Fiber Diffraction As a Screen for Amyloid Inhibitors

2 downloads 0 Views 5MB Size Report
1Biology Department, Boston College, Chestnut Hill, MA 02467, USA; ... 3Chemistry Department, Mount Holyoke College, South Hadley, MA 01075, USA.
288

Current Alzheimer Research, 2008, 5, 288-307

Fiber Diffraction As a Screen for Amyloid Inhibitors Daniel A. Kirschner1,*, Abby A. R. Gross1, Marla M. Hidalgo1, Hideyo Inouye1, Katherine A. Gleason1, George A. Abdelsayed1, Gerardo M. Castillo2, Alan D. Snow2, Angela Pozo-Ramajo3, Sarah A. Petty3 and Sean M. Decatur3 1 3

Biology Department, Boston College, Chestnut Hill, MA 02467, USA; 2ProteoTech, Inc., Kirkland, WA 98034, USA; Chemistry Department, Mount Holyoke College, South Hadley, MA 01075, USA Abstract: Targeting the initial formation of amyloid assemblies is a preferred approach to therapeutic intervention in amyloidoses, which include such diseases as Alzheimer’s, Parkinson’s, Huntington’s, etc., as the early-stage, oligomers that form before the development of -conformation-rich fibers are thought to be toxic. X-ray patterns from amyloid assemblies always show two common intensity maxima: one at 4.7 Å corresponding to the hydrogen-bonding spacing between the -chains, and the other at ~10 Å corresponding to the spacing between -pleated sheets. We report here the application of fiber x-ray diffraction to monitor these structural indicators of amyloid fiber assembly in the presence of small, aromatic molecules, some of which have been assessed by other techniques as being inhibitory. The compounds included butylated hydroxytoluene, chloramphenicol, cotinine, curcumin, diphenylalanine (FF), ethyl 3-aminobenzoate methane sulfonate, hexachlorophene, melatonin, methylpyrrolidine, morin, nicotine, phenolphthalaine, PTI-00703® (Cat's claw), pyridine, quinine, sulfadiazine, tannic acid, tetracaine, tetrachlorosalicylanilide, and tetracycline. Their effects on the aggregation of A1-40, A11-25, A12-28, A17-28, A16-22, and A16-22[methylated] analogues were characterized in terms of the integral widths and integrated intensities of the two characteristic reflections. Peptide A11-25 with or without small molecules showed varying relative intensities but similar coherent lengths of 28– 49 Å in the intersheet and 171–221 Å in the H-bonding directions. PTI-00703®, however, abolished the H-bonding reflection. Among previously reported aromatic inhibitors for A11-25, PTI-00703®, tannic acid, and quinine were more effective than curcumin, morin, and melatonin based on the criterion of crystallite volume. For the N-methylated and control samples, there were no substantial differences in spacings and coherent lengths; however, the relative volumes of the crystallites, which were calculated from the magnitude of the intensities, decreased with increase in concentration of A16-22Me. This may be accounted for by the binding of A16-22Me to the monomer or preamyloid oligomer of A1622. The fiber diffraction approach, which can help to specify whether an amyloidophilic compound acts by impeding hydrogen-bonding or by altering intersheet interactions, may help provide a rationale basis for the development of other therapeutic reagents.

Keywords: Alzheimer’s disease / amyloidosis / drug inhibition / protein folding / x-ray diffraction / N-methylated A16-22 / aromatic inhibitors. INTRODUCTION “Amyloid”, a 19th C. designation based on staining amylose with iodine/sulfuric acid, now refers to fibrillar deposits of misfolded proteins. Amyloids share the common features of a fibrillar ultrastructure, binding of Congo red and thioflavins, and a cross- arrangement of polypeptide chains. Over 30 different proteins have been identified as amyloid precursors [1, 2], including amyloid- (Alzheimer’s), synuclein (Parkinson’s), prions (transmissible spongiform encephalopathies), polyglutamine repeat proteins [3], and polyalanine repeat-containing proteins (e.g., PABP2 in oculopharyngeal muscular dystrophy [4,5]). Amyloid fibrils have widths ranging from ~50-100 Å or more, are nonbranching, fairly rigid, often twisting, and of indeterminate length. Amyloid can also manifest as amorphous, or as nonfibrillar, variable-width ribbons or plates, particularly when *Address correspondence to this author at the Biology Department, Boston College, 140 Commonwealth Avenue, Chestnut Hill, MA 02467-03811, USA; Tel: 617-552-0211; E-mail: [email protected] 1567-2050/08 $55.00+.00

derived from synthetic peptides with native sequences—e.g., A [6], amylin [7], SAA [8], and prion-related peptides [9]. Such assemblies may represent the core folding regions of the in vivo amyloid and relate to transition structures that occur during self-assembly [6, 10, 11]. “Aging” of amyloidogenic peptides to oligomers etc, may herald transition to a -conformation and cytotoxicity. The development of inhibitors that interfere with protein folding/assembly could act early in amyloidogenesis and help thwart formation of toxic intermediates [10]. By contrast, disruption of mature fibrils might result in a shower of toxic species. In Alzheimer brain, the major polypeptide components of amyloid plaque are A1-40 and A1-42. Neuronal death in Alzheimer’s disease has been ascribed to either or both the pre-fibrillar oligomers/protofilaments of A amyloid [12, 13] and the fibrillar form [14, 15]. The fibrillar form of amyloid, by contrast, may have a beneficial role if it sequesters abnormally-folded proteins and prevents their aggregation into toxic oligomers [16]. In support of the fibril’s toxic effect, however, is positron emission tomography (PET), which ©2008 Bentham Science Publishers Ltd.

Inhibition of Amyloid Formation

shows that the in vivo amount of fibrillar amyloid correlates with lower scores in memory performance [17, 18]. One strategy in therapy and diagnosis has been concerned with the use of small molecules to inhibit amyloid formation and, thereby, lower the amyloid burden. Since Alzheimer's  amyloid (A) contains a core structure of sheets constituted by residues 17-21 (sequence LVFFA; [6]), then this domain has been targeted with short peptides as a key site for inhibition or disruption. Such inhibitors include KLVFF, which is homologous to the hydrophobic core [19], and LPFFD [20], which is also similar to the core but contains a  breaker residue (proline). N-methylated peptides that lack some of the amide protons required for -sheet formation have also been tested [21, 22]. Another prospective target in A [23] is the putative binding site for Congo red and heparan sulfate [24-26] at residues 12-16 (sequence VHHQK). Because amyloid-specific dyes such as Congo red are aromatic [26,27], similar compounds have been extensively studied as possible inhibitors and as imaging agents for positron emission tomography (PET)—e.g., tetracycline [28, 29], a Cat's claw derivative (PTI-00703®; [30]), quinine [31], morin [32], melatonin [33], curcumin [34, 35], tannic acid [36], and cotinine [37]. Methods for assessing fibril assembly and disassembly by small molecules include electron microscopy (EM), light scattering, sedimentation, turbidity, thioflavine T fluorescence, Congo red UV absorbance, and atomic force microscopy (references cited in [26, 38]). Most of these methods, however, do not provide estimates of the size and quantity of the fibrils. In the current study we propose a fiber x-ray diffraction approach for the quantitative assessment of amyloid inhibition by a variety of compounds including the ones cited above and new ones, i.e., tetracaine, phenolphthalein, dipeptide FF [39, 40], hexachlorophene, chloramphenicol, butylated hydroxytoluene, sulfadiazine, tetrachlorosalicylanilide, and ethyl 3-aminobenzoate methane sulfonate. We examined the effect of small molecules on A aggregation by measuring the integral widths and integrated intensities of the two reflections that are characteristic of -crystallites— namely, the one at 4.7 Å and at ~10 Å Bragg spacing [6]. To estimate the binding constants for peptide monomer addition and inhibitor-peptide interaction we used the observed coherent length of the 4.7 Å reflection as applied to a linear assembly model. The crystallinity or volume of the -sheet assembly was also estimated from the integral intensities of the reflections. In addition to using the physiological peptide A1-40, we also examined A analogues 11-25, 12-28, 1728 and 16-22. Peptides A1-40, A11-25, and A12-28 contain the putative -sheet core domain (residues17-21) and heparan sulfate binding domain (residues 12-16), while peptides A17-28 and A16-22 contain only the -sheet core. By using these peptides the site of action of the prospective inhibitors could be studied. Among the previously reported analysis of inhibitors based on EM and thioflavin T fluorescence, our current x-ray diffraction assessment showed that Cat’s claw, tannic acid, and quinine were more effective than curcumin, morin, and melatonin in impeding amyloid fiber formation.

Current Alzheimer Research, 2008, Vol. 5, No. 3

289

MATERIALS AND METHODS Small Molecules and Peptides The following compounds were studied for their effects on fibril formation: tetracycline hydrochloride (Calbiochem, #58346; Lot B51586), tetracaine hydrochloride (Sigma, # T7508; Lot 41F-0218), quinine (Sigma; #Q-1625; Lot 78C0157), phenolphthalein (Sigma; #P-7570; Lot 072K1292), hexachlorophene (Sigma, #H-4625; Lot 28C-0375), ethyl 3aminobenzoate methanesulfonate salt (Sigma, #A5040), chloramphenicol (Calbiochem; #22051; Lot-B32807), butylated hydroxytoluene (Sigma, B-1378; Lot 70K0071), sulfadiazine (ICN Pharmaceuticals, # 102983; Lot 4867), 3,3,4,5 tetrachlorosalicylanilide (Kodak, #10677), tannic acid (Sigma, Lot 66F-0751), melatonin (Sigma, M5250; 065K1239), curcumin (Sigma, C1386-5G; 045K0933), morin (Sigma, M4008-2G; 114K2617), cotinine (Sigma, C5923-250MG, 055K4052), nicotine (Sigma, N0267100MG, 0355K4076), pyridine (Sigma, 270407-100ML, 10510PC) and methylpyrrolidine (Aldrich, M79204-5ML, 04730HC) (Fig. 1), and PTI-00703® (isolated from Cat’s claw) (ProteoTech, Inc., Kirkland, WA). For the first ten molecules listed above, lyophilized A11-25 (EVHH QKLVFFAEDVG; Dr. David B. Teplow) was dissolved at 10 mg/ml, either by itself in distilled water or added to a solution containing the test compound at a peptide:compound weight ratio of 3:1. Peptide A12-28 was obtained from Research Biochemicals International (SigmaAldrich, St. Louis, MO) and Bachem (King of Prussia, PA), and A17-28 was from Peninsula Laboratories, L.L.C. (San Carlos, CA). Lyophilized A1-40 (from Biosource, US Peptide, or Bachem Peptide) was dissolved at 1 mg/ml, either alone in distilled water or added to a Cat’s claw solution, at peptide:PTI-00703® weight ratio of 3:1 or 6:1. Each sample solution was aspirated into a 0.7 mm-diameter siliconized thin-wall glass capillary tube (Charles A. Supper, Co., South Natick, MA, USA), which was then flame-sealed at the narrow end. The wide-end of the capillary was sealed with wax through which a pin-hole was punched using a hot needle. The peptide solution in the capillary tube (which was vertical) was then allowed gradually to dry under ambient temperature and humidity. The dehydration was monitored by the height of the solution in the tube, and formation of oriented assemblies or liquid-crystalline domains was monitored using a polarizing microscope. The peptide solutions were dried to small, uniform disks. Some of the peptide samples were examined at different hydration states by X-ray diffraction. Protocol for Preparing A1-40 + FF Dipeptide Lyophilized phenylalanine dipeptide (FF) (gift from M. Reches and E. Gazit, Tel Aviv University) was dissolved in hexafluoroisopropanol (HFIP) at a concentration of 100 mg/ml to form a stock solution that was then diluted to 2 mg/ml in double distilled water (ddH2O). The peptide solution of 5:1 weight ratio A1-40:FF was then prepared by dissolving 0.10 mg into 10 μl of the FF solution. The final concentration of A1-40 was 10 mg/ml. Subsequent treatment of the sample was the same as above.

290 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (1). Chemical structures of small organic compound tested for inhibiting fibril formation. The chemical structure of tannic acid was reproduced from (Fig. 1 cited in Ref. [91]). Others are from ChemFinder at http://chemfinder.cambridgesoft.com/. Not included here is PTI00703® from Cat’s claw.

Concentration-Dependent Effect of Inhibitors on A1228 and A17-28 Fibril Formation The coherent domain size along the H-bonding direction was measured for dehydrated A12-28 (VHHQKLVFF AEDVGSNK) and A 17-28 (LVFFAEDVGSNK) as a function of concentration of the added small molecules. The concentration of solubilized A12-28 peptide was kept constant at 10 mg/ml (5.1 mM) and the compound's concentration varied in the range of 2.55–10.23 mM for tetracycline and cotinine, and for A17-28, the peptide concentration was 5 mg/ml (3.8 mM) and the molar ratio of peptide:compound was >1:6 for cotinine, pyridine, methylpyrrolidine, and nicotine in 8.8% formic acid.

N-Methylated A 16-22 Peptide The samples included: A16-22 (Ac-KLVFFAENH2;”A”); A16-22Me (Ac-KLVFFAE-NH2, where alternate residues (underlined) were methylated;”B”); a 1:1 molar ratio of A16-22 and A16-22Me (”C”); and a 1:5 ratio of A16-22 and A16-22Me (”D”). The peptides, synthesized in the Decatur lab [22], were purified by HPLC and lyophilized, then dissolved for 6 h in a 0.05 M DCl solution for H/D exchanging and re-lyophilized. The two distinct peptides (A16-22 and A16-22Me) were dissolved separately (1 mM potassium phosphate buffer, pH 7), and then mixed at 1:1 and 1:5 molar ratios of unmethylated to methylated peptide.

Inhibition of Amyloid Formation

Current Alzheimer Research, 2008, Vol. 5, No. 3

291

Peptides Used in Current Study A

1-40

DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV

11-25

EVHHQKLVFFAEDVG

12-28

VHHQKLVFFAEDVGSNK

17-28

LVFFAEDVGSNK

16-22

X-Ray Diffraction Measurements of x-ray diffraction patterns from the peptides were conducted at room temperature using the Oxford diffraction Xcalibur PX Ultra system (Oxford Diffraction Ltd., 130A Baker Avenue, Concord, MA 01742, USA) located in the laboratory of Dr. Andrew Bohm (Department of Biochemistry, Tufts University, Boston, MA) [41]. The CuK x-ray beam was generated using an Enhance Ultra, which is a sealed tube-based system incorporating confocal multilayer optics. The x-ray beam was monochromated and the K component was removed by means of the double bounce within the confocal optic. The x-ray beam was focused to 0.3 mm x 0.3 mm (full-width at half-maximum width at detector position). A two-dimensional Onyx CCD detector (Oxford Diffraction Inc., Concord, MA) was placed 85 mm from the sample position, covering the scattering range for Bragg spacings 1.8–54 Å. The sample-to-detector distance was calibrated using a spherical ylid crystal (mol formula C10H10SO4) or a cubic alum crystal according to the information given by manufacturer. For calibrating the specimen-to-film distance we used the Bragg peaks from silver behenate (58.38 Å period) [42, 43]. The active range of the detector was 165 mm, and the two-dimensional image (1024 x 1024 pixels; in 2  2 binning) was collected using the software package CrysAlis (CrysAlis CCD and RED, version 171 (2004), Oxford. UK) and stored in the compressed image format IMG. Exposure time was 150 seconds. The diffraction image was displayed by NIH image software (developed at the U.S. National Institutes of Health and available at http://rsb.info.nih.gov/nih-image/) or program FIT2D for powder diffraction [44]. From the known Bragg peaks from silver behenate, the calculated pixel size was 120.8 μm. Arc integration of the intensity distribution along the azimuthal angle was measured using program FIT2D. The integrated intensity and integral width of the intensity maxima were measured by fitting a Gaussian curve to the observed intensity after background subtraction, where the background was fit by a polynomial [45]. THEORY Measurement of Crystallinity To assess the amount of peptide in the assembled, ordered domain as a fraction of the total volume, the crystallinity was defined [27]. The observed intensity (Iobs) as a function of radial component in spherical coordinates was obtained after subtraction of the scattered intensity arising from the empty capillary tube and integrating along the azimuthal angle. The observed intensity was composed of the intensities from the Bragg peaks (IB) and diffuse bands (ID)—i.e.,

KLVFFAE

Iobs = IB + ID. The Bragg peaks come from from the lattice, while the diffuse bands are from lattice disorder and amorphous domains. The extent of lattice disorder was not considered here. The diffracting power P for powder diffraction was defined by

P=

 4 R I 2

obs

(R)dR ,

(1)

where R is the radial component of the spherical reciprocal coordinate. Parseval's equation relates this to the electron density distribution by 2

P =   (r ) d r .

(2)

From the average electron density and the total volume of the scattering object V, P = V2. P is the sum of the diffracting power from the assembled peptides that form a lattice (PB) and the diffracting power from the disordered amorphous peptides (PD), i.e., P = PB + PD.

(3)

The diffracting power is the product of the volume and electron density for the individual domains: V2 = VB2 + VD2,

(4)

where the first term, corresponding to the ordered domain, was measured as the integral area under the observed intensity after background subtraction, and the second term corresponded to the area under the background curve. The crystallinity was then defined as the fraction of total diffracting power that was due to the lattice-forming peptides, or PB/P. While different background tracing methods have been proposed [46-49], here we fit the intensity minima by a polynomial curve [45]. Coherent Domain Size and Lattice Disorder The breadth of the Bragg peak is determined by the interference function convoluted by the Fourier transform of the step-function expression for the extent of the lattice [45, 50, 51]. Assuming paracrystalline theory for the lattice statistics, the integral width of the Bragg peak for the (h,0,0) reflection, for example, is

w 2 = b 2 + (1/Nd) 2 + ( 4 h 4 /d 2 )( /d) 4 ,

(5)

where b is the integral width of the Gaussian form of the direct beam, h is the hth order reflection, d is the unit cell constant along the x-direction, and  is the lattice disorder as defined by paracrystalline theory. Assuming a small size for the direct beam and lattice disorder, then the inverse of the

292 Current Alzheimer Research, 2008, Vol. 5, No. 3

integral width of the Bragg reflection is equal to the coherent length of the lattice in the x-direction. Aggregation Model Fibril formation has been accounted for using a linear aggregation model [38, 52]. In the current study, the average number of monomers per fibril was measured by the coherent length of the 4.7 Å-reflection and was related to binding constants Ke for the peptide and Ki for the inhibitor (see Appendix). The effectiveness of inhibitor was estimated from the values of these constants. RESULTS A1-40 Showed Smaller -Crystallites than did A11-25 and A12-28. Water-solubilized A proteins (sequences 1-40, 11-25 and 12-28) gradually formed thin disks during dehydration in the glass capillaries. The disks showed the two reflections that are characteristic of -crystallites—i.e., ones at spacings of 4.7 Å and ~10 Å (Table 1 and Figs. 2,3). There were small variations in these spacings among different samples; and the greater variation for the intersheet separation was probably due to a greater sensitivity to ambient hydration of the sheetsheet interactions. Fibril size was characterized here by the coherent lengths in the hydrogen-bonding (4.7 Å) and intersheet stacking (~10 Å) directions (Table 1). The coherent lengths (±S.D.) of these reflections were 37 (±7) Å and 34 (±7) Å for A 1-40, 194 (±62) Å and 34 (±1) Å for A 11-25, and 213 Å and 38 Å for A 12-28. These values indicate that the A1-40 assemblies were significantly shorter in the hydrogen-bonding (fibril axis) direction compared with the others, whereas the extent of sheet stacking was similar among the three peptides. The spherically-averaged intensity distributions from A1-40 suggest that the shorter coherent domains along the fibril direction tend to promote a more random orientation of the fibrils. The diffraction patterns from the A11-25 and A12-28 assemblies, by contrast, showed strong orthogonal accentuation or arcing of reflections on the meridian and equator (Figs. 2,3) owing to alignment of the fibrils having more extensive coherent domains. Treating peptide assemblies with tetracycline increased fibril disorientation, as evident from the increased circularity of the reflections (Fig. 3). The volume of the  sheet crystallites was estimated by the total integrated intensities of these characteristic reflections. The relative scale of the integrated intensity was 100:110:112 for A1-40, 11-25 and 12-28, indicating a similar  crystallite volume for these peptides (Fig. 4). The measured coherent lengths for the 4.7 Å reflections and volumes for the crystallites were consistent in both indicating that A1-40 formed short fibrils while A11-25 and 12-28 formed long fibrils. Effect of Small Molecules on Fibril Formation by A140, A11-25, and A12-28 Lyophilized A peptides were added to solutions containing the putative inhibitor and the sample was gradually dried. The x-ray diffraction patterns were then recorded, and the integral widths of the 4.7 Å and ~10 Å reflections, and

Kirschner et al.

the total integrated intensities (Table 1-3 and Fig. 4) were compared. For A1-40, PTI-00703®, a compound purified from Cat’s claw (Uncaria tomentosa) [30] showed a significant inhibitory effect on fibril formation such that the diffraction patterns did not reveal any distinct Bragg peaks. This was consistent with the earlier finding that an aqueous extract from Cat’s claw inhibited fibril formation, whereas an herbal extract from Ginkgo biloba, reported to be antiamyloidogenic [53-55], did not prevent fibril formation but did alter the intersheet spacing (Fig. 4, top). By comparison, the dipeptide di-phenylalanine (FF) enabled fibril formation like that of the control, as indicated by the coherent length of the hydrogen-bonding reflection and the -crystallite volume (Fig. 4). For A11-25, a significant inhibitory effect was found for PTI-00703®, as no visible Bragg peaks were observed (Table 1 and Figs. 3,4). Among the small molecules tested previously using electron microscopy and fluorescence spectroscopy, PTI-00703®, quinine, and tannic acid reduced the volume of the -crystallites significantly, whereas curcumin actually increased the volume (Fig. 4, bottom). Using the volume of the -crystallites as a measure of inhibition, the effectiveness was in the order from most effective inhibitor to least: PTI-00703® > tannic acid > quinine > tetracycline ~ tetrachlorosalicylanilide ~ teracaine ~ hexachlorophene > sulfadiazine > melatonin ~ butylated hydroxytoluene > morin ~ A11-25 control > phenolphthaleine > chloramphenicol > curcumin > ethyl 3-aminobenzoate methanesulfonate. Ethyl 3-aminobenzoate methanesulfonate showed a crystalline diffraction pattern, indicating the largest crystallites (and hence least inhibition). Among this variety of small molecules, the coherent lengths for 4.7 Å and ~10 Å were similar, whereas the integrated intensities varied. For peptide A12-28 (Fig. 4, bottom), the effects of DMSO, curcumin dissolved in DMSO, and tetracycline were examined. Among them, DMSO alone most reduced the volume. The similar reduction by curcumin in DMSO, therefore, was likely a solvent effect. Tetracycline also reduced the volume, but not as much as DMSO. These small molecules, however, did not reduce the coherent lengths of the crystallites in either the H-bonding or intersheet directions. Dose-Dependent Effects for A12-28 and A17-28 The coherent domain size along the H-bonding direction was measured for dehydrated A12-28 (VHHQKLVF FAEDVGSNK) and A17-28 (LVFFAEDVGSNK) as a function of the concentration of added small molecules (Fig. 5). The concentration of solubilized peptide A12-28 was constant at 10 mg/ml (5.1 mM) and the concentration of compound varied from 2.56–10.23 mM for tetracycline and for cotinine (Table 3). The concentration of peptide A17-28 was constant at 5 mg/ml (3.8 mM) and the peptide solutions, which contained test compound (cotinine, pyridine, methylpyrrolidine, or nicotine at molar ratios of peptide: compound > 1:6) were dried (Fig. 6). A12-28 gave a cross- pattern with the cylindrical axis in the H-bonding direction, whereas the patterns from A17-28 showed both the 4.7 and 10 Å reflections on the equator, indicating a plate-like structure, where the cylindrical axis is on the me-

Inhibition of Amyloid Formation

Table 1.

Current Alzheimer Research, 2008, Vol. 5, No. 3

293

Effects of Small Molecules on A Fibril Formation

Peptide

Small molecule

d1

I1

D1

d2

I2

D2

I1+I2

A1-40 MW=4330

-

4.68 (.03)

327 (201)

37 (7)

9.50 (.17)

434 (285)

34 (7)

761 (457)

2.31mM

Cat’s claw 3:1

-

-

-

-

-

-

-



FF 5:1, MW=312, 6.41mM

4.65 (.01)

354 (259)

41 (14)

9.66 (.14)

409 (41)

30 (5)

763 (301)

A11-25 MW=1755

-

4.74 (.02)

157 (89)

194 (62)

10.80 (.36)

682 (357)

34 (1)

839 (446)

5.70mM

Cat’s claw 6:1

-

-

-

-

-

-

-



Tetracycline 3:1 MW=480.9, 6.93 mM

4.74

172

221

10.53

200

31

372



Tetracaine 3:1 MW=300.8, 11.08mM

4.75 (.01)

54 (23)

187 (51)

10.72 (.09)

335 (270)

46 (9)

389 (288)



Quinine 3:1 MW=324.4, 10.27 mM

4.75 (.01)

123 (7)

192 (21)

10.44 (.08)

173 (57)

37 (9)

296 (49)



Phenolphthaleine 3:1, MW=318.3, 10.47 mM

4.74 (.01)

248 (90)

171 (7)

10.30 (0.12)

692 (88)

28 (3)

940 (1)



Butylated hydroxytoluene 3:1, MW=220.4, 15.12 mM

4.75

184

180

10.30

423

32

607



Ethyl 3-aminobenzoate methanesulfonate 3:1, MW=261.3, 12.75 mM

-

-

-

10.92

-

184

-



Tetrachloro salicylanilide 3:1; MW=351.0, 9.49 mM

4.74

211

196

10.69

174

32

385



Sulfadiazine 3:1 MW=358.1, 9.31 mM

4.74

153

212

10.72

356

33

509



Chloramphenicol 3:1, MW=323.1, 10.31 mM

4.73

125

150

10.29

931

32

1056



Hexachlorophene 3:1, MW=406.9, 8.19 mM

4.76

119

219

10.77

295

44

414



Tannic acid 3:1, MW=1701.2, 1.96 mM

4.75

62

205

9.91

140

24

202



Melatonin 3:1, MW=232.3, 14.33 mM

4.72

161

219

10.83

429

38

590



Curcumin 3:1, MW=368.4, 9.04 mM

4.72

163

217

10.23

1542

35

1705



Morin 3:1, MW=302.2, 11.02 mM

4.76

217

185

10.56

618

49

835

A12-28 MW=1955

-

4.73

101

213

10.48

745

38

846

5.11mM

Tetracycline 3:1 MW=480.9, 6.93 mM

4.74

115

180

10.59

345

31

460



DMSO 3:1 MW=78.1, 42.67 mM

4.76

29

224

11.62

66

34

95



Curcumin 3:1, MW=368.4, 9.04 mM/DMSO

4.77

41

189

11.72

56

48

97

The effects were measured in terms of the integrated intensities (I1 and I2) and coherent lengths (D1 and D2 in Å) of H-bonding and intersheet spacings (d1 and d2 in Å). The coherent domain size is expressed here as the reciprocal of the integral width. The average values and standard deviation (in parenthesis) are shown. Weight ratio of peptide/small molecule, either 3:1, 5:1, 6:1 corresponding molar concentration, and molecular weights are indicated. For the highly ordered diffraction pattern from A11-25 in ethyl 3-aminobenzoate methanesulfonate, only the 10 Å-data is listed.

294 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (2). Selected X-ray diffraction patterns for A1-40, A11-25, and A12-28 peptides treated with small molecules, mostly at weight ratio 3:1 of peptide/compound. The weight ratio of A12-28 and PTI-00703® was 9:1. For all patterns shown, the background was subtracted. The exposure time (150 sec) and specimen-to-film distance (85 mm) were the same for all patterns.

ridian along the -chain direction [6]. For A12-28 the coherent length of the 4.7 Å Bragg peak along the H-bonding direction decreased with an increase in the concentration of all small molecules (Fig. 5), whereas for A 17-28 it varied little (Fig. 6). Initial Assembly Process of A 11-25 At higher hydration the characteristic  sheet reflection at 4.7 Å was not observed for A11-25 either with or without tetracycline (Fig. 7). The sample contained 5.70 mM A1125 and 6.93 mM tetracycline. This corresponded to peptide and tetracycline at a 3:1 weight ratio. A reflection at 11.0 Å with coherent length 40 Å was observed from hydrated A11-25, and a similar spacing having coherent length 44 Å was observed when tetracycline was added. At lower hydration (higher dehydration) the 4.7 Å, H-bonding reflections were observed for both samples (Fig. 2). These findings sug-

gest that the initial event in amyloid fibril assembly may involve sidechain interactions rather than hydrogen-bonding between -strands. N-Methylated A16-22 No Bragg reflections were recorded for the methylated peptide A16-22Me (“B”) when either hydrated or solubilized/dried; however, all of the other hydrated and solubilized/dried samples in this series (A16-22 (“A”); 1:1 A1622:A16-22Me (“C”); and 1:5 A16-22:A16-22Me (“D”)) gave sharp reflections at 4.7 and ~10 Å spacings on the equator, which are characteristic of plate-like -crystallites with a cylindrical axis along the -chain direction. There were no substantial differences among different types of samples with respect to spacings and coherent lengths. Based on the magnitude of the intensities, the relative volumes of the -crystallites were C > A > D >> B for the hydrated

Inhibition of Amyloid Formation

Current Alzheimer Research, 2008, Vol. 5, No. 3

295

Fig. (3). X-ray diffraction patterns from A11-25 before and after incubation with tetracycline or PTI-00703®. (A) Control peptide without compound. (B) Peptide with tetracycline. The initial solution contained 5.70 mM peptide and 6.93 mM tetracycline, and then the sample was gradually dehydrated at room temperature. (C) Peptide with PTI-00703®. The sharp reflection at 4.7 Å Bragg spacing (in A and B) arises from the -chain array in the H-bonding direction. With PTI-00703® treatment (C) the Bragg reflections from the -crystallites were nearly undetected. The sharp reflection at 2.8 Å from the PTI-00703®-treated A peptide was identified as coming from salt. The specimen-to-film distance and the exposure time were same for all specimens (see Fig. 2). (D, E) FIT2D scans of the diffraction patterns. (D) X-ray diffraction intensity as a function of reciprocal coordinate (1/Å) for (top) A11-25, (middle) A11-25 in the presence of tetracycline, and (bottom) PTI00703®. The intensities from the circularly-symmetric patterns were integrated and averaged over an azimuthal angle of 360° using program FIT2D. The intensity maxima at (10 Å)-1 and (4.7 Å)-1 correspond to intersheet and H-bonding distances for the -sheet structure. PTI-00703® significantly inhibited the -sheet formation. The intensity maximum for A11-25 in the presence of PTI-00703® was at (7.5 Å)-1. Tetracycline disordered the intersheet packing, as indicated by a larger breadth of the 10 Å-reflection. (E) X-ray intensity of the 4.7 Å-reflection plotted as a function of azimuthal angle for (top) A11-25 and (bottom) A11-25 in the presence of tetracycline. Note that the angular half-width of the latter (82°) was significantly larger than that of the former (65°), indicating that tetracycline increased the fibril disorientation.

samples, and A > C > D >> B for the dried samples. The integral widths of the reflections were used to calculate the coherent lengths in the H-bonding and intersheet directions (Fig. 8 and Table 3). DISCUSSION Numerous strategies are focused on discovering compounds that can prevent amyloid formation. Such targeted amyloids include amyloid- of Alzheimer’s [36, 56-61], prion [29, 62], -synuclein of Parkinson’s [63, 64], transthyretin [65], insulin [66], and Ig light chains [67]. Current screening techniques include a variety of physicalchemical and toxicity assays—e.g., turbidity and Congo red or thioflavin binding, morphological imaging (electron microscopy; atomic force microscopy), spectroscopy (circular dichroism; NMR), and cytotoxicity and biochemistry (cell

cultures; protease-resistance; immunoblotting). Irrespective of amyloid morphology, x-ray fiber diffraction shows that the two key characteristics of these structures are the 4.7 Ådistance between H-bonded -strands and the side chaindependent 5-15 Å-distance between pleated sheets [6]. Moreover, the extent of H-bonding and intersheet stacking can be monitored for an amyloidogenic peptide by comparing the widths of the x-ray reflections in patterns recorded from lyophilized material, vapor-hydrated peptide, and solubilized/dried samples [8]. This suggests a method for assessing whether a compound inhibits fibril formation and what structural feature a compound targets when it does inhibit. Here, we used x-ray fiber diffraction to test the inhibitory effect of a number of small aromatic molecules and Nmethylated A16-22 on the aggregation of different A analogues (sequences 1-40, 11-25, 12-28, 17-28, and 16-22).

296 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (4). Effects of small molecules on -sheet formation by A1-40, A11-25, and A12-28. (Top) Scans of x-ray patterns from lyophilized A1-40 (L) or dissolved in water and dried (S/D), or mixed at peptide:compound weight ratios of 3:1 and 6:1 with water extracts of Cat’s claw (CC) or Gingko biloba (GB) and then dried under ambient conditions. The position of the 4.7 Å-reflection from H-bonded chains in pleated sheets (arrows) and of the intersheet spacing at ~10 Å (arrowhead and horizontal bars) are shown. Note that the intersheet spacing in A:GB at 3:1 shifted (to ~7.7 Å-spacing). The lyophilized peptide contained a very small but detectable amount of residual secondary structure, as evidenced by the weak intensity maxima. When it was hydrated and dried, considerable hydrogen-bonding occurred, presumably as fibrils form. The pattern from peptide dissolved in CC at the 3:1 weight ratio resembled that from the lyophilized peptide alone, indicating inhibition of assembly; however, at lower CC concentration (6:1), the pattern from the mixture resembled that from the peptide alone, after solubilizing and drying. The pattern from peptide dissolved in GB at the 3:1 ratio showed both increased hydrogen-bonding and a shift in the intersheet reflection, indicating an effect on the -sheet structure but not inhibition of its formation. At lower GB concenration, the pattern from the mixture again resembled that from the peptide alone, after solubilizing and drying. (Bottom) Coherent domain sizes (lower x-axis) along intersheet (solid bars) and H-bonding (gray bars) directions. and the total intensity (upper x-axis; speckled red bars) for the peptides either alone or treated with compound. The coherent domain sizes were measured from the integral widths of the 4.7 Å and ~10 Å reflections from the  crystallite. The total intensity was measured from the sum of the integral intensities of the H-bonding and intersheet reflections. The total intensity for the A1-40 control was set to 100. For the highly-ordered diffraction pattern from peptide treated with Ethyl 3-, only the coherent length for the 10 Å-reflection is shown. PTI-00703® abolished the reflections from A1-40 and A11-25.

Inhibition of Amyloid Formation

Table 2.

Current Alzheimer Research, 2008, Vol. 5, No. 3

297

Binding Constants Between A12-28 and Tetracycline or Cotinine

Small molecule

Concentration (mM)

Ke (mM-1)

Ki (mM-1)

Dobs (Å)

Dcalc (Å)

A12-28 control

-

376

-

213

213

+ Tetracycline

2.56

530

330

227

230.7

5.12

500

350

185

189.9

10.23

500

320

124

122.5

510

333

2.56

170

40

139

138.8

5.12

130

140

71

70.5

10.23

110

140

24

23.4

140

110

Average + Cotinine

Average

Binding constants for addition of monomer (Ke) and inhibitor (Ki) for A12-28 at 5.11 mM concentration, as derived from the dose-dependent effects of the small molecule. Binding constant values were determined by varying the concentration of monomer to give a calculated coherent length (Dcalc) that was most similar to the observed one (Dobs) of the 4.7 Å reflection.

Summary of X-Ray Diffraction Data for Effect of Methylated A on A Assemblya

Table 3.

Intersheet Reflection Sample

b

H-Bonding Reflection

Spacing (Å)

Length (Å)

Intensity

Spacing (Å)

Length (Å)

Intensity

H

9.85

138

125

4.76

107

69

S/D

9.78

129

685

4.74

110

425



9.80

112

961

4.73

214

253

H

-

-

-

-

-

-

S/D

-

-

-

-

-

-



-

-

-

-

-

-

H

9.85

150

223

4.77

91

123

S/D

9.81

117

501

4.75

99

278



9.85

120

494

4.74

230

129

H

9.86

142

81

4.76

99

44

S/D

9.87

133

339

4.75

130

118



9.88

145

219

4.75

200

71

A16-22

A16-22Me

A:AMe 1:1

A:AMe 1:5

a

Measurements are from diffraction patterns recorded at Tufts University School of Medicine, Department of Biochemistry, using an Oxford Diffraction Xcalibur PX Ultra System in the laboratory of Dr. Andrew Bohm. Exposures were typically 150 sec. Background subtraction and data analysis was carried out as described in text. Samples: H, hydrated; S/D, solubilized then dried while in a 2 Tesla permanent magnet. One set of samples was diffracted from while hydrated, and then they were solubilized and dried; a second set of samples was examined only after drying. b

298 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (5). Dose-dependent X-ray diffraction patterns of solublized/dried (S/D) A 12-28 with or without PTI-00703®, tetracycline and cotinine were observed. (A) Dependence of coherent length (Å) of the 4.7 Å reflection on binding constant for monomer addition (mM-1) for 2.31 mM A 1-40 (circle), 5.70 mM 11-25 (square) and 5.11mM 12-28 (triangle). The best binding constants were 18 mM-1, 278 mM–1, and 376 mM–1, respectively. (B) Dependence of coherent length (Å) of the 4.7 Å reflection for A12-28 on the concentration of compound (small molecule; mM) i.e., tetracycline (square) and cotinine (triangle). The calculated lines were based on the linear aggregation model with inhibitor. The input values included 5.1 mM initial monomer peptide concentration, Ke=510 mM-1, Ki=333 mM-1 for tetracycline (dissociation constants 1.96 μM, 3.00 μM), and Ke=140 mM-1, Ki=110 mM-1 for cotinine (7.14 μM, 9.09 μM dissociation constants). (C) Concentration (mM) of monomer in fibrils Af (solid line) and inhibitor-bound monomer (dashed line) as a function of the concentration of tetracycline (mM). (D) Average number of monomers per fibril Af/Cf (solid line). The initial A12-28 monomer concentration was 5.1 mM, and binding constants Ke and Ki were 510 mM-1 and 333 mM-1, respectively.

Fig. (6). Selected X-ray diffraction patterns of A1728 with small molecules at peptide/compound weight content 3:1 for (A) control, (B) cotinine, (C) nicotine, (D) pyridine, and (E) methylpyrrolidine. (F) Dependence of coherent length (Å) for the 4.7 Å reflection on molar ratio of compound/A17-28 (5 mg/ml), i.e., peptide alone ( at zero molar ratio), cotinine (square), nicotine (diamond). pyridine (filled circle), and methylpyrrolidine (triangle) .

Inhibition of Amyloid Formation

The isoelectric pH’s (or pIs) calculated for these peptides were 5.4, 5.4, 8.4, 4.2, and 7.4, respectively [26]. The compounds tested included tetracycline, tetracaine, quinine, phenolphthalaine, PTI-00703® (Cat's claw), diphenylalanine (FF), hexachlorophene, chloramphenicol, butylated hydroxytoluene, sulfadiazine, tetrachlorosalicylanilide, ethyl 3aminobenzoate methane sulfonate, tannic acid, curcumin, melatonin, morin, cotinine, nicotine, pyridine, and methylpyrrolidine. For A11-25, PTI-00703® abolished the Hbonding reflection, and tannic acid and quinine decreased the -crystallite volume more than did curcumin, morin, and melatonin. The decrease in the coherent length of the 4.7 Åreflection of A12-28 was observed as a function of concentration of tetracycline and cotinine. The difference in the Hbonding coherent length, and the dose-dependency is discussed below in terms of the binding constants for monomer elongation, and inhibitor binding. Binding Constants of Monomer Addition for A1-40, A11-25, and A12-28 The molar concentrations for 10 mg/ml of A1-40, A11-25, and A12-28 were 2.31 mM, 5.70 mM, and 5.11 mM, respectively. The coherent lengths D1 along the 4.7 Å direction—i.e., fibril direction for the cross- conformation—were 37 Å, 194 Å, and 213 Å (Table 1 and Fig. 4), and these values can be related to the average number of monomers per fibril as defined by < n f >= D1 / d1 . Here D1 is the coherent length and d1 is the spacing for the 4.7 Å reflection. The average number of monomers within fibrils was derived by dividing the total concentration of monomers within fibrils by the concentration of the fibrils, according to < n f >= A f / C f (see Appendix). The right-hand-side terms are related to the binding constant of monomer addition Ke and to the total monomer concentration. The best Ke was searched by systematically changing the values to give a

Current Alzheimer Research, 2008, Vol. 5, No. 3

299

calculated coherent length similar to the observed one. The binding constants derived in this way were 18 mM-1, 278 mM-1, and 376 mM-1 for A1-40, A11-25, and A12-28, respectively (Fig. 5a). The corresponding dissociation constants were 55.6 μM, 3.6 μM, and 2.7 μM. The broader 4.7 Å reflection for A1-40, therefore, is accounted for by weaker binding between the monomers. A charge effect is unlikely to account for this weaker binding, as the calculated pI’s for A 1-40 and A11-25 are both 5.4, and that of A12-28 is 8.4. It may be that the side chains in A1-40 sterically hinder H-bonding formation between monomers, whereas the side chains in the short peptides (A 11-25 and A12-28) do not. Aggregation Model Accounts for Dose-Dependent Inhibitory Effect on A12-28 Based on the aggregation model, the binding constants for monomer addition and for the monomer-inhibitor interaction were calculated from the observed coherent length as a function of the concentration of the small molecule. The best values were taken from the average of the calculated constants that most closely fit the observed individual data at the specific concentration of the small molecule. The initial monomer concentration was 10 mg/ml (5.1 mM) for A1228. The binding constants for both Ke and Ki were systematiccally varied in order to fit the data points at different concentration of small molecules (Table 3; Fig. 5), and were determined to be Ke=510 mM-1, Ki=333 mM-1 for tetracycline (corresponding dissociation constants 2.0 μM and 3.0 μM), and Ke=140 mM-1, Ki=110 mM-1 for cotinine (7.1 μM and 9.1μM dissociation constants). The decrease in the coherent length of the reflection at 4.7 Å-spacing as a function of compound concentration was accounted for by this model (see solid line in Fig. 5B). With an increase in the inhibitor concentration, the concentration of the inhibitor-bound

Fig. (7). X-ray diffraction patterns after background subtraction for hydrated A11-25 with or without tetracycline at 3:1 weight ratio, which corresponds to 5.70 mM A11-25 and 6.93 mM tetracycline. The specimen-to-film distance was 85 mm, and the exposure time was 150 seconds for all samples. Brightness and contrast of the image were adjusted to enhance the ~10 Å reflections. Both samples showed a broad intersheet reflection at ~11.0 Å-spacing, however no hydrogen-bonding reflection was detected.

300 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (8). X-ray diffraction patterns after background subtraction for hydrated A16-22 (“A”) with or without N-methylated A16-22 (“Me”) at molar ratios of 1:1 and 1:5. As both the 4.7 Å and 10 Å reflections were on the equator, the assembly was plate-like where the -chain direction (the cylindrical axis) was along the meridian. In each row, the first two panels are from one sample (hydrated and then dried), and the third panel was from a second sample after drying.

monomer increased, while the concentration of the monomer within fibrils decreased (Fig. 5C). The measured dissociation constants for monomer addition for A12-28 was 2.7 μM, similar to the value 7 μM for A9-25 measured previously by fluorescence energy transfer [68]. These constants are similar to the range of 2.9-5.9 μM for the binding of Congo red to A fibrils [38]. Tetracycline at physiological pH is zwitterionic: on the A ring, there is a positive charge on the amino group (pKa 8.8-9.8) and a negative charge on the oxygen bonded to C3 (pKa 3.4), and on the B ring there would

be some deprotonation of the hydroxyl group bonded to C12 (pKa 7-8) [69-71]. Cotinine, by contrast, is uncharged, as neither its N-methylpyrrolidine nitrogen as part of an amide bond nor its pyridine nitrogen (pKa 3.12) are protonated at physiological pH [37]. As A12-28 is positively charged (pI 8.4), it is conceivable that the negatively charged oxygen on the C3 (or C12) in tetracycline interacts with the peptide. This correlates with the slightly stronger binding constant Ki for tetracycline than for cotinine.

Inhibition of Amyloid Formation

Current Alzheimer Research, 2008, Vol. 5, No. 3

Comparison with previous estimates of Ke and Ki In the one report that gives the dissociation constant of inhibitor from the A protein, the authors show the effect of A1-40 concentration on the initial rate of fibril extension with or without apoE3 (Fig. 5 in [72]). Their plot was accounted for using a linear aggregation model. According to our nomenclature (see Appendix) the free monomer concentration as a function of time is expressed as

1 (t) = (k /k + ) + exp(k + mt)(Atot  k /k + ) ,

(7)

where k+ is the forward kinetic constant of monomer addition, k- is that of monomer dissociation, k-/k+ is the monomer concentration at infinite time which is equivalent to the dissociation constant at equilibrium, and m is the constant polymer concentration indicating that the monomers are added on the template and that no spontaneous template is produced [38]. The monomer concentration within the aggregates is defined as a function of time by A f (t) = Atot  1 (t) . The first derivative is written as

dA f (t) /dt =  [1 exp(k+ mt)], where  = Atot  k /k + . The slope at t=0 is dA f (t) t= 0 /dt = k + mAtot  k m . This shows that a plot of the slope at t=0 gives a straight line as a function of initial concentration of monomer Atot. The slope of this straight line is k+m and the x-coordinate intercept is Atot=k-/k+. The observed x-coordinate in reference [72] was ~2.5 μM for A1-40. Read as a dissociation constant for monomer addition, this value is significantly smaller than our observed value of 56 μM for A1-40, but is similar to the values of 3.6 and 3.8 μM for peptides A11-25 and A12-28. A similar plot for A1-40 by the same authors [73], however, shows a different straight line that crosses the origin, indicating inconsistent experimental results that may be due to a varied response of thioflavin T fluorescence to the monomer concentration within the fibril. The dissociation constant of 0.37–0.39 μM for apoE [72] was nearly onetenth the values for our measurements for the small molecules examined here. Numerical Calculation and Structural Model The current study assumed a linear assembly of monomers along the fibril direction with period d1, which is the simplest model for the parallel -sheet assembly. Some modification may be needed to relate the parameters used in aggregation theory to a more sophisticated atomic model in the future. The initial monomer concentration Atot and the total inhibitor concentration itot were based on experiment. The peptide monomer concentration involved in the growth of the fiber direction, however, is not always the same as Atot, but may be written as

A˜ tot = Atot  /(D2 /d2 ) ,

(8)

where  is the number of chains in a single peptide molecule aligned in the intersheet direction, d2 is the period (~ 10 Å) in the intersheet direction, D2 is the coherent length of the Bragg reflection of d2 spacing,  is in the range of 0 and 1 and may be related to a ratio between

301

(I1 + I2 ) inhibitor /(I1 + I2 ) control , where I1 and I2 are the integrated intensities for the 4.7 Å and 10 Å reflections. Here, D2/d2 gives the number of chains aligned in the intersheet direction. If A1-40 contains two chains in this direction [74], then  may be 2, and for A11-25 and A12-28,  may be 1 if the peptides do not contain a reverse turn. The average number of monomers within the fibril is related to the coherent length of D1 by

D1 = 4.7 < n f >  ,

(9)

where  is the number of chains in a single peptide molecule aligned in the 4.7 Å direction. For a parallel -chain this number is 1, while for a peptide molecule having antiparallel -chains this number may be 2. In the current study for A140, A11-25, and A12-28 this number was 1. In the linear assembly model the total monomer concentration involved in fibril growth is written as A˜ tot = K i 1i q + 1 /(1 K e 1 ) 2 . When the binding constants Ke and Ki are known, the concentrations of free monomer and inhibitor-bound monomer 1, i can be determined. The values of Cf ,Af and which are the concentration of fibrils, monomer concentration within fibrils, and the average number of monomers within a fibril, are then calculated. The coherent length along the fibril direction was calculated from . The best values for Ke and Ki were based on a numerical search to give a calculated D1 similar to the observed one. Another widely known model of the -helical nanotube [75] gives a linear assembly of monomers that has a 4.7 Å-period for A1-40 in the fiber direction, and no adjustable parameter may be needed in this case. Therapeutic and Diagnostic Implications—Site of Inhibitory Action Inhibitors against amyloid fibril formation have been studied in the hope of finding therapeutic and diagnostic compounds. Therapeutic drugs are required to reduce the amyloid burden, while diagnostic probes interact with fibrils or preamyloid oligomers, and should be cleared from the site. The binding sites may be on the monomer, nucleus, protofilament, or fibril [38, 76], as well as on sites involved in the blood-brain barrier, clearance, and metabolic pathways [77-80]. These prospective drugs should not increase the concentration of the toxic components. Inhibitors that disrupt amyloid fibrils may become hazardous if soluble toxic aggregates were to be released. The drugs for therapeutic use, therefore, should inhibit the transition from monomer to oligomer [81], to increase the clearance of the monomer so that the equilibrium is shifted to higher monomeric concentration in brain [77, 82]. Effective drugs, therefore, should act like an antagonist of the monomer-monomer interaction. The energy involved in this binding is not large, given the μM levels for dissociation constants as shown in the current study. Sites of action include the amyloidogenic core sequence of A (residues 1721: Leu-Val-Phe-Phe-Ala) which is likely involved in the highly ordered H-bonding -sheet formation [6], and another site is at residues 12-16 (Val-His-His-Gln-Lys) which is ex-

302 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

Fig. (9). (A) Dependence of number of monomers in fibrils on elongation binding constant Ke. From the top, initial monomer concentration Atot was 100, 50, and 10, respectively (solid line). From the top, the same Atot with total inhibitor concentration ito t=50 and inhibitor binding constant Ki=50 (dashed line). (B) Dependence of number of monomers in fibrils on initial monomer concentration. From the top, binding constant Ke was 100, 50 and 10, respectively (solid line). From the top the same constant Ke, and inhibitor concentration 50 and Ki=50 (dashed line). (C) The monomer concentration in fibrils as a function of time (arbitrary units) according to the nucleation dependent model. The number of monomers in the nucleus was 4. From the top (solid lines), the initial monomer concentration and k+*k+ were (10, 4x10-4); (10, 4x10-5); (5, 4x10-4). With total inhibitor concentration of 10, dissociation constant of inhibitor binding Kdi=10, inhibitor stoichiometry q=1 and the same condition of the top curve without inhibitor (solid circle). (D) The monomer concentration in fibrils as a function of time (arbitrary units) according to linear aggregation model. Polymer concentration was 1 μM for all curves. From the top (solid lines), initial monomer concentration (μM), k+ ,k- are (100, 1, 0.1), (100, 0.5, 0.1), (50, 1,0.1). With 50 μM inhibitor and dissociation constant Kdi=10, Atot=100, k+=1, k-=0.1, q=1, the monomer concentration within fibrils (solid circle) is shown.

Fig. (10). Schematic proposing how pre-formed assemblies of A16-22 would be affected by subsequent incubation with methylated peptide.

Inhibition of Amyloid Formation

posed on the surface of the fibril and is likely involved in the electrostatic-driven, lateral aggregation of the fibril [6, 83]. The latter site is also thought to be a binding site for heparan sulfate proteoglycan [24, 84] and Congo red [25]. A recent clinical study confirms this site as a promising drug target [23]. Comparison here of the effect of extracts from Cat’s claw versus Ginkgo biloba showed that whereas the former prevented fibril formation, the latter did not but rather altered the intersheet spacing. These effects were readily distinguished using fiber x-ray diffraction and indicate that compounds can target different features of the amyloidogenic peptide or peptide assembly. Nonetheless, the chemical structure for an amyloid inhibitor that a priori has a definitive, specific effect on amyloidogenesis has not yet been designed. One reason may be that the effect of inhibition itself has not been quantified. Even though the aggregation model is widely used for interpreting amyloid formation, the binding constants and kinetic parameters have not been established from inhibition experiments. Our current study gave, for the first time, values of Ke and Ki for different peptides and small molecules. Among the compounds studied was the anionic molecule Congo red, which may interact strongly with histidine residues localized within A sequence 12-16 (VHHQK) [26]. This may also be the site for cotinine, pyridine, methylpyrrolidine, and nicotine, because A17-28 fibril formation was not inhibited by them whereas A12-28 was. It is likely that the A sequence 12-16 interacts with the small molecules via aromatic - and/or electrostatic interaction involving histidine and/or ionic interactions involving lysine. This notion is consistent with the binding, suggested from NMR measurements, between the pyrrolidine moieties of nicotine and histidine residues [37]. Initial Structural Event for Amyloid Formation The initial conformational change of the A monomer that involves residues in the sequence 11-25 may be crucial for amyloid formation. Designing molecules that target this region may help to arrest the monomeric species from selfassembly into toxic species. A number of different mechanisms for the initial amyloid-forming event have been proposed, including intersheet hydrophobic contact [85], reverse-turn formation [11, 86-88], H-bonding, -sheet, polar sheet formation [89], and folding into a polyproline II helix [90]. Our current x-ray results from peptide A11-25 (EVHHQKLVFFAEDVG) when it is still hydrated prior to fibril formation showed a reflection at ~10 Å, suggesting that the side chain interactions precede formation of the -sheets, which are constituted of H-bonded -strands. Aggregation of -amyloid, therefore, arises first from the interaction between side-chains followed by the H-bonding interaction between amides. Whether the intersheet and H-bonding interactions are intramolecular or intermolecular is unknown. If the intramolecular packing generates an 11 Å-reflection, then reverse-turn formation within the sequence 11-25 may be a first step leading to -sheet formation. No apparent difference in the observed spacings for A11-25 with tetracycline versus A11-25 without suggests that tetracycline does not interfere with the initial peptide folding. It may be that interaction of aromatic small molecules with the histidine residues within sequence 12-16 inhibits subsequent H-

Current Alzheimer Research, 2008, Vol. 5, No. 3

303

bonding stacking of A monomers or inter-fibril aggregation. N-Methylated A16-22 Because the dimensions of the -crystallites in the Hbonding and intersheet directions were similar among samples having different molar ratios of normal and Nmethylated peptides, then the observed difference in the volumes of the -crystallites must derive from the crystallite dimension in the -chain direction. Therefore, at lower concentration (i.e., equimolar with A16-22) the A16-22Me peptides could facilitate the packing of the plates in the chain direction, whereas at higher concentration (fivefoldmolar less A) the A16-22Me might inhibit the packing of plates. After solubilization and drying, the higher molar ratio of A16-22Me inhibited plate packing to an even greater extent. That no characteristic -sheet reflections were observed here for the methylated peptide supports previous findings that A16-22Me does not form -sheets [21, 22], owing to its lack of the amide hydrogens required for -sheet formation. As previous studies using different techniques found that the methylated peptide inhibited the initial aggregation of A, we were surprised here that A16-22Me did not inhibit the H-bonding of A16-22 (or the assembly of the plate-like structures). This finding suggests that the A16-22 peptides were already in a stable -sheet structure before mixing with A16-22Me. Thus, it would be unlikely for the “mature” A amyloid fibril to be disrupted by A16-22Me (Fig. 10). The reduction of the total diffracted intensities with increase in A16-22Me concentration, however, clearly indicated its binding to the monomer (or preamyloid form) of normal A16-22. CONCLUSION Our current x-ray diffraction study shows that the volume of the -crystallites formed by peptide A11-25 was reduced by PTI-00703®, tannic acid, and quinine, and that of A1622 by an N-methylated peptide. As these small molecules inhibit the transition from monomers to oligomers, they may be of therapeutic use by preventing formation of toxic and higher order species. Whether this holds true for the longer, more physiological peptides A1-40 and A1-42 will be addressed in future studies. Finally, this study demonstrates the usefulness of x-ray fiber diffraction in assessing the effectiveness of potential inhibitors in an in vitro screen. Developing and implementing a high-throughput screen with this technique should be straightforward using robotic sample preparation, sample movement through a high-intensity x-ray beam, automated data collection with fast electronic detectors, and automated analysis of the x-ray data. Moreover, application to a wide variety of amyloidogenic peptides that figure in protein-folding disorders, as enumerated in the introduction, can be anticipated. APPENDIX Thermodynamic and kinetic models for linear assembly and nucleus-dependent assembly have been published [38, 76]. Here, we derive some new equations to include the interaction between a small molecule and a peptide.

304 Current Alzheimer Research, 2008, Vol. 5, No. 3

Kirschner et al.

K i = i / 1i q ,

1. Mathematical expressions essential to the following N

GN (x) =  x j = j=1

N +1

xx . 1 x

where i is the concentration of the inhibitor-bound monomer, and q is a positive integer. The total concentration Ctot of the assembly including monomers and inhibitor-bound monomers is

When |x|= A f / C f .

The initial concentration of monomers is

Given

(A13)

The monomer concentration within the fibrils is

where x =Ke1.. When x = ( Atot  1 ) /(C tot  1 ) = A f / C f .

(A12)

j= 2

(A2)

j=1

Atot = 1 /(1 K e 1 ) 2 .

(A11)

Because the Ctot and Atot include the monomer concentration, then the total concentration of the fibrils Cf is

N

Atot =  j j = H N (x) /K e ;

(A10)

and

The total concentration of the polymers is, therefore,

Ctot = 1 /(1 K e 1 ) .

(A9)

The initial concentration of monomers Atot is

(A1)

Ctot =   j = GN (x) /K e ,

(A8)

j=1

When |x|