1. Neuroscience
Download icon

Elevated synaptic vesicle release probability in synaptophysin/gyrin family quadruple knockouts

  1. Mathan K Raja
  2. Julia Preobraschenski
  3. Sergio Del Olmo-Cabrera
  4. Rebeca Martinez-Turrillas
  5. Reinhard Jahn
  6. Isabel Perez-Otano
  7. John F Wesseling  Is a corresponding author
  1. Universidad de Navarra, Spain
  2. Max Planck Institute for Biophysical Chemistry, Germany
  3. Institute for Neurosciences CSIC-UMH, Spain
Research Communication
  • Cited 0
  • Views 354
  • Annotations
Cite this article as: eLife 2019;8:e40744 doi: 10.7554/eLife.40744

Abstract

Synaptophysins 1 and 2 and synaptogyrins 1 and 3 constitute a major family of synaptic vesicle membrane proteins. Unlike other widely expressed synaptic vesicle proteins such as vSNAREs and synaptotagmins, the primary function has not been resolved. Here, we report robust elevation in the probability of release of readily releasable vesicles with both high and low release probabilities at a variety of synapse types from knockout mice missing all four family members. Neither the number of readily releasable vesicles, nor the timing of recruitment to the readily releasable pool was affected. The results suggest that family members serve as negative regulators of neurotransmission, acting directly at the level of exocytosis to dampen connection strength selectively when presynaptic action potentials fire at low frequency. The widespread expression suggests that chemical synapses may play a frequency filtering role in biological computation that is more elemental than presently envisioned.

Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed (see decision letter).

https://doi.org/10.7554/eLife.40744.001

Introduction

Synaptophysin 1 and 2 and synaptogyrin 1 and 3 constitute a major family of synaptic vesicle membrane proteins expressed widely, possibly in all synaptic vesicles throughout the animal kingdom (Jahn et al., 1985; Stenius, 1995; Fernández-Chacón and Südhof, 1999; Takamori et al., 2006). Synaptogyrin 2, also known as cellugyrin, is non-neuronal (Janz and Südhof, 1998). The widespread expression of neuronal family members suggests a fundamental role in synaptic transmission, but what that might be is not known.

Family members bind to the vSNARE synaptobrevin 2/VAMP 2, which is a core component of the machinery that catalyzes membrane fusion during synaptic vesicle exocytosis (Söllner et al., 1993; Calakos and Scheller, 1994; Washbourne et al., 1995; Edelmann et al., 1995; Becher et al., 1999; Khvotchev and Südhof, 2004). And, overexpression of family members potently inhibited neurotransmitter release in a cell line (Sugita et al., 1999). Despite this, the hypothesis that the native function might involve negative regulation of exocytosis has not been pursued extensively, possibly because no clear evidence was found for increases in neurotransmitter release at synapses from synaptophysin 1 and synaptogyrin 1 single and double knockouts (McMahon et al., 1996; Janz et al., 1999; Abraham et al., 2006; Stevens et al., 2012). Instead, recent research has been focused on a variety of mechanisms that operate downstream of exocytosis, including: endocytosis of membrane; and/or recycling of proteins thought to be needed to catalyze subsequent rounds of exocytosis (Kwon and Chapman, 2011; Gordon et al., 2011; Rajappa et al., 2016). However, the previous studies, at least in mammals, involved exogenous expression or genetic deletion of synaptophysin 1 and synaptogyrin 1, whereas possible compensatory activity of synaptophysin 2 and synaptogyrin 3 has never been assessed (McInnes et al., 2018).

Here, we report that individual action potentials trigger exocytosis of a higher fraction of the readily releasable vesicles at a variety of synapse types from quadruple knockout mice (QKO) where all four neuronal family members have been deleted (see Materials and methods). No deficit was detected in other presynaptic parameters that control function such as the capacity of the readily releasable pool (RRP) for storing vesicles and the timing of vesicle recruitment to the RRP during light or heavy use. The results suggest strongly that family members play an inhibitory role at the level of exocytosis rather than the facilitatory downstream role that is currently envisioned. A follow-on analysis of double and triple knockouts showed that synaptophysin 1 and synaptogyrin 3 can compensate for missing family members, whereas synaptophysin 2 seemed to play a dominant negative role.

Results

QKO mice appeared to develop normally when housed in individually ventilated cages, and produced litters of normal size. However, adults were prone to convulsions, sometimes causing death, especially after being startled. We were not able to maintain the colony in a second facility where ventilated cages were not available, suggesting that the ventilation system aided survival, possibly by producing continuous white noise that limited startling. A quantitative western blot analysis of homogenized tissue and purified synaptosomes from QKO brains revealed a selective decrease in VAMP 2 levels, but no major changes in a wide variety of other synaptic proteins (Figure 1); a decrease in VAMP 2 levels was detected previously in synaptophysin 1 single knockouts (McMahon et al., 1996).

Figure 1 with 1 supplement see all
Selective decrease in VAMP 2 levels in synaptosomes of QKO mice.

Representative immunoblots and quantification of the indicated proteins from synaptosomes purified from brains of 3-month-old WT and QKO mice. Synaptosomes were prepared separately from cohorts of four male and four female individuals, but results were pooled because no substantial differences were detected between sexes. Horizontal lines are median values, boxes are the middle two quartiles. *p<0.05; Wilcoxon rank sum with Bonferonni correction for multiple comparisons; n 6 (2 independent preparations; samples were run at least 3 times).

https://doi.org/10.7554/eLife.40744.002

Elevated neurotransmitter release at calyx of Held synapses from QKO mice

In a first set of experiments to determine the primary functional deficit, we found that calyx of Held synapses from QKO mice were substantially stronger than WT when action potentials were evoked in the afferent axon at low frequency (Figure 2A–B). Excitatory postsynaptic currents (EPSCs) recorded in voltage clamped principal neurons of the medial nucleus of the trapezoid body (MNTB) had a greater quantal content (Figure 2C), indicating that the synapses were stronger because of exocytosis of transmitter from more presynaptic vesicles, in-line with the presynaptic locus of expression of synaptophysin family members. Spontaneous quantal release was elevated by a similar amount (Figure 2D–E), and the size of quantal events was elevated by a smaller amount (Figure 2F). No significant alterations were detected in the time courses of the EPSCs evoked with low-frequency stimulation or in spontaneous events (Figure 2—figure supplement 1).

Figure 2 with 1 supplement see all
Increased transmitter release at QKO calyces of Held.

(A) Diagram of calyx of Held preparation; MNTB is the medial nucleus of the trapezoid body. (B) Larger responses at QKO synapses after isolated/low frequency presynaptic action potentials (i.e., each after at least 1 min of rest). Traces are average responses across all preparations; n 11 calyces for both WT and QKO, each from a separate slice; experimenter was blind to genotype; extracellular 1mM kynurenic acid was used throughout. (C) Response sizes from (B) after normalizing by quantal size calculated as in Mahfooz et al. (2016). (D-F) Analysis of spontaneous responses recorded before adding kynurenic acid; n 13 calyces for both WT and QKO. (D) Black traces are overlay of all individuals from a single QKO neuron. The white trace is the mean that was used later for quantification. Mean traces are means of all individuals across all preparations. (E and F) Data points correspond to single preparations. Bars are mean ± s.e.m.; *p < 0.05; **p < 0.01; Wilcoxon rank sum.

https://doi.org/10.7554/eLife.40744.004

Elevated probability of release, with no alteration in RRP size

The number of vesicles that undergo exocytosis when action potentials are fired at low frequency is determined by two factors that seem to be controlled independently: the number of vesicles within a readily releasable pool, termed RRP content and possibly determined by the number of sites in the active zone area of the plasma membrane where vesicles dock; and the mean probability of release per vesicle (p¯v) within the RRP (Figure 3A, left); the abbreviation Pr has been used to denote the same concept in some other studies but is also sometimes used instead to denote the probability of release per synapse, which is the mathematical product of p¯v and RRP content, and thus a different concept. To determine which of the parameters was altered, we stimulated at 300Hz for 300ms (90 action potentials; Figure 3B). The difference in synaptic strength disappeared quickly, by the 4th action potential (Figure 3B–C), and there was no difference in the total number of quanta released during the first 150ms of the trains (Figure 3D). The result suggests that the RRP content was not altered at QKO synapses because 150ms is enough to nearly completely exhaust the RRP (Mahfooz et al., 2016).

Figure 3 with 1 supplement see all
Selective increase in probability of release at QKO calyces of Held.

(A) Diagram illustrating the RRP when nearly full during low-frequency stimulation (left) and when driven to a near-empty steady state by high-frequency stimulation (right). Gray circles represent vesicles, vesicles docked to release sites (squares) are readily releasable. By definition, the quantal content of individual synaptic responses is equal to the mean probability of release per vesicle within the RRP - p¯v - multiplied by the number of vesicles within the RRP. However, the precise value of p¯v ceases to be relevant when the RRP is driven to a near-empty steady state because vesicles undergo exocytosis soon after being recruited and, as a consequence, recruitment to the RRP (black arrows, right panel) supplants vesicle exocytosis as the rate-limiting mechanism. (B) Average response across all calyces during 300 ms of 300 Hz stimulation after blanking stimulus artifacts. Scale bars are 1 nA vs 100 ms (outer) and 1 nA vs 10 ms (inner, corresponding to the insets showing first 15 responses). (C) Mean number of quanta for each response vs time. Responses were first measured as the current integral after subtracting a baseline calculated from the 100 ms before stimulation began, and then normalized by mean quantal size. (D) Cumulative number of quanta. Theoretical curves are estimates of the cumulative response generated by release of transmitter recruited to the RRP during ongoing stimulation plus the offset needed to make the phase between 150 ms and 300 ms match the cumulative release; the value at Time = 0 equals the capacity of the RRP for storing vesicles. Lines marked ‘S’ are calculated using the method in Schneggenburger et al. (1999), whereas lines marked ‘M’ are calculated using Eqn (1) in Mahfooz et al. (2016), and describe the model illustrated in (A); note that both curves for both genotypes are plotted (magenta for QKO, blue for WT). The full 300 ms of 300 Hz stimulation elicited multiple rounds of exocytosis of readily releasable vesicles for both genotypes, including a total of 7076 ± 531 quanta at QKO synapses, which is more than double even the largest estimates of RRP content (ordinate intercept of ‘M’). (Ep¯v for calyces calculated using the theories in (D) to estimate vesicle recruitment (p < 0.01; rank sum; same preparations as Figure 2). (F) Unitary recruitment rate for individual calyces. The unitary recruitment rate is defined as the fraction of vacant space within the RRP replenished in a given amount of time; the concept is depicted by the black arrows in (A), right, and is analogous to a rate constant in first-order kinetics.

https://doi.org/10.7554/eLife.40744.006

Technically, the number of quanta released during trains that exhaust the RRP is not a perfect measure of RRP content because new vesicles are continually recruited and contents released during ongoing stimulation (Figure 3A, right). However, the amount of recruitment can be estimated by a variety of methods; two that have been proposed for this purpose are plotted in Figure 3D (i.e. curves marked ‘S’ and ‘M’). To our knowledge, the two span the full range of quantitative models that have been proposed; theory ‘S’ was proposed earlier, by Schneggenburger et al. (1999), whereas ‘M’ incorporates the conclusion of Mahfooz et al. (2016) that the RRP has a fixed capacity and vesicles are recruited to vacant spaces, such as empty release sites, as illustrated in Figure 3A (right panel). Although the various methods produced a variety of estimates for the amount of recruitment during the trains, all methods agreed that the amount was not different at QKO compared to at WT synapses (e.g. intercepts of curves ‘S’ and ‘M’ with ordinate-axis in Figure 3D). This result confirms that the initial RRP content was not altered at QKO synapses.

In contrast, p¯v is calculated by dividing the number of quanta released after isolated action potentials by the RRP content, and was approximately double at QKO synapses (Figure 3E). These results show that removing synaptophysin family members increased the value of p¯v without altering RRP content. Taken together, they suggest that endogenous synaptophysin family members inhibit neurotransmission downstream of vesicle recruitment to the RRP.

No alteration in the timing of vesicle priming

The conclusion is consistent with early experiments where exogenous expression of family members inhibited exocytosis (Sugita et al., 1999). However, one of the current hypotheses is that at least synaptophysin 1 plays a post-exocytosis role in clearing components of spent vesicles from the release machinery. The idea is that clearance determines the rate at which new vesicles can be recruited to the RRP during heavy use (Kwon and Chapman, 2011; Gordon et al., 2011; Rajappa et al., 2016). If so, removing family members would have produced a deficit in the rate at which vesicles are recruited to the RRP at later times during the trains of 300 Hz stimulation, which drove multiple rounds of exocytosis (see Legend of Figure 3D), and during subsequent rest intervals. But no such deficits were seen at QKO synapses.

That is, the methods used above to estimate the total amount of vesicle recruitment during trains additionally produce estimates of the ongoing rate of recruitment during heavy use (arrows in Figure 3A, right). All produced matching estimates for QKO and WT synapses (Figure 3F); see Figure 3—figure supplement 1 for control experiments verifying that the analyses were not confounded by postsynaptic mechanisms such as glutamate receptor desensitization.

The absence of a deficit in recruitment during extended stimulation could additionally be deduced without referencing any of the methods simply from the observation that steady state quantal output after the 150th ms of 300 Hz stimulation was similar at QKO synapses compared to WT (53.4 ± 4.7 quanta/action potential vs 59.2 ± 11.7). Nor did we find any alteration in the timing of RRP replenishment during rest intervals that followed 300 Hz stimulation (Figure 4). These results do not support the hypothesis that synaptophysin family members play a critical post-exocytosis role in facilitating ongoing vesicle priming and release, at least at calyx of Held synapses.

No alteration in timing of RRP replenishment at QKO calyces of Held.

(A) Experimental design. (B) Averaged recordings for trials with a rest interval of 0.5s between trains from single preparations. (C) RRP replenishment vs time estimated as in Mahfooz et al. (2016); n 20 trials from 7 calyces for QKO and 8 trials from 3 calyces for WT. The dashed line is RRPt=1eα^t with α^t the decaying exponential defined by Eqn (4) in Mahfooz et al. (2016), except α^0=5.4/s to match the value used to generate curve ‘M’ in Figure 3D.

https://doi.org/10.7554/eLife.40744.008

Elevated p¯v pertains to both high and low pv vesicles

The results so far suggest that synaptophysin family members ordinarily play an inhibitory role in neurotransmission downstream of vesicle recruitment to the RRP and upstream of neurotransmitter release, consistent with the possibility that the action is directly at the level of exocytosis. However, some current models include intervening priming steps between recruitment and exocytosis.

That is, the initial idea was that the RRP is a homogeneous pool (Elmqvist and Quastel, 1965; Vere-Jones, 1966), but there is now widespread agreement that some readily releasable vesicles are released more slowly than others during repetitive stimulation owing to a lower probability of release (Wu and Borst, 1999; Sakaba and Neher, 2001; Moulder and Mennerick, 2005). For clarity, we refer to the readily releasable vesicles with high and low probability of release as high and low pv vesicles, respectively, but note that it is possible that there are more than only two classes (Mahfooz et al., 2016; Taschenberger et al., 2016); elsewhere, low pv vesicles have been termed slow-releasing, or reluctantly releasable.

Both high and low pv vesicles seem to be immediately releasable, at least in the calyx of Held preparation examined here, and the timing of exocytosis is tightly synchronized to action potentials (Mahfooz et al., 2016). Our own models continue to treat recruitment to the RRP as the final vesicle trafficking step upstream of exocytosis (Mahfooz et al., 2016). the idea is that low and high pv vesicles are docked to distinct types of release sites (Figure 5A; Hu et al., 2013; Müller et al., 2015; Böhme et al., 2016). However, recent models of other research groups include additional mechanisms where individual vesicles already within the RRP reversibly transition between a variety of primed states distinguished by a range of release probabilities (Lee et al., 2013; Neher, 2017). If so, synaptophysin family members might either: limit the rate of transition of one of the forward steps, decreasing the fraction of vesicles in a high pv state; or directly inhibit exocytosis of low and high pv vesicles alike.

Figure 5 with 2 supplements see all
Enhanced release of vesicles with low pv remaining in RRP after 50 Hz or 100 Hz stimulation at QKO synapses.

(A) Diagram illustrating the steady state contents of the RRP for a range of stimulating frequencies. Release sites are depicted as stable, and are characterized by either a high (green squares) or a low (purple squares with yellow interior) probability of catalyzing the exocytosis of docked vesicles (i.e. high or low pv release sites). Alternative models where the readily releasable vesicles transition back and forth between high and low pv states would be the same except the release sites would not have a defined pv when empty, and the locations of the high and low pv vesicles would change over time. In either case, 50 Hz and 100 Hz stimulation is rapid enough to eliminate the vesicles with high pv from the RRP, but leaves a flow through pool of low pv vesicles that can then be released by subsequent 300 Hz stimulation. (B) Example recordings for trials where the stimulation frequency was increased from 50 Hz to 300 Hz; blue is WT, magenta is QKO. The insets are the last response during 50 Hz stimulation and responses during subsequent 300 Hz stimulation. The scale bars pertain to both sets of traces: the vertical bar is 1 nA and the horizontal is 250 ms for the full traces; and 500 pA and 50 ms for the insets. (C) Mean responses for the full data set quantified as the number of quanta released during sequential 20 ms segments, allowing direct comparison of the time-averaged rate of release when stimulation was 50 vs when 300 Hz; single segments contain the quantal content of responses to single action potentials for times when the stimulating frequency was 50 Hz, and of responses to 6 consecutive action potentials when the frequency was 300 Hz. Red symbols are for trials where stimulation was increased from 50 to 300 Hz, and gray are for trials where stimulation was maintained at 50 Hz throughout, as diagrammed at top. Boxes demarcate responses used to calculate the differential release in (D). (D) The additional release - termed differential release here - elicited by increasing the stimulation frequency to 300 Hz was calculated by subtracting the time-averaged values during continued 50 Hz stimulation from the corresponding values during 300 Hz stimulation. The theoretical curves are estimates of the fraction of the differential release produced by exocytosis of neurotransmitter that was recruited to the RRP during ongoing stimulation. ‘M’ and ‘S’ signify the same as in Figure 3D; ‘M’ describes the model in the illustrations at bottom and in (A). (Ep¯v values for individual calyces during steady state 50 and 100 Hz stimulation using the same methods ‘M’ and ‘S’ used to generate the theoretical curves in (D) ( is p < 0.1, * is p < 0.05; n 7; rank sum). (F) Cumulative version of the plot in (D), except after normalizing by the quantal content of the first differential response. Solid and dashed lines represent theories ‘M’ and ‘S’ as in (D) except offset to match the rightmost data points. Without the normalization, the lines would intersect the y-axis at the value that corresponds to the contents of the RRP at the start of 300 Hz stimulation (Schneggenburger et al., 1999). However, the normalization converts the estimate into multiples of the quantal content of the first response, making the intersection equal to 1/p¯v.

https://doi.org/10.7554/eLife.40744.009

To determine whether the action is at the level of exocytosis or upstream, we conducted frequency jump experiments where high pv vesicles are first eliminated from the RRP with submaximal stimulation of 50 or 100 Hz, leaving a standing population of low pv vesicles that can then be released by subsequent 300 Hz stimulation (Figure 5A). The experimental design was developed previously for isolating the kinetics of release of low pv vesicles at WT calyces of Held (Mahfooz et al., 2016). In the present case, the full experiment consisted of 6 types of interleaved trials: frequency jumps where 1000 ms of 50 Hz stimulation was followed by 200 ms of 300 Hz stimulation and a matched control where the stimulating frequency was maintained at 50 Hz throughout (Figure 5B–C); two additional types of frequency jumps where 500 ms or 750 ms of 100 Hz stimulation was followed by 200 ms of 300 Hz stimulation, along with a matched control where the stimulating frequency was maintained at 100 Hz throughout (Figure 5—figure supplement 1A); and an additional control where stimulation was 300 Hz for 300 ms, with no prior stimulation for at least 1 min (subset of data in Figures 23).

Figure 5B–C and Figure 5—figure supplement 1A show that 300 Hz stimulation transiently increased the rate of release following 50 Hz or 100 Hz stimulation, extending to QKO calyces of Held the previous finding that 100 Hz stimulation is not sufficiently frequent to exhaust the RRP at WT synapses (Mahfooz et al., 2016). To determine the p¯v value for the vesicles remaining in the RRP, we first isolated the increase in release by subtracting the time-averaged steady state response recorded during the matched trials when the stimulation frequency was not increased (gray data points in Figure 5C and Figure 5—figure supplement 1A; the amount of increase - termed the differential release - is plotted in Figure 5D and Figure 5—figure supplement 1B). The value for p¯v was then calculated by dividing the differential release after the first action potential during 300 Hz stimulation by the RRP content at the start of 300 Hz stimulation. The RRP content at the start of 300 Hz stimulation was estimated by subtracting the amount of recruitment to the RRP during 300 Hz stimulation from the total.

As above for Figure 3D–E, the estimate for the amount of recruitment during 300 Hz stimulation depended partly on assumptions about mechanism that continue to be debated (lines marked ‘M’ vs ‘S’ in Figure 5D and Figure 5—figure supplement 1B). However, all methods produced p¯v values for the vesicles remaining in the RRP that were higher for QKO synapses compared to WT (Figure 5E). The effect can be seen most clearly in the cumulative plots of the differential release measurements in Figure 5F, which are normalized so that the theoretical curves produced by the ‘M’ and ‘S’ theories intersect the y-axis at 1p¯v (see Legend). Note that the effect is not as readily apparent in Figure 5C–D because the standing state fullness of the RRP was lower at QKO synapses (Figure 5—figure supplement 2) - which is an expected consequence of the higher p¯v value (Mahfooz et al., 2016) - and because of the large bin size in Figure 5C.

The p¯v values for both WT and QKO were approximately 3-fold lower during 50 or 100 Hz stimulation compared to the corresponding values when the RRP was full (compare Figure 5E to Figure 3E), confirming that high pv vesicles were eliminated. And indeed, almost all of the vesicles remaining within the RRP during both 50 and 100 Hz stimulation must have been in the low pv state because 100 Hz stimulation depleted the RRP to a greater extent (Figure 5—figure supplement 2), but did not further lower the value for p¯v (Figure 5E).

These results do not distinguish between the classes of models with single and multiple sequential vesicle priming steps mentioned above, but do indicate that synaptophysin family members inhibit exocytosis of low and high pv vesicles alike. The results therefore strongly suggest that family members inhibit neurotransmission downstream of the final step in vesicle priming, which is consistent with the increase in spontaneous exocytosis seen in the absence of action potentials documented in Figure 2D–E. The analysis is based on the premise that low pv vesicles are immediately releasable, which seems likely because release of neurotransmitter continues to be tightly synchronized to action potentials during frequency jumps that are initiated after the high pv vesicles have been eliminated, and because synapses with more low pv vesicles express more paired pulse facilitation when the RRP is full (Figure 5B and Figure 5—figure supplement 1C; Mahfooz et al., 2016). Nevertheless, we did additionally considered the alternate scenario where the low pv vesicles are not immediately releasable (Miki et al., 2016; Gustafsson et al., 2019), but found it was not compatible with the results in Figure 5D and Figure 5—figure supplement 1B, as explained in the Legend of Figure 5—figure supplement 1.

Elevated p¯v at Schaffer collateral synapses from QKO mice

We next conducted experiments analogous to Figures 34 on synapses between Schaffer collaterals and CA1 pyramidal neurons of the hippocampus. We stimulated with 20 Hz trains instead of 300 Hz because 20 Hz is frequent enough to nearly completely empty the RRP owing to ∼20-fold slower recruitment of new vesicles during ongoing stimulation compared to the calyx of Held (Wesseling et al., 2002; Mahfooz et al., 2016).

Estimating the quantal content of responses from individual Schaffer collaterals is not as straightforward as for the calyx of Held because the number of afferent axons activated during a typical experiment is an unknown function of the strength of the individual pulses of stimulation (Figure 6A). Nevertheless, similar to at the calyx of Held, postsynaptic responses depressed more rapidly at QKO synapses compared to WT (Figure 6B–C). A kinetic analysis indicated that p¯v was approximately double (Figure 6D–E), whereas the timing of vesicle recruitment to the RRP during ongoing stimulation was not altered (Figure 6F). And, no difference was detected in the time course of RRP replenishment during subsequent rest intervals (Figure 6G). These results extend to hippocampal synapses the conclusion that the machinery that catalyzes synaptic vesicle exocytosis becomes more efficient after removing synaptophysin family proteins, whereas the timing of vesicle recruitment to the RRP is not altered. The comparison between calyces of Held and Schaffer collateral synapses is a good test for generality across synapse types because, in nature, Schaffer collateral synapses are typically used at frequencies that are ∼15-fold lower in addition to striking morphological and molecular differences and the ∼20-fold difference in the timing of vesicle trafficking already noted above (Ranck, 1973; Hermann et al., 2007; Borst and Soria van Hoeve, 2012).

Electrophysiological analysis of Schaffer collateral synapses.

(A) Diagram of ex vivo hippocampal slice preparation. (B-F) Increased p¯v at QKO synapses, but no differences in timing of vesicle trafficking during ongoing stimulation. (B) Responses during 20 Hz stimulation; shown are the first 300 ms of 6s-long trains from individual preparations; traces are the average of four trials. (C) Mean sizes of responses vs time. Individual trials were repeated at least three times for each preparation, and each preparation was allowed to rest at least 4 min before beginning each trial (n = 8 preparations per genotype). Responses were measured as the current integral and then normalized by the RRP contents at the start of stimulation calculated as in Wesseling et al. (2002); when normalized this way, the leftmost values are then equal to p¯v. (D) Cumulative responses, normalized as in (C). Theoretical curves are labeled as in Figure 3D, except here ‘M’ refers to the theory described in Wesseling et al. (2002), which is analogous to Mahfooz et al. (2016) but specific for hippocampal synapses. (Ep¯v values across preparations (p < 0.001; rank sum). Methods ‘M’ and ‘S’ are the same as for Figure 3E. (F) Values for the unitary recruitment rate across preparations. (G) RRP replenishment vs time; the dashed line is RRPt=1eαt with α^t the decaying exponential α(t) in Wesseling et al. (2002).

https://doi.org/10.7554/eLife.40744.012

Higher throughput assay in primary cell culture

To confirm that synaptic vesicle exocytosis is increased at QKO hippocampal synapses when action potentials are fired at low frequency, and to assess the contribution of each of the four synaptophysin family members, we then developed an optical imaging assay in primary cell culture with higher throughput than the electrophysiological assays (see Figure 7A and diagram atop Figure 7B). We first loaded the recycling synaptic vesicles with FM4-64 dye during 60 s of 20 Hz electrical stimulation (Gaffield and Betz, 2006; see Figure 7—figure supplement 1 for example images). We then monitored destaining with time-lapse fluorescence imaging during low frequency (0.2 Hz) stimulation in the absence of dye, followed by near complete destaining with a second 20 Hz train.

Figure 7 with 1 supplement see all
Higher throughput analysis of triple and double knockout synapses.

(A) Diagram of concepts. Recycling vesicles are first stained by driving synaptic vesicle exocytosis and subsequent recycling using electrical stimulation in the presence of FM4-64 in the extracellular fluid. Extracellular dye is then removed and dye stuck to the outside of the plasma membrane is washed off. Fluorescence levels are then monitored with time-lapse imaging as synapses are destained by triggering action potentials at low frequency (0.2 Hz). At such a low frequency, the RRP remains almost completely full because the time between action potentials (5 s) is enough for recruitment of new vesicles to replace the ones that undergo exocytosis. Since each action potential releases a higher fraction of the RRP contents at QKO synapses - i.e., because p¯v is higher - more vesicles undergo exocytosis, and destaining is faster as a consequence. In contrast, recruitment to the RRP becomes rate-limiting during high frequency stimulation that is fast enough to drive the RRP to a near empty steady state (20 Hz). As a consequence, the amount of destaining no longer depends on p¯v, and the synaptophysin family proteins no longer influence the timing. (B) Destaining during electrical stimulation for WT and QKO. Data points are mean ± s.e.m. of median values from each preparation; n 11 preparations, each with >250 ROIs. ΔF/F0 values in (C and E) are calculated as 1 minus the value indicated by the horizontal dashed line. (C) Comparison across genotypes of amount of destaining during the 0.2 Hz train of stimulation. Experimenter was blind to genotype. Boxes are middle two quartiles; horizontal lines are medians; notches signify 95% confidence intervals (***p<0.001, *p<0.05, compared to WT; ###p<0.001, ##p<0.01, compared to QKO; ANOVA followed by Tukey’s honest significant difference criterion; n 11 for each). (D) No difference between WT and QKO in time course of destaining when the frequency of stimulation was 20 Hz; n 3 preparations. (E) Follow-on paired tests indicate that synaptophysin 2 lessens the amount of compensation produced by synaptogyrin 3 or synaptophysin 1 when expressed alone (***p<0.001, *p<0.05; rank sum).

https://doi.org/10.7554/eLife.40744.013

We reasoned that destaining would be directly proportional to p¯v during low-frequency stimulation because the RRP would remain almost completely full. In contrast, we reasoned that destaining would not be influenced at all by p¯v during 20 Hz stimulation which empties the RRP in less than the 4 s interval between acquisition of successive images, after which transmitter release is no-longer influenced by p¯v and is instead rate-limited by vesicle recruitment to the RRP (Wesseling et al., 2002; see Figure 7A).

And indeed, QKO synapses destained almost 2-fold more than WT during the 0.2 Hz stimulation (Figure 7A–C), confirming that p¯v is elevated. And, no differences were detected during subsequent 20 Hz stimulation (not shown), or during 20 Hz stimulation when the 0.2 Hz train was omitted (Figure 7D), confirming that the timing of vesicle recruitment to the RRP was not altered. Furthermore, no differences between QKO and WT were detected in the amount of staining during loading, further supporting the conclusion of no alterations in rate-limiting steps in vesicle trafficking in QKO synapses.

Analysis of triple and double knockouts

Of the triple knockouts lacking all but one of the neuronal family members, synaptophysin 1 or synaptogyrin 3 alone largely compensated for the loss of the other three family members (Figure 7C).

Intriguingly, synapses from triple knockouts expressing only synaptophysin 2 were not noticeably different from QKO, but the combination of synaptophysin 2 with either synaptophysin 1 or synaptogyrin 3 was less effective at compensation than synaptophysin 1 or synaptogyrin 3 alone (Figure 7E). This result suggests that synaptophysin 2 may act as a competitive inhibitor of the function of other family members, or play a dominant negative role. And indeed, synaptophysin 2 is unique in that it lacks many of the sites for C-terminal tyrosine phosphorylation that are striking features of the other family members (Evans and Cousin, 2005).

Discussion

Synaptophysin family proteins are widely expressed in synaptic vesicle membranes, with more individuals per vesicle than the extensively studied synaptotagmins, although likely less than the vSNARE VAMP 2 and homologs (Takamori et al., 2006; Wilhelm et al., 2014). Despite the abundance, information about the role in presynaptic function has been elusive. Here, we show that the efficiency of the release machinery is elevated at a variety of synapse types from knockout mice where all four neuronal family members have been deleted, with no indication of any alteration in RRP content or the timing of vesicle recruitment to the RRP during light or heavy use. The new results strongly suggest that synaptophysin family members modulate function at the level of exocytosis. If so, the native action likely includes inhibition, especially when the results are taken together with an earlier study where exogenous synaptophysin 1 and synaptogyrin 1 potently inhibited exocytosis (Sugita et al., 1999; but see Alder et al., 1995).

The concern that the elevated p¯v seen at QKO synapses might instead reflect complicating developmental or compensatory mechanisms that are not directly related to the native function of the missing proteins is countered by the following. First, the alteration is unusually specific; p¯v was elevated without desynchronizing the relationship between action potentials and neurotransmitter release, or disrupting other vesicle trafficking mechanisms that control presynaptic function such as the capacity of the RRP for storing synaptic vesicles and the timing of vesicle recruitment to vacancies within the RRP. Second, the alteration was robust across a variety of synapse types embedded within neuronal networks that are subject to dissimilar developmental forces, including the networks formed by neurons grown in dissociated cell culture. Third, besides the decrease in VAMP 2, we did not detect substantial changes in levels of a variety of other proteins that have been implicated in exocytosis, including syntaxin 1, SNAP-25, RIM 1/2, Munc13-1, Rab 3a, and complexin 1/2. And fourth, synaptophysin 2 seemed to inhibit the function of other family members, which argues against the specific concern that removing the other family members increased the fusogenicity of synaptic vesicles mechanically, by simply exposing space on the surface vesicular membranes.

Notably, however, no elevation in p¯v was detected in neuromuscular junctions after deleting the family from C. elegans or Drosophila - and exocytosis of neurotransmitter was increased rather than decreased after expressing exogenous synaptophysin 1 in Xenopus - and it seems unlikely that an entire family of proteins would have an unrelated function in the different species (Alder et al., 1995; Abraham et al., 2006; Stevens et al., 2012). One possibility is that family members function as bi-directional regulators of exocytosis where the directionality is modulated by second messengers, possibly via phosphorylation of the tyrosine residues along the C-terminal tail. In any case, we anticipate that the cause of the discrepancies between species will become clear when more is known about the mechanism.

Indeed, even basic information about the mechanism remains to be elucidated. One possibility is that synaptophysin family members might interact directly with catalysis as outlined in Rothman et al. (2017); see also Adams et al. (2015). However, when bound to synaptophysin 1, VAMP 2 was excluded from the core SNARE complex consisting of VAMP 2, syntaxin 1, and SNAP25 that catalyzes exocytosis (Edelmann et al., 1995), and the combination of syntaxin 1 and SNAP25 could disrupt the binding between synaptophysin 1 and VAMP 2 in an enriched vesicle preparation (Siddiqui et al., 2007). A second possibility is that family members might lessen SNARE complex formation simply by restricting the availability of VAMP 2. However, we are not aware of evidence that exocytosis can be modulated in this manner, and indeed, lowering levels of the SNAP-25 component paradoxically increased p¯v in at least one study (Antonucci et al., 2013).

In any case, the relevance of the reduction in VAMP 2 levels in purified QKO synaptosomes to the increase in exocytosis seen at intact synapses in the functional assays is not known, and is counter-intuitive given that VAMP 2 is necessary for exocytosis. Intriguingly, the reduction is in-line with a previous study where exogenous VAMP 2 could be driven to synaptic vesicles by co-expressing synaptophysin 1 (Pennuto et al., 2003). The results do not suggest that synaptophysin family members are required for targeting VAMP 2 to synaptic vesicles, but are consistent with the possibility that VAMP 2 exists in two pools within vesicle membranes: one of which is stabilized by binding to synaptophysin family members; and the other by a different factor that remains to be identified. On the other hand, it is possible that the reduction in QKO synaptosomes resulted from depletion from plasma membrane rather than synaptic vesicles because significant amounts of VAMP 2 are consistently found in plasma membranes (Sankaranarayanan et al., 2000), and synaptosomes contain plasma membrane in addition to synaptic vesicles.

It is not clear how or if the small elevation in size of spontaneous responses seen both here and after deleting synaptogyrin from Drosophila is related to the elevated probability of release (Stevens et al., 2012). We cannot rule out a postsynaptic mechanism, but vesicles were larger in the Drosophila mutants, which might play a role. Possibly also relevant: Vesicles were also larger in synapses from knockouts of other presynaptic proteins involved in exocytosis, including VAMP 2, SNAP-25, and Munc13-1/2; the increase in radius was <10%, but even this small increase would translate to an increase in volume of 30%, which is more than the elevation in quantal size seen here (Imig et al., 2014).

Our results are in-line with previous reports of a role for synaptophysin family members in short-term synaptic plasticity (Janz et al., 1999; Kwon and Chapman, 2011; Rajappa et al., 2016). However, the decreased paired-pulse facilitation and increased paired-pulse depression seen in single, double and quadruple knockouts at a variety of synapse types do not necessarily indicate defects in mechanisms underlying short-term plasticity, but instead may result from the baseline elevation of p¯v. And indeed, paired-pulse facilitation could be unmasked at QKO synapses by lowering extracellular Ca2+, which lowers the baseline (Figure 8). In any case, the elevation in p¯v at QKO synapses cannot be attributed solely to the superpriming phenomenon proposed in Taschenberger et al. (2016) because the frequency jump experiments documented in Figure 5 indicated that the elevated release probability pertained to both low and high pv vesicles, whereas superpriming is thought to pertain exclusively to vesicles with high pv.

Reduced paired-pulse facilitation at QKO synapses is caused by occlusion.

(A) Schaffer collateral synapses of the hippocampus. Traces are the average of the first two responses during 20 Hz stimulation across the entire data set (inter-pulse interval was 50 ms; n = 8 preparations for WT in 2.6 mM Ca2+/1.3 mM Mg2+; n = 12 for QKO in 2.6 mM Ca2+/1.3 mM Mg2+; and n = 5 for QKO in 1.8 mM Ca2+/2.1 mM Mg2+; * signifies p < 0.05, *** is p < 0.001; Wilcoxon rank sum; bars are mean ± s.e.m.). 1.8 mM Ca2+/2.1 mM Mg2+ was chosen for these exeriments because the paired-pulse ratio at QKO synapses then matched WT synapses when bathed in 2.6 mM Ca2+/1.3 mM Mg2+. QKO synapses exhibited even more paired-pulse facilitation when Ca2+ was lowered and/or Mg2+ increased further. (B). Calyx of Held synapses also exhibited significantly less paired-pulse facilitation in the experiments documented in Figure 3 (** signifies p < 0.01; Wilcoxon rank sum; bars are mean ± s.e.m.). .

https://doi.org/10.7554/eLife.40744.015

The evidence against substantial deficits in the timing of vesicle recruitment to the RRP at QKO synapses seems to be strong. In particular, receptor desensitization mechanisms that could occlude differences between QKO and WT in the electrophysiological studies were ruled out with additional experiments in Figure 3—figure supplement 1 for calyx of Held and in Wesseling et al. (2002) for Schaffer collateral synapses. And, the evidence in Figure 7D relies on logic that avoids postsynaptic mechanisms altogether. However, our results do not conflict with the evidence for biochemical and cell biological alterations previously seen downstream of exocytosis in synaptophysin 1 knockouts (Kwon and Chapman, 2011; Gordon et al., 2011; Rajappa et al., 2016), although the results do indicate that any such downstream alterations would not affect the timing of vesicle recruitment to the RRP.

The selectivity of the increase in p¯v, with no change in the timing of vesicle recruitment or the size of the RRP, suggests that synaptophysin family members normally dampen synaptic connection strength when synapses are used at low frequencies. However, their effective role during the type of burst firing that occurs routinely in vivo would be more complex owing to slower depletion of the RRP as a direct consequence of reduced transmitter release. One way to characterize this sort of functional complexity is so called redistribution of synaptic efficacy where a decrease in synaptic strength at the beginning of a train of action potentials serves to enhance the strength later on (Markram and Tsodyks, 1996). Intriguingly, it seems that some activity-dependent form of redistribution of synaptic efficacy can be induced at a broad range of synapse types. For example, long-lasting bidirectional redistribution can be induced at cortical and hippocampal synapses by some of the same experimental protocols used to activate standard long-term potentiation and depression mechanisms at other synapse types (Markram and Tsodyks, 1996; Sjöström et al., 2003; Yasui et al., 2005; Monday et al., 2018). And, although the terminology was different, other reports have suggested that the mechanisms underlying post-tetanic potentiation and a third type of short-term plasticity termed augmentation have a similar re-distributive effect (Stevens and Wesseling, 1999; Garcia-Perez and Wesseling, 2008; Lee et al., 2010). Going forward it will be interesting to determine if synaptophysin family members or other selective regulators of p¯v such as GIT 1/2 or Mover are involved in any of these phenomena (Körber et al., 2015; Montesinos et al., 2015).

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Chemical
compound, drug
DL-APVAbcamCat# ab12027150 or 100 μM
Chemical
compound, drug
DNQXAbcamCat# ab12016910 μM
Chemical
compound, drug
kynurenic acidSigmaCat# K33751-4 mM
Chemical
compound, drug
picrotoxinSigmaCat# P167550 μM
Chemical
compound, drug
strychnineAbcamCat# ab1204160.5 μM
Chemical
compound, drug
Advasep-7CydexCat# ADV7-03A-031051 mM
Chemical
compound, drug
FM4-64BiotiumCat# BT7002115 μM
Strain, strain
background
(mouse)
RRID:IMSR_JAX:008454JacksonCat# 008454
CRE expressor
Strain, strain
background
(mouse)
RRID:
IMSR_JAX:008415
JacksonCat# 008415
Syp KO/Syngr1 KO/
Synpr KI/Syngr3 KI
AntibodiesAllsee Table 1

Knockout and WT control mice were obtained from two independent crosses. For the experiments in Figures 15 and 7, QKO, matched WT controls, and the ten logically possible double and triple knockouts were bred out in three or four generations by crossing a germline CRE expressing line (Jackson labs catalog number 008454) with Jackson line 008415, which carries targeted knockouts of synaptophysin 1 and synaptogyrin 1 genes and floxed conditional mutations of synaptophysin 2 and synaptogyrin 3; the CRE transgene was eliminated during the process. QKO and WT mice for experiments in Figure 6 were obtained directly from Dr. Thomas Südhof.

Western blotting 

Synaptosomes were prepared as in Kohansal-Nodehi et al. (2016). Six tissue preparations were analyzed in parallel from separate cohorts of five 3-month-old males and females, and a mixed cohort of six 17-day-old males and females. Samples from all six preparations were run on each blot, and each was replicated at least three times. Optical densities were first normalized by the mean density of all six samples, and then re-normalized so that the mean WT value was 1.0 before calculating the summary statistics. Hippocampal tissue homogenates were prepared as in Fiuza et al. (2013). Eight tissue preparations were analyzed in parallel from four males of each genotype. Samples from all eight were run on each blot, and each was replicated three times. Optical densities were normalized by the mean density of the WT samples on each blot, and then values for each sample were averaged across blots before calculating the summary statistics. See Table 1 for information about antibodies.

Table 1
Antibodies
https://doi.org/10.7554/eLife.40744.016
ProteinDesignationSourceSpeciesDilutionsynapto-someshipp. tissue
Synaptophysin 1RRID:AB_2313839Milliporemouse1:2000
Cl7.2Jahn et al., 1985
mouse1:1000
Synaptophysin 2RRID:AB_887841Synaptic Sys.rabbit1:1000
1:2000
Synaptogyrin 1RRID:AB_887818Synaptic Sys.rabbit1:2000
1:1000
Synaptogyrin 3RRID:AB_2619752Synaptic Sys.rabbit1:1000
Synaptotagmin 1RRID:AB_10622660Enzomouse1:2000
RRID:AB_11042457Synaptic Sys.rabbit1:1000
Synapsin 1RRID:AB_2619772Synaptic Sys.mouse1:5000
VAMP 2RRID:AB_887811Synaptic Sys.mouse1:1000
1:2000
RIM1/2RRID:AB_887774Synaptic Sys.rabbit1:1000
Munc13-1RRID:AB_887733Synaptic Sys.rabbit1:1000
Complexin1/2RRID:AB_887709Synaptic Sys.rabbit1:1000
vATPaseRRID:AB_887696Synaptic Sys.rabbit1:500
Syntaxin 1RRID:AB_887844Synaptic Sys.mouse1:1000
β-actinRRID:AB_11042458Synaptic Sys.rabbit1:1000
RRID:AB_476744Sigmamouse1:10000
SNAP-25RRID:AB_2315340Synaptic Sys.mouse1:1000
Rab3a,b,cCl42.1Matteoli et al., 1991
mouse1:1000
vGATSA5387Takamori et al., 2000
rabbit1:500
vGlut 1RRID:AB_2187690Santa Cruzgoat1:1000
Shigeo3Takamori et al., 2001
rabbit1:2000
vGlut 2Shigeo6Takamori et al., 2001
rabbit1:1000

Electrophysiology

Methods were the same as Mahfooz et al. (2016) for the calyx of Held and Wesseling et al. (2002) for Schaffer collateral synapses. All experiments were done in ex vivo slices from 13- to 21-day-old animals. Unless otherwise noted, n-values in Results and Figure Legends refer to number of preparations; multiple trials from individual preparations were averaged at the level of raw data before further analysis. Experiments in Figures 13, 5 and 7 were done blind to genotype.

Cell culture assay

Mice of the various genotypes became available sporadically over a period of seven months. Minimum desired sample sizes of 10 preparations were estimated beforehand based on results from pilot experiments conducted on WT and QKO synapses using 1 Hz rather than 0.2 Hz stimulation. However, actual sample sizes were larger in most cases owing to repetitions conducted to evaluate reproducibility over time.

FM4-64 fluorescence imaging

Hippocampal neurons were cultured from mice up to one day after birth and grown on glass coverslips coated with laminin and polyornithine as described in Chowdhury et al. (2013). Imaging was performed between 14–21 days after plating on an inverted microscope via a 25× oil immersion objective (Zeiss LD LCI Plan-APOCHROMAT 440842–9870; NA = 0.8) using a CCD camera (Photometrics CoolSNAP HQ; on chip binning by 2; pixel size was 0.5 μm × 0.5 μm). Illumination was <160lm for 25 ms with a green LED (520 nm; Luxeon LXHL-LM5C) via the XF102-2 filter set from Omega Optical. Time lapse imaging was at 0.25 Hz. The imaging chamber was low volume (∼35μl) and sealed on top and bottom. Flow was continuous during imaging (0.2-0.5 ml/min). Electrical stimulation was bipolar (0.5 ms at - 30 V then 0.5 ms at + 30 V; Falco Systems WMA 280) via two platinum wires (1 mm diameter, separated by ∼0.5 cm) that were glued within the chamber and flattened by milling so that the entire lower surface would make contact with the surface of the culture bearing coverslip. A thin layer of already hardened Sylgard 184 (Dow Corning; < 1 mm) was used instead of rubber or vacuum grease for sealing the chamber. FM4-64 was used at 15 μm and loaded with 60 s of 20 Hz stimulation followed by 2 min rest, and then at least 5 min wash in the absence of FM4-64 and presence of 1 mM Advasep-7. Advasep-7 was continuously present during the destaining phase of experiments. Other solutes were (in mM): NaCl (118); KCl (2); Ca2+ (2.6); Mg2+ (1.3); Glucose (30); and HEPES (25). Neurotransmitter receptors were blocked with (in μM): picrotoxin (50); DNQX (10); and DL-APV (50).

Processing

Images from time lapse experiments were aligned using the imagej plugin StackReg:Translation (Thévenaz et al., 1998) and in house software. Regions of interest (ROIs) were 2 × 2 pixels (1 μm X 1 μm) and were detected with in house software based on the change in contrast during the experiment (see Figure 7—figure supplement 1).

Normalization

For comparing images across preparations, median or individual ROI values were: (1) divided by the mean value of the background region; and then, (2) corrected for any rundown by subtracting the straight line fitting the values during the rest period immediately preceding the 20 Hz train stimulation. Next: (3) F - the residual fluorescence remaining after the final 20 Hz train - was subtracted; and (4) the values were normalized by dividing by F0, which was the mean value over the 2 Min preceding electrical stimulation.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Vesicle-associated membrane protein and synaptophysin are associated on the synaptic vesicle
    1. N Calakos
    2. RH Scheller
    (1994)
    The Journal of Biological Chemistry 269:24534–24537.
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
    The small and dynamic Pre-primed pool at the release site; A useful concept to understand release probability and Short-Term synaptic plasticity?
    1. B Gustafsson
    2. R Ma
    3. E Hanse
    (2019)
    Frontiers in Synaptic Neuroscience, 11, 10.3389/fnsyn.2019.00007, 30899219.
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69

Decision letter

  1. Graeme W Davis
    Reviewing Editor; University of California, San Francisco, United States
  2. Gary L Westbrook
    Senior Editor; Vollum Institute, United States
  3. Graeme W Davis
    Reviewer; University of California, San Francisco, United States
  4. Nils Brose
    Reviewer; Max Planck Institute of Experimental Medicine, Germany

In the interests of transparency, eLife includes the editorial decision letter, peer reviews, and accompanying author responses.

[Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed.]

Thank you for submitting your article "Synaptophysin/gyrin family proteins are selective negative regulators of exocytosis at mouse central synapses" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Graeme W Davis as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Gary Westbrook as the Senior Editor. The following individual involved in review of your submission has also agreed to reveal his identity: Nils Brose (Reviewer #2).

The Reviewing Editor has highlighted the concerns that require revision and/or responses, and we have included the separate reviews below for your consideration. If you have any questions, please do not hesitate to contact us.

Summary:

Raja et al. present an interesting story on tetraspan vesicle membrane proteins of the synaptophysin and synaptogyrin families. Synaptic vesicles (SVs) contain three types of tetraspan membrane proteins, synaptophysins, synaptogyrins, and SCAMPs. Although synaptophysins are among the most abundant SV proteins, their function – and the function of synaptophysins, synaptogyrins, and SCAMPs in general – has remained enigmatic. Indeed, genetic elimination of synaptophysin, synaptogyrin, and SCAMP in C. elegans, which expresses only one isoform of each family of tetraspan vesicle membrane proteins, has no effect on spontaneous or evoked synaptic transmission (Abraham et al., 2006). Raja et al. performed a related study on mammalian synaptophysins and synaptogyrins, focusing on quadruple KO mice lacking all synaptophysins and synaptogyrins. The key finding is that the loss of synaptophysins and synaptogyrins causes increased SV release probability and faster excitation-secretion coupling. This is clearly interesting and important for the synapse biology field. Raja et al. based their analysis on electrophysiological recordings at the calyx of Held and Schaffer collateral synapses in the hippocampus, and on FM4-64 loading/destaining experiments with cultured hippocampal neurons. Overall, the experiments were done in a stringent manner and the data look convincing. After extensive discussion, all three reviewers agree that the study is an important contribution, but there remain issues that should be addressed prior to publications, as cited below.

Major revisions:

1) The reviewers discussed the issue of interpreting synaptic vesicle fusion in the absence of four synaptic vesicle proteins that are among the most abundant synaptic vesicle proteins. The data shown in Figure S1 show trends regarding changes in SV proteins in the whole hippocampus of quadruple KOs. The issue of compensatory changes in synaptic vesicle proteins as a primary or secondary mediator of the observed effects on synaptic vesicle release is important, and is something that has been routinely addressed in similar knockout studies. As such, the authors could extend this analysis of synaptic vesicle protein abundance in purified synaptic vesicle fractions from the quadruple knockouts, including proteins directly involved in release probability such as RIM, Munc13, Rab3, and Complexin. Alternatively, the authors must provide additional discussion, citing the relevant literature, regarding how any observed changes may contribute, or not contribute, to the observed phenotype. In particular, discuss whether removal of a large fraction of synaptic vesicle protein might directly contribute to enhanced release through direct effects on vesicle fusogenicity.

2) The reviewers recommend either addressing synaptic vesicle endocytosis with direct assays, or altering the text of the manuscript to be more circumspect.

3) It is important that the authors cite and discuss the relevance of work in the C. elegans system (Abraham et al., 2006), and discuss more precisely their claim of having assessed a broad range of synapses. Several synapses have been assessed. How similar or distinct are these synapses?

4) It was recommended that the Discussion section be focused more clearly on the data presented. The authors are, of course, free to speculate. However, there was discussion regarding whether the proposed link between the present study and phosphorylation of synaptophysins and synaptogyrins, mover, GITs, ELKS, or tomosyn was a bit large a leap.

5) Authors estimate quanta/vesicle numbers from the charge of EPSCs. In high frequency trains, receptor desensitization, saturation and spill was reported. How does it influence the results?

6) The description of methods and analysis is very limited to the extent that it is e.g. not understandable how the cumulative plots are generated. Why are there steps of different sizes on the y-axes and multiple data points at the same y value?

7) The authors should discuss the following issues:

7A) It remains confusing what vesicles the authors refer to with "eager" and "reluctant". "Eager" is neither referenced nor explained. They claim that low Pr and high Pr vesicles are immediately releasable. Does this mean in parallel? How do the authors exclude that high frequency increases the rate of conversion from low to high Pr? Is it a sequential process before fusion?

7B) How does the reduced delay between AP and vesicle release affect the trains at 300Hz? Is e.g. desensitization, receptor saturation differently affected?

7C) The traces in Figure 4C from the end of the 50Hz train and the switch to 300Hz for the two genotypes look similar to me. It remains unclear how the authors derive a difference in the low Pr vesicles.

7D) Why would high Pr vesicles influence the time course of an EPSC?

7E) A similar amplitude after the 4th AP during a train is consistent with the description of superprimed vesicles (e.g. Taschenberger et al., 2016). Thus in the QKO, more vesicles could reside in the superprimed state. An estimate of the RRP from the steady state at the end of a train does not allow to distinguish between vesicle numbers in different Pr states.

Minor Comments:

1) In subsection "Higher throughput assay in primary cell culture" the authors refer the reader to a figure legend for the explanation of the authors' interpretation of the de-staining data. This explanation should be given in the main text.

2) The description of methods and analysis is very limited to the extent that it is e.g. not understandable how the cumulative plots are generated. Why are there steps of different sizes on the y-axes and multiple data points at the same y value?

Separate reviews (please respond to each point):

Reviewer #1:

The authors generate quod knockout mice, deleting Synaptophysin 1 and 2 and synaptogyrin 1 and 3. Subsequent analyses at Calyx and hippocampal synapses in the quad-KO demonstrate altered presynaptic vesicle release consistent with an increase in presynaptic release probability. The fact that this effect is observed at two very different synapses, in vitro and in vivo, argues for the generality of the observed effects. The paper ends with a modest amount of data on gene-specific contributions, highlighting a potentially interfering function for synaptophysin 2. In general, the experiments seem to have been performed to a high standard, are clearly presented and appropriately discussed. The findings are clear, and provide what could be considered a definitive set of experiments regarding the required action of this set of proteins for synaptic transmission. Given that very similar phenotypes are observed at Calyx and hippocampus, in vivo and in vitro, it seems unlikely that the effects will be caused by altered developmental trajectory of synapse development. However, this was never directly addressed. There is a curious effect on synaptic delay at the Calyx. It would be nice to know if this was altered at the hippocampal synapses. My only other comment is that an assessment of intrinsic excitability would be easily achieved in both systems, somatic in the hippocampus and presynaptically at the Calyx. Given that there is a shortened delay in the calyx and there is evidence of increased excitability, I am curious if the action potential waveform is altered. While this would be expected to alter the waveform of the EPSC, it remains a final piece of the puzzle that could be resolved. Given that these studies are carried out in quad KO animals, I am hesitant to require additional work unless it provides fundamental evidence in favor of the existing data, or has the potential to alter the fundamental conclusions in some manner.

Reviewer #2:

Raja et al. present an interesting story on tetraspan vesicle membrane proteins of the synaptophysin and synaptogyrin families. Synaptic vesicles (SVs) contain three types of tetraspan membrane proteins, synaptophysins, synaptogyrins, and SCAMPs. Although synaptophysins are among the most abundant SV proteins, their function – and the function of synaptophysins, synaptogyrins, and SCAMPs in general – has remained enigmatic. Indeed, genetic elimination of synaptophysin, synaptogyrin, and SCAMP in C. elegans, which expresses only one isoform of each family of tetraspan vesicle membrane proteins, has no effect on spontaneous or evoked synaptic transmission (Abraham et al., 2006). Raja et al. performed a related study on mammalian synaptophysins and synaptogyrins, focusing on quadruple KO mice lacking all synaptophysins and synaptogyrins. The key finding is that the loss of synaptophysins and synaptogyrins causes increased SV release probability and faster excitation-secretion coupling. This is clearly interesting and important for the synapse biology field.

Raja et al. based their analysis on electrophysiological recordings at the calyx of Held and Schaffer collateral synapses in the hippocampus, and on FM4-64 loading/destaining experiments with cultured hippocampal neurons. Overall, the experiments were done in a stringent manner and the data look convincing. Nevertheless, I have several comments regarding the data and their interpretation that the authors should address.

1) In principle, I like the mouse KO approach. However, when four proteins are removed from SVs – and among these some of the most abundant SV proteins – the issue of 'collateral damage' arises. In essence, I think the authors can at present not be sure that the quadruple KO SVs are properly equipped with all relevant protein components. The data shown in Figure S1 already show trends of changes in SV proteins in the whole hippocampus of quadruple KOs. To rule out substantial changes at the level of SV composition, it would be necessary to systematically assess protein levels in purified SV fractions from quadruple KOs. The problem here may be that not many labs can do this properly.… but it could be done within a reasonable time frame on the basis of a collaboration (e.g. with people like Reinhard Jahn). If this is out of the question, the authors should at least discuss this issue and tone down the manuscript text regarding collateral changes in SV composition after loss of all synaptophysins and synaptogyrins. I personally think that SV composition may well be changed, contributing to the phenotypical changes observed.

2) An issue related to the one above (point 1) concerns the intrinsic 'fusogenicity' of SVs after loss of all synaptophysins and synaptogyrins. Upon removal of these proteins, one has to assume that large SV surface areas are stripped of a 'protein coat', so that much more 'naked' lipid membrane is exposed than in wild-type SVs. What argues against the possibility that this alone changes the fusion characteristics of SVs in a manner that might explain the phenotypes observed (e.g. because no 'protein coat' is present to inhibit the interactions of the SV and target membrane, thereby facilitating fusion)? I think this issue should at least be discussed.

3) In subsection "No alteration in the timing of vesicle trafficking", the authors relate their findings to earlier claims that synaptophysin 1 may play a role in post-endocytic clearing of SV material from the presynaptic membrane. In this context, they cite three papers, two of which mainly addressed a role of Synaptophysin in endocytosis without focusing on release-site clearance, which are related but different processes. I think the data presented in the present study cannot be taken as strong evidence against a role of synaptophysins and synaptogyrins in endocytosis. To make this claim, the authors would have to assess endocytosis directly. If this is out of the question, the corresponding text should be toned down.

4) In the Abstract and the main text, the authors state that they studied "a broad range of synapse types". I find this a bit hyperbolic. They studied calyx of Held synapses, Schaffer collateral synapses, and synapses of cultured hippocampal neurons. This is nice – I do not want to be misunderstood here – but not a "broad range". After all, some calyx experts have repeatedly made the casual statement that the calyx of Held is like a parallel array of hippocampal-like synapses, and Schaffer collateral synapses in slices and synapses in cultured hippocampal neurons are not so dissimilar either.

5) Based on the argument above, I am not convinced that the authors' "comparison between calyx of Held and Schaffer collateral synapses is a good test for the generality [of the phenotypes they observe] across synapse types". On the contrary, I think the general relevance of the present findings remains questionable. This is particularly critical in view of a corresponding study on C. elegans (Abraham et al., 2006), which showed that the complete loss of synaptophysin, synaptogyrin, and SCAMP has no effect on spontaneous or evoked synaptic transmission and which is not even cited in the present study. I think that not citing the study by Abraham et al., 2006, is a problematic omission that must be rectified. Further, I think that in the course of discussing the Abraham-paper the whole 'generality' discussion should be toned down in view of the arguments I made above.

6) The discussion part of the present paper is too speculative for my taste. The link between the present study and phosphorylation of synaptophysins and synaptogyrins, mover, GITs, ELKS, or tomosyn feels a bit constructed and arbitrary. After all, several other mutants show increased vesicular release probability. Likewise, the discussion of "activity-dependent redistribution of synaptic efficacy" goes too far for my taste; at present, there is no connection. Instead, I suggest to discuss the issues mentioned above, which directly relate to the actual findings of the present study.

Minor Comments:

In subsection "Higher throughput assay in primary cell culture" the authors refer the reader to a figure legend for the explanation of the authors' interpretation of the de-staining data. I think this explanation should be given in the main text.

Reviewer #3:

This is an interesting study by Raja and colleagues to revisit the function of the synaptic vesicle protein families of synaptogyrins and synaptophysins. Using a combination of multiple KO mice, they achieved a step forward in uncovering one of their physiological functions.

The function of these protein families remained largely illusive and controversial. Overexpression previously showed an inhibitory role of synaptogyrins and synaptophysin on release in PC12 cells. Here, authors confirm this function and report that these proteins act at different synapses in the CNS in a similar way. Authors report that isoforms function partially redundantly and a full functional loss is only achieved with certain deletion of multiple isoforms. They describe an interesting effect on the delay between action potentials and vesicle release as well as provide a robust analysis of changes in release probability, while vesicle pools sizes appear unchanged. However, the further dissection of the mechanism remains inconclusive.

– that bind via VAMP 2 to the machinery that catalyzes exocytosis." Misleading statement. They might bind to VAMP 2 before it engages into the fusion machinery.

– It remains confusing what vesicles the authors refer to with "eager" and "reluctant". "Eager" is neither referenced nor explained. They claim that low Pr and high Pr vesicles are immediately releasable. Does this mean in parallel? How do the authors exclude that high frequency increases the rate of conversion from low to high Pr? Is it a sequential process before fusion?

– Why would high Pr vesicles influence the time course of an EPSC?

– A similar amplitude after the 4th AP during a train is consistent with the description of superprimed vesicles (e.g. Taschenberger et al., 2016). Thus in the QKO, more vesicles could reside in the superprimed state. An estimate of the RRP from the steady state at the end of a train does not allow to distinguish between vesicle numbers in different Pr states.

– The traces in Figure 4C from the end of the 50Hz train and the switch to 300Hz for the two genotypes look similar to me. Not clear to me how the authors derive a difference in the low Pr vesicles.

– Some proteins were analyzed, but not proteins previously shown to regulate release probability, such as RIM, Munc13, Rab3, Complexin. Are those proteins upregulated and mediate the increase in Pr?

– The description of methods and analysis is very limited to the extent that it is e.g. not understandable how the cumulative plots are generated. Why are there steps of different sizes on the y-axes and multiple data points at the same y value?

– Authors estimate quanta/vesicle numbers from the charge of EPSCs. In high frequency trains, receptor desensitization, saturation and spill was reported. How does it influence the results?

– How does the reduced delay between AP and vesicle release affect the trains at 300Hz? Is e.g. desensitization, receptor saturation differently affected?

https://doi.org/10.7554/eLife.40744.021

Author response

Reviewer #1:

The authors generate quod knockout mice, deleting Synaptophysin 1 and 2 and synaptogyrin 1 and 3. Subsequent analyses at Calyx and hippocampal synapses in the quad-KO demonstrate altered presynaptic vesicle release consistent with an increase in presynaptic release probability. The fact that this effect is observed at two very different synapses, in vitro and in vivo, argues for the generality of the observed effects. The paper ends with a modest amount of data on gene-specific contributions, highlighting a potentially interfering function for synaptophysin 2. In general, the experiments seem to have been performed to a high standard, are clearly presented and appropriately discussed. The findings are clear, and provide what could be considered a definitive set of experiments regarding the required action of this set of proteins for synaptic transmission. Given that very similar phenotypes are observed at Calyx and hippocampus, in vivo and in vitro, it seems unlikely that the effects will be caused by altered developmental trajectory of synapse development. However, this was never directly addressed. There is a curious effect on synaptic delay at the Calyx. It would be nice to know if this was altered at the hippocampal synapses. My only other comment is that an assessment of intrinsic excitability would be easily achieved in both systems, somatic in the hippocampus and presynaptically at the Calyx. Given that there is a shortened delay in the calyx and there is evidence of increased excitability, I am curious if the action potential waveform is altered. While this would be expected to alter the waveform of the EPSC, it remains a final piece of the puzzle that could be resolved. Given that these studies are carried out in quad KO animals, I am hesitant to require additional work unless it provides fundamental evidence in favor of the existing data, or has the potential to alter the fundamental conclusions in some manner.

We agree with the reviewer on all of these points and appreciate the input very much.

The fact that we analyzed a variety of synapse types embedded within neuronal networks that are subject to dissimilar developmental forces (including networks formed by neurons grown in dissociated cell culture) is now mentioned in the second paragraph of the Discussion.

We thought the comments about the synaptic delay at the calyx of Held (Figure 1D of the initial submission) were especially insightful and we agree that this merits a better analysis. Instead, though, we have now removed the panel. The original experiments were done blind to genotype. However, the effect was not reproduced in a second data set designed specifically to address the present comments, possibly owing to an unexpected amount of variation between preparations. At this point, we would like to be cautious because we believe the decreased time – if truly occurring downstream of Ca2+ influx – may have important implications for understanding the coupling between Ca2+ and how exocytosis is catalyzed. We do not have the resources to do this well in a short amount of time, but plan to follow this up later. In any case, we have measured action potential conduction times along the axon in the calyx of Held preparation, and did not find any difference between WT and QKO.

Reviewer #2:

Raja et al. present an interesting story on tetraspan vesicle membrane proteins of the synaptophysin and synaptogyrin families. Synaptic vesicles (SVs) contain three types of tetraspan membrane proteins, synaptophysins, synaptogyrins, and SCAMPs. Although synaptophysins are among the most abundant SV proteins, their function – and the function of synaptophysins, synaptogyrins, and SCAMPs in general – has remained enigmatic. Indeed, genetic elimination of synaptophysin, synaptogyrin, and SCAMP in C. elegans, which expresses only one isoform of each family of tetraspan vesicle membrane proteins, has no effect on spontaneous or evoked synaptic transmission (Abraham et al., 2006). Raja et al. performed a related study on mammalian synaptophysins and synaptogyrins, focusing on quadruple KO mice lacking all synaptophysins and synaptogyrins. The key finding is that the loss of synaptophysins and synaptogyrins causes increased SV release probability and faster excitation-secretion coupling. This is clearly interesting and important for the synapse biology field.

Raja et al. based their analysis on electrophysiological recordings at the calyx of Held and Schaffer collateral synapses in the hippocampus, and on FM4-64 loading/destaining experiments with cultured hippocampal neurons. Overall, the experiments were done in a stringent manner and the data look convincing. Nevertheless, I have several comments regarding the data and their interpretation that the authors should address.

We thank the reviewer for insightful comments, and pointing out key areas where we were naive. We have now followed essentially all of the recommendations except a single one about “redistribution of synaptic efficacy” as outlined below.

1) In principle, I like the mouse KO approach. However, when four proteins are removed from SVs – and among these some of the most abundant SV proteins – the issue of 'collateral damage' arises. In essence, I think the authors can at present not be sure that the quadruple KO SVs are properly equipped with all relevant protein components. The data shown in Figure S1 already show trends of changes in SV proteins in the whole hippocampus of quadruple KOs. To rule out substantial changes at the level of SV composition, it would be necessary to systematically assess protein levels in purified SV fractions from quadruple KOs. The problem here may be that not many labs can do this properly.… but it could be done within a reasonable time frame on the basis of a collaboration (e.g. with people like Reinhard Jahn). If this is out of the question, the authors should at least discuss this issue and tone down the manuscript text regarding collateral changes in SV composition after loss of all synaptophysins and synaptogyrins. I personally think that SV composition may well be changed, contributing to the phenotypical changes observed.

We have now collaborated with Dr Jahn to analyze a larger variety of synaptic proteins. We used purified synaptosomes instead of purified synaptic vesicles to save time and to simultaneously analyze presynaptic proteins that are not integrated into vesicle membranes. However, purified synaptosomes are thought to provide an accurate readout for the relative abundance of synaptic vesicle proteins (Wilhelm et al., 2014).

We did see ~50% reduction in VAMP 2 levels, in-line with a trend in our original analysis, and consistent with a significant reduction seen previously for synaptophysin 1 knockouts (McMahon et al., 1996). This is intriguing and may be related to the phenotype. If so, however, the mechanism is not obvious because VAMP 2 is required for exocytosis. This is now discussed in the Discussion section (fifth paragraph).

Besides the decrease in VAMP 2, we did not see alterations in: other synaptic vesicle proteins including synaptotagmin 1, vGlut 1&2, and vATPase; or in other proteins previously implicated in regulating release as recommended by the third reviewer (i.e., SNAP25, syntaxin1, RIM1/2, Munc13-1, Complexin, and Rab3).

2) An issue related to the one above (point 1) concerns the intrinsic 'fusogenicity' of SVs after loss of all synaptophysins and synaptogyrins. Upon removal of these proteins, one has to assume that large SV surface areas are stripped of a 'protein coat', so that much more 'naked' lipid membrane is exposed than in wild-type SVs. What argues against the possibility that this alone changes the fusion characteristics of SVs in a manner that might explain the phenotypes observed (e.g. because no 'protein coat' is present to inhibit the interactions of the SV and target membrane, thereby facilitating fusion)? I think this issue should at least be discussed.

The results in Figure 7E suggesting that synaptophysin 2 competitively inhibits the function of synaptophysin 1 and synaptogyrin 3 would seem to argue against the possibility. This is now mentioned explicitly in the Discussion (end of second paragraph).

3) In subsection "No alteration in the timing of vesicle trafficking", the authors relate their findings to earlier claims that synaptophysin 1 may play a role in post-endocytic clearing of SV material from the presynaptic membrane. In this context, they cite three papers, two of which mainly addressed a role of Synaptophysin in endocytosis without focusing on release-site clearance, which are related but different processes. I think the data presented in the present study cannot be taken as strong evidence against a role of synaptophysins and synaptogyrins in endocytosis. To make this claim, the authors would have to assess endocytosis directly. If this is out of the question, the corresponding text should be toned down.

We agree and have now adjusted the text accordingly. We continue to acknowledge that our results do not contradict the results of endocytosis experiments on synaptophysin 1 knockouts, but we removed the speculation about an alternative interpretation (eighth paragraph of Discussion).

4) In the Abstract and the main text, the authors state that they studied "a broad range of synapse types". I find this a bit hyperbolic. They studied calyx of Held synapses, Schaffer collateral synapses, and synapses of cultured hippocampal neurons. This is nice – I do not want to be misunderstood here – but not a "broad range". After all, some calyx experts have repeatedly made the casual statement that the calyx of Held is like a parallel array of hippocampal-like synapses, and Schaffer collateral synapses in slices and synapses in cultured hippocampal neurons are not so dissimilar either.

We have now replaced “broad range” with “variety”.

We agree that calyx of Held synapses are similar at the level of first principles to others with active zones. In fact, we published evidence supporting this in Mahfooz et al., 2016. However, our point then was not that there are no differences between synapses. Indeed, we saw order of magnitude sized differences in vesicle recruitment rates, and baseline mean pv was approximately triple at calyx of Held compared to Schaffer collaterals. Instead, our point was that the different types of synapses shared the same first principles and that the differences in function would be caused by differences in parameter values rather than by qualitative differences in underlying mechanisms. We believe that comparing Schaffer collateral synapses to calyces of Held is a particularly good test for generality across synapse types in the present context because of the large differences in the timing of vesicle recruitment, which is particularly relevant because a current hypothesis predicts that the timing would be altered in the knockout, and that the defect would be more prominent during heavier use.

5) Based on the argument above, I am not convinced that the authors' "comparison between calyx of Held and Schaffer collateral synapses is a good test for the generality [of the phenotypes they observe] across synapse types". On the contrary, I think the general relevance of the present findings remains questionable. This is particularly critical in view of a corresponding study on C. elegans (Abraham et al., 2006), which showed that the complete loss of synaptophysin, synaptogyrin, and SCAMP has no effect on spontaneous or evoked synaptic transmission and which is not even cited in the present study. I think that not citing the study by Abraham et al., 2006, is a problematic omission that must be rectified. Further, I think that in the course of discussing the Abraham-paper the whole 'generality' discussion should be toned down in view of the arguments I made above.

We agree that we were naive on this point. We now cite Abraham et al., 2006, and another study where synaptogyrin was knocked out of Drosophila, and discuss both in the third paragraph of the Discussion. Claims about generality are toned down throughout the manuscript.

6) The discussion part of the present paper is too speculative for my taste. The link between the present study and phosphorylation of synaptophysins and synaptogyrins, mover, GITs, ELKS, or tomosyn feels a bit constructed and arbitrary. After all, several other mutants show increased vesicular release probability. Likewise, the discussion of "activity-dependent redistribution of synaptic efficacy" goes too far for my taste; at present, there is no connection. Instead, I suggest to discuss the issues mentioned above, which directly relate to the actual findings of the present study.

We now limit mention of GITs and Mover to the last sentence of the Discussion, with no mention at all of ELKS and tomosyn.

We have preserved mention of activity-dependent re-distribution of synaptic efficacy. We would like to keep this because we think the term is an elegant way to describe the phenotype, and it would seem like an oversight to avoid making the connection.

Minor Comments:

In subsection "Higher throughput assay in primary cell culture" the authors refer the reader to a figure legend for the explanation of the authors' interpretation of the de-staining data. I think this explanation should be given in the main text.

The explanation is now included in the main text as recommended.

Reviewer #3:

This is an interesting study by Raja and colleagues to revisit the function of the synaptic vesicle protein families of synaptogyrins and synaptophysins. Using a combination of multiple KO mice, they achieved a step forward in uncovering one of their physiological functions.

The function of these protein families remained largely illusive and controversial. Overexpression previously showed an inhibitory role of synaptogyrins and synaptophysin on release in PC12 cells. Here, authors confirm this function and report that these proteins act at different synapses in the CNS in a similar way. Authors report that isoforms function partially redundantly and a full functional loss is only achieved with certain deletion of multiple isoforms. They describe an interesting effect on the delay between action potentials and vesicle release as well as provide a robust analysis of changes in release probability, while vesicle pools sizes appear unchanged. However, the further dissection of the mechanism remains inconclusive.

We thank the reviewer for raising these points, some of which invoke what we believe are key outstanding issues for understanding presynaptic function in general. We believe that we have well-reasoned responses, which are outlined below.

– that bind via VAMP 2 to the machinery that catalyzes exocytosis." Misleading statement. They might bind to VAMP 2 before it engages into the fusion machinery.

We agree that this was poorly worded. It seems that there is a range of thoughts about this, which is now referenced and discussed in the fourth paragraph of the Discussion.

– It remains confusing what vesicles the authors refer to with "eager" and "reluctant". "Eager" is neither referenced nor explained. They claim that low Pr and high Pr vesicles are immediately releasable. Does this mean in parallel? How do the authors exclude that high frequency increases the rate of conversion from low to high Pr? Is it a sequential process before fusion?

a) The term “eager” has now been replaced by the concept of vesicles belonging to a high pv subdivision of the RRP;

b) We use pv and mean pv instead of Pr to avoid a separate sort of confusion;

c) We do think that low pv vesicles are immediately releasable, in parallel with high pv vesicles;

d) We do not exclude conversions between low to high pv states, or frequency-dependent acceleration of the mechanism (Legend of Figure 5A);

e) We believe that the evidence against key alternate scenarios where synaptophysin family members would modulate transition from an un-releasable to releasable state is strong.

The explanations are below.

We think that low pv vesicles are immediately releasable because release of neurotransmitter continues to be tightly synchronized to action potentials during frequency jumps that are initiated after the high pv vesicles have been eliminated (Figure 5B and Figure 5—figure supplement 1C). Moreover, the low pv vesicles likely are released in parallel with high pv vesicles during low frequency stimulation when the RRP is full because synapses with more low pv vesicles express more paired-pulse facilitation whereas synapses with fewer express more paired-pulse depression (Figure 9 of Mahfooz et al., 2016).

We are aware that conversion from low to high pv states plays a key role in models from other groups, and we do not rule this out – so long as vesicles in the low pv subdivision can be released directly to account for the synchronicity of release immediately after the frequency jump (see Results under heading “Elevated probability of release.…”, and Legend of Figure 5A). Nor do we rule out the idea that high frequency stimulation might accelerate the rate of transfer. In fact, we were able to elucidate a surprising quantitative constraint in Mahfooz et al., 2016; specifically, the transfer between subdivisions would have to be ∼9/s during 300Hz stimulation, at room temperature – not substantially faster or slower – and would have to be reversible. This constraint is not altered at QKO synapses

Notably, our own WT and QKO data could be modeled equally well by eliminating the transfer in both directions. This eliminates a degree of freedom and has interesting consequences for how synapses might be modulated at the level of mechanism (last paragraph of Results of Mahfooz et al., 2016). In addition, at least one of the recent articles proposing a serial scheme acknowledged that the authors could not rule out alternatives where low and high pv vesicles are processed in parallel (Miki et al., 2016). And finally, there seems to be a growing body of evidence that stable high and low release probability release sites do co-exist at some synapse types (Hu et al., 2013; Muller et al., 2015; Böhme et al., 2016).

Nevertheless, we have considered alternate scenarios where the low pv vesicles are not actually immediately releasable and the ongoing synchronous transmitter release in the absence of high pv vesicles is explained by fast transfer of a tiny fraction of the un-releasable vesicles to an immediately releasable state in the time between action potentials. The fraction would be the value of mean pv we estimated for the low pv vesicles. Our understanding is that the reviewer is asking about the subset of models in this category where the essentially equivalent value for mean pv at 50 and 100Hz in Figure 5E is explained by activity dependent acceleration of the rate of transfer that is close to directly proportional to the frequency of stimulation.

We found that models of this class are not compatible with the results of the frequency jump experiments. We arrived at this conclusion as follows. We reasoned that the acceleration mechanism could not influence the rate of transfer from the un-releasable to immediately releasable state until after the first 3.33 ms interstimulus interval during 300Hz stimulation. In contrast, the rate of transfer during the first 3.33 ms interstimulus interval would have to be equivalent to the rate during the first 3.33ms of the 20 or 10ms interstimulus intervals during the preceding 50 or 100Hz stimulation. Therefore, one would expect 6-fold or 3-fold fewer new immediately releasable vesicles at the end of the first 3.33ms interstimulus interval compared to before the preceding action potential. If so, the second pulse during 300Hz stimulation would elicit paired-pulse depression rather than the facilitation seen in Figure 5D and Figure supplement 1B. And, indeed, one would expect twice as much paired-pulse depression in Figure 5D compared to Figure supplement 1B (i.e., because the transfer rate during 50Hz stimulation would be half the rate when stimulation is 100Hz). If the acceleration was very fast, the third pulse could then elicit facilitation.This reasoning is now included in the Legend of Figure 5—figure supplement 1.

For completeness: We are aware that it would be possible to rescue the alternate scenario by replacing the premise that activity accelerates the transfer from un-releasable to immediately releasable states with the premise that single action potentials – at any frequency – always induce transfer of a similar fraction of un-releasable vesicles to the immediately releasable state. If so, synaptophysin family members would have two functional roles:

a) Negative regulation of the releasablility of immediately releasable vesicles; and,

b) Negative regulation of the fraction of un-releasable vesicles that are transferred to the immediately releasable state after each action potential.

However, we do not include a discussion of models of this final class in the manuscript because our understanding is that none have yet been proposed, and it seems unlikely to us because the two roles would have analogous functional effects on the releasability of vesicles in the high and low pv subdivisions of the RRP via dissimilar types of mechanisms.

– Why would high Pr vesicles influence the time course of an EPSC?

We are not certain what is meant here. We would not necessarily expect an influence, but such a factor could explain the results in Fedchyshyn and Wang, 2007.

– A similar amplitude after the 4th AP during a train is consistent with the description of superprimed vesicles (e.g. Taschenberger et al., 2016). Thus in the QKO, more vesicles could reside in the superprimed state.

We agree that the plot in Figure 3C is reminiscent of the effect produced by PdBu in Taschenberger et al., 2016. However, our understanding is that the superpriming idea proposed in Taschenberger et al., 2016, supposes that vesicles are converted from the high pv state to the superprimed state, whereas mean pv at QKO synapses seems to be elevated for both the low and high pv subdivisions at QKO synapses; this is now noted in the seventh paragraph of the Discussion, which now includes a reference to Taschenberger et al., 2016. Note, however, that the plot in Figure 3C is equally reminiscent of Figure 2B of Mahfooz et al., 2016, where mean pv was increased by raising extracellular Ca2+, which presumably would not alter the number of superprimed vesicles. Also relevant: Ongoing experiments in the lab suggest that post-tetanic potentiation is intact in QKO synapses; the results will be included in a separate article (we haven’t tried PdBu yet).

An estimate of the RRP from the steady state at the end of a train does not allow to distinguish between vesicle numbers in different Pr states.

We agree and did not mean to suggest otherwise. A key point is that we define mean pv to be equal to the number of release events following an action potential divided by the quantal content of the RRP immediately beforehand; mean pv now replaces some instances of pv in the previous version to make this more clear. Thus, mean pv is the mean probability of release of all vesicles within the RRP. Our reasoning does not rely on any assumptions about the distribution of release probabilities of the individuals within the RRP.

We reasoned that the mean pv-value for the vesicles remaining within the RRP during steady state 50 or 100 Hz stimulation can be calculated by dividing the number of quanta released by each action potential during steady state stimulation by the steady state quantal content of the RRP (Figure 5E) just as mean pv for the vesicles within the RRP after long rest intervals can be calculated by dividing the number of quanta released by the first action potential in a train by the quantal content of the RRP when completely replenished (Figure 3E).

We then found that the mean pv-value for vesicles remaining in the RRP during steady state 50 or 100 Hz stimulation was lower than the value when the RRP was full for both genotypes, confirming that at least some of the high pv vesicles were eliminated. Next, the mean pv value during 50 Hz stimulation was essentially equivalent to during 100 Hz stimulation, even though the standing fullness of the RRP was substantially less during 100 Hz stimulation (Figure 5—figure supplement 2), indicating that almost all of the high pv vesicles were eliminated during both frequencies of stimulation. Finally, the mean pv-values for QKO synapses during 50 and 100 Hz stimulation were higher than the corresponding values for WT synapses (Figure 5E). We interpret these results as indicating that synaptophysin family proteins negatively regulate the probability of release of both low and high pv vesicles, and as arguing against scenarios where only vesicles in the high pv subdivision are affected.

– The traces in Figure 4C from the end of the 50Hz train and the switch to 300Hz for the two genotypes look similar to me. Not clear to me how the authors derive a difference in the low Pr vesicles.

We have clarified this in the Results and Figure 5F where the data are normalized so that both QKO and WT results can be plotted on the same graph. Most of the difference between QKO and WT is not actually visible in Figure 5C – Figure 4C in the previous version is Figure 5C in the current version – because of how the data are binned; each plot on the graph is the sum of 6 EPSCs. However, one can still see that – even when binned – responses did decay faster at QKO synapses, and reached a new steady state sooner.

– Some proteins were analyzed, but not proteins previously shown to regulate release probability, such as RIM, Munc13, Rab3, Complexin. Are those proteins upregulated and mediate the increase in Pr?

We have now looked at levels of RIM1/2, Mun13-1, Raba,b,c, and Complexin in purified synaptosomes in collaboration with Dr. Reinhard Jahn. The results are in Figure 1 and Figure 1—figure supplement 1. We did not find clear evidence for upregulation of any.

– The description of methods and analysis is very limited to the extent that it is e.g. not understandable how the cumulative plots are generated. Why are there steps of different sizes on the y-axes and multiple data points at the same y value?

The cumulative plots of summary statistics have been replace with standard bar graphs; individual values are plotted on top of the bars. The cumulative plots of electrophysiological responses during repetitive stimulation have not been altered, but these are standard in the field and our understanding is that it was the plots of summary statistics which were seen as difficult to interpret.

– Authors estimate quanta/vesicle numbers from the charge of EPSCs. In high frequency trains, receptor desensitization, saturation and spill was reported. How does it influence the results?

Responses:

a) Desensitization: Our conditions were different from the conditions of studies reporting postsynaptic receptor desensitization and saturation in two regards that we believe are relevant: (1) we used tissue from older animals; and (2), experiments where quantal content is estimated from responses during repetitive stimulation were conducted in the presence of 1mM kynurenic acid. We have now added Figure 3—figure supplement 1A-E to show that receptor desensitization does not play a substantial role under these conditions.

b) Saturation: Likewise, receptor saturation was not a concern in the presence of kynurenic acid because responses were reduced greatly in size, by 85% ± 2% for QKO and 85% ± 3% for WT. Saturation (or voltage clamp errors) could affect estimates of EPSC size recorded in the absence of kynurenic acid when measured as the peak or current integral. However, the estimates of amount of block by 1mM kynurenic acid were likely not affected substantially because we measured the slope of the rising phase (20 to 60%) instead of the peak or current integral for this purpose.

c) Spill: We have had personal conversations with calyx of Held experts who have raised concerns that part of the response we see during steady state 300Hz stimulation might arise from glutamate that spills over from nearby synaptic clefts. Newly added Figure 3—figure supplement 1F now addresses this possibility.

In any case:

i) Our estimates of quantal content and unitary rates of vesicle recruitment during repetitive use agree with estimates from other studies that used different techniques (Neher, 2010);

ii) We think the estimates are accurate, but even if not, the key conclusions are drawn from either the presence or absence of differences between WT and QKO synapses, rather than from the precise values of quantal content or recruitment rates; and,

iii) The key conclusions that involve quantal estimates are additionally demonstrated in two other ways using different techniques and preparations. In particular, the experiments in cell culture in Figure 7D avoid completely the involvement of postsynaptic receptors

– How does the reduced delay between AP and vesicle release affect the trains at 300Hz? Is e.g. desensitization, receptor saturation differently affected?

We have removed the evidence for a reduced delay between AP and vesicle release as explained in the response to the first reviewer. The results in Figure 3—figure supplement 1AE show that receptor desensitization does not seem to play a role in 1mM or more kynurenic acid

References:

Fedchyshyn MJ & Wang LY (2007). Activity-dependent changes in temporal components of neurotransmission at the juvenile mouse calyx of Held synapse. Journal of Physiology 581, 581–602.

Neher E (2010). What is Rate-Limiting during Sustained Synaptic Activity: Vesicle Supply or the Availability of Release Sites. Frontiers in Synaptic Neuroscience 2, 144.

https://doi.org/10.7554/eLife.40744.022

Article and author information

Author details

  1. Mathan K Raja

    Department of Neuroscience, Universidad de Navarra, Pamplona, Spain
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7727-6994
  2. Julia Preobraschenski

    Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Sergio Del Olmo-Cabrera

    Institute for Neurosciences CSIC-UMH, San Juan de Alicante, Spain
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Rebeca Martinez-Turrillas

    Department of Neuroscience, Universidad de Navarra, Pamplona, Spain
    Contribution
    Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Reinhard Jahn

    Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany
    Contribution
    Formal analysis, Supervision, Investigation, Methodology, Writing—review and editing
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1542-3498
  6. Isabel Perez-Otano

    1. Department of Neuroscience, Universidad de Navarra, Pamplona, Spain
    2. Institute for Neurosciences CSIC-UMH, San Juan de Alicante, Spain
    Contribution
    Formal analysis, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  7. John F Wesseling

    1. Department of Neuroscience, Universidad de Navarra, Pamplona, Spain
    2. Institute for Neurosciences CSIC-UMH, San Juan de Alicante, Spain
    Contribution
    Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    johnfwesseling@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7565-2594

Funding

Ministerio de Ciencia y Tecnología (BFU2009-12160)

  • John F Wesseling

Ministerio de Ciencia y Tecnología (SEV-2013-0317)

  • Isabel Perez-Otano
  • John F Wesseling

Universidad de Navarra

  • Isabel Perez-Otano
  • John F Wesseling

Ministerio de Educación, Cultura y Deporte (Salvador de Madariaga Visiting Scholarship)

  • John F Wesseling

Jeronimo de Ayanz program

  • John F Wesseling

Ministerio de Ciencia y Tecnología (BFU2016-80918R)

  • John F Wesseling

Ministerio de Ciencia y Tecnología (SAF2013-48983R)

  • Isabel Perez-Otano
  • John F Wesseling

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Daniela Urribarri for genotyping the animals and other technical assistance, Dr. Thomas Südhof for providing the equipment, animals, and reagents for the experiments conducted in hippocampal slices, Dr. Santiago Canals for help with the statistical analysis, and Drs. Joan Galcerán, Juan Lerma, Donald Lo, Rafa Fernández-Chacón, and Dani Gitler for suggestions about how to write the manuscript and Dr. David Litvin for suggestions about the writing and help with illustrations.

Ethics

Animal experimentation: Animal protocols were approved by the Universidad de Navarra and Universidad Miguel Hernandez Institutional Animal Care and Use Committees (2017/VSC/PEA/00196) and conformed to the guidelines of Spanish Royal Decree 1201/2005.

Senior Editor

  1. Gary L Westbrook, Vollum Institute, United States

Reviewing Editor

  1. Graeme W Davis, University of California, San Francisco, United States

Reviewers

  1. Graeme W Davis, University of California, San Francisco, United States
  2. Nils Brose, Max Planck Institute of Experimental Medicine, Germany

Publication history

  1. Received: August 15, 2018
  2. Accepted: April 18, 2019
  3. Version of Record published: May 15, 2019 (version 1)

Copyright

© 2019, Raja et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 354
    Page views
  • 75
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)