Cell Surface Membrane Remodelling and Mitochondrial Remodelling ...

4 downloads 0 Views 1MB Size Report
reorganized in response to stimuli that cause cell swelling and shrinking. ..... Note: In bath labelling experiments, the capacity of the confocal microscope to.
Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

Cell Surface Membrane Remodelling and Mitochondrial Remodelling in Live Neurons: Practical Approaches for Studying Dynamic Processes Using Confocal Microscopy N. Shulyakova, P. Kaifosh, D. Diec and L.R. Mills*,1,2 1 Genetics and Development Division, Toronto Western Research Institute, University Health Network 11-430 Toronto Western Hospital 399 Bathurst St Toronto ON M5T 2S8 2 Department of Physiology, University of Toronto, Toronto, ON, Canada

Neurite outgrowth and growth cone motility are dynamic processes that involve a consequent remodelling of the cell surface membrane. Plasma membrane remodelling in neurons is also essential for successful recovery from insults that trigger cell swelling e.g., ischemia and excitotoxicity. Confocal microscopy, by facilitating high resolution studies on live neurons in dissociated cultures and in slice cultures, has had a major impact on our understanding of these events. During previous studies on plasma membrane remodelling induced by osmotic perturbation, we also observed striking and reversible changes in mitochondrial morphology. This article describes experimental protocols we have developed, and practical solutions for problems encountered when examining cell surface membrane remodelling and mitochondrial remodelling in live neurons. Keywords neurons; PC12 cells; slice cultures: plasma membrane remodelling; mitochondria; VLDs confocal microscopy; biolistic transfection; mtGFP; lck-GFP; lipophilic dyes; Lucifer Yellow

1. Introduction Remodelling of the cell surface membrane occurs whenever neurons change their shape. As neurons grow during development such activity is virtually continuous. In mature neurons remodelling occurs during regeneration, and in response to insults that trigger neuronal swelling. Our previous studies experimentally induced plasma membrane remodelling by using transient osmotic shocks to trigger cell surface area regulation (SAR) [1-4]. (SAR) is the homeostatic process by which the plasma membrane is reorganized in response to stimuli that cause cell swelling and shrinking. Swelling (induced by hyposmotic medium), recruits internal membrane stores to the plasma membrane. During shrinking (induced by an abrupt return to normosmotic medium), the now excess plasma membrane is retrieved, generating transient vacuole–like dilations (VLDs) at the adherent surface. VLD recovery, the process by which VLDs are re-internalized, involves cortical F-actin [2], and is partially dependant on ATP [2,3,4]. Neurons in hippocampal slice cultures demonstrate SAR, suggesting it is an intrinsic property of neurons, and may be a key element in the brain’s response to ischemic insults [3]. During the course of our studies on SAR, we observed concomitant changes in mitochondrial morphology. Unlike the mitochondrial swelling observed in response to stresses associated with apoptosis [5], or necrosis (e.g., Ca2+ overload [6], glutamate toxicity [7], Cd2+ toxicity [8], loss of membrane potential [9], complex-I inhibition [10]), the changes we observed were rapid, global, and reversible within minutes. Given the complex morphology of neurons, this ‘mitochondrial remodelling’ may be essential for mitochondrial trafficking in neurites, particularly under conditions of increased bioenergetic demands (e.g. during regeneration, or at synaptic sites). This chapter describes some practical approaches for imaging cell membrane remodelling and mitochondrial remodelling in live neurons.

*

Corresponding author: e-mail: [email protected], Phone: 416-603-5420

49

Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

2. Plasma membrane remodelling in live neurons 2.1

Experimental models

Confocal microscopy is the method of choice for monitoring dynamic events in live cells [11,12] To examine remodelling of the plasma membrane we use osmotic perturbations to evoke neuronal swelling and shrinking. Swelling is induced by replacing culture medium with a hyposmotic solution (medium to water ratio 1:5 to 1:9) for 2-4 minutes; replacing this solution with normosmotic medium (290 mOsm) causes rapid shrinking. To monitor these events we use fluorescent dyes to label the plasma membrane, or the bath. Our previous experience with live cell imaging encompasses studies on molluscan neurons [1,2], differentiated PC12 cells [4,13], sympathetic neurons [14], acute hippocampal slices [15, acute spinal cord slices [16], and hippocampal slice cultures [3]. 2.2 Optimizing imaging conditions To reduce complications associated with cell overlap, PC12 cells and primary neurons are plated at low density (5×104 cells/ 35 mm dish). Undifferentiated PC12 cells are treated with NGF for 4-6 days to induce differentiation or can be are differentiated in tissue culture flasks for up to 10 days prior to plating. Hippocampal slice cultures are cultured on porous membrane inserts (0.4µm, Millicell-CM), for at least 7 days, and preferably longer, during which time they gradually ‘thin out’, optimizing transfection and imaging conditions [17]. High resolution imaging requires high numerical aperture objectives, in our cases a 40X fluor or a 60X planApo. The short working distances of these objectives effectively limit imaging to the central region of a 35 mm dish. To maximize cell attachment in this area, we plate neurons in a 75-100 µL drop of medium in the centre of the dish, and allow them to attach for several hours before topping up the medium. Our confocal is on an inverted microscope platform so in order to achieve high resolution images we must grow cells on glass. Glass-bottomed dishes can be made by drilling a 10-20 mm diameter hole in the centre of a 35 mm plastic dish and gluing a #1 coverslip (0.17 mm), to the bottom surface of the dish, effectively creating a small well. The choice of glue, coverslip glass, and cleaning agents, can be problematic, especially for hippocampal neurons, and we frequently use a commercial source (WillcoWells). Alternatively, neurons can be cultured on #1 coverslips in plastic dishes, and the coverslips transferred to specialized chambers for imaging; this approach is not ideal for experiments that require cultures be transferred back to the incubator. Since PC12 cells and primary neurons do not attach/grow as well on glass compared to plastic, the glass is coated with a double layer of 25% collagen or 1.5 µg/ml poly-L-ornithine (PLO). PC12 cell growth is optimal on collagen, but for frequent media exchanges PLO offers greater adhesivity. Laminin, while a good substrate for primary neurons, is not optimal for PC12 cell differentiation. Note: Confocal imaging of live cells grown on plastic can be done on an inverted microscope, using 10X and 20X fluor objectives (e.g. [14]), but there is considerable loss of signal. For fixed cells however, high resolution imaging is possible on an inverted microscope using the following modifications. A #1 coverslip is mounted over the culture (using clear nail polish), and the dish is inverted face-down over the objective, attaching it to the underside of the microscope stage by double-sided tape. Paraformaldehyde (2%) is the fixative of choice as glutaraldehyde is autofluorescent. Where an up-right microscope platform is available, a water immersion lens permits high resolution imaging of live cells growing on plastic. However, rapid medium exchanges are more difficult in this configuration, the continued culture of now contaminated cells is not an option, and fluorescent reagents necessitate very careful cleaning of the objective. For imaging hippocampal slices the feet on the culture inserts are removed (use scissors), and the inserts transferred to 50mm glass bottomed dishes. Under these conditions, organotypic cultures growing on the inserts can be imaged using high numerical aperture objectives, in our case a 40X fluor, and with some care, a 60X planApo [3,4,23]. Note: Intact inserts can used for long-term studies by transferring

50

Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

them to new culture dishes using pieces of capillary tube (diameter 1.5 mm) placed underneath the membrane insert to substitute for the membrane feet (which normally serve to prevent flooding of the tissue). When imaging in intact inserts is difficult, a section of the membrane (with the slice), can be cut out, inverted in a glass bottomed dish, and immobilized by reducing the medium volume and partially covering the central well with a coverslipPrior to imaging, culture dishes are affixed to the microscope stage to eliminate movement; blue tack, or double-sided tape are inexpensive and highly effective. Relocating an individual neuron is difficult if a culture dish has been moved even slightly, so it is useful to mark the original orientation with a permanent marker pen. Note: These pens should not be used on the bottom of the dish since, despite their name, the ink is not indelible after contact with emersion oil. Photo-etched gridded coverslips have limited use for tracking individual cells since they are too thick (1.5mm). For repeated imaging fiduciary marks can be made prior to plating by very gently scratching the glass surface with a diamond knife. Note: Deep scratches can fatally weaken the glass. The requirements of live neurons for perfusion with oxygenated media vary considerably. In our studies the duration of each cycle of osmotic perturbations was relatively short and continuous perfusion was not required. Since rapid medium changes were essential and our perfusion system has a relatively slow clearance rate (2 ml/min), exacerbated by the presence of the central well in glass bottomed dishes, we typically removed the medium by pipette attached to a vacuum suction device. Medium was reintroduced as a bolus, also by pipette. We usually elected to work at a (warm) room temperature (28 C), since it slows plasma membrane remodelling slightly, facilitating the capture of fast events. Note: In previous studies of calcium dynamics in acute spinal cord preparation [16], sympathetic neurons [14] and acute hippocampal slices [15] imaging at room temperature improved cell survival. We use an inverted (Nikon based) scanning confocal microscope (Bio–Rad MRC-600) equipped with an argon-ion laser. To minimize photo-bleaching and photo-damage, potential cells are initially selected in epi-fluorescence mode and confocal imaging parameters optimized with cells outside this field. To reduce photobleaching we use a neutral density filter of 3 (1% transmission), a zoom factor below 3, and single scans (rates ranging from 1/s to 1/30s). Simultaneous fluorescent and phase contrast images can be taken using the second channel for phase (non-confocal, transmitted light) images. VLDs can also be monitored in 3-D by optically sectioning neurons; optical series are taken using one single scan per section and step sizes (the distance between two optical sections) of 0.5–3.0 µm. Note: Optical series are only practical if the object of interest does not move and/or change shape during series collection. 2.3

Imaging plasma membrane remodelling

VLD formation and recovery were monitored by confocal microscopy using a variety of reagents that label the plasma membrane, and by adding dyes to the bath. In studies on SAR high resolution phase images, are also crucial to establish and maintain the focal plane, and to interpret the confocal images. FM 4-64 (T13320, Molecular Probes, Invitrogen) is an lipophilic styryl dye (MW 60, excitation 558nm, emission 734nm). This water-soluble dye is non-toxic, and virtually nonfluorescent in aqueous media, but becomes intensely fluorescent upon insertion into the outer leaflet of the surface membrane. FM4-64 is reasonably photostable, after 50 frames in cells (pulsed-labelled with FM4-64), less than 10% of the signal is lost to photobleaching [18]. We used 10µl/ml of a 1mg/ml FM 4-64 stock in methanol, maintaining this concentration during solution exchanges and imaged with the BHS (fluorescein) filter blocks. FM4-64 effectively labelled the plasma membrane (Fig.1). Images taken at the level of the substratum showed that VLDs appeared as fluorescent rings that corresponded closely with VLDs visible in phase mode (Fig. 1b,c). During recovery VLDs gradually shrank and then disappeared as they were absorbed into the plasma membrane (see ref [1] for details). While easy to use FM4-64 was rapidly internalized by endocytosis and labelled other membranes. For this reason, it is most useful for imaging single cycles of SAR. Trials of other membrane dyes indicated that if a dye stained both the plasma membrane and endomembranes, distinguishing VLD membranes was difficult. The carbocyanine dye 1,18-dioctadecyl3,3,38,38-tetramethylindocarbocyanine perchlorate (DiI) in ethanol or in DMSO (final concentration 2.5

51

Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

µM) proved to be the best choice for selective and rapid labelling of the plasma membrane. DiI labelled neurons were imaged using the GHS (rhodamine) filter block. Since DiI is relatively insoluble and forms a fine, intensely fluorescent precipitate in the bath, it was added for only 30 sec in the swelling phase,

Fig. 1 Osmotic changes in PC12 cells labelled with FM 4-64: Swelling in hyposmotic medium for 4 min. (a) and subsequent shrinking in normosmotic medium (b) induces formation of VLDs (arrow) to accommodate the decrease in cell surface area. VLDs form as membrane invaginations at the adherent surface (see [1]). (c) phase contrast image of the cell in (b). (d,e) VLDs spontaneously recover in normosmotic medium.

and cultures were washed repeatedly during the shrinking phase (triggered by normosmotic medium), to eliminate the bath signal. Note: This protocol affords no opportunity to image the plasma membrane during the initial swelling phase, but swelling can be monitored in subsequent cycles. DiI labelled VLDs formed within 4 sec of the onset of cell shrinking [1], initially appearing as small fluorescent ‘points’. During the next minute, VLDs rapidly expanded and then gradually shrank. Comparison of Figs. 2a and 2b shows that some VLDs visible in confocal mode were obscured in

Fig. 2 VLDs in snail neurons labelled with DiI. (a) Shrinking in normosmotic medium induces numerous VLDs (arrow) visible in phase contrast mode. (b) Confocal image shows numerous DiI labelled VLDs some of which are undetected by phase optics. (c) VLDs at different stages of recovery. Magnification, 60x.

phase images by the nucleus. At the level of the substratum DiI revealed a remarkable degree of complexity in the morphology of VLDs during recovery. While some VLDs appeared to collapse into the plasma membrane (flattened circles, Fig. 2c), others appeared as irregular annuli of motile, adherent, processes (arrows in Fig. 2c). Such profiles were in marked contrast to the smooth membrane profiles observed for the same VLDs at focal planes deeper in the cytoplasm (not shown). Strikingly, VLDs formed in second or third cycles were similar in size, shape and site of origin to those formed in the first cycle.

52

Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

Fig. 3 Bath dyes taken up by VLDs as they form label VLD contents. (a) Profiles of LY filled VLDs in PC12 cells, image taken at the level of the adherent surface. (b) RD3000 filled VLDs in a snail neuron. This image, taken 5.0 µM above the substratum shows that VLDs extend part way into the cytoplasm.

2.4

Bath dyes and VLD dynamics

Lucifer yellow (L-453, Molecular Probes, Invitrogen) is a water-soluble, plasma membrane impermeable dye (excitation/emission 428/536 nm) that we used to label the bath contents. While it is commonly believed that Lucifer yellow (LY), due to its low absorbance at 488 nm, is insufficiently excited with the argon-ion laser, our experience shows otherwise [1,15]. Rhodamine-Dextran 3000 (RD3000), was also used to label the bath. Note: In bath labelling experiments, the capacity of the confocal microscope to reject out-of-focus fluorescence is essential (in epi-fluorescence mode the bath signal predominates and cells are not visible). In confocal images neurons appear black objects against an intense bath signal. Within seconds of adding normosmotic medium to previously swollen cells, numerous VLDs of diverse size and shape were detected (see Figure 3a,b). VLDs imaged via their fluorescent aqueous contents behaved like VLDs imaged via membranes labelled with DiI and FM4-64; they formed rapidly (2.0µM), hyposmotic medium caused mitochondria to release a small amount of Rhod-123 into the cytoplasm. The mitochondrial signal itself appeared unchanged. These results suggest that our swelling regime caused only a modest and transient depolarization of the inner mitochondrial membrane. They also emphasize the importance of selecting appropriate loading protocols for Rhod-123 (see [22], for an excellent discussion of these issues). Western blots of cytoplasmic fractions collected 5 min and 2hrs post 2 cycles of swelling and shrinking showed zero to trace amounts of cytochrome c, confirming that our protocols did not result in a sustained depolarization of mitochondria, or activate apoptotic pathways.

3. Summary Neurons in disociated cultures display a considerable capacity for cell surface remodelling [1,2,4,25]. Experiments using biolistic transfection of membrane targeted plasmids have shown similar events ocurr in the complex 3-D environment of the hippocampal slice [3]. Studies on SAR using differentiated PC12 cells transfected with a mitochondrially targeted mtGFP revealed dramatic and reversible changes in mitochondrial morphology [23, and unpublished observations]. These rapid morphological transformations suggest that mitochondria have a hitherto unappreciated capacity to remodel in response to stress. While the mechanisms that underlie this mitochondria plasticity are unknown, it is clear that such transitions are not associated with decreased neuronal survival. Further, while the role of mitochondrial remodelling in physiological conditions remains to be elucidated, the capacity to remodel may play a key role in neuronal ability to recover from insults that target mitochondria. Acknowledgements. NSERC and CIHR to LRM; and OGS to NS and MSA to NS. The authors thank Alexandra Haggard for her help with the mitochondrial experiments and Chloe Shniffer for her help in preparing this manuscript.

References [1]L.R. Mills and C.E. Morris, Journal of Membrane Biology, 166, 223 (1998) [2] T.L. Herring, C.S. Cohan, E.A. Welnhofer, and L.R. Mills, Journal of Membrane Biology, 171, 151 (1999) [3] N. Shulyakova and L.R. Mills, Microscopy and Microanalysis, 12 (Supp 2), 242 (2006) [4] N. Shulyakova, J.A. Wang, C.E. Morris and L.R. Mills, Canadian Journal of Neurological Sciences, 34 (Supp3), 106 (2007), Annual Meeting Canadian Association for Neuroscience, Toronto, Canada, May 23-25 [5] E.C. Cheung, H.M. McBride, and R.S. Slack, Apoptosis, 12, 979 (2007)

58

Modern Research and Educational Topics in Microscopy. ©FORMATEX 2007 A. Méndez-Vilas and J. Díaz (Eds.) _______________________________________________________________________________________________

[6] S. Kahlert and G. Reiser, FEBS Letters, 529, 351 (2002) [7] N. Shalbuyeva, T. Brustovetsky, A. Bolshakov, and N Brustovetsky, Journal of Biological Chemistry, 281, 37547 (2006) [8] W.K. Lee, U. Bork, and F. Thévenod, Toxicology Mechanisms and Methods, 14, 67 (2004) [9] T. Minamikawa, D.A. Williams, D.N. Bowser, and P. Nagley, Experimental Cell Research, 246, 26 (1999) [10] N.A. Pham, T. Richardson, J. Cameron, B. Chue, and B.H. Robinson, Microscopy and Microanalysis, 10, 247 (2004) [11] L.R. Mills, C.E. Niesen, and R.G. Kerr, in: J.K. Stevens, L.R. Mills, J. Trogadis (eds.) (Academic Press Inc., Orlando, Florida, 1994), chap. no.14, pp. 325-351 [12]D.J. Stephens and V.J. Allan, Science, 300, 82 (2003) [13]D.P. Sirk, Z. Zhu, J.S. Wadia, and L.R. Mills, Cytometry, 56, 15 (2003) [14]R.K. Amer, C.R. Pace-Asciak, and L.R. Mills, Neuroscience, 116, 935 (2003) [15] L.R. Mills, C.E. Niesen, A.P. So, P.L. Carlen, I. Spigelman, and O.T. Jones, Journal of Neuroscience, 14, 6815 (1994) [16]L.R. Mills, A.A. Velumian, S.K. Agrawal, E. Theriault, and M.G. Fehlings, Neuroimage, 21, 1069 (2004) [17] Y. Adamchik, M.V. Frantseva, M. Weisspapir, P.L. Carlen, and J.L. Perez Velazquez, Brain Research Protocols, 5, 153 (2000) [18] A.M. Benediktsson, S.J. Schachtele, S.H. Green, and M.E. Dailey, Journal of Neuroscience Methods, 141, 41 (2005) [19]J. Grutzendler, J. Tsai, and W.B. Gan, Methods, 30, 79 (2003) [20] J R. Nakajima, T. Nakamura, M. Ogawa, H. Miyakawa, and Y. Kudo, Journal of Neuroscience Research, 76, 723 (2004) [21] A. Kaasik, D. Safiulina, A. Zharkovsky, and V. Veksler, American Journal of Physiology – Cell Physiology, 292, C157 (2007) [22]D.G. Nicholls and M.W. Ward, TINS, 23, 166 (2000) [23] P. Kaifosh, A, Haggard, and LR. Mills The Canadian Journal of Neurological Sciences, 34 (Supp3), 79, (2007), 1st Annual Meeting of the Canadian Association for Neuroscience, Toronto, Canada, 23-25 May [24]N. Shulyakova, N. Phan, D.P. Sirk, J. Fong, and L.R. Mills, submitted to Mitochondrion (2007) [25]CE. Morris, UJ. Homann, Journal of Membrane Biology 15; 179, 79 (2001).

59