Hexasubstitutedhexa-peri-hexabenzocoronenes (HBCs) are particularly promising members of this family of compounds because they have shown unusually high charge-carrier mobility along the 𝜋-stacked columns in the bulk mesogenic phases, both for n-alkyl and for phenylenealkyl derivatives [197]. Further functionalization of the alkyl side groups by the termination of either one or all alkyl chains with a carboxylic acid group allows complex formation with polymeric cations thus providing new structures in which hydrogen bonding and 𝜋 − stacking are the main driving forces for the organization of the bulk material [198,199]. In order to understood the principles of self-organization and molecular 𝜋 − stacking the possibilities of nanoscale manipulation and processing of monoacid derivative of HBC (Fig. 62A) organized in Langmuir and LB films are considered (Appendix G). The LB method complemented with some structural studies helps to establish new strategies for design of these molecules showing that the competition between alkyl-chain packing and 𝜋 −stacking in functionalized conjugated surfactants allows the frequently competing packing motives of alkyl chains and 𝜋 −stacks to co-exist without distorting the desired order in the 𝜋 − stack.[200,201,202,203,204,205]. These competing packing modes are also characteristic in HBC molecules in which the alkyl chains are connected to the periphery of a disk rather than to the edges of board-like molecules. In this particular geometry, the packing competition gives rise to two distinct phases: one driven by 𝜋 − stacking, the other by alkyl-chain packing. By applying grazing-incidence X-ray diffraction (GIXD) and AFM imaging in combination with Kelvin Force Microscopy (KFM) is demonstrated that the HBC amphiphilic molecules self-assembles into lamellae on the water surface. In one phase, the 𝜋 −stacks are very well ordered and tilted in a similar manner to fallen domino bricks. This high degree of order leads to band formation in the one-dimensional columns. At higher values of the lateral pressure the HBC Langmuir film undergoes a phase transition in which the alkyl chains are crystalline and the coherence in the 𝜋 −stack is lost. This results in a measurable change in electronic properties of the film. 118

Figure 62. (A) a. Molecular structure of the studied molecule, with approximate dimensions according to standard bond lengths and angles; b. Schematic end-view and topview of the molecule that has crystalline alkyl chains. The top-view shows the alkyl chains as gray circles and the core 𝜋-conjugated system as a black line with protruding p orbitals; c. Schematic end-view and top-view of the molecule having disordered alkyl chains. The top-view representations in b. and c. is used throughout. (B) Compression isotherm of HBC at 21°C marked with arrows B1, B2 and B3 the monolayer states where X-ray measurements are performed. At (B1÷ B3) are presented the electron-density profile arising from the three-step model after fitting the reflectivity data from X-ray diffraction measurements at different surface pressures which correspond to areas per molecule 100 Å2, 80 Å2 and 60 Å2 , respectively. For B1 and B3 is also shown the physical interpretation of the reflectivity data as a schematic drawing of the molecules in sideview on the water subphase.

119

The compression isotherm of a HBC Langmuir film is shown at Fig. 62B. The crystal structure of the Langmuir monolayer is determined by GIXD during compression at three different mean areas per molecule: 100 80

Å2 𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒

(Fig. 62B2), and 60

Å2 𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒

Å2 𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒

(Fig. 62B1),

(Fig. 62B3). The obtained GIXD experi-

mental data show the coexistence of two different crystallographic phases. The high-pressure phase (Fig. 62B3), measured at mean area per molecule of 60

Å2 𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒

, shows the existence of three peaks at 2𝜃𝑥𝑦 with values of 5.47°,10.95°

and 16.46°, respectively (Table 5.1). Table 5.1. Summary of GIXD data for the high-pressure phase.[a] 𝟐𝜽𝒙𝒚 [°] 5.47 10.92 16.46

𝑸𝒙𝒚

[b]

0.46 0.92 1.38

𝒅[c] [Å] 13.66 6.83 4.55

Peak intensity strong weak strong

Index {ℎ𝑘} {10} {20} {01} (or {30})

Coherence Length [d] [Å] 280 150 120

[a] The tabulated values correspond to the X-ray diffractogram shown at the top of (Fig. 62B3) and measured at 60 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒 . [b] Scattering vector. [c] Repeat distance. [d] The coherence lengths were calculated with the Scherrer formula.

The peak with the highest angle ( 2𝜃𝑥𝑦 = 16.46° ) corresponds to a spacing of 𝑑 = 4.55 Å and is similar to a peak found in reported GIXD data for other amphiphiles with alkyl chains packed in a hexagonal lattice, for example, alcohols or carboxylic acids [75]. It should be notices that if the structure was non-hexagonal then the peak would be split [206]. The two other peaks at lower angles are believed to result from a trimerized superstructure of the alkyl chains linked to the HBC core. With six alkyl groups linked covalently to the HBC core, this would mean that three alkyl chains are pointing up from the HBC core and three are pointing down. Because no mixed-indexed {ℎ𝑘} reflections are observed the unit cell parameters of the high-pressure phase cannot be uniquely determined, however the most likely structure is the hexagonal with unit cell parameters 𝑎 = 3𝑏 = 15.78 Å, 𝑏 = 5.26 Å, 𝛾 = 120°, as it shown in Fig. 63B. This is in agreement with the AFM data shown below. The unit cell area is 𝐴 = 71.9 Å2 , which indicates that the measurement taken at a mean molecular area of 60 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒 is performed in the collapse area of the closely packed HBC monomolecular film. At a smaller surface pressure of 50 𝑚𝑁/𝑚, which corresponds to a nominal area 80 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒, the coexistence of two phases is observed. At a further reduction in surface pressure to a nominal area per molecule

120

of 80 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒, mainly diffraction from a low-pressure phase is seen. The diffraction from this low-pressure phase comes from a crystal structure with a uniquely identified rectangular unit cell: 𝑎 = 22.95 Å, 𝑏 = 4.94 Å, 𝛾 = 90°, 𝐴 = 3 × 37.8 = 113 Å2 (Fig. 63A).

Figure 63. Structures of Langmuir films of amphiphilic HBC. (A) Above: unit cell parameters of the low-pressure phase with: 𝑎 = 22.95 Å, 𝑏 = 4.94 Å, 𝛾 = 90°, 𝐴 = 113 Å2 . Below: the schematic packing with the dark parts representing a top view of p-stacking conjugated parts of the molecules. (B) Above: unit cell parameters of the high-pressure phase: 𝑎 = 15.78 Å, 𝑏 = 5.26 Å, 𝛾 = 120°, 𝐴 = 71.9 Å2 . Below: the schematic packing with circles showing the hexagonal close-packed alkyl chains that are trimerized on account of their covalent linkage to the conjugated core of the molecule shown in black with protruding p orbitals. Compared to the low-pressure phase (A), the  overlap in the highpressure phase (B) is diminished.

Table 5.2. Summary of GIXD data for the low-pressure phase.[a] 𝟐𝜽𝒙𝒚 [°] 3.26 5.4 6.51 9.78 13.05 15.17 15.51 16.3 18.05 20.0 22.4

𝑄𝑥𝑦

[b]

0.27 0.45 0.55 0.82 1.10 1.27 1.30 1.37 1.51 1.68 1.87

𝑑[c] [Å]

Peak intensity

Index[d] {ℎ𝑘}

22.95 13.8 11.48 7.66 5.74 4.94 4.83 4.60 4.15 3.74 3.36

Very strong Medium Strong Very weak Weak Medium Medium Mid-strong Weak Mid-weak Very weak

{10}} ℎ𝑝{10} {20} {30} {40} {01} {11} ℎ𝑝{10}/{50} {31} {41} {51}

Coherence Length [e] [Å] 323 303 426 253 252 175 180 284 208 259

[a] The tabulated values correspond to the X-ray diffractogram shown at the top of (Fig. 62B 3) and measured at 100 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒 . [b] Scattering vector. [c] Repeat distance. [d] The {ℎ𝑘} index corresponds to the unit cell parameters 𝑎 = 22.95 Å, 𝑏 = 4.94 Å, 𝛾 = 90°, 𝐴 = 3 × 37.8 = 113 Å2 , with the hp indices arising from high-pressure peaks coexisting with the low-pressure phase. [e] The coherence lengths were calculated with the Scherrer formula. 121

All of the unusually high number of observed reflections can be indexed according to this unit cell (except two, marked with ℎ𝑝 in Table 5.2, which result from a reminiscent high-pressure phase). The molecules form columnar stacks at the water surface, with an inter-columnar spacing along the 𝑎 − 𝑎𝑥𝑖𝑠 and with the repeat distance 𝑏 = 4.94 Å along the stack. As evidenced by X-ray reflectivity the HBC core is in a tilted orientation in the low-pressure phase. This is in agreement with the notion that the normal 𝜋 − 𝜋 contact distance between the HBC cores is 3.5 Å [193] which implies that the MAHBC core is tilted by 45° with respect to the surface normal to allow the projected repeat distance of 4.94 Å. The coherence lengths along the stack are estimated from the peak-widths to be 200 ÷ 400 Å (Table 2). It is important to notice that the coherence of the conjugated 𝜋 stack (as evidenced by the peaks at 2𝜃𝑥𝑦 values between 15° and 23°) is lost in the transition from the low- to the high-pressure phase. This is most likely a consequence of the incommensurability between the crystal structure of the closely packed alkyl chains (being closely packed by the applied surface pressure) and the structure of the 𝜋 stack. The loss of coherence in the 𝜋 stack gives rise to a change in the electronic potential, as shown by Kelvin force microscopy (KFM) measurements on LB transferred films. It is not directly evident from the compression isotherm that two phases coexist, since no plateau region in the compression isotherm is observed, as it is typically found in e.g. phospholipid systems [192]. The absence of such a horizontal region can be due to the high crystallinity (compared to lipids) of the monolayer resulting in very long equilibration times. The vertical electron-density profile of the Langmuir monolayer is determined by X-ray reflectivity measurements performed at the same surface pressures as the diffraction experiments. The experimental reflectivity data are presented with the circles in the graphs at the middle row of Figs. 62B1-63B3 along with the data fits (solid lines). The reflectivity curves show that as the surface pressure increases, as expected the thickness of the monolayer also increases. The fit to the reflectivity data of the low-pressure phase gives a monolayer thickness of 30 Å. This thickness is too small which means that the molecules must be in a tilted conformation, since the length of the fully extended molecule as it shown at Fig. 62A should be 40 Å. When the surface pressure increases, the electron-rich conjugated center of the molecules is clearly seen as a protrusion of the electrondensity curve, with the electron-poor alkyl layers below and above. The fit to the reflectivity data of the high-pressure phase shows a vertical tailing of the electron density (up to 80 Å). This indicates that a portion of the molecules have been pushed out of the monolayer indicating the existence of collapse. The reflectivity fits are in agreement with the model described above: at low pressure the molecules are tilted and efficient and coherent packing is obtained, which is dominated by 𝜋 −stacking. This coherence is then lost on compression.

122

AFM measurements on HBC films LB- transferred to solid supports are performed in Deflection-mode of imaging in order to be captured AFM images of the high-pressure phase with atomic resolution.

Figure 64. (A) Deflection mode AFM image of an LB monolayer of HBC transferred from the air/water interface at a mean area 80 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒 (high-pressure phase) to atomically flat mica. The cell parameters are 𝑎 = 5.1 Å, 𝑏 = 4.5 Å, and 𝛾 = 108°. A cartoon of HBC molecules stacked along a horizontal direction is shown, in which each molecule has three protruding alkyl chains; (left upper corner) 2D Fast Fourier Transform (FFT) of the image also proves the same molecular pattern (B) AFM tapping-mode topography image of a LB monolayer of HBC transferred to Si-wafer with natural oxide layer at mean area per molecule 100 Å2 ⁄𝑚𝑜𝑙𝑒𝑐𝑢𝑒𝑙𝑒 and a cross-section profile from the middle of the scan, showing the height difference between the high-pressure and the low-pressure phases. The height difference is 5 Å. (C) KFM image showing the potential landscape of the same area as that shown in a). The potential of the low-pressure phase is 20 𝑚𝑉 higher than the potential of the high-pressure phase, as also can be seen from the cross-section profile.

123

Fig. 64A shows an AFM image of a monolayer transferred onto mica by horizontal lifting i.e. by LS method. The HBC monolayer is compressed to 80 Å and then transferred by having the substrate in the aqueous subphase prior to spreading of the monolayer. The AFM images of the high-pressure phase with atomic resolution show that it consists of closely packed (crystalline) alkyl chains that protrude from the surface. The repeat distances are 𝑎 = 5.1 Å, 𝑏 = 4.5 Å, while 𝛾 = 108°. The area of the unit cell is 𝐴𝐴𝐹𝑀 = 21.8 Å2 . One possible orientation of the molecules as seen from the top is sketched in Fig. 63. The deviation from the observed X-ray data parameters ( 𝑎 = 3 × 𝑏 = 15.78 Å, 𝑏 = 5.26 Å, 𝛾 = 120° and 𝐴 = 71.8 Å2 = 3 × 23.9 Å2 ) could be the result of relaxation during transfer to a solid support, although it is more likely to be a consequence of nonlinearity in the XYpiezoes of the AFM. At Fig.64B is presented a tapping mode AFM image showing the topography of a HBC monolayer transferred by LS method on Si wafer with a natural hydrophilic oxide coating. The LS transfer is performed after compressing the HBC Langmuir monolayer the monolayer to a mean molecular area 𝐴 = 100 Å2 and then by horizontal lifting up. This area corresponds to a point in the phase diagram that is dominated by the low-pressure phase (Fig. 62B2). Together with the topography of the sample (Fig.64B) an image obtained by Kelvin force microscopy (KFM) (Fig. 64C) visualizes the electrical potential of the same HBC film. The presence of two phases is clearly visible as height differences in the topography image shown in Fig.64B. The height difference between the two phases is of the order of 5 Å, which agrees with the height differences seen in the reflectivity data (Fig. 62). The height difference suggests that the high areas are domains of molecules in the high-pressure phase, with surroundings that consist of molecules in the low-pressure phase. This is also supported by the fact that the alkyl packing seen by AFM in Fig. 64A could only be found in the high areas. The KFM micrograph (Fig. 64C) shows that the surface potential is inversely correlated to the topography graph: the domains which are high in topography have low potential and vice versa. The measured nominal potential difference of 20 𝑚𝑉 is in the range generally observed for bandwidths of 𝜋 stacks [207]. An estimate of the number of charge carriers can be given by examining the almost circular high-pressure domain seen under the rightmost marker in the KFM image (Fig. 64C). Assuming that this domain and its immediate surroundings can be modeled as two coaxial cylinders with a radius of 𝑏 = 220 𝑛𝑚 and 𝑎 = 219.5 𝑛𝑚 and height of 𝑑 = 3 𝑛𝑚, the capacitance is given by the equation [208]: 𝐶=

2𝜋𝜀𝑑 𝑏 𝑎

ln

(6.1)

124

In (6.1) 𝑑 is the thickness of the monolayer and 𝜀 is the dielectric constant. The number of charges on the inner domain can then be found from the basic equation of capacitance: (6.2)

𝑄 = 𝐶𝑉

In vacuum dielectric constant is 𝜀 = 8.854 × 10−12 , 𝐹𝑚−1 , thickness 𝑑 ≈ 3 𝑛𝑚, and voltage drop 𝑉 = 20 𝑚𝑉, the number of charges is calculated as given in (6.1): 𝑛=

𝑄 𝑒

=

2𝜋𝜀𝑑 𝑏 ln 𝑎

𝑉 ⁄𝑒 =

2𝜋×8.85×10−12 ×20×10−3 𝑏 𝑎

(ln )×1.6×10−12

=9

(6.3)

This calculation of the order of magnitude indicates that the number of electrons is very low. The number of molecules inside the domain is about 212 000. Despite the fact that the film is undoped, the nine holes in the high-pressure phase to be filled by the nine electrons from the low-pressure phase can easily be explained by impurities, corrupted molecules etc. In summary: A prerequisite of high charge-carrier mobility in thin films of discotic molecules is coherent order in the 𝜋 stack. HBC with solubilizing alkyl substituents proves to be a good candidate for balancing some often counteracting requirements for building 𝜋 stacked molecular architectures. For practical applications the transfer of HBC on solid supports e.g. mica or Si wafer from Langmuir films seems to be a promising route for obtaining highly ordered thin films with good charge-carrier mobilities. The examined HBC films by AFM and KFM imaging show the coexistence of two phases with different electronic properties persisted during the transfer from liquid to solid surface. The KFM experiment can be explained on the basis of a PN junction, which has a built-in potential drop across the junction which will have a negative P side and a positive N side because electrons flow from the N side to the P side to recombine with the holes on the P side. The potential of the N side will therefore rise above the P side in the KFM image, as described elsewhere [209,210,211,212,213]. On the basis of the obtained structural data, is assumed that the HBC electrons are delocalized along the 𝜋 stack in the low-pressure areas, and, hence, the formation of a band structure that gives rise to a reduction in the HOMO-LUMO gap relative to the localized HBC core is expected while in the high-pressure domains is assumed that the electrons are highly localized on the molecules (Fig. 65A). These assumptions are supported by the obtained X-ray data, which indicate that coherence between the graphite disks exists in the low-pressure phase, leading to a better 𝜋 − 𝜋 overlap than in the high-pressure phase (Fig. 62), in which the coherence in the 𝜋 stack is lost. When the two phases are brought into contact, the more loosely bound electrons lying high in the “conduction” band of the low-pressure phase will be attracted to the more tightly bound molecular orbitals in the high-pressure phase. 125

As a result a small amount of electrons (