Membrane retrieval following exocytosis in synaptic terminals is fast and compensatory, however, little is known about the factors that regulate or contribute to this special form of endocytosis. We used whole-terminal capacitance measurements to examine the effect of hydrostatic pressure on compensatory endocytosis in single synaptic terminals of retinal bipolar neurons. We report that a small increase in hydrostatic pressure reversibly inhibits compensatory endocytosis. Elevation in hydrostatic pressure does not block all membrane retrieval, however. A small, fast component of endocytosis persists, while a slower component is inhibited. When the hydrostatic pressure is then stepped back to a near-neutral setting, an even slower form of endocytosis is observed that restores the resting membrane capacitance to baseline. Thus even when endocytosis is temporally uncoupled from calcium entry and exocytosis, it can still be compensatory, indicating that presynaptic surface area is highly regulated. Our results suggest that at least two distinct mechanisms of membrane retrieval contribute to compensatory endocytosis. Given its dramatic inhibitory effect on membrane retrieval, we suggest that hydrostatic pressure be carefully controlled when studying endocytosis in the whole cell recording configuration.
In synaptic terminals of retinal bipolar neurons, membrane retrieval rapidly follows exocytosis and restores the membrane surface area, and thus membrane capacitance, to resting levels. This form of endocytosis, termed “compensatory endocytosis,” is typically completed within seconds following calcium-dependent fusion of the releasable pool of synaptic vesicles (von Gersdorff and Matthews 1994a). Despite the importance of compensatory endocytosis for maintenance of nerve terminal surface area, relatively little is known about the mechanisms that underlie this particular form of membrane retrieval. For example, it is not clear to what extent the multiple kinetic components of retrieval observed under various experimental conditions represent mechanistically distinct pathways of membrane retrieval or the activity-dependent modification of a single mechanism.
Compensatory endocytosis in synaptic terminals can be inhibited by an elevation in cytosolic calcium (von Gersdorff and Matthews 1994b) or by interference with ATP function (Heidelberger 2001a). However, to resolve the issue of whether multiple mechanisms of membrane retrieval contribute to compensatory endocytosis, an approach is needed that allows the kinetic components of compensatory endocytosis to be separated in a manner that is independent of activity. Increased membrane tension has been reported to impair the ability to retrieve membrane in other types of animal cells and plant cells (for review, see Apodaca 2002; Dai and Sheetz 1995; Morris and Homann 2001). Therefore we investigated whether intracellular hydrostatic pressure alters the ability of synaptic terminals to rapidly and completely retrieve membrane following exocytosis. We employed time-resolved membrane capacitance measurements to monitor membrane addition and retrieval in isolated synaptic terminals of retinal bipolar neurons.
Compensatory endocytosis was rapidly and reversibly inhibited by a subtle increase in intracellular pressure, produced by exogenous pressure applied through a whole cell patch pipette or by elevated internal osmolarity. Although capacitance failed to return completely to baseline when internal pressure was elevated, a fast component of endocytosis remained, resulting in rapid but partial recovery after each bout of exocytosis. Thus at least two components of endocytosis contribute to compensatory membrane retrieval after single bouts of exocytosis: a slower component that is blocked by elevation of internal hydrostatic pressure and a faster component that is unaffected by increased pressure. These results, combined with the similarity in intraterminal calcium parameters at low and high internal hydrostatic pressure, suggest that in synaptic terminals, the kinetically distinct components of compensatory endocytosis that participate in rapid restoration of membrane surface area following calcium-triggered exocytosis reflect functionally distinct mechanisms.
Cell isolation and solutions
Single bipolar neurons were isolated by mechanical trituration after enzymatic digestion of goldfish retina as described previously (Heidelberger and Matthews 1992). Recordings were made from isolated synaptic terminals of type Mb1 bipolar cells whose axons were severed during the isolation procedure. Terminals were selected on the basis of their characteristic appearance and size (8–12 μm diam), and identification was confirmed by their distinct electrophysiological profile (von Gersdorff and Matthews 1994a). In pipette pressure experiments, the intracellular solution contained the following (in mM): 76.5 or 100 Cs-gluconate, 9–10 tetraethylammonium (TEA)-Cl, 3 MgCl2, 2 Na2ATP, 0.5 GTP, 5 EGTA, 2.5 CaCl2, 0.2 fura-2, and 30 or 58.5 HEPES (pH = 7.2, 260–265 mOsm). No difference in results was noted between solutions. For osmolarity experiments, the internal solution contained the following (in mM): 95–120 Cs-glutamate, 5 TEA-Cl, 30–35 HEPES, 2 MgATP, 0.5 GTP, 0 or 0.5 EGTA, and 0.5 fura-2 (pH = 7.2, 246–290 mOsm). The 265-mOsm internal solution was a dilution of the 275-mOsm solution. The external recording solution typically contained the following (in mM): 115 NaCl, 2.5 KCl, 1.6 MgCl2, 1 CaCl2, 10 glucose, and 10 HEPES (pH = 7.3, 255–260 mOsm). This external solution was selected to be consistent with the large base of prior physiological experiments on goldfish retinal neurons.
Electrophysiological and [Ca2+]imeasurements
Whole cell recordings were performed at 21–24°C using 7–12 MΩ patch pipettes coated with Sylgard, which were pulled from 1.5-mm tubing with an internal diameter of 1.2 mm. For all experiments, the angle of the pipettes was held constant at 30° from horizontal, and the pipettes were back-filled with a fixed volume of fluid (12 μl). The wire for electrical contact, which affects capillary action by reducing the effective internal diameter of the pipette shank, had a diameter of 0.3 mm. The fluid column in the pipette provided a positive pressure, while capillary action provided a negative pressure. Under our conditions, the net result was negative when the pipette interior was open to the atmosphere, as evidenced by the movement of floating debris in the external fluid toward the pipette. To regulate the pressure within a patch pipette, the perfusion port of the pipette holder was connected to a feedback-regulated pipette pressure controller from Lorenz (Katlenburg-Lindau, Germany), forming a closed-loop system. This device could be bypassed to allow delivery of a brief suction pulse used to achieve the whole cell recording configuration. The pressure controller allowed us to toggle between two preset pressure settings. Visual inspection of fluid entering or leaving the pipette tip showed that with our pipette configuration, an empirical pressure setting of 2.8 cm H2O applied by the controller resulted in a near-neutral, but slightly negative net pipette pressure. This pressure setting was adopted as our baseline condition.
Electrophysiological measurements were performed with an EPC-9 patch-clamp amplifier controlled by E9screen or Pulse software (HEKA Electronik, Lambrecht, Germany). In some experiments, the automatic capacitance compensation of the EPC-9 amplifier was used to track changes in membrane capacitance. In other experiments, capacitance measurements were made using the software lock-in extension of the Pulse software, which implements the Lindau-Neher sine + DC technique (Gillis 1995). In the latter case, a 1600-Hz (linear frequency) sinusoidal stimulus with a 30-mV peak-to-peak amplitude was superimposed on the holding potential of −60 mV. For all experiments, the holding potential was −60 mV. Bipolar cells are tonically active neurons that produce sustained depolarizations in response to illumination, and depolarizations of 250–500 ms are required to exhaust the releasable pool at their ribbon-type synapses (von Gersdorff and Matthews 1994a; Mennerick and Matthews 1996). Therefore depolarizations to 0 mV for 250–1,000 ms were used to ensure the fusion of the entire releasable pool during depolarization. No consistent differences in exocytosis or endocytosis were observed between the pulse durations, and results were combined unless otherwise noted. Terminals with leak current >30 pA were excluded from the data set.
For measurement of [Ca2+]i, alternating excitation at 345 and 390 nm was provided by a two-flash-lamp system (T.I.L.L. Photonics) or by a computer-controlled monochrometer-based system (T.I.L.L. Photonics; Messler et al. 1996). Intraterminal Ca2+ was calculated from the ratio of the emitted light at the two wavelengths (Grynkiewicz et al. 1985), using calibration constants determined by dialyzing cells with highly buffered, known concentrations of Ca2+ (Heidelberger and Matthews 1992). Previous work has shown that intraterminal Ca2+ concentrations below 500 nM are permissive for compensatory endocytosis (von Gersdorff and Matthews 1994b). Therefore only terminals with resting Ca2+ <500 nM were included in the analysis. Resting Ca2+ in the majority of terminals analyzed was <300 nM. In undialyzed bipolar cell synaptic terminals loaded with membrane-permeant fura, the resting intraterminal calcium is ≈150 nM, a value somewhat higher than that measured in bipolar cell somata (Heidelberger and Matthews 1992).
Capacitance, conductance, fluorescence, calcium, and pressure traces acquired with Pulse were exported into Igor from Wavemetrics (Lake Oswego, OR) for analysis. The rate constants of endocytosis were obtained using a curve-fitting routine for a single or double exponential decay that does not require initial user guesses. Goodness of fit was determined by analysis of the χ2value and confirmed by eye. Ensemble analyses were performed in Microsoft Excel. For this purpose, capacitance records were normalized, and traces were aligned according to the peak of the capacitance response. To include two terminals in the osmotic pressure ensemble average that had a sloping capacitance baseline, a linear correction of the baseline slope was performed in Igor prior to importation into Excel. Corrections were not applied to other data. Ensemble averages of the calcium records were obtained simply by aligning the calcium records according to when the change in pipette pressure occurred. To estimate terminal diameter, images of bipolar cell terminals and a calibration graticule were taken with an USB AlphaData video TurboAdapter AD-VDO1 Video Capture (Alpha Data) and software AVRE Capture Tool (Nogatech) and analyzed in software ImageJ 1.26t (Wayne Rasband, NIMH, Bethesda, MD). Statistical analyses were performed with Microsoft Excel and SAS software (SAS Institute, Cary, NC). All pooled data are expressed as mean ± SE.
Compensatory endocytosis is reversibly inhibited by elevated hydrostatic pressure
In synaptic terminals of retinal bipolar neurons, capacitance measurements reveal that a bout of exocytosis is normally followed by endocytosis. This “compensatory endocytosis” restores the resting surface area of the terminal within seconds following fusion of synaptic vesicles, as illustrated in Fig.1 A. In this experiment, hydrostatic pressure within the whole cell patch pipette was adjusted by a feedback-regulated pressure controller connected to the pipette interior. Shortly after break-in to begin whole cell recording, the applied pressure was set to 2.8 cm H2O. At this nominal value, the net pipette pressure was near-neutral, but slightly negative (the source of negative pressure being capillary action). At the time indicated by the arrow, the membrane voltage was stepped from a holding potential of −60 to 0 mV for 500 ms to activate calcium influx through voltage-gated calcium channels. This resulted in an increase in the average intraterminal calcium, ratiometrically calculated from the change in fura-2 fluorescence (Fig. 1 B), and a 120-fF increase in membrane capacitance (Fig.1 A), indicative of the fusion of the releasable pool of synaptic vesicles (Heidelberger 2001b). Both internal calcium and membrane capacitance quickly recovered to baseline with a time constant of a few seconds after the stimulus (τendo ≈ 4.4 s). Thus neither exocytosis nor endocytosis was affected by the slightly negative pipette pressure.
Approximately 20 s later, the hydrostatic pressure of the patch pipette was increased from 2.8 to 5.0 cm H2O, causing the net pressure within the pipette to become positive. The switch from slightly negative pressure to positive pressure did not alter the resting membrane capacitance or intraterminal calcium, nor did it detectably change the appearance of a synaptic terminal (Fig.1 C). The average diameter of terminals was 8.29 ± 0.58 μm at near-neutral pressure and 8.31 ± 0.58 μm at positive pressure (mean ± SE; n = 8). Furthermore, the pressure jump was not associated with a significant change in the access or membrane conductance or in internal perfusion, as evidenced by the constancy in fura-2 fluorescence (not shown). After approximately 14 s at positive pressure, a second 500-ms depolarizing voltage step from −60 to 0 mV was given. As before, membrane depolarization triggered calcium influx and an increase in membrane capacitance (72fF). However, in marked contrast to what was observed at the near-neutral pressure, at positive pressure the membrane capacitance failed to completely return to baseline after exocytosis (Fig. 1 A). Endocytosis was still incomplete some 35 s after the depolarization. At this time, the pipette pressure was then stepped back to 2.8 cm H2O. When challenged for the third time with the identical stimulus, exocytosis (93 fF) was once again followed by full compensatory endocytosis (τendo ≈ 6.3 s). Similar results were observed in all six terminals examined. Ensemble averages of the normalized capacitance responses at near-neutral pressure and positive pressure from four of these terminals are shown in Fig.2. As with the terminal depicted in Fig.1, the ensemble averages demonstrate that positive pipette pressure inhibits the complete restoration of membrane surface area that usually follows exocytosis. In all terminals, the inhibition of compensatory endocytosis by positive pipette pressure was reversible on switching to a lower pressure (e.g., Fig. 1 A).
Average cytosolic calcium and the magnitude of calcium entry have both been suggested to regulate the rate of membrane recovery in synaptic terminals (Cousin and Robinson 2000; Hsu and Jackson 1996; Neves et al. 2001; Rouze and Schwartz 1998; von Gersdorff and Matthews 1994b). Figure 3 Ademonstrates that the change in pipette pressure from a near-neutral to a positive setting had no effect on the average intraterminal calcium, nor did it affect the resting membrane capacitance, suggesting that calcium did not substantially rise in the vicinity of active zones, where calcium channels cluster (Morgans 2001;Morgans et al. 2001; Raviola and Raviola 1982). In addition, we found that calcium entry was not altered by the change in pipette pressure; the average peak amplitude of the calcium current was virtually identical between the two conditions (near-neutral pressure: I Ca = −133.5 ± 10.7 pA, n = 6; positive pressure:I Ca = −130.8 ± 26.1 pA,n = 11; P = 0.64). Consistent with the lack of effect of pressure on calcium current, the average peak calcium concentration following depolarization was not significantly different at neutral and elevated pressure (Fig. 3 B; black bars;n = 4, P = 0.55). In addition, the rate of return of calcium to baseline after depolarization was similar under the two pressure conditions (Fig. 3 B; gray bars;n = 4, P = 0.35). Thus it is unlikely that a pressure-related change in calcium entry or calcium handling can account for the failure of compensatory endocytosis when the pipette pressure is positive.
Compensatory endocytosis is inhibited by mild osmotic swelling
We assessed the generality of the inhibitory effect of positive hydrostatic pressure on compensatory endocytosis by increasing hydrostatic pressure via a change in osmotic pressure. Figure4 shows the capacitance and calcium records from a single synaptic terminal in which the osmolarity of the internal recording solution was 30 mOsm higher than the standard internal recording solution. Each depolarization from −60 to 0 mV triggered an increase in membrane capacitance. The average capacitance increase was 116 ± 8 fF (n = 7), suggesting that this amount of osmotic pressure does not block fusion of the releasable pool of synaptic vesicles. However, unlike what is seen under standard recording conditions, exocytosis was not followed by the rapid retrieval of an equivalent amount of membrane. Instead, only ≈32% of the total membrane added by exocytosis was retrieved. This resulted in a cumulative net gain of ≈550 fF and a 27% increase in the resting membrane capacitance of the terminal during successive stimuli (Fig.4 A). Figure 4 B indicates that the incompleteness of membrane retrieval cannot be attributed to a sustained elevation in internal calcium because intraterminal calcium was rapidly restored to baseline following closure of the calcium channels. Similar results were observed in three other terminals dialyzed with this hyperosmotic internal solution.
We next compared the rate and completeness of membrane retrieval in four groups of terminals that differed from each other in the osmolarity of the internal recording solution. Figure5 shows the ensemble averages of the normalized capacitance responses for each group. When the internal osmolarity was 246 mOsm (Fig. 5 A), endocytosis was relatively fast (τendo ≈ 5.4 s) on average, and following a bout of exocytosis, the membrane capacitance was restored approximately to baseline. When the internal osmolarity was elevated to 265 mOsm (Fig. 5 B), the average rate of membrane retrieval was considerably slower (τendo ≈ 16.7 s), but retrieval of membrane was substantially complete (approximately 70%). By contrast, when the osmolarity of the internal solution exceeded that of the bath solution by ≥20 mOsm, endocytosis was dramatically impaired (Fig. 5,C and D).
As in the pipette pressure experiments, we examined the various calcium parameters to determine whether the effect of osmotic pressure on endocytosis could be attributed to changes in internal calcium or calcium influx. There was no difference in the average peak amplitude of the calcium current among the groups (1-way ANOVA, P= 0.66). The mean calcium current for terminals dialyzed with the 246-mOsm internal solution was −194 ± 37 pA (n = 6) compared with −176.6 ± 8.5 pA (n = 10) for the 265-mOsm solution, −217 ± 45 pA (n = 4) for the 275-mOsm solution and −217 ± 6 pA (n = 4) for the 290-mOsm solution. Although differences in the resting calcium existed among the groups, there was no trend that could account for the inhibition of endocytosis (1-way ANOVA with testing for linear trend,P = 0.54). For example, whereas the 246-mOsm group and 265-mOsm group had different average resting calcium concentrations (246 mOsm: 365 ± 25 nM, n = 6; 265 mOsm: 196 ± 78 nM, n = 10; P < 0.001), both of these groups exhibited compensatory endocytosis. A similar disparity in resting calcium was observed between the two groups of terminals in which endocytosis was greatly impaired (275 mOsm: 132 ± 10 nM,n = 4; 290 mOsm: 300 32 ± 16 nM,n = 4; P < 0.001). In addition, no consistent trend was observed among the different internal osmolarities in the peak calcium concentration after depolarization (Fig.3 B, black bars; regression analysis, P = 0.36) or in the rate of recovery of calcium to baseline after depolarization (Fig. 3 B, gray bars; regression analysis,P = 0.06). From these data, we conclude that the effect of osmolarity on compensatory endocytosis is independent of changes in resting cytosolic calcium concentration, the magnitude of voltage-dependent calcium entry, or calcium handling.
Elevated hydrostatic pressure isolates a small, fast component of endocytosis
Although compensatory endocytosis was severely impaired when hydrostatic pressure was elevated by exogenous or osmotic pressure, membrane retrieval was not completely abolished in most cases. Inspection of both the individual capacitance records (Figs. 1 and 4) and ensemble averages (Figs. 2 and 5) reveal that a small, fast component of endocytosis often persisted. This pressure-resistant component was observed in 16 of 19 experiments. Superposition of the normalized ensemble averages of the capacitance responses at normal and elevated hydrostatic pressure indicate that the pressure-resistant component of endocytosis retrieved approximately 45% of the previously added membrane in the pipette pressure experiments (Fig.6 A, ●) and 28% percent of added membrane in the osmotic experiments (Fig. 6 B, ●). These percentages correspond to the retrieval of 25 and 31 fF of membrane, respectively, matching the estimates of the amount of membrane retrieved by the pressure-resistant component estimated from the individual capacitance traces (positive pipette pressure: 25 ± 2 fF, n = 10; elevated osmotic pressure: 33 ± 4 fF, n = 6). On average, the component of endocytosis resistant to an increase in hydrostatic pressure retrieved 28 ± 2 fF (n = 16) of membrane, a value similar to the total number of docked vesicles at the active zones of the average bipolar neuron synaptic terminal (von Gersdorff et al. 1996).
Membrane retrieval via the pressure-resistant component typically proceeded with a time constant of a few seconds. In experiments in which the pipette pressure was elevated, estimates from both the ensemble average and from analysis of the individual capacitance traces, yielded a time constant of approximately 3 s (ensemble average: τ ≈ 3.4 s (Fig. 6 A); individual capacitance traces: τ = 3.1 ± 0.5 s,n = 10). In experiments in which the osmotic pressure was elevated, the time constant of membrane retrieval was approximately 2.5 s (Fig. 6 B), similar to the estimate obtained from analysis of the individual traces (2.2 ± 0.4 s,n = 6). Overall, the mean time constant for the component of endocytosis that was resistant to elevated hydrostatic pressure was 2.7 ± 0.4 s (n = 16).
In bipolar neuron synaptic terminals, at least two kinetic components of membrane retrieval have been reported to participate in compensatory endocytosis in both the whole cell (von Gersdorff et al. 1994b) and perforated-patch recording configurations (Neves and Lagnado 1999; Neves et al. 2001). The faster component retrieves membrane with a time constant of a few seconds, similar to that of the pressure-resistant component, while the slower component has a time constant that is approximately 10-fold slower. This raises the possibility that modest elevations in hydrostatic pressure, which preserved a fast component of endocytosis, prevented compensatory endocytosis by selectively blocking a slow component of membrane retrieval. We therefore determined whether two kinetic components of membrane retrieval contributed to compensatory endocytosis under our control conditions following a single 500-ms depolarization. At near-neutral pressure, the time course of endocytosis was best fit in six of eight terminals, as judged by eye and χ2 analysis, by a double exponential with an average τfast = 3.3 ± 1.0 s and τslow = 20.9 ± 6.9 (n = 6). These two kinetic components are readily apparent in the ensemble average (Fig. 6 A, ○). The slower of these two components was lost when the hydrostatic pressure was elevated (Fig. 6 A, ●). Two kinetic components of compensatory endocytosis were also evident in the osmotic control ensemble average (Fig. 6 B, ○). Again, the slower component had a time constant of approximately 23 s, although in these cells the faster component was not as fast as in the pipette pressure experiments. As with an elevation in hydrostatic pressure by mechanical means, the slower component was lost when the osmotic pressure was elevated (Fig. 6 B).
When temporally uncoupled from exocytosis, endocytosis is slow but compensatory
Following the return to permissive pressure, a relatively slow retrieval of membrane was often noted (e.g., see Fig. 1 A). In each instance, the onset of this slow endocytosis, which is separated in time from exocytosis, coincided with the change in pipette pressure from the positive setting to the near-neutral setting. The ensemble average of the capacitance records that demonstrate this form of endocytosis is depicted in Fig.7 A. The time course of membrane recovery could be fitted by a single exponential with a mean τendo of 38.9 ± 5.3 s (n = 5). This is approximately an order of magnitude slower than the pressure-resistant component of endocytosis and the faster component of compensatory endocytosis, and it is nearly double the time constant of the slower component of compensatory endocytosis. In five of five instances where this form of endocytosis was not interrupted by subsequent test depolarizations, it restored the membrane capacitance to prestimulus levels (Fig. 7 A). This suggests that even though exocytosis and endocytosis were uncoupled in time, the balance between the amount of membrane added and the amount retrieved was maintained.
The onset of this slow component of membrane retrieval did not require the elevation of intraterminal calcium. On average, the start of the slow endocytosis occurred 29 ± 7 s after the depolarization (n = 5), at which point calcium microdomains are expected to have collapsed (reviewed in Neher 1998) and the average cytosolic calcium concentration to have returned to resting levels. The ensemble average of intraterminal calcium shown in Fig.7 B confirm that intraterminal calcium returned to basal concentrations prior to the decrease in pipette pressure and the onset of endocytosis. Together, these observations suggest that this slow, temporally uncoupled component of membrane retrieval does not require the elevation of cytosolic calcium at the time of its initiation.
Multiple components of compensatory endocytosis
A variety of kinetically and functionally distinct modes of endocytosis may contribute under different conditions to membrane retrieval following exocytosis in secretory cells such as chromaffin cells (Artalejo et al. 1995; Chan and Smith 2001; Engisch and Nowycky 1998; Smith and Neher 1997). In synaptic terminals, multiple kinetic components of compensatory endocytosis have also been described (Hsu and Jackson 1996; Neves and Lagnado 1999;Rouze and Schwartz 1998; von Gersdorff and Matthews 1994a,b; Wu and Betz 1996), with time constants ranging from approximately 1 s to several tens of seconds. Furthermore, the rate of compensatory retrieval in synaptic terminals is regulated in an activity-dependent manner (von Gersdorff and Matthews 1994a,b; Wu and Betz 1996). Retrieval after single depolarizations or after low-frequency trains of action potentials proceeds more rapidly than retrieval after repetitive stimuli or high-frequency trains of action potentials. In part, this slowing of retrieval reflects calcium-dependent inhibition of endocytosis (Cousin and Robinson, 2000; Hsu and Jackson 1996;Rouze and Schwartz 1998; von Gersdorff and Matthews 1994b). However, elevated internal calcium does not entirely account for activity-dependent slowing of endocytosis in either bipolar-cell synaptic terminals (Heidelberger 2001a; von Gersdorff and Matthews 1994b) or the neuromuscular junction (Wu and Betz 1996). Thus the kinetically distinct components of endocytosis observed under different conditions at synapses may reflect the operation of functionally distinct mechanisms.
The results presented here show that multiple components of endocytosis also exist within the comparatively rapid compensatory endocytosis that follows single depolarizations in bipolar-cell synaptic terminals. This was seen most clearly under elevated internal hydrostatic pressure, which selectively eliminated the slower of the two components (τ ≈ 20 s), leaving a rapid component (τ ≈ 3 s). The rapid component retrieved only 30–50% of the membrane added by vesicle fusion, suggesting that it may be limited in capacity. The differential effect of pressure suggests that the two components of retrieval may represent distinct forms of endocytosis that utilize different mechanisms. We note that this selective action of pressure differs from the more global suppression of endocytosis produced by blocking ATP hydrolysis, which prevents all endocytosis after a bout of exocytosis (Heidelberger 2001a).
When retrieval is delayed in bipolar cell synaptic terminals by exogenously applied intracellular pressure, the rate of endocytosis slows when the applied pressure is subsequently removed. For instance, the slower time constant of endocytosis was approximately 20 s when endocytosis was allowed to proceed immediately after a bout of exocytosis, but the time constant of retrieval slowed further to approximately 40 s when endocytosis was delayed by a pulse of elevated pressure and then allowed to proceed after stepping pressure back to normal. Elevated calcium has been reported to accelerate a component of endocytosis in hippocampal neurons (Klingauf et al. 1998), raising the possibility that when endocytosis occurs after recovery of intraterminal calcium, the rate of retrieval may appear prolonged. However, in bipolar neurons, the rate of a slow component of retrieval is not altered by calcium (Neves et al. 2001). In addition, when endocytosis in synaptic terminals is delayed by prolonged trains of stimuli, which elevate intraterminal calcium, the rate of endocytosis is also slowed (von Gersdorff and Matthews 1994a,b). This implies that there may be yet another form of endocytosis that occurs under these delayed conditions. The existence of several functionally distinct forms of endocytosis within synaptic terminals is a reasonable possibility, given the large number of different endocytosis mechanisms known in both plant and animal cells. These include clathrin-dependent and clathrin-independent endocytosis, caveolae, macropinocytosis, and phagocytosis, all of which have distinct underlying biochemical mechanisms (see Mukherjee et al. 1997, for review).
How does applied hydrostatic pressure inhibit endocytosis?
The mechanism by which internal hydrostatic pressure inhibits membrane retrieval is not understood in detail, but an increase in membrane tension would likely increase the energetic cost of forming invaginations prior to retrieval (Dai and Sheetz 1995). In our experiments, the applied pressure required to inhibit endocytosis was relatively small and any changes in cell shape were submicroscopic, as evidenced by constant diameter of the synaptic terminal when pressure was varied. An increase in cell diameter with elevated pressure is thought to reflect the straightening out of membrane infoldings (Solsona et al. 1998). Bipolar-cell synaptic terminals lack such infoldings, and their membrane surface is smooth and uncrenulated when viewed with electron microscopy (C. Paillart, G. Matthews, P. Sterling, unpublished observations). This probably explains why their diameter does not increase with internal pressure, but it also means that any applied pressure is translated directly into increased membrane tension. The applied pressure of 1.0–3.3 cm H2O would increase membrane tension by 0.2–0.7 mN/m, as calculated from Laplace's law for a uniform, thin-walled sphere with a diameter of 8.3 μm (T = Pd/4; where T is tension, P is hydrostatic pressure in Newtons per square meter, and d is diameter; 1 cm H2O = 98 N/m2). By comparison, the tension required to lyse mast cells by applying pressure via a whole cell patch pipette is 5–8 mN/m (Solsona et al. 1998), or about 10-fold higher than the tensions we applied. Nevertheless, the increase in membrane tension per se may be sufficient to prevent membrane invagination during membrane retrieval and hence to inhibit at least some forms of endocytosis.
Because exocytosis was not inhibited by the applied hydrostatic pressure in our experiments, we assume that the relationship between the plasma membrane and cytoskeletal elements, such as synaptic ribbons, was not grossly distorted. This is also consistent with the lack of a discernible increase in terminal diameter. However, miniscule swelling of the terminal during application of pressure might still affect the interaction between the plasma membrane and cytoskeleton, which could in turn affect the ability to form and/or pinch off infoldings during compensatory endocytosis. It is not yet clear whether such spatial disruption of membrane-cytoskeleton interactions contributes to the inhibition of endocytosis in bipolar-cell synaptic terminals in addition to or in place of the effect of the global increase in membrane tension.
How might the pressure-sensitive and pressure-insensitive components of endocytosis differ? One possibility is that the pressure-sensitive form requires the formation of membrane infoldings, whereas the pressure-insensitive form does not. The former could represent mechanisms that retrieve fully-collapsed vesicles, such as clathrin-dependent endocytosis. The latter could be accounted for by a mechanism such as reversibility of the fusion pore, or kiss-and-run endocytosis, for example. If this view is correct, then the time constant of approximately 3 s could be taken as an upper limit for the time course of fusion pore reversibility in the bipolar cell terminal under our experimental conditions. The apparently limited capacity of the pressure-insensitive endocytic pathway might then represent the fraction of vesicles (30–50%) that fail to undergo full collapse during the first few seconds after a stimulus.
Some practical considerations for patch-clamp studies of endocytosis
The rate of endocytosis can be dramatically altered by slight changes in intracellular pressure that produce no discernible changes in cell diameter. Such small pressure changes can readily arise from seemingly insignificant variations in the details of experimental technique, including such factors as the osmolarity of internal and external solutions, patch pipette geometry, pipette angle, and fluid volume in the patch pipette. Differences in these details may contribute to variability in results reported by different laboratories in studies of endocytosis. Attention to the hydrostatic pressure (including osmotic pressure) applied via the patch pipette is of particular importance in experiments in which endocytosis is compared before and after entering whole cell recording mode (e.g., Neves et al., 2001; Smith and Neher 1987), because intracellular pressure may well differ between the two conditions. These considerations may be especially important in bipolar-cell synaptic terminals, which lack an irregular, folded surface that can accommodate pressure changes by unfolding, with the result that pressure is readily translated into increased membrane tension.
Whole cell recordings are often performed with the interior of the pipette open to the atmosphere, with suction applied only to form seals and perhaps to rupture the membrane to enter whole cell mode. In this situation, whether the pipette applies positive or negative pressure to the cell interior is governed by several competing factors. Negative pressure arises from capillary action of the fluid within the pipette, while positive pressure arises from the action of gravity on the fluid column. The net outcome depends on the internal diameter of the glass tubing used to manufacture pipettes, the size of wire used to make electrical connection, the height of the fluid column within the pipette, and the angle of the pipette from vertical during recording. The relative osmolarities of the pipette solution and external solution are also relevant. Rather than depend on these competing factors to determine the outcome, we prefer in studies of exocytosis and endocytosis to use a closed system for the pipette interior, and to monitor the pressure with an inexpensive digital manometer (in the absence of a feedback-regulated pressure controller). The interior hydrostatic pressure can then be set empirically as appropriate to attain a fixed level of fluid flow through the open tip of the pipette. Of course, it is still necessary to use consistent procedures for filling the pipette with fluid and for setting the angle of the pipette during recording. For studies of endocytosis, it is probably preferable to avoid positive pipette pressure, which could retard membrane retrieval.
We thank N. Waxham and J. O'Brien for fruitful discussions, and A. Chuang and the Core Grant for Vision Research (EY-10608) for providing statistical support.
This work was supported by National Eye Institute Grants EY-12128 to R. Heidelberger and EY-03821 to G. Matthews, and by the Esther A. and Joseph Klingenstein Fund (RH) and the Alfred P. Sloan Foundation to R. Heidelberger.
Address for reprint requests: R. Heidelberger, Dept. of Neurobiology and Anatomy, UT-Houston Medical School, MSB 7.046, 6431 Fannin, Houston, TX 77030 (E-mail:).
- Copyright © 2002 The American Physiological Society