The Lack of Synapsin Alters Presynaptic Plasticity at Hippocampal Mossy Fibers in Male Mice

Synthesis

The authors have address most major concerns abot data analysis in the revision and the mabuscript is much improved.

One issue that remains is a comprehensive assessment of the differences between the single KO in prior studies and the double knockout presented here. Specifically, the Discussion needs to clarify how the findings of this study differ from the results of earlier studies in single KOs. For example, changes in excitability in glutamatergic synapses (Fig. 1 A,B) were also found in Syn II KO. The reduced facilitation (Fig. 1 C) was observed in Syn I KO. The increased depression (Fig. 2 B) and SV depletion (Fig. 4 C) were observed in Syn II KO. The authors should explain which of the observed phenotypes were observed at single KO/DKO, which of them are additive for different Syn genes, and which are not additive. A table summarizing differences between genotypes would be helpful.

Author Response

eNeuro eN-TNWR-0330-23X "The lack of synapsin alters presynaptic plasticity at hippocampal mossy fibers in male mice" Dear Prof. Grubb, Thank you for dedicating your time to reviewing our manuscript and for considering our work of potential interest.

We are grateful for the reviewers' comments and have followed their suggestions.

In summary, we performed new electrophysiological recordings to improve figures 1, 1-2, 2, 2-1 and 3 (new data points from 2 additional WT mice). This especially addressed some uneven datasets in figures 1-2 and 2, as pointed out by reviewer 1. We also performed new analysis of the electron microscopy images to deepen the understanding of morphological differences between genotypes and conditions. For this, we analyzed the active zone area of single active zones in boutons from WT and SynTKO animals, in treated and untreated conditions (Figure 5). We also analyzed the number of vesicles in vicinity to the active zone membrane (correlate for the readily-releasable pool) as well as the overall distribution of vesicle distances to the membrane (Figure 4-1). Furthermore, we improved the data visualization in figures 1, 1-2, 2, 4 and 5, as suggested by the reviewers. Our efforts resulted in the substantially revised version of our manuscript, that we are happy to submit today with the updated title: "The lack of synaptic alters presynaptic plasticity at hippocampal mossy fibers in male mice". The changes to the text, mainly in the results section and in the discussion, are marked in blue.

Below we enclose a point-by-point response to the reviewers' concerns.

Synthesis of Reviews:

Both Reviewers see potential value in the study, but they both raise serious concerns regarding different elements of the MS. Please address all of these concerns fully before resubmitting a substantially revised version. ___ Reviewer 1:

This is an interesting study, which reports new observations on functional and ultrastructural phenotypes in Syn TKO mossy fiber boutons. However, several technical and conceptual issues need to be addressed. Although electrophysiology data is solid, the EM part needs clarification and possibly additional work. Also, some aspects in the interpretation of Syn III function need to be clarified.

Major points:

1. Fig. 4. The panel A suggests that there may be problems with TEM quality: the vesicle do not seem clearly visibly, until they are outlined by the detection software. This raises the question of whether the software outlines and detects the vesicles reliably. Showing high magnifications micrographs with clearly seen vesicles could resolve this issue.

We thank Reviewer 1 for the careful revision of our manuscript and for the suggestion to improve the visualization of our data. We improved panel a) in figure 4, which now includes high magnification micrographs from WT and SynTKO synapses with clearly visible synaptic vesicles, without the detection software outlines (two images at the bottom of the panel). We also included two high magnification images from WT and SynTKO mossy fibers showing vesicles and active zones in the new supplementary figure 4-1. The striking reduction in vesicle density is clearly noticeable in the images from the SynTKO mossy fiber boutons.

2. Fig. 4 reports the reduced vesicle density in TKO. This was reported for every synapse studied for SynI/SynII DKO and also for SynII KO, and therefore this result is expected. The Discussion make implications about RRP and its relations to different forms of plasticity, however the authors do not try to assess RRP by EM. The EM results would have more merit if the authors quantify vesicles in the vicinity of AZs and make implications about RRP from their EM analysis.

We thank Reviewer 1 for raising this interesting point. We analyzed the number of vesicles and their distance to specific active zones. We plotted the number and proportion of vesicles within 40 nm from the active zone (bona fide correlate of the readily-releasable pool) to gain more insights into the general vesicle distribution as well as possible changes in the RRP.

Our data confirm once again the expected reduced overall vesicle density in the whole synapse (figure 4-1). Interestingly, this analysis further reveals a reduction in the absolute vesicle number within 40 nm from the active zone in boutons from SynTKO animals. When chemically potentiated with forskolin, though, both genotypes showed an increase in the vesicle number within 40 nm. These findings could indicate that parallel synapsin-independent mechanisms are in place that modulate the RRP during potentiation.

We did not include this analysis in our previous version of the manuscript because we are aware of technical limitations in the chemical fixation methodology as well as in the detection algorithm (we discussed these limitations in the updated method section and discussion). For these reasons, we now present the additional findings not in a main figure, but in supplemental figure 4-1. We believe that, while the high-pressure-freezing cryo-fixation method would have been a better choice, the analysis we present here gives us valuable insights for understanding the SynTKO phenotype. The changes we observed, are in line with previous work done with cryo fixation (Orlando et al., 2021; Kim et al., 2023). We revised the manuscript methods, results and discussion section to incorporate these additional findings.

3. Fig. 5. This figure needs to show unedited micrographs with identifiable AZs. It also needs clear description of AZ hallmarks and the evidence that they can be seen in unedited micrographs. As is, the figure only shows the results of the image analysis, and this is not acceptable.

We thank the reviewer for pointing out the need to clearly show active zones in raw images. We now improved figure 5 by adding unedited EM images from both genotypes and conditions (Figure 5a). There, active zones are indicated by black arrows. Briefly, we identified active zones as the area of membrane opposite to a postsynaptic density/spine profile and with vesicles attached and accumulated next to the membrane.

4. More detail is needed on how the mossy fiber boutons were identified. The methods section states: "The stratum lucidum of the hippocampal region CA3 was easily distinguishable for the presence of big mossy fiber boutons and for its localization just above the pyramidal cell layer." What magnification was used to determine the localization? How heterogeneous is the stratum lucidum? What criteria were used for the identification of mossy fiber boutons? This needs to addressed in detail and preferably illustrated.

We added a more detailed description of how we identified the mossy fiber boutons within the stratum lucidum to the method section. It reads like this: "A magnification of 7K x was used to determine the localization of mossy fiber boutons. Subsequently, synapses were imaged at 20 kx using a EM 900 Transmission Electron Microscope (Carl Zeiss, RRID:SCR_021364) operated at 80 keV and equipped with a Proscan 2K Slow-Scan CCD-Camera (Carl Zeiss). The stratum lucidum of the hippocampal region CA3 was easily distinguishable for the presence of big mossy fiber boutons and for its localization just above the pyramidal cell layer. Mossy fiber boutons were recognized by their large size, their position close to CA3 pyramidal cell bodies and a large number of presynaptic vesicles (Rollenhagen et al., 2007), including large clear and large dense-core vesicles." 5. The authors make parallels between DKO and TKO phenotypes in PTP and LTP and conclude that Syn III makes a major contribution here. What is known about the phenotype of single SynIII KO in PTP and LTP and how doe this agree with the conclusion of this manuscript? Cultured autaptic hippocampal neurons from SynIII KO animals have been investigated with electrophysiology (Feng et al. 2002). It was shown that while the transmission was normal in excitatory cells, inhibitory autapses from SynIII KO had significantly smaller currents. Paired-pulse facilitation was unchanged in excitatory neurons, but synaptic depression in response to 20 Hz trains was substantially lower in SynIII KO neurons.

In another study, cultured neurons from SynTKO showed reduced augmentation and potentiation in miniature excitatory postsynaptic currents when compared to WT. This phenotype could only be rescued via the SynIIIa isoform and was mediated via the cAMP-PKA pathway (Cheng et al. 2018). However, to our knowledge there is no study investigating the SynIII KO in regard to PTP or LTP.

PTP and LTP at the mossy fiber synapse have been investigated in SynDKO and SynI KO animals, but not in SynTKO. While PTP was suggested to be lower in mossy fibers of SynDKO animals, LTP was unchanged in these animals (Spillane et al. 1995). PTP and LTP were also described to be unchanged in SynI KO animals (Takei et al. 1995).

In light of our SynTKO characterization we conclude that it is likely that all synapsins participate in PTP, hence we see more severe impairment in this form of plasticity in SynTKO compared to findings from SynDKO. The fact that LTP is not altered in mossy fibers from SynDKO animals, instead, suggests that the phenotype that we describe might be due to the additional deletion of SynIII. This hypothesis is also supported by evidence that, in the absence of SynIII, axonal growth cones are structurally reorganized similarly to the mossy fibers. Also the proposed role for SynIII in the negative regulation of the recycling pool of vesicles supports our finding that the RRP seems reduced (Fig 4-1) (Piccini et al., 2015; Feng et al. 2002).

6. The datasets in figs 1-2 and 2 are very uneven: there are considerably fewer data points for WT We now performed additional recordings from 2 male WT mice, leading to more balanced data sets between SynTKO and WT in figure 1-2 and figure 2, with 11 recordings from 5 WT and 12 recordings from 6 SynTKO animals. To improve transparency, we additionally added a sentence to the study design section in the methods, which reads: "Due to the pooled data from the two presymptomatic SynTKO cohorts, there is an imbalance between numbers from WT and numbers from SynTKO animals in high-frequency recordings".

Minor:

Fig 1-2 Panels (b) and (c) have the same Y scale, and the scale in the panel (a) is different. Why? We adjusted the y-axis in Figure 1-2 panel a) to make the comparison between panels easier.

Fig. 2 It would be worthwhile to add WT versus KTO comparison at intermediate points, for example around 10th and 20th stimulus.

We added a comparison at 25th stimulus for the first and fourth stimulation train in figure 2 c) and d). We also added the description of those values and the statistics in table 7.

Reviewer 2:

The study by Bruentgens et al. addresses the physiological function of synapsin 3 at the mossy fiber/CA3 synapse of the mouse hippocampus. This is an important question because this synapse is one of the few places where synapsin 3 is expressed in the adult brain. The authors used electrophysiological recordings in hippocampal slices to compare synaptic transmission at the mossy fiber (MF) synapse in wild-type (WT) and synapsin triple knock out (TKO) mice. These measurements were complemented by EM analysis of the morphology of MF presynaptic terminals. Because synapsin TKO mice have seizures, the authors were commendably careful to compare WT and TKO mice at 2 different ages, before and after the onset of seizure activity.

The results described in the paper show that deletion of all synapsins increases fEPSP amplitude, impairs synaptic facilitation and early post tetanic potentiation (PTP). In addition, there is an increase in late PTP and LTP. Structurally, at MF synapses there is a large loss of synaptic vesicles - as previously reported for many other synapsin-deficient synapses - as well as an increased density of presynaptic active zones.

While I acknowledge the efforts of the authors, at the end of the day we still do not really know what synapsin 3 is doing at the mossy fiber synapse because of two fundamental issues with the paper, one technical and one more conceptual:

1. The technical issue is evident in the enhancement of fEPSP amplitude shown in Fig. 1b (and 1.1b). This would indicate a huge increase in glutamate release from the MFs in TKO mice, a result that is at odds with decades of research on the effects of synapsins on glutamate release (as well as the absence of effects on synaptic facilitation, as shown in Fig. 1.2 of the paper). In lines 574-584 of the text, the authors themselves point out that this paradoxical result is likely due to the (unmeasured) contributions of inhibitory synapses, which in fact are preferentially affected by loss of synapsins. Given this confound, it becomes almost impossible to interpret the results shown in Figures 1-3 because the authors are measuring some ill-defined mixture of effects on excitatory and inhibitory synaptic transmission. The fact that DCG-IV blocks at least 75% of the synaptic response (e.g. Figure 3a,c) is not able to solve this problem, because of the possible feedback inhibition between mossy fibers/CA3 neurons and interneurons. I really do not understand why the authors did not sidestep this serious problem by performing their measurements in the presence of GABA receptor blockers.

We thank reviewer 2 for pointing out this technical issue.

For this study we did not perform experiments with GABA receptors blockers. The reasoning behind this decision was that GABA receptor blockers in the excitable SynTKO slices would, in our experience, induce epileptic activity in the slice. A possible solution to this problem could have been the addition of both GABA receptor blockers and AMPA receptor blockers to prevent epileptic activity and to subsequently measure NMDAR currents. We did not perform these experiments as they were beyond the scope of our current investigation.

We would like to clarify that we did not claim that the input/output curve is due to a massive increase in glutamate release. We see similar increases in excitability as Farisello and coworkers did see in area CA1 and we discussed the contribution of inhibition as a major mechanism for excitability and epileptogenesis (see also Farisello et al., 2013). Furthermore, an increase in glutamate release would not be the only option leading to the effects we see. A change in pre- and/or postsynaptic receptor composition could for example be involved and it has indeed been described that glutamate receptor levels increased in area CA3 of SynDKO mice (Owe et al. 2005). In fact, with our new finding of reduced RRP size in SynTKO animals we rather see a structural correlate for a decrease in release probability. This finding also fits to our observed trend in the increase of PPR (Figure 1-2), because a smaller RRP is associated with a lower release probability, as is an increased PPR. We tried to be careful to not over interpret those tendencies, because with field recordings this measure can be confounded.

2. Beyond this important technical shortcoming, the conceptual issue is that the results in the paper do not allow us to make any real conclusions about the function of synapsin 3 because the mossy fibers of TKO mice have lost all synapsins (1,2 and 3 genes), rather than synapsin 3 alone. The authors attempt to work around this problem by comparing their results to those published previously on synapsin double knockout (or single knockout) mice (e.g. Spillane et al., 1995; Owe et al., 2009). While such comparisons seem reasonable at first, in fact the experimental conditions differ substantially between the results reported in each paper. To give just one example, the protocol used by the authors to evoke activity-dependent facilitation of synaptic transmission (0.05 Hz stimulation followed by 1 Hz stimulation for 20 min) is rather different from the protocol used by Owe et al. (2009): 0.1 Hz stimulation followed by 2 Hz stimulation for 15 min. Different patterns of activity surely will elicit differing amounts of plasticity and may even engage different plasticity mechanisms, making a direct comparison difficult. Similar arguments apply to the striking change in LTP observed in TKO slices (Figure 3c). To yield definitive conclusions, the authors should perform side-by-side comparisons between different genotypes in identical experimental conditions.

We thank the reviewer 2 for his or her suggestions. We agree that the interpretation of the SynTKO results is confounded by the deletion of SynI and SynII. As we mention in the discussion, in this exploratory study we wanted to exclude possible compensatory effects due to the presence of SynI and II in SynIII single KO animals. With this work we gained knowledge regarding frequency facilitation, PTP and LTP in the absence of all synapsins. We agree that future work aimed at characterising SynIII KO mossy fiber structure and function will be important to specifically investigate SynIII role in adult mossy fiber synapses. We now removed all conclusions specifically regarding SynIII and, in general, toned down our claims.

Reviewer 2 points also out, that we use different protocols compared to Owe et al. (2009) as well as Spillane et al. (1995). We want to elaborate on our choice of protocols and thereby underline the robustness of our approach.

Owe and co-workers used a 2 Hz stimulation paradigm while we made use of a 1 Hz stimulation paradigm. We believe that both are suited to evoke mossy fiber facilitation due to the unique physiology of this synapse. These synapses are able to facilitate in a wide range of frequencies, described by Salin and others (Salin et al., 1996; Nicoll and Schmitz, 2005). At frequencies as low as 0.33 Hz, they can facilitate 5- fold (Salin et al., 1996, Figure 4), comparable to the levels we observe in WT animals at 1 Hz (6.5 fold) and Owe et al. at 2 Hz (4 fold). Owe et al. used a baseline frequency of 0.1 Hz and increased it to 2 Hz as the facilitating stimuli (Owe et al., 2009). The ratio between our baseline and stimulation frequency was the same, with a baseline frequency of 0.05 Hz and a stimulation frequency of 1 Hz. Although Owe et al. stimulated substantially longer at 2 Hz than we did at 1 Hz, they show a zoom-in to the first three minutes of their recording, making it feasible to compare their initial facilitation levels within a similar time frame to ours.

Spillane and coworkers used an induction protocol with 100 Hz trains to induce PTP and LTP (Spillane et al., 1995), while we decided to use a 25 Hz protocol in our study. This latter protocol is used for decades as an alternative to induce mossy fiber LTP and is comparable to a 100 Hz induction protocol (Mellor and Nicoll, 2001).

In addition to these fundamental problems, here are a few minor issues for the authors to consider:

1. In the long-stimulation experiments (Figures 2a and b), the amplitude of the first fEPSP in synTKO is smaller than that of WT, which is inconsistent with the effect shown in Figure 1. The authors need to explain this discrepancy.

We agree with reviewer 2 that it seems contradictory that the example traces in Figure 2a seem to have smaller fEPSP amplitudes in the SynTKO, although the example traces in Figure 1b show the opposite. However, this is the result of a necessary recording paradigm: the choice of stimulation strength. In our input-output recordings we varied the stimulation strength and saw a steeper increase in the relation between PFV and fEPSP for SynTKO than for WT animals when increasing the stimulation strength (Figure 1b). However, most data points from SynTKO are situated in the lower left corner, indicating a small fEPSP amplitude. To avoid epileptiform activity during our slice recordings and to allow for activity-dependent increase in the responses, we chose a non-saturating stimulation strength for the recordings after the input/output curve. In figure 1b our main goal was to show that with a similar number of stimulated fibers (measured via the PFV), the fEPSP amplitude is larger for SynTKO animals. To make it clearer, we now marked the data points referring to the example traces in Figure 1b with a black arrow.

To further explain the apparent conundrum, we compared averaged absolute amplitudes from the baseline before the LTP recording, resulting from the moderate, non-saturating stimulation strength. We found that the average baseline fEPSP amplitude was more variable in SynTKO animals than in WT, with a coefficient of variation of 61.3% for SynTKO and 30.84% for WT. We also found a tendency for the median baseline amplitude to be smaller in SynTKO than in WT animals with a median amplitude of 0.202 [0.136; 0.241] mV in WT and of 0.16 [0.09; 0.21] mV for SynTKO (p = 0.065, Mann Whitney U test). However, when actually measuring the first fEPSP amplitudes from the given two example traces in Figure 2a, we find comparable amplitudes with 0.121 mV for the WT trace and 0.118 mV for the SynTKO trace. We realized that our example traces in Figure 2 might have been a bit too small to detect such details, so we increased their size. Regarding Figure 2b we want to mention that due to the normalization of the fEPSPs the first amplitudes of both genotypes almost co-localize, making it hard to spot them.

2. The authors showed an interesting increase in active zone density in TKO mice compared to WT. However, because the total area of the AZ is also a factor that affects glutamate release, the authors should measure AZ area.

We thank the reviewer for pointing out this aspect. We measured the AZ area in both conditions and found it to be unchanged, in line with results from previous work in control versus potentiated WT boutons (Orlando et al., 2021). We now added the results of this analysis to figure 5 panel d) as well as to the text: "the forskolin-induced increase in docked and tethered vesicles was not accompanied by a change in the active zone area of mossy fiber boutons (Orlando et al., 2021). To test if this holds true in SynTKO animals, we analysed the area of individual active zones from partial 3D reconstructions of treated and untreated WT and SynTKO boutons. Mean active zone areas per bouton were comparable for both genotypes and treatments (Figure 5d). We fitted a generalized linear mixed model (Gamma family with log link) to estimate active zone area given the genotype and treatment, including the individual boutons as random effects. When comparing the model to nested null models, we found no evidence for an effect of treatment (p = 0.5846), genotype (p = 0.593) or the interaction of both (p = 0.8113) on the active zone areas. Thus, the active zone density increases while the active zone area is unchanged." 3. One (very minor) annoyance is that the authors italicize boutons throughout the text. This definitely is not the usual practice, because boutons is now considered to be a standard English word, despite its French origins! We now changed all "boutons" to regular font.