JOURNAL OF ELECTRON MICROSCOPY TECHNIQUE 14:6-1%( 1990)
Time-Resolved Cryotransmission Electron Microscopy YESHAYAHU TALMON, JANET L. BURNS, MATTHEW H. CHESTNUT, AND DAVID P. SIEGEL Department of Chemical Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel (Y.T.1; Miami Valley Laboratories, T h e Procter & Gamble Company, Cincinnati, Ohio 45239-8707 (J.L.B., M.H.C., D.P.S.)
KEY WORDS
Direct imaging, Time-resolved cryo-TEM, Transient microstructures, Phospholipid mesophases
ABSTRACT We describe a new technique, time-resolved cryotransmission electron microscopy (TRC-TEM), that can be used to study changes in microstructure occurring during dynamic processes such as phase transitions and chemical reactions. The sample is prepared on a n electron microscope grid maintained a t a fixed temperature in a controlled atmosphere. The dynamic process is induced on the grid by a change in pH, salt, or reactant concentration by rapid mixing with appropriate solutions. Alternatively, induction is by rapid change of specimen temperature, or by controlled evaporation of a volatile component. We call such procedures on-the-grid processing. The dynamic process is permitted to run for a defined time and then the thin-film specimen is thermally fixed by plunging into liquid ethane a t its freezing point, producing a cryotransmission electron microscopy specimen. By repeating this procedure with varying delays between induction and sample fixation, we can observe transient microstructures. We demonstrate the use of TRC-TEM to study the intermediate structures that form during the transitions between L,,, 111, and HI1 liquid crystalline phases in phospholipid systems. We also identify several other possible applications of the technique.
INTRODUCTION It is now well accepted that thermal fixation of thin specimens is the least artifact-prone technique for preparing transmission electron microscopy (TEM) specimens from aqueous systems of biological or synthetic origin. Imaging of the thermally fixed, usually vitrified, specimen provides the most direct, highresolution microstructural information from systems like viruses (Adrian et al., 1984), microtubules (Mandelkow et al., 19861, vesicles (Miller et al., 19871, and micelles (Burns and Talmon, 1987; Burns et al., 1988). Most of these systems are sensitive to temperature and concentration fluctuations. The specimen must be prepared in a n environment of controlled temperature and saturation to prevent evaporation of volatiles from the specimen. A controlled environment vitrification system (CEVS) for the preparation of vitrified TEM specimens of sensitive systems has been recently described by Bellare et al. (1988). The CEVS has been used to investigate the microstructure of systems a t equilibrium (Burns and Talmon, 1987; Burns et al., 1989). Here we describe new sample preparation techniques using the CEVS that make it possible to use cryo-TEM to image transient changes in microstructure during chemical or physical transformations. These techniques involve the induction of dynamic processes in samples maintained within the CEVS. Observation of a series of samples vitrified a t fixed times after induction reveals the evolution of microstructure during the dynamic process. We refer to such a series of experiments as time-resolved cryo-TEM (TRC-TEM), and to the procedures used to induce the dynamic process a s onthe-grid processing. In order to demonstrate the utility
'c, 1990 ALAN
R. LISS. INC
of on-the-grid processing and TRC-TEM, we have used TRC-TEM to image the evolution of microstructure during transitions between lamellar (La),inverted cubic (IlI),and inverted hexagonal (HIJ phases in phospholipid dispersions. A more detailed report of the results and mechanistic inferences will appear elsewhere (Siege1 et al., 1989a).
TECHNIQUE CEVS The CEVS is a n environmental chamber in which the temperature can be maintained within 0.1"C between -5" to + 95°C. The atmosphere inside, air or a n inert gas, can be saturated with one or two volatiles. Manipulation of the specimen, prepared on a holey carbon-film supported on a TEM grid, is possible from outside the chamber via two rubber septa, one on each side. A third opening in the bottom of the chamber is fitted with a shutter that opens simultaneously with the triggering of the spring-loaded plunging mechanism. This propels the specimen (held with a tweezer attached to the plunging rod) into a reservoir of cryogen (typically liquid ethane a t its freezing point) that is placed under the environmental chamber. After thermal fixation the specimen is transferred under liquid nitrogen to the TEM cold-stage where i t is maintained and examined a t about -170°C. Just prior to fixation, specimens consist of very thin
Received September 1, 1988; accepted in revised form January 23, 1989. Address reprint requests to Yeshayahu Talmon, Dept. of Chemical Ehgineering, Technion-Israel lnstitute of Technology, Haifa 32000, Israel.
TRC-TEM
(2 niis) into a cryogen at its freezing point. The cryogm is chosen to have a large temperature range between its melting point and its boiling point to avoid formalion of a vapor envelope around the specimen during thermal fixation. The nature of the cryogen (in our case, ethane), the sample geometry, and the high plunging velocity result in a n ultrahigh cooling rate in the specimen, vitrifying its liquid components. This prevents possible structural rearrangements associated with thermotropic phase transitions, e.g., formation of ice crystals. The vitreous state of water and many organic liquids is thermodynamically unstable. For this reason, the sample is maintained below - 150'C during manipulation and examination to prevent reversion to the stable crystalline phases.
On-the-grid processing and time-resolved cryo-TEM It is simple to prepare vitrified TEM specimens using the CEVS (Bellare e t al., 1988). A micropipet is inserted into the CEVS through one of the rubber septa, and a 5 ~1 drop of sample is placed on the grid. Most of the liquid is then blotted away, using filter paper wrapped around a flat metal strip inserted through the other septum. The plunging mechanism is then triggered. When necessary, the specimen can be left in the CEVS for hours prior to vitrification, if the temperature and vapor saturation of the atmosphere within the unit a re controlled. It is often desirable to reequilibrate samples in the CEVS prior to thermal fixation, for example, to relax stresses that were induced in the sample by pipetting and blotting. However, conditions in the chamber can be changed in several different ways to induce physical or chemical processes that lead to microstructural changes in the specimen. The simplest such operation is changing the temperature. Tenierature changes may be used to induce processes such as phase transitions (e.g., from one mesomorphic phase to another), gelation, and crystallization. We call this kind of procedure on-the-grid processing. In the range of 20-70"C, the heating rate obtained in the CEVS is about 20"C/min. Cooling is slower: 8"C/min in the range of 70-30°C and 3"Cimin in the 30-20°C range. At least three other methods of on-the-grid processing are feasible. First, the dynamic process can be induced by controlled evaporation using partial saturation in the atmosphere around the specimen. Partial saturation can be maintained by equilibration with .;tandardized saturated salt solutions. Second, a chemical reaction can be triggered on the grid using a UV source mounted inside the chamber. In systems that i'orm gels or high viscosity phases, on-the-grid processing is the only way to prepare thin specimens for cryo-TEM without microtomy or replication. Third, two drops of different liquids can be mixed on the grid. This rnay be used to change pH, ionic strength, and salt concentration, parameters that in many systems lead to changes in the microstructure. Solutions of two reactants can also be mixed on the grid, thus initiating
7
a chemical reaction that leads to a new microstructured phase. In each of these applications one can compare the microstructure of similar specimens before and after the induced change. Any type of on-the-grid processing used to initiate a dynamic event, can be combined with timed cryofixation of sequential specimens. We call such a series of experiments time-resolved cryotransmission electron microscopy (TRC-TEM). Sequential samples are vitrified at different times after initiation of the event of interest on the grid. Using this approach, one can directly visualize the evolution of microstructure during a dynamic process.
APPLICATION: La/III/HII PHASE TRANSITIONS The L, phase consists of stacks of phospholipid bilayers, and the HII phase of hexagonally packed inverted rod-like micelles. 111 phases and the so-called "isotropic" phases are complex structures composed of curved bilayers (Siegel, 1 9 8 6 ~ )"Isotropic" . phases are thought to represent poorly ordered 111 phase. The mechanism of the transitions between L,, 111,and HI1 phases has been the subject of debate (Siegel, 1984, 1986a-c; Verkleij, 1984). Recently, a kinetic theory of the phase transition mechanisms was developed (Siegel, 1986a-c) predicting the structure of the intermediates that form during these phase transitions. The transition to 111or HII phases is initiated by the formation of labile intermediates between apposed L, phase bilayers. In one class of systems, the labile intermediates rapidly convert the apposed bilayers into the HI1 phase. In the other class of systems, the labile intermediates are predicted to form a second, much more stable intermembrane structure known as an interlamellar attachment (ILA), as described by Siegel (1986b,c). ILAs are semitoroidal (automobile tire rimshaped) bilayer connections between apposed bilayer sheets; the diameter of the structures is a t least 12 nm. ILAs are predicted to assemble into the 111phase under appropriate conditions (Siegel, 1 9 8 6 ~ ) . In this study, we examined liposomal dispersions of dioleoylphosphatidylethanolamine (DOPE) and Nmethylated dioleoylphosphatidylethanolamine(DOPEMe) executing these phase transitions. DOPE is a system that is predicted to form very few ILAs, whereas DOPE-Me should form them in large numbers (Siegel, 1986b,c). MATERIALS AND METHODS Materials All phospholipids were purchased from Avanti Polar Lipids (Birmingham, AL), and were more than 99% pure as determined by two-dimensional thin-layer chromatography criteria; they were used without further purification. All salts and buffers were reagent grade. The water used for solutions was purified with a Milli-Q system (Millipore Co.) and had a n initial resistivity of ca. 18 MWcm. The lipids were suspended in buffers of the following compositions: DOPE in 100 mM NaC1, 20 mM glycine (pH 9.91, 0.1 mM EDTA; DOPEMe in 100 mM NaC1, 20 mM glycine (pH 9.5), 0.1 mM EDTA. Buffers used to produce the pH or magnesium
8
Y. TALMON ET AL
proceeded to different extents in different areas of the same grid. It was often necessary to delay blotting in order to observe later stages in the transition process. To study the transitions in the slower DOPE-Me system, for example, we sometimes had to wait 10 seconds before blotting to ensure the formation of lipid aggregates large enough to show nascent 111 domains. Vitrified specimens were transferred under liquid nitrogen to the “work-station’’ of a Gatan 626 coldstage, loaded into the stage, transferred and examined in a Hitachi H-500 TEM operating a t 100 kV. Holder Methods temperature was typically - 170°C. Images were reLipids were purchased as lyophilized powders and corded on Kodak SO-163 film processed for highest were dispersed in the relevant buffer by frequent electron speed in full strength D-19 developer, for 12 vortex mixing a t a temperature above the chain- minutes. To enhance contrast, images were taken a t melting transition of the lipid. This was followed by defocus of about 4 pm. Typical electron exposures were three freeze-thaw cycles (dry icei40”C water bath). in the range of 1-10 ke-/nm2. Liposomal dispersions were produced by extrusion RESULTS AND DISCUSSION through 0.1 or 0.2 pm Nuclepore filters (Nuclepore Figure 1is a cryo-TEM image of a vitrified specimen Corp., Pleasanton, CA) in a Lipex Bilmembranes (Vancouver, B.C.) extruder a t a nitrogen gas pressure of of DOPE liposomes a t 25°C and pH 9.9, conditions 200-600 psi. Some DOPE dispersions were produced by where the lipid is anionic and stable in the L, phase. detergent dialysis using octyl glucoside (Mimms et al., Embedded in the vitreous ice are unilamellar and 1981) in order to investigate the effect of high pressure oligolamellar liposomes, some of which are invaginated, and a few long tubular structures. These latter extrusion on liposome morphology. Specimens for cryo-TEM were prepared on holey- structures appear to form during high pressure extrucarbon films mounted on 200 mesh copper grids. Hole sion of the lipid dispersions to produce liposomes, as diameter was 1 to 10 pm. We used the CEVS as liposomes produced via detergent dialysis are all described by Bellare et al. (1988). Experiments were nearly spherical (data not shown). When the pH is less than ca. 8.5, DOPE can form the conducted below, above, or a t room temperature. Liquid nitrogen was used as the refrigerant in the cooling HII phase a t temperatures above ca. 5°C (e.g., Gruner module of the CEVS. (Dry ice-acetone is more conve- et al., 1988). Figure 2 shows the earliest structural nient as it lasts longer, but acetone dissolved the change visible within these dispersions when the pH is Styrofoam insulation very rapidly when splashed out- lowered by mixing drops of liposome dispersion and the side the cooling reservoir.) Air in the CEVS chamber appropriate buffer: the liposomes aggregate into was kept a t 90-95% relative humidity, as determined clumps. Figure 3 is a n image obtained later in this by a General Eastern model 800B hygrometer process, showing the formation of small HII domains equipped with a type 814 sensor. within these aggregates. The HIr domains are identiFor TRC-TEM, the CEVS was set to the appropriate fied by their enhanced optical-density, and by the fact temperature, and the air inside was saturated with that the lines in images of these domains are 8 nm water vapor using sponge wicks in the CEVS sample apart (the HI1 tube width in this sytem) (Gruner et al., chamber and a flow of humidified, filtered compressed 1988). Occasionally, bundles of HII tubes are visible in air. The liposome dispersions and buffer solutions were cross section. At later times, one finds areas consisting maintained a t the experimental temperature in a sep- only of large HI1 phase domains (Fig. 4). Qualitatively arate water bath. A 5 pl drop of liposome suspension the same results are achieved with bilayer-buffer (acwas placed on the grid mounted on the CEVS plunger. etate, pH 4.5) or nonpermeating buffer (TES, pH 7.4; or Then a second 5 pl drop of the appropriate pH or Mg2 bicine, pH 8.3).It is important to note that nascent HII concentration-jump buffer was mixed with the first tubes were never observed in isolated, nonapposed drop on the same side of the EM grid. This was found to bilayers: this indicates that the obligatory intermedibe superior to applying the drops to opposite sides of ates in this transition are interlamellar structures, as the grid. The grid was then blotted immediately, or predicted (Siege], 1984, 1986a). blotted after a time delay of several seconds. The DOPE-Me systems incubated a t temperatures ca. shortest interval between drop mixing and fixation 10°C below the equlibrium L,/HII phase transition that we could achieve was 3 seconds. In order to do this, temperature, TH. give rise to isotropic and inverted it is best to have an assistant standing by, to hand the cubic (111)phase (Ellens et al., 1989). I t was proposed pipets to the operator, trigger the plunger, and time the (Siege], 1986a-c) that this is due to the formation of operation. After thermal fixation, the procedure is metastable ILAs under these circumstances. Therefore, identical to other CEVS experiments. we examined DOPE-Me dispersions by TRC-TEM in an Although some steps in the phase transitions in attempt to observe these structures. these systems are quite rapid, the 3 second preparation The predicted ILA geometry is depicted in Figure 5A: time did not preclude observation of earlier stages in the structure is semitoroidal and cylindrically symmetthe process. Mixing did not appear to be uniform across rical about the vertical dashed axis. The lower limit to the grid, since the transitions often appeared to have the outer diameter of the “waist” of the structure is
ion concentration-jumps were made iso-osmotic to the suspending buffer in each case, using a vapor pressure osmometer (Wescor model #5500, Wescor Corp.). The pH-jump buffers had the following compositions for DOPE: ca. 80 mM NaCl plus either 40 mM bicine (pH 8.3), 40 mM TES (pH 7.41, or 40 mM acetate (pH 4.5). The Mg2 concentration-jump buffer for DOPE-Me was 20 mM NaC1, 60 mM MgCI2, 2 mM TES (pH 7.4). Liquid ethane (chemically-pure, Matheson, Inc.) a t its freezing point was used for specimen vitrification.
+
TRC-TEM
Fig. 1. TRC-TEM image of a DOPE liposomal dispersion at pH 9.9 and 25°C The liposomes were produced by high-pressure extrusion through 0.2 pm filters. This seems to produce numerous invaginated and spherical forms. Liposomes produced by detergent dialysis are all nearly spherical (data not shown). Bar equals 100 nm. Fig. 2. DOPE liposomal dispersion ca. 5 seconds after mixing with bicine buffer (pH 8.3) a t 25°C. The liposomes have aggregated. Bar equals 100 nm. Fig. 3. Aggregates of DOPE liposomes forming inclusions of HI[ phase ca 5 seconds after mixing with bicine buffer a t 15°C. The HII
9
domains are distinguished by their optical density, the 8 nm periodicity of the dark lines within them (approximately the HII tube diameter in this system), and by the occasional appearance of HII tubes in cross section (arrow). Bar equals 100 nm. Fig. 4. Large domain of HII phase in a DOPE dispersion 3 seconds after acidification with TES buffer (pH 7.4) at 25°C. The system forms these domains faster than a t 10.5"C because the temperature is higher, and thus further above the equilibrium L,/HII phase transition temperature of ca. 5°C. Bar equals 100 nm.
Y. TALMON ET AL
10
Figure 6 were obtained ca. 6 seconds after magnesium addition. ILAs are observed in the DOPE-Me a t the same temperature, solution conditions, and on the same time scale (seconds) a s membrane fusion is detected in the same system with fluorescence assays (Ellens et al., 1989). These results, taken together with the direct visualization of the fusion process (Fig. 6A), suggest that ILAs are the structures that mediate membrane fusion under these circumstances (Ellens et al., 1986; Siegel, 1986b; Siegel et al., 1989b). Systems like DOPE do not ordinarily form isotropic or inverted cubic phases. They are observed to do so when certain types of detergents are added (Wieslander et al., 1986). According to the kinetic theory (Siegel, 1986b,c), the detergents should greatly increase the ILA formation rate. If the 111 phase really is produced from ILAs, then these systems should exhibit Fig. 5. (A) Predicted geometry of an ILA (Siegel, 1986c), viewed ILA morphology. We did observe ILA morphology in in cross-section. The stippled-edge slabs represent lipid bilayers. (B) such a system. Figure 7 is a TRC-TEM image of a Expected appearance in transmission microscopy of an ILA when viewed with the electron beam perpendicular to the bilayer stack DOPE liposomal dispersion produced by dialysis of a (top) and parallel to the bilayer stack (bottom).The appearance of a DOPEioctyl glucoside mixture after the pH was refold in one of the bilayer membranes is also depicted. C,D: Flow in the duced to 7.4 a t 25°C. The octyl glucoside detergent was aqueous film caused by capillary film thinning prior to fixation can not completely removed. An HII phase domain (lower move folds in membranes. lLAs connecting adjacent membranes are tilted almost into the plane of the sample film when the fold reaches left), L, phase bilayers (liposomes, upper right), and them, resulting in the depicted morphology (as viewed from the top; incipient 111phase (large arrow; compare with Fig. 6A) upper part of figure). are shown. Individual ILAs are barely visible in profile at left (small arrows). A
estimated at 12 nm (Siegel, 1 9 8 6 ~ ) .The expected appearance of ILAs within a bilayer stack is indicated schematically in Figure 5B-D. When the bilayers are perpendicular to the electron beam, ILAs should appear as light-centered rings (top of Fig. 5B). When the bilayers are parallel to the beam, they should appear as reversed parentheses (bottom of Fig. 5B). Aqueous films in TRC-TEM samples undergo thinning under the influence of capillary forces. This induces a radially directed flow in the films. This flow can move the folded edges of membrane sheets as indicated in Figure 5B-D. ILAs that connect adjacent membranes will be reoriented by the edge of such a moving fold: the ILAs will be tilted into the plane of the sample film, yielding the morphology shown schematically in Figure 5C,D. La/III/HIItransitions in DOPE-Me can be induced by addition of magnesium ion a t pH 9.5 (Ellens et al., 1989). TRC-TEM images of liposomal samples mixed with magnesium buffer a t 50°C yield clear examples of the expected ILA morphology depicted in Figure 5. Examples of these TRC-TEM images are displayed in Figure 6. ILAs are visible from the top (light-centered rings approximately 15 nm in diameter; small arrows in Fig. 6A); from the side (reversed parentheses figures; arrows in Fig. 6B); and a t the margins of membrane folds (large arrow in 6A; compare with Fig. 5D.) Areas with numerous ILAs (superposition of rings in many successive stacked bilayers, viewed in projection) are also visible (arrowhead in Fig. 6A). The accumulation of large numbers of ILAs is thought to be the first step in 111phase formation (Siegel, 1 9 8 6 ~ )Elsewhere, . it is shown that 111 phases are assembled via formation of a n ordered lattice of ILAs (Siegel et al., 1988). In Figure 6A, liposomes seem to be fusing into larger structures via ILA formation. The TRC-TEM images in
CONCLUSIONS We have demonstrated the ability of time-resolved cryotransmission electron microscopy (TRC-TEM) to capture intermediate microstructures in dynamic processes, using the L, + HII phase transformation associated with phospholipid membrane fusion. The basic idea is to trigger a physical or chemical transformation on the grid and freeze the process a t different elapsed times from its onset. The results of a series of such experiments are images of the microstructure corresponding to the various steps in the process. TRC-TEM is one of the applications of on-the-grid processing. In the example presented here, the process was initiated by a pH or Mg+ concentration-jump induced by mixing a drop of the vesicular dispersion with a drop of a n appropriate buffer. The concept of TRC-TEM is currently used by one of us (Y.T.) and his colleagues to study three other systems. The first is the process of polymer gelation (Cohen, 1987; Cohen et al., 1989) where the trigger is a temperature drop or a concentration change by evaporation. The time scale in that case is several seconds. The solvents in the system studied so far have been benzene and benzyl alcohol. We have also begun a study of nucleation and growth of crystals from solution by this technique (Addadi, Weissbuch and Talmon, unpublished results), using supersaturation as the driving force for the formation of the new phase. In the latter systems the process is slower and, in some cases, may take up to 5 minutes. A third example involves U.V.-induced vesicle polymerization using a U.V. light source mounted within the CEVS chamber to polymerize the surfactant in vesicles, thus producing microstructural changes. The time scale in this case is seconds to minutes. At present TRC-TEM can be used to study processes
TRC-TEM
11
Fig. 6. DOPE-Me liposomal dispersions ca. 6 seconds after addition of 30 mM Mg” at 50°C. ILAs are visible viewed from the top (light-cer tered rings indicated by the small arrows in A, from the side (“reverset-l parentheses” indicated by the arrows in B)); and at the margins if membrane folds (indicated by the large arrow in A. The arrowhead in A indicates a region of many ILAs in a multibilayer stack. Liliosomes are fusing into larger structures via ILA formation at upper right. Bar equals 100 nm.
Fig. 7. TRC-TEM image of a liposomal dispersion produced by dialysis of a DOPEioctyl glucoside detergent mixture 3-5 seconds after the pH was reduced to 7.4 at 25°C. An HI1phase is visible at lower left (asterisk),liposomes fusing via ILA formation at the upper right and at arrowhead, and incipient isotropic or 111phase at the center (large arrow). Individual ILAs (small arrows) are also visible in a band of apposed bilayers. Bar equals 100 nm.
with time constants of several seconds or longer. Many processes of fundamental interest fall in this range. Paramc’ters such as temperature’ concentration’ and pH can be changed to bring the time Of most systems into the range approachable by the technique.
Bellare, J.R., Davis, H.T., Scriven, L.E., and Talmon, Y. (1988) Controlled environment vitrification system: An improved sample preparation technique. J. Electron Microsc. Tech., 10:87-111. Burns, J.L., and Talmon, Y. (1987) Cryo-TEM of micellar solutions. In: Proceedings of the 45th Annual EMSA Meeting. G.W. Bailey, ed. (San Francisco Press, San Francisco), pp. 498-499. Burns, J.L., Cohen, Y., and Talmon, Y. (1989) Determination of the Pm3n cubic phase in a surfactant-water system by cryo-TEM and SAXS. Science, submitted. Cohen, Y. (1987) Structure Formation in Solutions of Rigid Polymers
REFERENCES Adrian, M., Dubochet, J., Lepault, J., and McDowell, A.W. (1984) Cryoelwtron microscopy of viruses. Nature, 308:32-36.
12
Y. TALMON ET AL
Undergoing a Phase Transition. Ph.D. Dissertation, University of Massachusetts, Amherst. Cohen, Y., Talmon, Y., and Thomas, E.L. (1989) On the structure of poly(y-benzyl-L-glutamate) gels. In: Physical Networks. W. Burchard and S.B. Ross-Murphy, eds. Elsevier Applied Science Pub., Amsterdam, in press. Ellens, H., Bentz, J., and Szoka, F.C. (1986) Fusion of phosphatidylethanolamine-containing liposomes and the mechanism of the La/HIIphase transition. Biochemistry, 25:4141-4147. Ellens, H., Siegel, D.P., Alford, D,, Yeagle, P.L., Boni, L.T., Lis, L.J., Quinn, P.J., and Bentz, J. (1989) Membrane fusion and inverted phases. Biochemistry, in press. Gruner, S.M., Tate, M.W., Kirk, G.L., So, P.T.C., Turner, D.C., and Keane, D.T. (1988) X-ray diffraction study of the polymorphic behavior of N-methylated dioleoylphosphatidylethanolamine.Biochemistry, 27:2853 -2866. Mandelkow, E.-M., Rapp, R., and Mandelkow, E. (1986) Microtubule structure studied by quick freezing: Cryo-electron microscopy and freeze fracture. J. Microsc., 141:361-377. Miller, D.D., Bellare, J.R., Evans, D.F., Talmon, Y., and Ninham, B.W. (1987) On the meaning and structure of amphililic phases: Inference from video-enhanced microscopy and cryo-transmission electron microscopy. J. Phys. Chem., 91:674-685. Mimms, L.T., Zampighi, G., Nozaki, Y., Tanford, C., and Reynolds, J.A. (1981) Phospholipid vesicle formation and transmembrane protein incorporation using octyl glucoside. Biochemistry, 202333840.
Siegel, D.P. (1984) Inverted micellar structures in bilager membranes. Formation rates and half-lives. Biophys. J., 45:399-420. Siegel, D.P. (1986a) Inverted micellar intermediates and the transitions between lamellar, cubic and inverted hexagonal lipid phases. I. Mechanism of the L
Time-Resolved Cryotransmission Electron Microscopy YESHAYAHU TALMON, JANET L. BURNS, MATTHEW H. CHESTNUT, AND DAVID P. SIEGEL Department of Chemical Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel (Y.T.1; Miami Valley Laboratories, T h e Procter & Gamble Company, Cincinnati, Ohio 45239-8707 (J.L.B., M.H.C., D.P.S.)
KEY WORDS
Direct imaging, Time-resolved cryo-TEM, Transient microstructures, Phospholipid mesophases
ABSTRACT We describe a new technique, time-resolved cryotransmission electron microscopy (TRC-TEM), that can be used to study changes in microstructure occurring during dynamic processes such as phase transitions and chemical reactions. The sample is prepared on a n electron microscope grid maintained a t a fixed temperature in a controlled atmosphere. The dynamic process is induced on the grid by a change in pH, salt, or reactant concentration by rapid mixing with appropriate solutions. Alternatively, induction is by rapid change of specimen temperature, or by controlled evaporation of a volatile component. We call such procedures on-the-grid processing. The dynamic process is permitted to run for a defined time and then the thin-film specimen is thermally fixed by plunging into liquid ethane a t its freezing point, producing a cryotransmission electron microscopy specimen. By repeating this procedure with varying delays between induction and sample fixation, we can observe transient microstructures. We demonstrate the use of TRC-TEM to study the intermediate structures that form during the transitions between L,,, 111, and HI1 liquid crystalline phases in phospholipid systems. We also identify several other possible applications of the technique.
INTRODUCTION It is now well accepted that thermal fixation of thin specimens is the least artifact-prone technique for preparing transmission electron microscopy (TEM) specimens from aqueous systems of biological or synthetic origin. Imaging of the thermally fixed, usually vitrified, specimen provides the most direct, highresolution microstructural information from systems like viruses (Adrian et al., 1984), microtubules (Mandelkow et al., 19861, vesicles (Miller et al., 19871, and micelles (Burns and Talmon, 1987; Burns et al., 1988). Most of these systems are sensitive to temperature and concentration fluctuations. The specimen must be prepared in a n environment of controlled temperature and saturation to prevent evaporation of volatiles from the specimen. A controlled environment vitrification system (CEVS) for the preparation of vitrified TEM specimens of sensitive systems has been recently described by Bellare et al. (1988). The CEVS has been used to investigate the microstructure of systems a t equilibrium (Burns and Talmon, 1987; Burns et al., 1989). Here we describe new sample preparation techniques using the CEVS that make it possible to use cryo-TEM to image transient changes in microstructure during chemical or physical transformations. These techniques involve the induction of dynamic processes in samples maintained within the CEVS. Observation of a series of samples vitrified a t fixed times after induction reveals the evolution of microstructure during the dynamic process. We refer to such a series of experiments as time-resolved cryo-TEM (TRC-TEM), and to the procedures used to induce the dynamic process a s onthe-grid processing. In order to demonstrate the utility
'c, 1990 ALAN
R. LISS. INC
of on-the-grid processing and TRC-TEM, we have used TRC-TEM to image the evolution of microstructure during transitions between lamellar (La),inverted cubic (IlI),and inverted hexagonal (HIJ phases in phospholipid dispersions. A more detailed report of the results and mechanistic inferences will appear elsewhere (Siege1 et al., 1989a).
TECHNIQUE CEVS The CEVS is a n environmental chamber in which the temperature can be maintained within 0.1"C between -5" to + 95°C. The atmosphere inside, air or a n inert gas, can be saturated with one or two volatiles. Manipulation of the specimen, prepared on a holey carbon-film supported on a TEM grid, is possible from outside the chamber via two rubber septa, one on each side. A third opening in the bottom of the chamber is fitted with a shutter that opens simultaneously with the triggering of the spring-loaded plunging mechanism. This propels the specimen (held with a tweezer attached to the plunging rod) into a reservoir of cryogen (typically liquid ethane a t its freezing point) that is placed under the environmental chamber. After thermal fixation the specimen is transferred under liquid nitrogen to the TEM cold-stage where i t is maintained and examined a t about -170°C. Just prior to fixation, specimens consist of very thin
Received September 1, 1988; accepted in revised form January 23, 1989. Address reprint requests to Yeshayahu Talmon, Dept. of Chemical Ehgineering, Technion-Israel lnstitute of Technology, Haifa 32000, Israel.
TRC-TEM
(2 niis) into a cryogen at its freezing point. The cryogm is chosen to have a large temperature range between its melting point and its boiling point to avoid formalion of a vapor envelope around the specimen during thermal fixation. The nature of the cryogen (in our case, ethane), the sample geometry, and the high plunging velocity result in a n ultrahigh cooling rate in the specimen, vitrifying its liquid components. This prevents possible structural rearrangements associated with thermotropic phase transitions, e.g., formation of ice crystals. The vitreous state of water and many organic liquids is thermodynamically unstable. For this reason, the sample is maintained below - 150'C during manipulation and examination to prevent reversion to the stable crystalline phases.
On-the-grid processing and time-resolved cryo-TEM It is simple to prepare vitrified TEM specimens using the CEVS (Bellare e t al., 1988). A micropipet is inserted into the CEVS through one of the rubber septa, and a 5 ~1 drop of sample is placed on the grid. Most of the liquid is then blotted away, using filter paper wrapped around a flat metal strip inserted through the other septum. The plunging mechanism is then triggered. When necessary, the specimen can be left in the CEVS for hours prior to vitrification, if the temperature and vapor saturation of the atmosphere within the unit a re controlled. It is often desirable to reequilibrate samples in the CEVS prior to thermal fixation, for example, to relax stresses that were induced in the sample by pipetting and blotting. However, conditions in the chamber can be changed in several different ways to induce physical or chemical processes that lead to microstructural changes in the specimen. The simplest such operation is changing the temperature. Tenierature changes may be used to induce processes such as phase transitions (e.g., from one mesomorphic phase to another), gelation, and crystallization. We call this kind of procedure on-the-grid processing. In the range of 20-70"C, the heating rate obtained in the CEVS is about 20"C/min. Cooling is slower: 8"C/min in the range of 70-30°C and 3"Cimin in the 30-20°C range. At least three other methods of on-the-grid processing are feasible. First, the dynamic process can be induced by controlled evaporation using partial saturation in the atmosphere around the specimen. Partial saturation can be maintained by equilibration with .;tandardized saturated salt solutions. Second, a chemical reaction can be triggered on the grid using a UV source mounted inside the chamber. In systems that i'orm gels or high viscosity phases, on-the-grid processing is the only way to prepare thin specimens for cryo-TEM without microtomy or replication. Third, two drops of different liquids can be mixed on the grid. This rnay be used to change pH, ionic strength, and salt concentration, parameters that in many systems lead to changes in the microstructure. Solutions of two reactants can also be mixed on the grid, thus initiating
7
a chemical reaction that leads to a new microstructured phase. In each of these applications one can compare the microstructure of similar specimens before and after the induced change. Any type of on-the-grid processing used to initiate a dynamic event, can be combined with timed cryofixation of sequential specimens. We call such a series of experiments time-resolved cryotransmission electron microscopy (TRC-TEM). Sequential samples are vitrified at different times after initiation of the event of interest on the grid. Using this approach, one can directly visualize the evolution of microstructure during a dynamic process.
APPLICATION: La/III/HII PHASE TRANSITIONS The L, phase consists of stacks of phospholipid bilayers, and the HII phase of hexagonally packed inverted rod-like micelles. 111 phases and the so-called "isotropic" phases are complex structures composed of curved bilayers (Siegel, 1 9 8 6 ~ )"Isotropic" . phases are thought to represent poorly ordered 111 phase. The mechanism of the transitions between L,, 111,and HI1 phases has been the subject of debate (Siegel, 1984, 1986a-c; Verkleij, 1984). Recently, a kinetic theory of the phase transition mechanisms was developed (Siegel, 1986a-c) predicting the structure of the intermediates that form during these phase transitions. The transition to 111or HII phases is initiated by the formation of labile intermediates between apposed L, phase bilayers. In one class of systems, the labile intermediates rapidly convert the apposed bilayers into the HI1 phase. In the other class of systems, the labile intermediates are predicted to form a second, much more stable intermembrane structure known as an interlamellar attachment (ILA), as described by Siegel (1986b,c). ILAs are semitoroidal (automobile tire rimshaped) bilayer connections between apposed bilayer sheets; the diameter of the structures is a t least 12 nm. ILAs are predicted to assemble into the 111phase under appropriate conditions (Siegel, 1 9 8 6 ~ ) . In this study, we examined liposomal dispersions of dioleoylphosphatidylethanolamine (DOPE) and Nmethylated dioleoylphosphatidylethanolamine(DOPEMe) executing these phase transitions. DOPE is a system that is predicted to form very few ILAs, whereas DOPE-Me should form them in large numbers (Siegel, 1986b,c). MATERIALS AND METHODS Materials All phospholipids were purchased from Avanti Polar Lipids (Birmingham, AL), and were more than 99% pure as determined by two-dimensional thin-layer chromatography criteria; they were used without further purification. All salts and buffers were reagent grade. The water used for solutions was purified with a Milli-Q system (Millipore Co.) and had a n initial resistivity of ca. 18 MWcm. The lipids were suspended in buffers of the following compositions: DOPE in 100 mM NaC1, 20 mM glycine (pH 9.91, 0.1 mM EDTA; DOPEMe in 100 mM NaC1, 20 mM glycine (pH 9.5), 0.1 mM EDTA. Buffers used to produce the pH or magnesium
8
Y. TALMON ET AL
proceeded to different extents in different areas of the same grid. It was often necessary to delay blotting in order to observe later stages in the transition process. To study the transitions in the slower DOPE-Me system, for example, we sometimes had to wait 10 seconds before blotting to ensure the formation of lipid aggregates large enough to show nascent 111 domains. Vitrified specimens were transferred under liquid nitrogen to the “work-station’’ of a Gatan 626 coldstage, loaded into the stage, transferred and examined in a Hitachi H-500 TEM operating a t 100 kV. Holder Methods temperature was typically - 170°C. Images were reLipids were purchased as lyophilized powders and corded on Kodak SO-163 film processed for highest were dispersed in the relevant buffer by frequent electron speed in full strength D-19 developer, for 12 vortex mixing a t a temperature above the chain- minutes. To enhance contrast, images were taken a t melting transition of the lipid. This was followed by defocus of about 4 pm. Typical electron exposures were three freeze-thaw cycles (dry icei40”C water bath). in the range of 1-10 ke-/nm2. Liposomal dispersions were produced by extrusion RESULTS AND DISCUSSION through 0.1 or 0.2 pm Nuclepore filters (Nuclepore Figure 1is a cryo-TEM image of a vitrified specimen Corp., Pleasanton, CA) in a Lipex Bilmembranes (Vancouver, B.C.) extruder a t a nitrogen gas pressure of of DOPE liposomes a t 25°C and pH 9.9, conditions 200-600 psi. Some DOPE dispersions were produced by where the lipid is anionic and stable in the L, phase. detergent dialysis using octyl glucoside (Mimms et al., Embedded in the vitreous ice are unilamellar and 1981) in order to investigate the effect of high pressure oligolamellar liposomes, some of which are invaginated, and a few long tubular structures. These latter extrusion on liposome morphology. Specimens for cryo-TEM were prepared on holey- structures appear to form during high pressure extrucarbon films mounted on 200 mesh copper grids. Hole sion of the lipid dispersions to produce liposomes, as diameter was 1 to 10 pm. We used the CEVS as liposomes produced via detergent dialysis are all described by Bellare et al. (1988). Experiments were nearly spherical (data not shown). When the pH is less than ca. 8.5, DOPE can form the conducted below, above, or a t room temperature. Liquid nitrogen was used as the refrigerant in the cooling HII phase a t temperatures above ca. 5°C (e.g., Gruner module of the CEVS. (Dry ice-acetone is more conve- et al., 1988). Figure 2 shows the earliest structural nient as it lasts longer, but acetone dissolved the change visible within these dispersions when the pH is Styrofoam insulation very rapidly when splashed out- lowered by mixing drops of liposome dispersion and the side the cooling reservoir.) Air in the CEVS chamber appropriate buffer: the liposomes aggregate into was kept a t 90-95% relative humidity, as determined clumps. Figure 3 is a n image obtained later in this by a General Eastern model 800B hygrometer process, showing the formation of small HII domains equipped with a type 814 sensor. within these aggregates. The HIr domains are identiFor TRC-TEM, the CEVS was set to the appropriate fied by their enhanced optical-density, and by the fact temperature, and the air inside was saturated with that the lines in images of these domains are 8 nm water vapor using sponge wicks in the CEVS sample apart (the HI1 tube width in this sytem) (Gruner et al., chamber and a flow of humidified, filtered compressed 1988). Occasionally, bundles of HII tubes are visible in air. The liposome dispersions and buffer solutions were cross section. At later times, one finds areas consisting maintained a t the experimental temperature in a sep- only of large HI1 phase domains (Fig. 4). Qualitatively arate water bath. A 5 pl drop of liposome suspension the same results are achieved with bilayer-buffer (acwas placed on the grid mounted on the CEVS plunger. etate, pH 4.5) or nonpermeating buffer (TES, pH 7.4; or Then a second 5 pl drop of the appropriate pH or Mg2 bicine, pH 8.3).It is important to note that nascent HII concentration-jump buffer was mixed with the first tubes were never observed in isolated, nonapposed drop on the same side of the EM grid. This was found to bilayers: this indicates that the obligatory intermedibe superior to applying the drops to opposite sides of ates in this transition are interlamellar structures, as the grid. The grid was then blotted immediately, or predicted (Siege], 1984, 1986a). blotted after a time delay of several seconds. The DOPE-Me systems incubated a t temperatures ca. shortest interval between drop mixing and fixation 10°C below the equlibrium L,/HII phase transition that we could achieve was 3 seconds. In order to do this, temperature, TH. give rise to isotropic and inverted it is best to have an assistant standing by, to hand the cubic (111)phase (Ellens et al., 1989). I t was proposed pipets to the operator, trigger the plunger, and time the (Siege], 1986a-c) that this is due to the formation of operation. After thermal fixation, the procedure is metastable ILAs under these circumstances. Therefore, identical to other CEVS experiments. we examined DOPE-Me dispersions by TRC-TEM in an Although some steps in the phase transitions in attempt to observe these structures. these systems are quite rapid, the 3 second preparation The predicted ILA geometry is depicted in Figure 5A: time did not preclude observation of earlier stages in the structure is semitoroidal and cylindrically symmetthe process. Mixing did not appear to be uniform across rical about the vertical dashed axis. The lower limit to the grid, since the transitions often appeared to have the outer diameter of the “waist” of the structure is
ion concentration-jumps were made iso-osmotic to the suspending buffer in each case, using a vapor pressure osmometer (Wescor model #5500, Wescor Corp.). The pH-jump buffers had the following compositions for DOPE: ca. 80 mM NaCl plus either 40 mM bicine (pH 8.3), 40 mM TES (pH 7.41, or 40 mM acetate (pH 4.5). The Mg2 concentration-jump buffer for DOPE-Me was 20 mM NaC1, 60 mM MgCI2, 2 mM TES (pH 7.4). Liquid ethane (chemically-pure, Matheson, Inc.) a t its freezing point was used for specimen vitrification.
+
TRC-TEM
Fig. 1. TRC-TEM image of a DOPE liposomal dispersion at pH 9.9 and 25°C The liposomes were produced by high-pressure extrusion through 0.2 pm filters. This seems to produce numerous invaginated and spherical forms. Liposomes produced by detergent dialysis are all nearly spherical (data not shown). Bar equals 100 nm. Fig. 2. DOPE liposomal dispersion ca. 5 seconds after mixing with bicine buffer (pH 8.3) a t 25°C. The liposomes have aggregated. Bar equals 100 nm. Fig. 3. Aggregates of DOPE liposomes forming inclusions of HI[ phase ca 5 seconds after mixing with bicine buffer a t 15°C. The HII
9
domains are distinguished by their optical density, the 8 nm periodicity of the dark lines within them (approximately the HII tube diameter in this system), and by the occasional appearance of HII tubes in cross section (arrow). Bar equals 100 nm. Fig. 4. Large domain of HII phase in a DOPE dispersion 3 seconds after acidification with TES buffer (pH 7.4) at 25°C. The system forms these domains faster than a t 10.5"C because the temperature is higher, and thus further above the equilibrium L,/HII phase transition temperature of ca. 5°C. Bar equals 100 nm.
Y. TALMON ET AL
10
Figure 6 were obtained ca. 6 seconds after magnesium addition. ILAs are observed in the DOPE-Me a t the same temperature, solution conditions, and on the same time scale (seconds) a s membrane fusion is detected in the same system with fluorescence assays (Ellens et al., 1989). These results, taken together with the direct visualization of the fusion process (Fig. 6A), suggest that ILAs are the structures that mediate membrane fusion under these circumstances (Ellens et al., 1986; Siegel, 1986b; Siegel et al., 1989b). Systems like DOPE do not ordinarily form isotropic or inverted cubic phases. They are observed to do so when certain types of detergents are added (Wieslander et al., 1986). According to the kinetic theory (Siegel, 1986b,c), the detergents should greatly increase the ILA formation rate. If the 111 phase really is produced from ILAs, then these systems should exhibit Fig. 5. (A) Predicted geometry of an ILA (Siegel, 1986c), viewed ILA morphology. We did observe ILA morphology in in cross-section. The stippled-edge slabs represent lipid bilayers. (B) such a system. Figure 7 is a TRC-TEM image of a Expected appearance in transmission microscopy of an ILA when viewed with the electron beam perpendicular to the bilayer stack DOPE liposomal dispersion produced by dialysis of a (top) and parallel to the bilayer stack (bottom).The appearance of a DOPEioctyl glucoside mixture after the pH was refold in one of the bilayer membranes is also depicted. C,D: Flow in the duced to 7.4 a t 25°C. The octyl glucoside detergent was aqueous film caused by capillary film thinning prior to fixation can not completely removed. An HII phase domain (lower move folds in membranes. lLAs connecting adjacent membranes are tilted almost into the plane of the sample film when the fold reaches left), L, phase bilayers (liposomes, upper right), and them, resulting in the depicted morphology (as viewed from the top; incipient 111phase (large arrow; compare with Fig. 6A) upper part of figure). are shown. Individual ILAs are barely visible in profile at left (small arrows). A
estimated at 12 nm (Siegel, 1 9 8 6 ~ ) .The expected appearance of ILAs within a bilayer stack is indicated schematically in Figure 5B-D. When the bilayers are perpendicular to the electron beam, ILAs should appear as light-centered rings (top of Fig. 5B). When the bilayers are parallel to the beam, they should appear as reversed parentheses (bottom of Fig. 5B). Aqueous films in TRC-TEM samples undergo thinning under the influence of capillary forces. This induces a radially directed flow in the films. This flow can move the folded edges of membrane sheets as indicated in Figure 5B-D. ILAs that connect adjacent membranes will be reoriented by the edge of such a moving fold: the ILAs will be tilted into the plane of the sample film, yielding the morphology shown schematically in Figure 5C,D. La/III/HIItransitions in DOPE-Me can be induced by addition of magnesium ion a t pH 9.5 (Ellens et al., 1989). TRC-TEM images of liposomal samples mixed with magnesium buffer a t 50°C yield clear examples of the expected ILA morphology depicted in Figure 5. Examples of these TRC-TEM images are displayed in Figure 6. ILAs are visible from the top (light-centered rings approximately 15 nm in diameter; small arrows in Fig. 6A); from the side (reversed parentheses figures; arrows in Fig. 6B); and a t the margins of membrane folds (large arrow in 6A; compare with Fig. 5D.) Areas with numerous ILAs (superposition of rings in many successive stacked bilayers, viewed in projection) are also visible (arrowhead in Fig. 6A). The accumulation of large numbers of ILAs is thought to be the first step in 111phase formation (Siegel, 1 9 8 6 ~ )Elsewhere, . it is shown that 111 phases are assembled via formation of a n ordered lattice of ILAs (Siegel et al., 1988). In Figure 6A, liposomes seem to be fusing into larger structures via ILA formation. The TRC-TEM images in
CONCLUSIONS We have demonstrated the ability of time-resolved cryotransmission electron microscopy (TRC-TEM) to capture intermediate microstructures in dynamic processes, using the L, + HII phase transformation associated with phospholipid membrane fusion. The basic idea is to trigger a physical or chemical transformation on the grid and freeze the process a t different elapsed times from its onset. The results of a series of such experiments are images of the microstructure corresponding to the various steps in the process. TRC-TEM is one of the applications of on-the-grid processing. In the example presented here, the process was initiated by a pH or Mg+ concentration-jump induced by mixing a drop of the vesicular dispersion with a drop of a n appropriate buffer. The concept of TRC-TEM is currently used by one of us (Y.T.) and his colleagues to study three other systems. The first is the process of polymer gelation (Cohen, 1987; Cohen et al., 1989) where the trigger is a temperature drop or a concentration change by evaporation. The time scale in that case is several seconds. The solvents in the system studied so far have been benzene and benzyl alcohol. We have also begun a study of nucleation and growth of crystals from solution by this technique (Addadi, Weissbuch and Talmon, unpublished results), using supersaturation as the driving force for the formation of the new phase. In the latter systems the process is slower and, in some cases, may take up to 5 minutes. A third example involves U.V.-induced vesicle polymerization using a U.V. light source mounted within the CEVS chamber to polymerize the surfactant in vesicles, thus producing microstructural changes. The time scale in this case is seconds to minutes. At present TRC-TEM can be used to study processes
TRC-TEM
11
Fig. 6. DOPE-Me liposomal dispersions ca. 6 seconds after addition of 30 mM Mg” at 50°C. ILAs are visible viewed from the top (light-cer tered rings indicated by the small arrows in A, from the side (“reverset-l parentheses” indicated by the arrows in B)); and at the margins if membrane folds (indicated by the large arrow in A. The arrowhead in A indicates a region of many ILAs in a multibilayer stack. Liliosomes are fusing into larger structures via ILA formation at upper right. Bar equals 100 nm.
Fig. 7. TRC-TEM image of a liposomal dispersion produced by dialysis of a DOPEioctyl glucoside detergent mixture 3-5 seconds after the pH was reduced to 7.4 at 25°C. An HI1phase is visible at lower left (asterisk),liposomes fusing via ILA formation at the upper right and at arrowhead, and incipient isotropic or 111phase at the center (large arrow). Individual ILAs (small arrows) are also visible in a band of apposed bilayers. Bar equals 100 nm.
with time constants of several seconds or longer. Many processes of fundamental interest fall in this range. Paramc’ters such as temperature’ concentration’ and pH can be changed to bring the time Of most systems into the range approachable by the technique.
Bellare, J.R., Davis, H.T., Scriven, L.E., and Talmon, Y. (1988) Controlled environment vitrification system: An improved sample preparation technique. J. Electron Microsc. Tech., 10:87-111. Burns, J.L., and Talmon, Y. (1987) Cryo-TEM of micellar solutions. In: Proceedings of the 45th Annual EMSA Meeting. G.W. Bailey, ed. (San Francisco Press, San Francisco), pp. 498-499. Burns, J.L., Cohen, Y., and Talmon, Y. (1989) Determination of the Pm3n cubic phase in a surfactant-water system by cryo-TEM and SAXS. Science, submitted. Cohen, Y. (1987) Structure Formation in Solutions of Rigid Polymers
REFERENCES Adrian, M., Dubochet, J., Lepault, J., and McDowell, A.W. (1984) Cryoelwtron microscopy of viruses. Nature, 308:32-36.
12
Y. TALMON ET AL
Undergoing a Phase Transition. Ph.D. Dissertation, University of Massachusetts, Amherst. Cohen, Y., Talmon, Y., and Thomas, E.L. (1989) On the structure of poly(y-benzyl-L-glutamate) gels. In: Physical Networks. W. Burchard and S.B. Ross-Murphy, eds. Elsevier Applied Science Pub., Amsterdam, in press. Ellens, H., Bentz, J., and Szoka, F.C. (1986) Fusion of phosphatidylethanolamine-containing liposomes and the mechanism of the La/HIIphase transition. Biochemistry, 25:4141-4147. Ellens, H., Siegel, D.P., Alford, D,, Yeagle, P.L., Boni, L.T., Lis, L.J., Quinn, P.J., and Bentz, J. (1989) Membrane fusion and inverted phases. Biochemistry, in press. Gruner, S.M., Tate, M.W., Kirk, G.L., So, P.T.C., Turner, D.C., and Keane, D.T. (1988) X-ray diffraction study of the polymorphic behavior of N-methylated dioleoylphosphatidylethanolamine.Biochemistry, 27:2853 -2866. Mandelkow, E.-M., Rapp, R., and Mandelkow, E. (1986) Microtubule structure studied by quick freezing: Cryo-electron microscopy and freeze fracture. J. Microsc., 141:361-377. Miller, D.D., Bellare, J.R., Evans, D.F., Talmon, Y., and Ninham, B.W. (1987) On the meaning and structure of amphililic phases: Inference from video-enhanced microscopy and cryo-transmission electron microscopy. J. Phys. Chem., 91:674-685. Mimms, L.T., Zampighi, G., Nozaki, Y., Tanford, C., and Reynolds, J.A. (1981) Phospholipid vesicle formation and transmembrane protein incorporation using octyl glucoside. Biochemistry, 202333840.
Siegel, D.P. (1984) Inverted micellar structures in bilager membranes. Formation rates and half-lives. Biophys. J., 45:399-420. Siegel, D.P. (1986a) Inverted micellar intermediates and the transitions between lamellar, cubic and inverted hexagonal lipid phases. I. Mechanism of the L
