In Situ Imaging of Mitochondrial Outer- Membrane Pores Using Atomic Force Microscopy

BIOIMAGING Research Report In situ imaging of mitochondrial outer- membrane pores using atomic force microscopy

Bradley E. Layton, Ann Marie Sastry, Christian M. Lastoskie, Martin A. Philbert, Terry J. Miller, Kelli A. Sullivan, Eva L. Feldman, and Chia-Wei Wang

BioTechniques 37:564-573 (October 2004)

Here we describe a technique for imaging of the outer contours of the mitochondrial membrane using atomic force microscopy, subsequent to or during a toxic or metabolic challenge. Pore formation in both glucose-challenged and 1,3-dinitrobenzene (DNB)- challenged mitochondria was observed using this technique. Our approach enables quantification of individual mitochondrial membrane pore formations. With this work, we have produced some of the highest resolution images of the outer contours of the in situ mitochondrial membrane published to date. These are potentially the first images of the component protein clusters at the time of formation of the mitochondrial membrane transition pore in situ. With the current work, we have extended the application of atomic force microscopy of mitochondrial membranes to fluid imaging. We have also begun to correlate 3-D surface features of mitochondria dotted with open membrane pores with features previously viewed with electron microscopy (EM) of fixed sections.

INTRODUCTION tion of a large, multiprotein porin com- channel size, and permeable substrates plex mediates either energy production for each membrane protein are shown Selective passage of solute mol- capability (closed probability state) or in Table 2 (14–22). It should be noted ecules into and out of the cell is ac- cell death (open probability state) by that while certain transport proteins complished by several classes of pore- preventing or permitting the exchange such as gramicidin form cylindrical forming proteins that are integral to the of matrix components with matrix and/ channels that are well described by a cell membrane. An important subset of or intermembrane components. Such single measure of internal pore width, these channel-forming proteins are the exchange is critical to secondary cel- many channel-forming proteins have porins (1), proteins that adopt a charac- lular processes (6–8) and thus largely internal pore cross sections that vary teristic barrel-shaped configuration and determines cell fate (9–13). Dozens of along the channel axis, with one or form channels spanning the extracellu- membrane channels have been identi- more internal cavities separated from lar and intracellular environments (2,3) fied to date, as summarized in Table 1 extracellular and intracelluar vestibules as shown in Figure 1. Porins are found (Reference 1; a database of these chan- by pore constrictions or gating regions. in the outer membranes of Gram-nega- nel-forming proteins may be found at In such instances, effective pore radii tive bacteria and in the membranes of http://saier-144-164.ucsd.edu/tcdb/). for the “cavity” and “gate” regions have mitochondria and chloroplast organ- Selected α-helix and β-barrel pore- been reported (Table 2) as obtained via elles in eukaryotic cells (1); typical forming proteins, with occurrence, crystallography (20) or from estimates interior diameters at the constriction obtained from permeation mea- regions within porin channels are about surements (21) or homology stud- 2 nm, which are sufficiently large to A B ies (22). allow the passage of a number of low One much studied porin is the molecular weight solutes via diffusive voltage-dependent, anion-selec- transport (4). Nonetheless, a large num- tive channel porin (VDAC). This ber of membrane porins are required porin is found in the mitochon- to accomplish significant transport. A drial outer membranes of animal typical bacterium, for example, is es- cells (23,24). It is selectively per- timated to have 105 porins in its outer meable to anions and can also un- membrane (5). Figure 1. OmpF, a trimeric β-barrel porin. (A) Top dergo a transition to be selective view, and (B) side view. In panel B, the top of the protein Porins play a critical role in cellular opens to the extracellular domain, and the bottom of the to cation transport (9). The VDAC and organellar transport processes. For protein opens to the intracellular domain. β strands are is also thought to play a key role example, in mitochondria, the forma- shown as ribbons and α-helices as red cylinders. in the regulation of mitochondrial-

University of Michigan, Ann Arbor, MI, USA

564 BioTechniques Vol. 37, No. 4 (2004) Table 1. Classifications of Channel-Forming Membrane Proteins with Their Occurrence, The three-dimensional (3-D) shape Function, and a Representative Protein from Each Class of the VDAC pore has been determined Channel Type Cell/Organelle Type Function Example Protein to approximately 0.1 nm resolution (26),

+ using sequence analysis and circular di- α-helix bundle bacteria, animal, and communication, K channel chroism to find that the protein forms plant tissue cells, homeostasis mitochondria a structure similar to a bacterial porin- like β-barrel. Measured conductance of β-barrel porin gram-negative bacteria, general or specific OmpF porin a membrane consisting of VDAC pores mitochondria, transport across in a bilayer in the presence of molecules chloroplasts membrane with diameters on the order of the pore holin bacteria, viral induce apoptosis autolysin inner diameter (27) was used to deter- pathogens mine an inner diameter of 1.2–1.5 nm protein toxin bacteria, yeast, nutrient leakage colicin for the Neurospora crassa mitochon- protozoa drial VDAC pore. An overall molecular peptide toxin bacteria, animal cells disrupt membrane bee venom weight of 34 kDa has also been obtained via Western blot analysis (28,29). The mediated apoptosis. Specifically, it has preceding the known release of cyto- membrane has an approximate density been hypothesized to either coalesce chrome c and the resulting activation of of several thousand pores per square mi- into a megapore, the permeability tran- irreversible cascades of apoptotic, in- crometer, as measured in bovine heart sition pore (PTP), or alternatively, to tracellular “killer proteases” (e.g., Ref- with enzyme-linked immunosorbent as- close entirely, presumably resulting erence 25). Thus, both the opening and say (ELISA) (30), and in the neurospora in heightened internal mitochondrial closure of the pore and the dimensions fungus using conductance measure- pressure. Either hypothesis would be associated with each have been the sub- ments across membranes and using sin- consistent with mitochondrial rupture ject of intense study. gle-channel recordings (31). One of the BIOIMAGING Research Report

Table 2. Selected α-Helix and β-Barrel Pore-Forming Proteins, with Occurrence, Channel Glucose Challenge Size, and Permeable Substrates for Each Membrane Protein Channel Organism Permeates Radius Reference Russell et al. (8) found that mitochon- (nm) dria from rat primary dorsal root ganglia

+ + + + (DRG) neurons challenged with 45 mM gramicidin bacteria H , Li , Na , K 0.2 Smart et al. (14) glucose for 6 h showed an increase in KcsA bacteria K+, Rb+, Cs+ 0.1 (gate) Doyle et al. (15) cross-sectional area of approximately 0.6 (cavity) Allen et al. (16) 50% (0.43 ± 0.03 μm2 vs. 0.66 ± 0.08 acetylcholine eukaryotes glycine, serotonin 0.3 (gate) Stein (17) μm2) by measuring maximal cross-sec- receptor 2.0 (cavity) Miyazawa et al. (18) tional area through a z-series of a fluo- OmpF bacterial outer nonselective for cat- 0.5 (gate) Jap et al. (19) rescent signals under confocal microsco- membrane ions, anions, sugars, 1.8 (cavity) Schirmer (20) py. We can estimate that this comprises others an 80% increase in volume, assuming an VDAC eukaryotic similar to OmpF, but 0.8 (gate) Benz (21) ellipsoidal shape for mitochondria (V ∝ mitochondrial voltage-dependent 1.3 (cavity) Casadio et al. (22) A3/2). Mitochondrial swelling was found anion selectivity to be dose- and time-dependent, with in- creased glucose molarities resulting in VDAC, voltage-dependent, anion-selective channel outer membrane porin. greater swelling rates, and longer expo- functional implications of its inherent cinoma cells. Several reports (43–45) sure resulting in greater volumetric in- pore geometry is that VDAC is readily have described 5 nm resolution images, creases (at 150 mM, 130% ΔA → 350% permeable to glucose, the latter of which obtained by imaging a relatively thick ΔV; at 300 mM, 166% ΔA → 440% has a diameter of <1 nm (32). (approximately 1 μm) section with ΔV) after 6 h. A plateau in cross-sec- Atomic force microscopy (AFM) a transmission electron microscope tional area (ΔA ~ 120% = ΔV ~ 225%) has been successfully applied to the (TEM) at a series of angles and recon- as measured by confocal microcopy was imaging of biological surfaces, as dem- structing the 3-D shape from the 2-D apparently reached after 24 and 48 h of onstrated in the characterization of the projections. excessive (45 mM) glucose exposure. general surface morphology of mito- In the present study, we use a re- Swelling appeared to be coupled with chondria (33–35), spermatozoa (36), fined AFM approach to investigate mi- the complete membrane depolarization and neurons (37). With its sub-nanome- tochondrial pore formation subsequent (loss of ΔΨmt) at 24 h, as measured by ter resolution and its capability to image to two specific types of challenge: ex- a change in wavelength of a voltage-de- features in aqueous environments, AFM cess glucose and 1,3-dinitrobenzene pendent fluorophore (8). is also well suited to imaging organelle (DNB) exposure. Under both condi- The precise mechanism for the dis- membrane proteins in situ (38) or in tions, mitochondria swell as a result of ruption of excess glucose on the viabil- monolayers (39,40). Examples of pre- water flux into the hypertonic matrix. ity of mitochondria is not yet known, but vious imaging experiments on the mi- Specific rationales follow for each likely involves disruption of the normal tochondrial outer-membrane structure case. This work combines a refined cycle of several proton-pumping in- with AFM include the use of immuno- mitochondrial isolation protocol, fol- ner-membrane proteins responsible for logically labeled 8 nm gold particles lowed immediately by either a meta- maintaining a negative voltage gradient conjugated to anti-rabbit immunoglob- bolic or toxic challenge, while being between the intermembrane space and ulin G (IgG) anti-PBR (peripheral-type tracked by AFM to determine structural the matrix (46). Thus, it is likely that a benzodiazepine receptor) (29,38) and changes in mitochondrial outer-mem- change in the metabolic cycle results in direct verification of the presence of brane topology as evidenced by porin a change in the electronegativity of the mitochondria through the cell mem- proliferation or reorganization. Two matrix, thus causing a change in osmotic brane of rat mammary carcinoma cells protocols are described: one in which pressure and producing swelling (47). (34). Two other groups (41,42) have im- mitochondria were isolated from em- aged mitochondrial-associated proteins bryonic rat primary dorsal root gan- 1,3-Dinitrobenzine/Cyclosporin-A directly, but did so in reconstituted lipid glion cells and subjected to a glucose films, rather than in situ. The inner mi- challenge, and one in which adult rat It has been observed by several tochondrial structure has been viewed liver mitochondria were isolated and separate groups that toxic challenge to via electron microscope (EM) tomogra- subjected to a DNB challenge. These mitochondria (e.g., exposure to DNB) phy (43–45). One group (29) found that two preparations were imaged directly triggers mitochondrial permeability the hCG hormone caused rapid reorga- with AFM to a resolution sufficient to transition (MPT), as marked by forma- nization of the peripheral-type benzodi- image mitochondrial membrane pro- tion of a megapore, the mitochondrial azepine receptor of unfixed mitochon- teins in situ. To our knowledge, ours is permeability transition pore (mtPTP) at dria of Leydig tumor cells with AFM in the first work to image unlabeled mito- the inner membrane (48–50). This pore air-contact mode using gold particles chondrial membrane pores in situ with has not been directly imaged, although as labels. Another group (34) imaged AFM subsequent to isolation from the its presence has been verified using spa- fixed mitochondria in air-contact mode cell and in conjunction with a toxic or tial distribution of fluorescent dyes, cal- in partially fractured rat mammary car- metabolic challenge. cein and tetramethylrhodamine methy-

566 BioTechniques Vol. 37, No. 4 (2004) lester (TMRM) (51). Inner-membrane mode, we used DNPS tips, employing proteins such as adenine nucleotide the shortest cantilever of the four on translocase (ANT) and outer-membrane the chip (approximately 100 μm) with proteins such as VDAC are putative core the narrowest legs (approximately 10 components. Cyclosporin-A (CsA) is μm) of spring constant 0.32 N/m. This a well-known inhibitor of MPT pore allowed for an imaging frequency of formation (52) The mtPTP has been approximately 10 kHz. Scan rates of 1 partially reconstituted in solution to Hz were typically used. Initially, pro- test the hypothesis that the permeabil- portional gain was set to 0.6 and the in- ity transition pore may be formed from tegral gain to 0.8. In tapping mode, to hexokinase, the enzyme responsible for minimize contact forces once engage- catalyzing the transfer of phosphate to ment occurred, set point amplitude was glucose, an outer-membrane porin, and increased toward drive amplitude until the adenylate nucleotide translocator, an disengagement occurred, and then de- inner-membrane integral protein (53). creased until the tip just regained con- tact. Typical error signals were on the Novelty of Technique order of 1 nm, once the tip force had been minimized. With this refinement Previously published AFM images of previous AFM imaging work (e.g., of mitochondria used indirect meth- Reference 33), individual membrane ods (29,34,38), such as artificial mem- features were resolvable. branes (41,42) or proteins outside of a membrane (33,37,54). Furthermore, Glucose Challenge the specific combination of our refine- ment of (55) technique for two separate A timed pregnant Sprague-Dawley tissue types combined with AFM is the rat (E15) was euthanized via sodium first such application to the knowledge pentobarbital (100 mg/kg) overdose. of the investigators. Approximately 30 DRG were harvest- ed per embryo, yielding approximately 500 DRG. Mitochondria were isolated MATERIALS AND METHODS using a protocol based on Verweij et al. (55) and modified as described. DRG In both studies, mitochondria were were placed for 10 min in 0.25 M tryp- isolated based on the mitochondrial sin and centrifuged for 5 min at 2200× isolation protocol of Verweij et al. g. The pellet was resuspended in a 1 (55). All procedures were executed in mL ice-cold isotonic buffer (250 mM accordance with university, state, and sucrose, 0.5 M EDTA, 20 mM HEPES, federal standards in accord with the 500 μM Na3VO4, pH 7.4) with protease National Research Council’s “Guide inhibitors (aprotinin and leupeptin and for the Care and Use of Laboratory An- phenylmethanesulfonyl fluoride). The imals.” These two isolation procedures solution was homogenized with a glass are tissue-specific due to the relatively Dounce homogenizer (Fisher Scientif- mitochondria high density and volume ic, Pittsburgh, PA, USA) on ice for 20 of tissue available in liver used in the strokes. One mL ice-cold isotonic buf- DNB study compared to that of em- fer and protease inhibitor were added, bryonic DRG cells used in the glucose- and the resulting solution was centri- challenge study. fuged at 900× g for 5 min at -4°C. Su- All imaging of preparations of mi- pernatant containing mitochondria was tochondria was performed using a carefully transferred into an ultracentri- NanoScope® III Dimension 3000 se- fuge tube. Care was taken to avoid any ries BioScope® (Veeco Instruments, nonmitochondrial contaminants from Woodbury, NY, USA) in air-contact, the pellet to be pipetted. The superna- air-tapping, or fluid-tapping modes. For tant containing mitochondria was then air-contact mode, Nanoprobe™ DNPS centrifuged at 12,000× g for 8 min. A tips were used (Veeco Instruments). For small visible pellet was resolvable at air tapping mode, Metrology Probes™ this point. The supernatant was poured (NanoDevices, Santa Barbara, CA, off and the pellet resuspended in 1 mL USA) were used at a frequency of ap- of ice-cold isotonic buffer (0.1 M sodi- proximately 290 kHz. For fluid-tapping um cacodylate in distilled water, of pH

Vol. 37, No. 4 (2004) BioTechniques 567 BIOIMAGING Research Report

7.4) without protease inhibitors. The morphology prior to examination by RESULTS solution was recentrifuged at 12,000× AFM. Samples of 1 μL were depos- g for 8 min. A pellet was again visible. ited on mica (air) poly-l-lysine (fluid) The DNB/CsA liver preparation pro- This final pellet was resuspended in Biocoat®-coated glass slides (Becton- duced a greater number and more dense- 150 μL of ice-cold isotonic buffer (250 Dickinson, Franklin Lakes, NJ, USA). ly clustered mitochondria than that of mM sucrose, 0.5 M EDTA, 20 mM Samples prepared for fluid tapping the DRG preparation on both mica and HEPES, pH 7.4, 500 μM Na3VO4). were immersed throughout prepara- poly-l-lysine surfaces (Figure 2). Four 5 μL samples were pipetted into tion and imaging. 1 mL tubes. The final yield was on the During imaging, regions contain- Glucose-Challenged order of approximately 100 mitochon- ing features resembling mitochondria dria per microliter. Two of these were or mitochondrial clusters were selected Approximately 20 mitochondria glucose-challenged with 40 mM glu- and imaged at a resolution of 256 × 256 were imaged from a total of 4 samples. cose for 6 h. At the end of 6 h, one of pixels over a region of approximately In fluid-tapping mode, 10 mitochondria the glucose-challenged and one of the 40 × 40 μm at a scan rate of approxi- were imaged. Those which revealed nonchallenged preparations were each mately 1 Hz. Within these regions, in- the best-preserved overall morphology fixed with 4% paraformadehyde in dividual mitochondria were selected were the fixed; the unfixed, glucose- preparation for AFM. and imaged at a resolution of 512 × challenged mitochondria lost their mor- 512, also at a rate of approximately 1 phology over the course of imaging. 1,3-Dinitrobenzine/Cyclosporin-A Hz. This imaging speed was found to At the whole-mitochondria scale, minimize thermal drift caused by ex- surface topology was visible (Figure Mitochondria from the liver of a male pected pore mobility within the outer 3A). The image shown is representa- Fisher rat were prepared in a manner membrane, as well as deformation and tive of those obtained; artificial lighting similar to that of Verweij et al. (55); the displacement of the mitochondrion be- and coloring were used to enhance to- resulting mitochondrial pellet was resus- ing imaged. This particular aspect of pological features. The exact shape of pended in 1 mL of 250 mM sucrose so- imaging membrane proteins in situ is the interface between the mitochondria lution. From this, 100 μL of mitochon- critical for obtaining the best possible and the substrate is not known, so a de- dria were withdrawn and distributed into image because a series of images col- convolution is not reliable in this case, several treatment tubes [DNB alone, lected for offline autocorrelation is not but approximate mitochondria volume DMSO (dimethyl sulfoxide; vehicle), possible due to the dynamic nature of has been estimated to be on the order of and CsA pretreated/DNB]. The final the outer membrane. Specifically, with 100 aL (1 aL = 100 nm3). concentration of DNB used was 100 μM in situ imaging, when returning to the On mitochondrial surfaces in con- (60 min treatment), and final concentra- same coordinates or when rastering trol preparations, no finely structured tion of CsA was 1 μM (30 min pretreat- back-and-forth over the same region, features were evident (Figure 4A). Fea- ment). Mitochondria were resuspended subsequent images invariably differ tures of the same scale of VDAC were in the high sucrose solution to ensure from previous images. apparent, however, in the glucose-chal- retention of viability. A portion was re- served and fixed with 4% paraformal- dehyde in 0.1 M cacodylate buffer. The presence of mitochondria was verified by the addition of 100 nM MitoTracker® Green (Molecular Probes, Eugene, OR, USA) at 490 nm and detection at 516 nm on a Microphot microscope (Nikon, Melville, NY, USA) within 10 min of staining. Images were captured on a SPOT-RT digital camera (Diagnostic In- struments, Sterling Heights, MI, USA). Three 1 mL samples were pipet- ted into 15 mL tubes. The final yield was greater than that of the glucose- challenged: approximately 1000 mi- tochondria per microliter were seen. One tube containing mitochondria was pretreated for 30 min with 1 μM Figure 2. Clustering of fixed mitochondria on mica resulting from two preparation techniques. (A) CsA prior to treatment as previously A 5-μm air-contact atomic force microscopy (AFM) image of glucose-challenged mitochondria from described (48). After 1 h, each of the embryonic Sprague-Dawley rat dorsal-root ganglion neurons, showing a relatively sparse concentration of mitochondria. (B) A 20-μm air-tapping AFM image of 1,3-dinitrobenzene (DNB)-challenged mito- three preparations was fixed in 4% chondria from Fisher rat liver, showing significant clustering of mitochondria. The pyramidal shape of paraformaldehyde in cacoldylate buf- the mitochondria in both panels is a result of the interaction with the pyramidal tip with four 35˚ sides, a fer to ensure retention of organelle nominal height of 10 μm, and a tilt of 10˚ from horizontal.

568 BioTechniques Vol. 37, No. 4 (2004) lenged mitochondria. Some imaging ar- in Figure 4B). To the authors’ knowl- tifacts arose (e.g., horizontal streaking edge, this is the first such imaging of a within scan lines), but the organization possible VDAC on a mitochondrial sur- of membrane proteins appeared visible. face. Due to our limited sample sizes, Periodic invaginations of approximately however, full quantification of pore-like 5 nm were present (typical image shown features has not yet been made.

Figure 3. Air atomic force microscopy (AFM) images of fixed, glucose-, and 1,3-dinitrobenzene (DNB)-challenged mitochondrial surfaces. (A) Air-contact AFM image of glucose-challenged, Sprague- Dawley rat embryonic dorsal root ganglion neuronal mitochondrion. (B) Air-tapping AFM image of three adjacent, fixed, DNB-challenged, Fisher rat liver mitochondria. Surface features of approximately 10 nm are evident. The invaginations in the outer membrane may reveal the topography of the inner membrane.

Figure 4. Fluid-tapping atomic force microscopy (AFM) images of fixed, control, and glucose-chal- lenged Sprague-Dawley embryonic DRG mitochondria. (A) The 256 × 256 pixel (approximately 4 nm per pixel) image of an unchallenged outer membrane, revealing features of 10–100 nm. The indenta- tion in the middle of the upper image may reveal topography of the inner membrane; a cropped portion of the upper image is shown in the lower image. (B) The 512 × 512 (approximately 2 nm per pixel) im- age of a glucose-challenged outer membrane, revealing invaginated features that may be VDAC pores; a cropped portion of the upper image is shown in the lower image. DNB, 1,3-dinitrobenzene; VDAC, voltage-dependent, anion-selective channel porin.

Vol. 37, No. 4 (2004) BioTechniques 569 BIOIMAGING Research Report

1,3-Dinitrobenzene-Challenged pore. When contrasted to the control ing optimally resolved AFM images is image (Figure 5A), similar features to balance the competing effects of the In general, the surface roughness are found but are not as pronounced. amount of time spent collecting data of DNB-challenged mitochondria was A complete statistical analysis of data from a given location (slow scanning) greater than that of either control or was not performed due to small sample and the amount of thermal drift that can DNB-CsA treated mitochondria. The sizes, but typical images are shown. occur over the course of a single scan single DNB-challenged mitochondrion or even between sequential fast-raster surface shown in Figure 3B represents movements. The challenge becomes a typical image. Clustering in the rat DISCUSSION greater when imaging on the top of liver preparation was common, perhaps deformable adsorbed organelles. The due to the higher yield relative to the With this work, we have presented challenge may be likened to trying to DRG preparation. some of the highest resolution images resolve the individual nubs on a par- In the unchallenged preparations, of the outer contours of the mitochon- tially flattened basketball without mov- few features were resolvable on the mi- drial membrane published to date. ing the ball itself. Though individual tochondrial surface; Figure 5A repre- These are potentially the first images biomolecules have been successfully sents a typical image. The single bulge of the component protein clusters at the imaged in membranes or other sub- evident in this image was typical and time of formation of the mitochondrial strates (57–59), imaging of membrane may have resulted from the protrusion membrane transition pore (Figure 5B). proteins directly on cells presents an of the inner membrane through the out- This structure could form via radial additional challenge. er membrane. It seems likely that if the expansion, wherein concentric rings Convolution artifact plays a role in tip had traversed a pore during imag- form and successively collapse, until AFM by distorting the image of the ing, that a 2–3 nm inner pore diameter a “megapore” of approximately 30 nm structures one views. In the present would be evident due to the interaction. results. This may involve N terminus to study, this effect can be seen in the Challenged preparations revealed al- C terminus bindings between adjacent large scan as evidenced by the pyra- tered mitochondrial surfaces; a typi- porins, as illustrated in Figure 1. An midal appearance of the mitochondria. cal image is shown as Figure 5B. This alternative hypothesis is that the struc- There is also some artifact due to the fi- surface appears to contain a pore-like tures seen in Figure 5B are those similar nite radius of the tip; however, accurate structure of dimensions consistent with to those seen by Kiselyova et al. (56). horizontal information is obtainable if a mitochondrial permeability transition One particular challenge for obtain- one is measuring the distance between two peaks because the lowest point of the tip must touch the highest point(s) of the sample. From this work, we may begin to approximate transport rates that must be present during a pore opening, given variables such as pore diameter and molecular species of transport. Poten- tially, the rate at which mitochondrial swelling occurs as a precursor to apop- tosis and correlates with loss of outer- membrane potential can be measured with this method. While this method is somewhat artificial in that it does not allow observation of mitochondrial behavior as it might occur within a liv- ing cell (60), it has the advantage of viewing the membrane directly during a toxic challenge and the potential for the identification of specific proteins involved in the stabilization/destabili- zation of the pore complex.

Clustering on Preparations Figure 5. Fluid-tapping atomic force microscopy (AFM) images of fixed, control, and DNB-chal- lenged Fisher rat liver mitochondria. (A) The 512 × 512 pixel (approximately 2 nm per pixel) image This clustering behavior could be re- of a control Fisher rat liver mitochondrial outer membrane, revealing surface features 10–20 nm. (B) lated to that seen by Yaffe (61), whose The 512 × 512 pixel (approximately 2 nm per pixel) image of a DNB-challenged mitochondrial outer membrane, showing large openings with diameters of approximately 10–20 nm, suggesting presence work describes the many specialized of permeability transition pores; an enlargement is shown in the lower image. These larger features are and reticulated shapes that mitochon- absent in panel A. DNB, 1,3-dinitrobenzene. dria or groups of mitochondria may

570 BioTechniques Vol. 37, No. 4 (2004) BIOIMAGING Research Report form in various cells, such as sperm ACKNOWLEDGMENTS mitochondrial membrane permeabilization: cells and muscle cells, as apparently an open-and-shut case? Cell Death Differ. dictated by the dynamin protein Dnm1 We are grateful for the assistance in 10:485-487. 11.Bernardi, P., R. Colonna, P. Costantini, O. and the fuzzy onion protein Fzo1, both mitochondrial preparations performed Eriksson, E. Fontaine, F. Ichas, S. Massari, outer-membrane associated proteins by Hui Wang, Carey Backus, and John A. Nicolli, et al. 1998. The mitochondrial per- with still unknown control mechanisms. Hayes. Major funding for this work meability transition. Biofactors 8:273-281. At this large scale, taken at a resolution was provided by the W.M. Keck Foun- 12.Halestrap, A.P., P.M. Kerr, S. Javadov, and K.Y. Woodfield. 1998. Elucidating the molec- of 512 × 512 nm, no surface features dation. Additional funding from NSF ular mechanism of the permeability transition can be seen, and tip artifact consisting PECASE (A.M.S.) and NSF CAREER pore and its role in reperfusion injury of the of a pyramidal appearance is present. (C.M.L.) grants, along with the Nation- heart. Biochim. Biophys. Acta 1366:79-94. With the current work, we have ex- al Institutes of Health grant no. NIH- 13.He, L. and J.J. Lemasters. 2002. Regulated tended the work of atomic force mi- R01 ES08846 (M.A.P.) and the Juvenile and unregulated mitochondrial permeabil- ity transition pores: a new paradigm of pore croscopists by imaging in fluid as well Diabetes Research Foundation Center structure and function? FEBS Lett. 512:1-7. as in air and have begun to interpret the for the Study of Complications in Dia- 14.Smart, O.S., J.M. Goodfellow, and B.A. physical landscape of mitochondria betes (E.F. and K.A.S.) also supported Wallace. 1993. The pore dimensions of gram- dotted with open membrane pores, in this work. icidin A. Biophys. J. 65:2455-2460. 15.Doyle, D.A., J. Morais Cabral, R.A. relation to those imaged using EM to- Pfuetzner, A. Kuo, J.M. Gulbis, S.L. Cohen, mography of fixed sections. B.T. Chait, and R. MacKinnon. 1998. The COMPETING INTERESTS structure of the potassium channel: molecular STATEMENT basis of K+ conduction and selectivity. Sci- FUTURE WORK ence 280:69-77. The authors declare no competing 16.Allen, T.W., A. Bliznyuk, A.P. Rendell, S. interests. Kuyucak, and S.H. Chung. 2000. The potas- The qualification of the pore type sium channel: structure, selectivity and diffu- will hopefully give a snapshot of the sion. J. Chem. Phys. 112:8191-8204. condition of the pore upon fixation and 17.Stein, W.D. 1990. Channels, Carriers, and REFERENCES Pumps: an Introduction to Membrane Trans- support one of three different possible port, p. 80,94. Academic Press, San Diego. pore-formation mechanisms. These 1.Saier, M.H., Jr. 2000. Families of proteins 18.Miyazawa, A., Y. Fujiyoshi, M. Stowell, and include: first, a more or less random forming transmembrane channels. J. Membr. N. Unwin. 1999. Nicotinic acetylcholine re- “pore flicker” event during which the Biol. 175:165-180. ceptor at 4.6 A resolution: transverse tunnels inner and outer pore components are 2.Brdiczka, D., G. Beutner, A. Ruck, M. in the channel wall. J. Mol. Biol. 288:765- Dolder, and T. Wallimann. 1998. The molec- 786. briefly aligned, implying a completely ular structure of mitochondrial contact sites. 19.Jap, B.K. and P.J. Walian. 1996. Structure entropy-driven process; second, a semi- Their role in regulation of energy metabolism and functional mechanism of porins. Physiol. random, or partially controlled forma- and permeability transition. Biofactors 8:235- Rev. 76:1073-1088. tion, implying transitory conforma- 242. 20.Schirmer, T. 1998. General and specific po- 3.Schatz, G. and B. Dobberstein. 1996. Com- rins from bacterial outer membranes. J. Struct. tional changes driven by either abrupt mon principles of protein translocation across Biol. 121:101-109. changes in electrochemical gradients membranes. Science 271:1519-1526. 21.Benz, R. 1994. Permeation of hydrophilic sol- or changes in mechanical states such 4.Schulz, G.E. 1996. Porins: general to specific, utes through mitochondrial outer membranes: as intermembrane pressure or relative native to engineered passive pores. Curr. Opin. review on mitochondrial porins. Biochim. Struct. Biol. 6:485-490. Biophys. Acta 1197:167-196. strains between membranes; or third, 5.Buechner, M., A.H. Delcour, B. Martinac, J. 22.Casadio, R., I. Jacoboni, A. Messina, and a more permanent pore with perhaps a Adler, and C. Kung. 1990. Ion channel activ- V. De Pinto. 2002. A 3D model of the volt- spoke-and-hook structure. ities in the Escherichia coli outer membrane. age-dependent anion channel (VDAC). FEBS In general, membrane pore geom- Biochim. Biophys. Acta 1024:111-121. Lett. 520:1-7. etry is critical to understanding both 6.Kroemer, G. and J.C. Reed. 2000. Mito- 23.Dihanich, M. 1990. The biogenesis and func- chondrial control of cell death. Nat. Med. tion of eukaryotic porins. Experientia 46:146- the types of molecules that may pass 6:513-519. 153. through the outer membrane as well 7.Petersen, K.F., D. Befroy, S. Dufour, J. Dzi- 24.Muller, D.J., H. Janovjak, T. Lehto, L. Kue- as the evolutionary origins of porins ura, C. Ariyan, D.L. Rothman, L. DiPietro, rschner, and K. Anderson. 2002. Observing (1). The putative symbiotic evolution G.W. Cline, et al. 2003. Mitochondrial dys- structure, function and assembly of single function in the elderly: possible role in insulin proteins by AFM. Prog. Biophys. Mol. Biol. of bacterial cells as mitochondrial or- resistance. Science 300:1140-1142. 79:1-43. ganelles within eukaryotic cells pro- 8.Russell, J.W., D. Golovoy, A.M. Vincent, P. 25.Martinou, J.C., S. Desagher, and B. An- vides an explanation for the structural Mahendru, J.A. Olzmann, A. Mentzer, and tonsson. 2000. Cytochrome c release from homology between bacterial and mi- E.L. Feldman. 2002. High glucose-induced mitochondria: all or nothing. Nat. Cell Biol. tochondrial porin channels (21). This oxidative stress and mitochondrial dysfunc- 2:E41-E43. tion in neurons. FASEB J. 16:1738-1748. 26.Mannella, C.A. 1998. Conformational similarity can be exploited to infer 9.Vander Heiden, M.G., N.S. Chandel, X.X. changes in the mitochondrial channel protein, mitochondrial membrane transport Li, P.T. Schumacker, M. Colombini, and VDAC, and their functional implications. J. characteristics from computational and C.B. Thompson. 2000. Outer mitochondrial Struct. Biol. 121:207-218. experimental investigations of diffusive membrane permeability can regulate coupled 27.Rostovtseva, T.K., A. Komarov, S.M. Bez- respiration and cell survival. Proc. Natl. Acad. rukov, and M. Colombini. 2002. VDAC transport in bacterial outer-membrane Sci. USA 97:4666-4671. channels differentiate between natural me- porins, which will also be part of future 10.Robertson, J.D., B. Zhivotovsky, V. tabolites and synthetic molecules. J. Membr. work. Gogvadze, and S. Orrenius. 2003. Outer Biol. 187:147-156.

572 BioTechniques Vol. 37, No. 4 (2004) 28.Garnier, M., A. Dimchev, N. Boujrad, M.J. Res. Commun. 298:744-749. lution images of protein surfaces. Biophys. J. Price, N.A. Musto, and V. Papadopoulos. 43.Frey, T.G. and C.A. Mannella. 2000. The in- 77:1150-1158. 1994. In vitro reconstitution of a functional ternal structure of mitochondria. Trends Bio- 58.Muller, D.J., M. Amrein, and A. Engel. peripheral-type benzodiazepine receptor. Mol. chem. Sci. 25:319-324. 1997. Adsorption of biological molecules to a Pharmacol. 45:201-211. 44.Mannella, C.A., K. Buttle, B.K. Rath, and solid support for scanning probe microscopy. 29.Boujrad, N., B. Vidic, and V. Papadopoulos. M. Marko. 1998. Electron microscopic to- J. Struct. Biol. 119:172-188. 1996. Acute action of choriogonadotropin on mography of rat-liver mitochondria and their 59.Muller, D.J., C.A. Schoenenberger, F. Scha- Leydig tumor cells: changes in the topography interaction with the endoplasmic reticulum. bert, and A. Engel. 1997. Structural changes of the mitochondrial peripheral-type benzodi- Biofactors 8:225-228. in native membrane proteins monitored at azepine receptor. Endocrinology 137:5727- 45.Perkins, G., C. Renken, M.E. Martone, subnanometer resolution with the atomic 5730. S.J. Young, M. Ellisman, and T. Frey. 1997. force microscope: a review. J. Struct. Biol. 30.De Pinto, V., G. Prezioso, F. Thinnes, T.A. Electron tomography of neuronal mitochon- 119:149-157. Link, and F. Palmieri. 1991. Peptide-spe- dria: three-dimensional structure and organi- 60.Petronilli, V., G. Miotto, M. Canton, R. Col- cific antibodies and proteases as probes of the zation of cristae and membrane contacts. J. onna, P. Bernardi, and F. Di Lisa. 1998. Im- transmembrane topology of the bovine heart Struct. Biol. 119:260-272. aging the mitochondrial permeability transi- mitochondrial porin. Biochemistry 30:10191- 46.Lamson, D.W. and S.M. Plaza. 2002. Mito- tion pore in intact cells. Biofactors 8:263-272. 10200. chondrial factors in the pathogenesis of diabe- 61.Yaffe, M.P. 1999. Dynamic mitochondria. 31.Freitag, H., W. Neupert, and R. Benz. 1982. tes: a hypothesis for treatment. Altern. Med. Nat. Cell Biol. 1:E149-E150. Purification and characterisation of a pore Rev. 7:94-111. protein of the outer mitochondrial membrane 47.Vander Heiden, M.G., N.S. Chandel, E.K. from Neurospora crassa. Eur. J. Biochem. Williamson, P.T. Schumacker, and C.B. Received 14 April 2004; accepted 123:629-636. Thompson. 1997. Bcl-xL regulates the mem- 7 July 2004. 32.Nitzan, Y., K. Orlovsky, and I. Pechatnikov. brane potential and volume homeostasis of 1999. Characterization of porins isolated from mitochondria. Cell 91:627-637. the outer membrane of Serratia liquefaciens. 48.Phelka, A.D., M.J. Beck, and M.A. Phil- Address correspondence to: Curr. Microbiol. 38:71-79. bert. 2003. 1,3-Dinitrobenzene inhibits mito- Professor Ann Marie Sastry 33.Wagner, O.I., J. Lifshitz, P.A. Janmey, M. chondrial complex II in rat and mouse brain- 2250 G.G. Brown Building Linden, T.K. McIntosh, and J.F. Leterrier. stem and cortical astrocytes. Neurotoxicology 2350 Hayward Street 2003. Mechanisms of mitochondria-neurofila- 24:403-415. Department of Mechanical Engineering/ ment interactions. J. Neurosci. 23:9046-9058. 49.Haworth, R.A. and D.R. Hunter. 1979. The Department of Biomedical Engineering 34.Pietrasanta, L.I., A. Schaper, and T.M. Jo- Ca2+-induced membrane transition in mito- University of Michigan vin. 1994. Imaging subcellular structures of chondria. II. Nature of the Ca2+ trigger site. Ann Arbor, MI 48109-2125 rat mammary carcinoma cells by scanning Arch. Biochem. Biophys. 195:460-467. e-mail: [email protected] force microscopy. J. Cell Sci. 107(Pt 9):2427- 50.Hunter, D.R. and R.A. Haworth. 1979. The 2437. Ca2+-induced membrane transition in mito- 35.Kordylewski, L., D. Saner, and R. Lal. 1994. chondria. I. The protective mechanisms. Arch. Atomic force microscopy of freeze-fracture Biochem. Biophys. 195:453-459. replicas of rat atrial tissue. J. Microsc. 173(Pt 51.Lemasters, J.J., A.L. Nieminen, T. Qian, 3):173-181. L.C. Trost, and B. Herman. 1997. The mi- 36.Joshi, N.V., H. Medina, C. Colasante, and tochondrial permeability transition in toxic, A. Osuna. 2000. Ultrastructural investigation hypoxic and reperfusion injury. Mol. Cell. of human sperm using atomic force micros- Biochem. 174:159-165. copy. Arch. Androl. 44:51-57. 52.Friberg, H. and T. Wieloch. 2002. Mitochon- 37.Parpura, V., P.G. Haydon, and E. Hender- drial permeability transition in acute neurode- son. 1993. Three-dimensional imaging of generation. Biochimie 84:241-250. living neurons and glia with the atomic force 53.Beutner, G., A. Ruck, B. Riede, and D. microscope. J. Cell Sci. 104(Pt 2):427-432. Brdiczka. 1998. Complexes between porin, 38.Papadopoulos, V., N. Boujrad, M.D. Iko- hexokinase, mitochondrial creatine kinase and nomovic, P. Ferrara, and B. Vidic. 1994. adenylate translocator display properties of Topography of the Leydig cell mitochondrial the permeability transition pore. Implication peripheral-type benzodiazepine receptor. Mol. for regulation of permeability transition by the Cell Endocrinol. 104:R5-R9. kinases. Biochim. Biophys. Acta 1368:7-18. 39.Neff, D., S. Tripathi, K. Middendorf, H. 54.Davis, J.J., H.A. Hill, and T. Powell. 2001. Stahlberg, H.J. Butt, E. Bamberg, and N.A. High resolution scanning force microscopy Dencher. 1997. Chloroplast F0F1 ATP syn- of cardiac myocytes. Cell. Biol. Int. 25:1271- thase imaged by atomic force microscopy. J. 1277. Struct. Biol. 119:139-148. 55.Verweij, B.H., J.P. Muizelaar, F.C. Vinas, 40.Schillers, H., T. Danker, H.J. Schnittler, F. P.L. Peterson, Y. Xiong, and C.P. Lee. 1997. Lang, and H. Oberleithner. 2000. Plasma Mitochondrial dysfunction after experimental membrane plasticity of Xenopus laevis oocyte and human brain injury and its possible rever- imaged with atomic force microscopy. Cell sal with a selective N-type calcium channel Physiol. Biochem. 10:99-107. antagonist (SNX-111). Neurol. Res. 19:334- 41.Ding, T.T., S.J. Lee, J.C. Rochet, and P.T. 339. Lansbury, Jr. 2002. Annular alpha-synuclein 56.Kiselyova, O.I., I.V. Yaminsky, Y.D. Ivanov, protofibrils are produced when spherical pro- I.P. Kanaeva, V.Y. Kuznetsov, and A.I. Ar- tofibrils are incubated in solution or bound chakov. 1999. AFM study of membrane pro- to brain-derived membranes. Biochemistry teins, cytochrome P450 2B4, and NADPH-cy- 41:10209-10217. tochrome P450 reductase and their complex 42.Epand, R.F., J.C. Martinou, S. Montessuit, formation. Arch. Biochem. Biophys. 371:1-7. R.M. Epand, and C.M. Yip. 2002. Direct 57.Moller, C., M. Allen, V. Elings, A. Engel, evidence for membrane pore formation by and D.J. Muller. 1999. Tapping-mode atomic the apoptotic protein Bax. Biochem. Biophys. force microscopy produces faithful high-reso-

Vol. 37, No. 4 (2004) BioTechniques 573