a stopped-flow/rapid-freezing machine with millisecond
TRANSCRIPT
A Stopped-flow/rapid-freezing machine with millisecond time
resolution to prepare intermediates in biochemical reactions for
electron microscopyJOURNAL OF ELECTRON MICROSCOPY TECHNIQUE
16:160-166 11990)
A Stopped-Flow/Rapid-Freezing Machine With Millisecond Time Resolution to Prepare Intermediates in Biochemical Reactions for Electron Microscopy THOMAS D. POLLARD, PAMELA MAUPIN, JOHN SINARD, AND HUGH E. HUXLEY Department of Cell Biology and Anatomy, The Johns Hopkins Medical School, Baltimore, Maryland 21205 (T.D.P., P.M., J.S.); Medical Research Council Laboratory of Molecular Biology, Cambridge, England CB2 2QH (H.E.H.)
KEY WORDS Stopped-flow, Rapid-freezing, Freeze-fracture, Electron microscopy, Rapid reactions
ABSTRACT We have developed an instrument capable of freezing transient intermediates in rapid biochemical reactions for subsequent freeze-fracturing, replication, and viewing by transmis- sion electron microscopy. The machine combines a rapid mixing unit similar to one widely used in chemical kinetics (Johnson, 1986) with a propane jet freezing unit previously used to prepare static samples for freeze-fracturing (Gilkey and Staehelin, 1986). The key element in the system is a unique thin-walled flow cell of copper that allows for injection and aging of the sample, followed by rapid freezing. During freeze-fracturing, a tangential cut is made along the wall of the flow cell to expose the sample for etching and replication. The dead time required for mixing and injection of the reactants into the flow cell is less than 5 ms. Electronic controls allow one to specify, on a millisecond time scale, any time above 5 ms between initiation of the reaction and quenching by rapid freezing.
INTRODUCTION There are many important chemical reactions, par-
ticularly in biology, in which mechanisms might be established if i t were possible to start and stop the re- actions on a millisecond time scale and to view the intermediates in the electron microscope. Some exam- ples include the assembly of viruses, fusion of mem- branes, and the production of motion by myosin and actin or by dynein and microtubules. In the classic work of Heuser et al. (1978), intermediates in exocytosis at the neuromuscular junction were captured by electrical stimulation of the nerve, followed by precisely timed impact freezing. The frozen samples were prepared for electron microscopy by freeze-fracturing or by freeze substitution, embedding, and thin sectioning.
More generally, i t would be useful to initiate such reactions by rapid mixing of the reactants, since rela- tively few processes can be initiated electrically. An example is a quenched-flow method pioneered by Rand et al. (1985) to investigate the fusion of lipid vesicles mixed with Ca2 to promote fusion. After mixing, the reactants traversed an aging line of variable length before the intermediates were sprayed into liquid pro- pane to stop the reaction. After freeze-fracturing and replication, the intermediates were studied by electron microscopy.
We have combined conventional rapid mixing stopped-flow procedures with propane jet rapid-freez- ing to improve the time resolution in the preparation of transient intermediates for electron microscopy. Fro- zen samples are freeze-fractured, etched, and repli- cated for viewing by transmission electron microscopy. The apparatus has a dead time of about 5 ms and mil- lisecond time resolution thereafter.
MATERIALS AND METHODS Overview
Reactions are initiated by driving the two reactants through a Berger-type mixer and into a thin-walled copper flow cell (Fig. 1). After a predetermined delay, the contents of the flow cell are rapidly frozen by a high-pressure jet of liquid propane cooled to -190°C. The frozen flow cell is removed quickly and stored in liquid nitrogen until the time for freeze-fracturing.
Mixing unit The mixing unit (Figs. 1, 2) is similar to one widely
used in chemical kinetics (Johnson, 1986). It consists of two Pressure-lok Series C gas syringes with Teflon plungers (Pierce Chemical Corp, Rockford, IL) con- nected to a 1.5-mm Mark I Berger-type mixer (Re- search Instruments and Manufacturing Co., San Diego, CA). The syringes are backloaded from external sy- ringes through 1.5-mm channels in the base. The tem- perature of the reactants is maintained by thermo- stated water circulating through the plastic base and the water jacket surrounding the internal syringes. The syringes are driven by a brass plate attached to a pneumatic cylinder (Dees Hydraulics Co., Zelienople, PA) pressurized to about 10 psi by nitrogen gas. Flow is initiated by opening the inlet solenoid valve of the
Received July 12, 1989; accepted in revised form October 12, 1989. Hugh E. Huxley's current address is Kosenstiel Center, Brandeis University,
Address reprint requests to Dr. Thomas D. Pollard, Department of Cell Biology Waltham, MA 02254.
and Anatomy, The Johns Ilopkins Medical School, Baltimore, MD 21205.
e 1990 WILEY-LISS. INC
RAPID-FREEZING MACHINE 161
Fig. 1. Mechanical drawings of the mixing unit. Left: Frontal view of a center section. The position of the mixer is superimposed. The photodiode assembly and water circulating channels are omitted for clarity. Right: Lateral view of a center section with several superimposed parts, including the internal syringes and the photodiode assembly. For the sake of clarity, the plastic support backing up the flow cell is not illustrated. See Figure 2G for illustration.
pneumatic cylinder and terminated without a mechan- ical stop by closing the solenoid when a pin that travels with the plunger driving the syringes triggers a pho- todiode detector (Fig. 2 0 . The resistance downstream from the syringes is sufficient to achieve an abrupt stop without detectable backflow. The distance between the start position and the optical stop determines the vol- ume delivered. The two syringes containing the reac- tants can be any size from 1 to 5 ml, so that various proportions can be mixed. Usually we mix equal vol-
umes from 2-ml syringes. The syringes are connected to the mixer by 1.5-mm channels drilled in the plastic base of the chamber. The mixer itself is pressure seated in a cylindrical cavity milled into the front of the plas- tic base (Figs. 1, 2E,F). The outlet of the mixer is ex- posed on the surface of the plastic base t o minimize the volume between the mixer and the flow cell. The whole mixing unit is attached to the freezing unit on hinges that allow one to tilt the mixing unit away from the freezing jets (Fig. 2A).
Fig. 2. Photographs of the mixing and freezing unit. A: Overview with mixing unit tilted back in the standby position with the flow cell away from the freezing jet. Solenoid (a) controls the level of liquid nitrogen cooling the propane cylinder. Solenoid (b) controls the pres- surization of the propane cylinder. Ports (c) are the exhaust from the liquid nitrogen dewar. Other details are illustrated more clearly in other photographs. H: Overview with the mixing unit in the vertical operational position. C: Back view of the mixing unit showing (d) the photodiode switch, ie) the activating pin on (f) the plunger, and (g) the screw that adjusts the height of the photodiode and thereby controls the volume of sample mixed and injected into the flow cell. D: Front
view of the mixing unit showing (h) the external syringes, (i) internal syringes, (j) quick-release flow cell holder, (k) flow cell, and (1) pro- pane jet with angle of spray marked on the top. E: Exit from the mixer showing (m) the O-ring that seals the flow cell against the mixer and (n) the complementary side of the quick-release holder showing the flow cell flange in the center. F: Quick-release holder with a flow cell (k) partially disengaged from the mixer. G: Closeup view of the quick release holder with the flow cell (k) in position in relationship to the propane jet (1). The right-hand jet is capped, so that the liquid pro- pane impacts the flow cell from one side only.
RAPID-FREEZING MACHINE 163
Fig. 3. Photographs of the flow cell. A: Alone. B Mounted on the modified Balzers freeze-fracture table. The flow cell is held in place by a metal clip and by virtue of the fact that the flange is inserted in a narrow slit cut into the table (arrow).
Flow cell The flow cell (Fig. 3A) is a flanged copper tube with
an internal diameter of 1.5 mm and a length of 10 mm. It is manufactured by the Servometer Corp. (Cedar Grove, NJ) by a proprietary process that involves elec- troplating the copper onto the surface of an electrically conductive plastic mandril. The mandril is then dis- solved in an organic solvent. The average wall thick- ness is 10 pm. The flow cells are labeled with a mark- ing pen.
Flow cell holder The flow cell is supported by a quick-release holder
made of plastic (Figs. 1, 2D-GI. The flow cell is in- serted from the back into a 1.5-mm hole in the holder. When the quick-release holder is attached to the mix- ing unit, the flange on the proximal end of the flow cell is pressed between an O-ring seated around the outlet of the mixer and the plastic back of the holder. This forms a reliable seal for the delivery of liquid from the mixer into and through the flow cell. The part of the flow cell projecting through the quick release holder is sup- ported on one side by a plastic block with a close-fitting cylindrical groove. This support prevents deformation of the flow cell when it is impacted by the propane jet. Once freezing has been completed, one can rapidly re- move the holder and flow cell by simply pulling the handle of the holder away from the mixing unit and immersing all but the handle of the holder in a liquid nitrogen bath. One can then push the flow cell back through the holder into the liquid nitrogen. With ex- perience, the interval between freezing and transfer of the sample to liquid nitrogen can be <2 s.
Freezer The propane jet freezer (Fig. 2A,B) was purchased
from the University of Colorado. It is a modification of a design of Mueller et al. (1980) by Gilkey and Staehe- lin (1986). A commercial version of the freezer is now available from the Research Manufacturing Corp. (Tucson, AZ). The propane tank is cooled with liquid nitrogen and pressurized for freezing with nitrogen gas
a t 150 psi. Upon pressurizing the liquid propane tank, the propane flows through a brass pipe and a 1-mm orifice onto the object to be frozen. The freezer was modified in several ways. One of the two jets is blocked off by a rubber cap (Fig. 2E). The functional jet is turned a t an angle of about 45" toward the mixing unit, so that i t impacts the middle of the labeled flow cell. One can thereby be assured of fracturing the most rap- idly frozen area, which is at the point of impact of the single jet. A ballast tank consisting of a 15-cm length of steel pipe 11.5 cm in diameter with caps and connectors on both ends is inserted between the gas supply and the freezing unit to minimize drops in pressure during freezing. Neoprene rubber insulation 2 cm thick is glued to the top of the freezer to avoid icing during long experiments in humid weather.
Control units The control unit (Fig. 4) was designed and manufac-
tured to our specifications by Biomedical Engineering Systems, Inc. (Baltimore, MD). The four modules that comprise the unit open and close the various solenoid valves in a precisely timed sequence and monitor the temperature in the propane reservoir. Audible count- down timers allow one person to operate the machine without having to watch the timer. The sample injec- tion and freezing sequence is initiated manually by pressing a start switch that activates an adjustable timer (allowing the operator to get in place a t the mixerifreezer). Then the solenoid valving the pneu- matic cylinder opens, driving the syringes to mix the reactants and fill the flow cell. When a preset volume is mixed, the pin attached to the plunger activates the photodiode, closing the solenoid valving the pneumatic cylinder and triggering a timer. After a preset delay of 1 ms to 9.999 s for the sample in the flow cell to age, a second solenoid valve between the high-pressure nitro- gen tank and the propane tank opens for a selected time, propelling the cold propane through the jet. Man- ual override switches allow one to bypass mixing, to freeze static samples, or to clear or fill the propane lines.
164 T.D. POLLARD ET AL.
Fig. 4. Control unit. Electronic components orchestrate the timing and duration of mixing of the reactants, injection into the flow cell, aging in the flow cell, and freezing. The temperature of the propane is also monitored
Operation Place the whole freezing unit in a well ventilated
fume hood to vent the propane and to preclude explo- sions. To prepare for a freezing run, charge the freezer with liquid propane and cool i t to -190°C with liquid nitrogen in the surrounding Dewar. Isolate the freez- ing jets with a thermal blanket consisting of several layers of flexible plastic foam sheeting. Load the reac- tants into the internal syringes. With the mixing unit tilted away from the freezing jet and no flow cell or its holder in place, carry out a trial mixing run. This will fill the lines with reactants. Then set the desired aging time and the volumes to be mixed, usually about 100 p1 per syringe. Load a clean flow cell in the quick-release holder, and attach i t to the mixer. Activate the jet for about 0.5 s under the thermal blanket to clear the pro- pane line of frost and precool the jet. Press the start button. During the short delay, usually 3 s, remove the thermal blanket and tilt the mixing unit into its ver- tical (operational) position with the flow cell next to the freezing jet. Observe the automatic cycle of mixing, injection, aging, and freezing. We usually set the jet to flow for 1 s. Just as the jet stops, quickly remove the flow cell holder (still wet with liquid propane), im- merse i t in liquid nitrogen, and push the distal end of the flow cell with a cooled probe to dislodge the flow cell into the liquid nitrogen. For samples with very long aging times, inject the sample into the flow cell with the mixing unit tilted back and the jet covered with the thermal blanket. With about 1 s remaining before freezing, remove the blanket and position the flow cell adjacent to the jet. This procedure prevents the sample from cooling or freezing during the delay time due to the proximity of the cold jet.
Freeze-fracturing The frozen flow cell is mounted under liquid nitrogen
on a modified Balzers table (see Fig. 3B) and attached to the precooled rotary stage of a Balzers freeze frac- ture machine. After evacuation of the chamber, the temperature of the stage is raised to about -107°C. A
tangential cut is carefully made along the long axis of the flow cell using a commercial, stainless steel razor blade as the knife. With practice, a section of copper wall and underlying ice about 5 mm long and 0.7 mm wide can be removed with one cut. This exposes the outer layers of frozen contents. After etching a t ~ 107°C for 4 min, the exposed surface is rotary shadowed with platinum a t an angle of 30" and backed with a layer of carbon applied a t an angle of 80". The flow cell is re- moved from the stage and the replica floated onto the surface of water. The replica is recovered and viewed in a transmission electron microscope.
RESULTS AND DISCUSSION The flow cell presented the major technical difficulty
in designing the stopped-flowhapid-freezing unit. We tested a number of different designs and materials be- fore settling on the flanged copper flow cell. Two-part flow cells were ideal for freeze-fracturing, since one side could be removed to expose the frozen contents, but most of these cells leaked very slightly, so that air was sucked inside during the injection of the reactants. Any air bubbles in the frozen specimen make i t difficult or impossible to freeze-fracture successfully. We initially experimented with flanged flow cells made of nickel, since the manufacturer was not confident that copper would be strong enough. The nickel proved next to im- possible to cut a t -100°C. The copper tubes with 10-pm walls are, in fact, more than strong enough for this application. Furthermore, the high thermal conductiv- ity of copper should favor rapid freezing.
To date most of our work has been with actin fila- ments and myosin head fragments, in a effort to estab- lish the structure of the weakly bound transient inter- mediates in the actomyosin ATPase cycle (Pollard et al., 1990). The actin filaments and attached myosin heads are preserved well in the superficial layers of the frozen samples (Fig. 5). High-quality freezing, indi- cated by the absence of visible ice crystals, is confined to the outermost 1 p m of the sample, as expected from theoretical considerations (Kopstad and Elgsaeter,
RAPID-FREEZING MACHINE 165
Fig. 5 . Electron micrograph of a replica of mixture of actin filaments and muscle myosin subfrag- ment-1 in 0.5 mM ATP reacted for 50 ms, frozen in a flow cell, and freeze-fractured. The random small bumps on the actin filaments are individual myosin heads captured during the active hydrolysis of ATP. Scale bar is 100 nm.
1982). Areas deeper than 1 pm from the surface have large ice crystals that destroy the architecture of the specimen. The limitation of good preservation to this outer layer is not a serious problem, because there is a large area of replica along the edges of the cut flow cell, where the well-preserved superficial layer is exposed.
The adequacy of cryopreservation by any freezing and replication method is always a concern, so we have carefully evaluated the fidelity of the molecular struc- ture retained in our replicas. As established by others (Heuser and Kirschner, 1980; Heuser, 1983a,b) and confirmed by our results, it is possible to preserved identifiable protein molecules by rapid freezing, freeze- fracturing, etching, and replication. Individual myosin heads and the 5.5-nm subunit repeat along actin fila- ments are routinely imaged in our specimens (Fig. 5). This means that any ice crystals formed during freez- ing are too small to image and that any recrystalliza- tion of the ice during fracturing and etching does not produce crystals large enough to distort the specimen. We are confident that we could detect such problems with ice crystals, since such crystals are obvious deep
in these specimens, where they undoubtedly arise due to slower freezing away from the surface. We routinely use 10% methanol (Heuser and Kirschner, 1980) in our sample buffer as a cryoprotectant, but the freezing in aqueous buffers without methanol is adequate to ob- tain very similar results.
We hope this method will be useful for other inves- tigators of rapid reactions in which the intermediates are large enough to be visualized by electron micros- copy. Some users may require very high time resolu- tion. This can be achieved by freezing without a delay immediately after filling the flow cell. Then the linear distance from the proximal to distal end of the frozen flow cell will represent reaction times of 5-10 ms.
ACKNOWLEDGMENTS This work was started a t the MRC Laboratory of
Molecular Biology during 1984 while TDP was a Guggenheim Fellow at Churchhill College, Cambridge. Our colleague, Dr. Alan G. Weeds, contributed valu- able advice a t that time. The work was supported by the Laboratory of Molecular Biology and by research
166 T.D. POLLARD ET AL
grants to TDP from the Muscular Dystrophy Associa- tion and the National Institutes of Health (GM-26132). We are indebted to Dr. Kenneth Johnson of Pennsyl- vania State University for advice on the design of the mixing unit, to Chris Rayburn of the Laboratory of Molecular Biology for helping with the design and fab- rication of the apparatus, and to Howard Conner of the Johns Hopkins Medical School Department of Biomed- ical Engineering for helping with the design and fab- rication of the flow cell holder. We also thank Dr. Ri- chard Johns of the Johns Hopkins Medical School for referring us to the Servometer Corporation, who finally made a workable flow cell for the unit.
REFERENCES Gilkey, J.C., and Staehelin, L.A. (1986) Advances in ultrarapid freez-
ing for the preservation of cellular ultrastructure. J. Electron Mi- crosc. Technique, 3:177-210.
Heuser, J.E. (1983a) Structure of the myosin crossbridge lattice in insect flight muscle. J. Mol. Biol., 169:123-154.
Heuser, J.E. i 1983b) Procedure for freeze-drying molecules adsorbed to mica flakes. J . Mol. Biol., 169:155-195.
Heuser, J.E., and Kirschner, M. (1980) Filament organization re- vealed in platinum replicas of freeze-dried cytoskeletons. J. Cell Biol., 86:212-234.
Heuser, J.E., Reese, T.S., Dennis, M.J., Jan , L.Y., and Evans, L. (1978) Synaptic vesicle exocytosis captured by quick-freezing and correlated with quanta1 transmitter release. J. Cell Biol., 81275- 300.
Johnson, K.A. (1986) Rapid kinetic analysis of mechanochemical ATPases. Methods Enzymol., 134:677-705.
Kopstad, G., and Elgsaeter, A. (1982) Theoretical analysis of speci- men cooling rate during impact freezing and liquid-jet freezing of freeze-etch specimens. Biophys. J., 40:163-170.
Muelier, M., Meister, N., and Moor, H. (19801 Freezing in a propane jet and its application to freeze-fracturing. J. Microsc. (Wein), 36: 129-140.
Pollard, T.D., Bhandari, D., Maupin, P., Weeds, A.G., and Zot, H. (1990) Electron microscopic visualization of the weakly bound in- termediates in the actomyosin ATPase cycle. Submitted for publi- cation.
Rand, R.P., Kachar, B., and Reese, T.S. (1985) Dynamic morphology of calcium-induced interactions between phosphatidylserine vesicles. Biophys. J., 47:483-490.
A Stopped-Flow/Rapid-Freezing Machine With Millisecond Time Resolution to Prepare Intermediates in Biochemical Reactions for Electron Microscopy THOMAS D. POLLARD, PAMELA MAUPIN, JOHN SINARD, AND HUGH E. HUXLEY Department of Cell Biology and Anatomy, The Johns Hopkins Medical School, Baltimore, Maryland 21205 (T.D.P., P.M., J.S.); Medical Research Council Laboratory of Molecular Biology, Cambridge, England CB2 2QH (H.E.H.)
KEY WORDS Stopped-flow, Rapid-freezing, Freeze-fracture, Electron microscopy, Rapid reactions
ABSTRACT We have developed an instrument capable of freezing transient intermediates in rapid biochemical reactions for subsequent freeze-fracturing, replication, and viewing by transmis- sion electron microscopy. The machine combines a rapid mixing unit similar to one widely used in chemical kinetics (Johnson, 1986) with a propane jet freezing unit previously used to prepare static samples for freeze-fracturing (Gilkey and Staehelin, 1986). The key element in the system is a unique thin-walled flow cell of copper that allows for injection and aging of the sample, followed by rapid freezing. During freeze-fracturing, a tangential cut is made along the wall of the flow cell to expose the sample for etching and replication. The dead time required for mixing and injection of the reactants into the flow cell is less than 5 ms. Electronic controls allow one to specify, on a millisecond time scale, any time above 5 ms between initiation of the reaction and quenching by rapid freezing.
INTRODUCTION There are many important chemical reactions, par-
ticularly in biology, in which mechanisms might be established if i t were possible to start and stop the re- actions on a millisecond time scale and to view the intermediates in the electron microscope. Some exam- ples include the assembly of viruses, fusion of mem- branes, and the production of motion by myosin and actin or by dynein and microtubules. In the classic work of Heuser et al. (1978), intermediates in exocytosis at the neuromuscular junction were captured by electrical stimulation of the nerve, followed by precisely timed impact freezing. The frozen samples were prepared for electron microscopy by freeze-fracturing or by freeze substitution, embedding, and thin sectioning.
More generally, i t would be useful to initiate such reactions by rapid mixing of the reactants, since rela- tively few processes can be initiated electrically. An example is a quenched-flow method pioneered by Rand et al. (1985) to investigate the fusion of lipid vesicles mixed with Ca2 to promote fusion. After mixing, the reactants traversed an aging line of variable length before the intermediates were sprayed into liquid pro- pane to stop the reaction. After freeze-fracturing and replication, the intermediates were studied by electron microscopy.
We have combined conventional rapid mixing stopped-flow procedures with propane jet rapid-freez- ing to improve the time resolution in the preparation of transient intermediates for electron microscopy. Fro- zen samples are freeze-fractured, etched, and repli- cated for viewing by transmission electron microscopy. The apparatus has a dead time of about 5 ms and mil- lisecond time resolution thereafter.
MATERIALS AND METHODS Overview
Reactions are initiated by driving the two reactants through a Berger-type mixer and into a thin-walled copper flow cell (Fig. 1). After a predetermined delay, the contents of the flow cell are rapidly frozen by a high-pressure jet of liquid propane cooled to -190°C. The frozen flow cell is removed quickly and stored in liquid nitrogen until the time for freeze-fracturing.
Mixing unit The mixing unit (Figs. 1, 2) is similar to one widely
used in chemical kinetics (Johnson, 1986). It consists of two Pressure-lok Series C gas syringes with Teflon plungers (Pierce Chemical Corp, Rockford, IL) con- nected to a 1.5-mm Mark I Berger-type mixer (Re- search Instruments and Manufacturing Co., San Diego, CA). The syringes are backloaded from external sy- ringes through 1.5-mm channels in the base. The tem- perature of the reactants is maintained by thermo- stated water circulating through the plastic base and the water jacket surrounding the internal syringes. The syringes are driven by a brass plate attached to a pneumatic cylinder (Dees Hydraulics Co., Zelienople, PA) pressurized to about 10 psi by nitrogen gas. Flow is initiated by opening the inlet solenoid valve of the
Received July 12, 1989; accepted in revised form October 12, 1989. Hugh E. Huxley's current address is Kosenstiel Center, Brandeis University,
Address reprint requests to Dr. Thomas D. Pollard, Department of Cell Biology Waltham, MA 02254.
and Anatomy, The Johns Ilopkins Medical School, Baltimore, MD 21205.
e 1990 WILEY-LISS. INC
RAPID-FREEZING MACHINE 161
Fig. 1. Mechanical drawings of the mixing unit. Left: Frontal view of a center section. The position of the mixer is superimposed. The photodiode assembly and water circulating channels are omitted for clarity. Right: Lateral view of a center section with several superimposed parts, including the internal syringes and the photodiode assembly. For the sake of clarity, the plastic support backing up the flow cell is not illustrated. See Figure 2G for illustration.
pneumatic cylinder and terminated without a mechan- ical stop by closing the solenoid when a pin that travels with the plunger driving the syringes triggers a pho- todiode detector (Fig. 2 0 . The resistance downstream from the syringes is sufficient to achieve an abrupt stop without detectable backflow. The distance between the start position and the optical stop determines the vol- ume delivered. The two syringes containing the reac- tants can be any size from 1 to 5 ml, so that various proportions can be mixed. Usually we mix equal vol-
umes from 2-ml syringes. The syringes are connected to the mixer by 1.5-mm channels drilled in the plastic base of the chamber. The mixer itself is pressure seated in a cylindrical cavity milled into the front of the plas- tic base (Figs. 1, 2E,F). The outlet of the mixer is ex- posed on the surface of the plastic base t o minimize the volume between the mixer and the flow cell. The whole mixing unit is attached to the freezing unit on hinges that allow one to tilt the mixing unit away from the freezing jets (Fig. 2A).
Fig. 2. Photographs of the mixing and freezing unit. A: Overview with mixing unit tilted back in the standby position with the flow cell away from the freezing jet. Solenoid (a) controls the level of liquid nitrogen cooling the propane cylinder. Solenoid (b) controls the pres- surization of the propane cylinder. Ports (c) are the exhaust from the liquid nitrogen dewar. Other details are illustrated more clearly in other photographs. H: Overview with the mixing unit in the vertical operational position. C: Back view of the mixing unit showing (d) the photodiode switch, ie) the activating pin on (f) the plunger, and (g) the screw that adjusts the height of the photodiode and thereby controls the volume of sample mixed and injected into the flow cell. D: Front
view of the mixing unit showing (h) the external syringes, (i) internal syringes, (j) quick-release flow cell holder, (k) flow cell, and (1) pro- pane jet with angle of spray marked on the top. E: Exit from the mixer showing (m) the O-ring that seals the flow cell against the mixer and (n) the complementary side of the quick-release holder showing the flow cell flange in the center. F: Quick-release holder with a flow cell (k) partially disengaged from the mixer. G: Closeup view of the quick release holder with the flow cell (k) in position in relationship to the propane jet (1). The right-hand jet is capped, so that the liquid pro- pane impacts the flow cell from one side only.
RAPID-FREEZING MACHINE 163
Fig. 3. Photographs of the flow cell. A: Alone. B Mounted on the modified Balzers freeze-fracture table. The flow cell is held in place by a metal clip and by virtue of the fact that the flange is inserted in a narrow slit cut into the table (arrow).
Flow cell The flow cell (Fig. 3A) is a flanged copper tube with
an internal diameter of 1.5 mm and a length of 10 mm. It is manufactured by the Servometer Corp. (Cedar Grove, NJ) by a proprietary process that involves elec- troplating the copper onto the surface of an electrically conductive plastic mandril. The mandril is then dis- solved in an organic solvent. The average wall thick- ness is 10 pm. The flow cells are labeled with a mark- ing pen.
Flow cell holder The flow cell is supported by a quick-release holder
made of plastic (Figs. 1, 2D-GI. The flow cell is in- serted from the back into a 1.5-mm hole in the holder. When the quick-release holder is attached to the mix- ing unit, the flange on the proximal end of the flow cell is pressed between an O-ring seated around the outlet of the mixer and the plastic back of the holder. This forms a reliable seal for the delivery of liquid from the mixer into and through the flow cell. The part of the flow cell projecting through the quick release holder is sup- ported on one side by a plastic block with a close-fitting cylindrical groove. This support prevents deformation of the flow cell when it is impacted by the propane jet. Once freezing has been completed, one can rapidly re- move the holder and flow cell by simply pulling the handle of the holder away from the mixing unit and immersing all but the handle of the holder in a liquid nitrogen bath. One can then push the flow cell back through the holder into the liquid nitrogen. With ex- perience, the interval between freezing and transfer of the sample to liquid nitrogen can be <2 s.
Freezer The propane jet freezer (Fig. 2A,B) was purchased
from the University of Colorado. It is a modification of a design of Mueller et al. (1980) by Gilkey and Staehe- lin (1986). A commercial version of the freezer is now available from the Research Manufacturing Corp. (Tucson, AZ). The propane tank is cooled with liquid nitrogen and pressurized for freezing with nitrogen gas
a t 150 psi. Upon pressurizing the liquid propane tank, the propane flows through a brass pipe and a 1-mm orifice onto the object to be frozen. The freezer was modified in several ways. One of the two jets is blocked off by a rubber cap (Fig. 2E). The functional jet is turned a t an angle of about 45" toward the mixing unit, so that i t impacts the middle of the labeled flow cell. One can thereby be assured of fracturing the most rap- idly frozen area, which is at the point of impact of the single jet. A ballast tank consisting of a 15-cm length of steel pipe 11.5 cm in diameter with caps and connectors on both ends is inserted between the gas supply and the freezing unit to minimize drops in pressure during freezing. Neoprene rubber insulation 2 cm thick is glued to the top of the freezer to avoid icing during long experiments in humid weather.
Control units The control unit (Fig. 4) was designed and manufac-
tured to our specifications by Biomedical Engineering Systems, Inc. (Baltimore, MD). The four modules that comprise the unit open and close the various solenoid valves in a precisely timed sequence and monitor the temperature in the propane reservoir. Audible count- down timers allow one person to operate the machine without having to watch the timer. The sample injec- tion and freezing sequence is initiated manually by pressing a start switch that activates an adjustable timer (allowing the operator to get in place a t the mixerifreezer). Then the solenoid valving the pneu- matic cylinder opens, driving the syringes to mix the reactants and fill the flow cell. When a preset volume is mixed, the pin attached to the plunger activates the photodiode, closing the solenoid valving the pneumatic cylinder and triggering a timer. After a preset delay of 1 ms to 9.999 s for the sample in the flow cell to age, a second solenoid valve between the high-pressure nitro- gen tank and the propane tank opens for a selected time, propelling the cold propane through the jet. Man- ual override switches allow one to bypass mixing, to freeze static samples, or to clear or fill the propane lines.
164 T.D. POLLARD ET AL.
Fig. 4. Control unit. Electronic components orchestrate the timing and duration of mixing of the reactants, injection into the flow cell, aging in the flow cell, and freezing. The temperature of the propane is also monitored
Operation Place the whole freezing unit in a well ventilated
fume hood to vent the propane and to preclude explo- sions. To prepare for a freezing run, charge the freezer with liquid propane and cool i t to -190°C with liquid nitrogen in the surrounding Dewar. Isolate the freez- ing jets with a thermal blanket consisting of several layers of flexible plastic foam sheeting. Load the reac- tants into the internal syringes. With the mixing unit tilted away from the freezing jet and no flow cell or its holder in place, carry out a trial mixing run. This will fill the lines with reactants. Then set the desired aging time and the volumes to be mixed, usually about 100 p1 per syringe. Load a clean flow cell in the quick-release holder, and attach i t to the mixer. Activate the jet for about 0.5 s under the thermal blanket to clear the pro- pane line of frost and precool the jet. Press the start button. During the short delay, usually 3 s, remove the thermal blanket and tilt the mixing unit into its ver- tical (operational) position with the flow cell next to the freezing jet. Observe the automatic cycle of mixing, injection, aging, and freezing. We usually set the jet to flow for 1 s. Just as the jet stops, quickly remove the flow cell holder (still wet with liquid propane), im- merse i t in liquid nitrogen, and push the distal end of the flow cell with a cooled probe to dislodge the flow cell into the liquid nitrogen. For samples with very long aging times, inject the sample into the flow cell with the mixing unit tilted back and the jet covered with the thermal blanket. With about 1 s remaining before freezing, remove the blanket and position the flow cell adjacent to the jet. This procedure prevents the sample from cooling or freezing during the delay time due to the proximity of the cold jet.
Freeze-fracturing The frozen flow cell is mounted under liquid nitrogen
on a modified Balzers table (see Fig. 3B) and attached to the precooled rotary stage of a Balzers freeze frac- ture machine. After evacuation of the chamber, the temperature of the stage is raised to about -107°C. A
tangential cut is carefully made along the long axis of the flow cell using a commercial, stainless steel razor blade as the knife. With practice, a section of copper wall and underlying ice about 5 mm long and 0.7 mm wide can be removed with one cut. This exposes the outer layers of frozen contents. After etching a t ~ 107°C for 4 min, the exposed surface is rotary shadowed with platinum a t an angle of 30" and backed with a layer of carbon applied a t an angle of 80". The flow cell is re- moved from the stage and the replica floated onto the surface of water. The replica is recovered and viewed in a transmission electron microscope.
RESULTS AND DISCUSSION The flow cell presented the major technical difficulty
in designing the stopped-flowhapid-freezing unit. We tested a number of different designs and materials be- fore settling on the flanged copper flow cell. Two-part flow cells were ideal for freeze-fracturing, since one side could be removed to expose the frozen contents, but most of these cells leaked very slightly, so that air was sucked inside during the injection of the reactants. Any air bubbles in the frozen specimen make i t difficult or impossible to freeze-fracture successfully. We initially experimented with flanged flow cells made of nickel, since the manufacturer was not confident that copper would be strong enough. The nickel proved next to im- possible to cut a t -100°C. The copper tubes with 10-pm walls are, in fact, more than strong enough for this application. Furthermore, the high thermal conductiv- ity of copper should favor rapid freezing.
To date most of our work has been with actin fila- ments and myosin head fragments, in a effort to estab- lish the structure of the weakly bound transient inter- mediates in the actomyosin ATPase cycle (Pollard et al., 1990). The actin filaments and attached myosin heads are preserved well in the superficial layers of the frozen samples (Fig. 5). High-quality freezing, indi- cated by the absence of visible ice crystals, is confined to the outermost 1 p m of the sample, as expected from theoretical considerations (Kopstad and Elgsaeter,
RAPID-FREEZING MACHINE 165
Fig. 5 . Electron micrograph of a replica of mixture of actin filaments and muscle myosin subfrag- ment-1 in 0.5 mM ATP reacted for 50 ms, frozen in a flow cell, and freeze-fractured. The random small bumps on the actin filaments are individual myosin heads captured during the active hydrolysis of ATP. Scale bar is 100 nm.
1982). Areas deeper than 1 pm from the surface have large ice crystals that destroy the architecture of the specimen. The limitation of good preservation to this outer layer is not a serious problem, because there is a large area of replica along the edges of the cut flow cell, where the well-preserved superficial layer is exposed.
The adequacy of cryopreservation by any freezing and replication method is always a concern, so we have carefully evaluated the fidelity of the molecular struc- ture retained in our replicas. As established by others (Heuser and Kirschner, 1980; Heuser, 1983a,b) and confirmed by our results, it is possible to preserved identifiable protein molecules by rapid freezing, freeze- fracturing, etching, and replication. Individual myosin heads and the 5.5-nm subunit repeat along actin fila- ments are routinely imaged in our specimens (Fig. 5). This means that any ice crystals formed during freez- ing are too small to image and that any recrystalliza- tion of the ice during fracturing and etching does not produce crystals large enough to distort the specimen. We are confident that we could detect such problems with ice crystals, since such crystals are obvious deep
in these specimens, where they undoubtedly arise due to slower freezing away from the surface. We routinely use 10% methanol (Heuser and Kirschner, 1980) in our sample buffer as a cryoprotectant, but the freezing in aqueous buffers without methanol is adequate to ob- tain very similar results.
We hope this method will be useful for other inves- tigators of rapid reactions in which the intermediates are large enough to be visualized by electron micros- copy. Some users may require very high time resolu- tion. This can be achieved by freezing without a delay immediately after filling the flow cell. Then the linear distance from the proximal to distal end of the frozen flow cell will represent reaction times of 5-10 ms.
ACKNOWLEDGMENTS This work was started a t the MRC Laboratory of
Molecular Biology during 1984 while TDP was a Guggenheim Fellow at Churchhill College, Cambridge. Our colleague, Dr. Alan G. Weeds, contributed valu- able advice a t that time. The work was supported by the Laboratory of Molecular Biology and by research
166 T.D. POLLARD ET AL
grants to TDP from the Muscular Dystrophy Associa- tion and the National Institutes of Health (GM-26132). We are indebted to Dr. Kenneth Johnson of Pennsyl- vania State University for advice on the design of the mixing unit, to Chris Rayburn of the Laboratory of Molecular Biology for helping with the design and fab- rication of the apparatus, and to Howard Conner of the Johns Hopkins Medical School Department of Biomed- ical Engineering for helping with the design and fab- rication of the flow cell holder. We also thank Dr. Ri- chard Johns of the Johns Hopkins Medical School for referring us to the Servometer Corporation, who finally made a workable flow cell for the unit.
REFERENCES Gilkey, J.C., and Staehelin, L.A. (1986) Advances in ultrarapid freez-
ing for the preservation of cellular ultrastructure. J. Electron Mi- crosc. Technique, 3:177-210.
Heuser, J.E. (1983a) Structure of the myosin crossbridge lattice in insect flight muscle. J. Mol. Biol., 169:123-154.
Heuser, J.E. i 1983b) Procedure for freeze-drying molecules adsorbed to mica flakes. J . Mol. Biol., 169:155-195.
Heuser, J.E., and Kirschner, M. (1980) Filament organization re- vealed in platinum replicas of freeze-dried cytoskeletons. J. Cell Biol., 86:212-234.
Heuser, J.E., Reese, T.S., Dennis, M.J., Jan , L.Y., and Evans, L. (1978) Synaptic vesicle exocytosis captured by quick-freezing and correlated with quanta1 transmitter release. J. Cell Biol., 81275- 300.
Johnson, K.A. (1986) Rapid kinetic analysis of mechanochemical ATPases. Methods Enzymol., 134:677-705.
Kopstad, G., and Elgsaeter, A. (1982) Theoretical analysis of speci- men cooling rate during impact freezing and liquid-jet freezing of freeze-etch specimens. Biophys. J., 40:163-170.
Muelier, M., Meister, N., and Moor, H. (19801 Freezing in a propane jet and its application to freeze-fracturing. J. Microsc. (Wein), 36: 129-140.
Pollard, T.D., Bhandari, D., Maupin, P., Weeds, A.G., and Zot, H. (1990) Electron microscopic visualization of the weakly bound in- termediates in the actomyosin ATPase cycle. Submitted for publi- cation.
Rand, R.P., Kachar, B., and Reese, T.S. (1985) Dynamic morphology of calcium-induced interactions between phosphatidylserine vesicles. Biophys. J., 47:483-490.