Synaptopodin regulates denervation-induced homeostatic synaptic plasticity

Denervation-Induced Homeostatic Synaptic Strengthening Is Not Observed in SP-Deficient Dentate Granule Cells.

Slice cultures prepared from SP-deficient mice were used to study the role of SP/SA in denervation-induced synaptic strengthening. First, the innervation of the dentate gyrus by EC fibers was evaluated using anterograde tracing of the entorhino-hippocampal projection (Fig. 1B). Furthermore, we verified that EC fibers form functional excitatory synapses, by stimulating the EC electrically while recording evoked excitatory postsynaptic currents (EPSCs) from dentate granule cells (Fig. 1C). On the basis of these observations, we concluded that a functional entorhino-hippocampal projection is present in SP-deficient cultures.

Subsequently, entorhinal denervation was performed and changes in excitatory synaptic strength were assessed at 3–4 d postlesion (dpl) by recording miniature EPSCs (mEPSCs) from dentate granule cells of wild-type and SP-deficient cultures (Fig. 1D). At this time point the denervation-induced increase in synaptic strength has reached a plateau (31). Although no significant differences in mean mEPSC frequencies were observed between the groups (wild-type, control 2.3 ± 0.2 Hz and 3–4 dpl 2.9 ± 0.4 Hz, P = 0.21; SP-deficient, control 2.3 ± 0.2 Hz and 3–4 dpl 3.0 ± 0.3 Hz, P = 0.11; no difference between wild-type and SP-deficient cultures), a significant increase in mEPSC amplitudes was only observed in denervated wild-type cultures and not in SP-deficient granule cells (Fig. 1E). This finding suggested a role for SP/SA in denervation-induced homeostatic synaptic plasticity.

Tetrodotoxin-Induced Homeostatic Synaptic Strengthening Is Impaired in SP-Deficient Dentate Granule Cells.

To confirm these results and to exclude that the inability of SP-deficient granule cells to express synaptic scaling following entorhinal denervation is the result of subtle alterations in innervation, which may have escaped our detection, a classical pharmacological approach, tetrodotoxin (TTX) treatment (13, 14, 35–37), was used to induce synaptic scaling. In contrast to TTX-treated (2 µM; 2–3 d) wild-type dentate granule cells, prolonged TTX treatment of SP-deficient granule cells failed to significantly increase mEPSC amplitudes (Fig. S1 A and B). Together, these results demonstrated an important role for SP/SA in homeostatic synaptic plasticity of dentate granule cells in vitro, both under conditions of partial denervation and pharmacological blockade of neuronal activity.

Expression of GFP/SP Rescues the Ability of SP-Deficient Dentate Granule Cells to Form SAs and to Homeostatically Increase Excitatory Synaptic Strength.

To corroborate the above findings and to follow the dynamics of SP in individual living neurons, a transgenic mouse strain, which expresses GFP-tagged SP under the control of the Thy1.2 promoter, was generated (Fig. S2). By crossing this strain with SP-deficient animals we obtained mice that lack endogenous SP and only express GFP/SP (Thy1-GFP/SP × SP-KO mice; details are given in SI Materials and Methods). In cultures prepared from these mice GFP/SP expression was regularly seen in the dentate gyrus (Fig. 1F). Of note, the ability of SP-deficient granule cells to form SAs was rescued in these cultures by the expression of GFP/SP (Fig. 1G).

After confirming the presence of a functional entorhino-hippocampal projection, we applied the two protocols (denervation, Fig. 1H; TTX, Fig. S1C) to test for homeostatic synaptic strengthening in slice cultures of Thy1-GFP/SP × SP-KO mice. Indeed, expression of GFP/SP was sufficient to rescue the ability of SP-deficient granule cells to increase their excitatory strength under both conditions.

Early Phase of Denervation-Induced Synaptic Scaling Requires the Presence of SP/SA.

Because it has been indicated that synaptic scaling may be divided into molecularly distinct phases (38), the role of SP was also assessed during the earlier phase of denervation-induced synaptic scaling. At 1 dpl no significant increase in mEPSC amplitudes was detected in SP-deficient granule cells, whereas transgenic expression of GFP/SP rescued the scaling response of denervated dentate granule cells (Fig. S3), similar to what was observed at 3–4 dpl. Taken together, we concluded that the ability of cultured dentate granule cells to express homeostatic synaptic plasticity depends on the presence of SP and the SA organelle.

Denervation-Induced Homeostatic Synaptic Plasticity Is Accompanied by an Enlargement of SP Clusters.

To provide further experimental evidence that SP is linked to denervation-induced homeostatic plasticity, we next addressed the question of whether homeostatic synaptic strengthening is accompanied by compensatory structural changes of SP/SA. In this set of experiments we exploited the fact that entorhinal denervation leads to a layer-specific increase in synaptic strength at 3–4 dpl, that is, to homeostatic strengthening of excitatory synapses predominantly in the OML (31). Denervated (3–4 dpl) and age-matched nondenervated wild-type cultures were fixed and stained for endogenous SP (Fig. 2A; specificity of antibody tested in tissue from SP-deficient mice, Fig. S2F) and the sizes of immunofluorescent endogenous SP clusters were assessed in the denervated OML and the nondenervated inner molecular layer (IML) (39–41). The results of these experiments disclosed a significant increase in SP cluster size of ∼35% in the denervated OML in comparison with nondenervated control cultures, whereas in the IML only a small (∼5%) increase in SP cluster sizes was observed (Fig. 2B). We concluded that denervation-induced homeostatic synaptic plasticity is accompanied by an increase in SP cluster sizes.

Fig. 2.

Denervation-induced homeostatic synaptic scaling is accompanied by structural changes of SP clusters and SA organelles. (A) Control and denervated wild-type cultures were stained for endogenous SP and mean cluster sizes were determined in the OML and IML. (Scale bar: 2 µm.) (B) A considerable increase in mean SP cluster sizes was detected in the OML following denervation. In the IML the effect was less pronounced (n = 6–10 cultures per group, three visual fields per culture). (C) Quantitative evaluation of GFP/Sp cluster fluorescent recovery after photobleaching (FRAP). Statistical comparisons performed with averaged values between 65–80 min per cluster (details given in SI Materials and Methods; n = 86 clusters from 12 control cultures; n = 70 clusters from 9 denervated cultures; six to nine clusters bleached per culture). (D) No significant difference in FRAP of GFP/SP clusters was observed in the IML of control and denervated cultures. (n = 54 clusters from eight control cultures; n = 52 clusters from seven denervated cultures; six to nine clusters bleached per culture). (E and F) The mean stack number of SAs was increased in the OML after denervation (n = 120–137 SAs in six cultures each; 20–25 SAs per culture). **P < 0.01; ***P < 0.001; NS, not significant.

Denervation-Induced Homeostatic Synaptic Plasticity Is Accompanied by an Increase in SP Cluster Stability.

In an attempt to evaluate whether the dynamics of SP clusters were affected by deafferentation, fluorescence recovery after photobleaching (FRAP) of GFP/SP clusters was determined in the OML and IML of control and denervated cultures (3–4 dpl, Fig. 2 C and D). Following a 15-min baseline recording (∆T = 5 min), individual GFP/SP clusters were bleached (<5% of initial fluorescence). The reappearance of GFP fluorescence was then followed for up to 80 min at 5-min intervals. The results revealed a highly significant difference in the FRAP of GFP/SP clusters in the OML. In control cultures, a comparatively fast recovery was observed such that 30% of the averaged prebleach GFP/SP fluorescence recovered within 20–30 min after photobleaching (Fig. 2C). In contrast, following the induction of homeostatic synaptic plasticity 30% recovery was reached ∼80 min after the bleach (biexponential fitting of FRAP curves is given in SI Materials and Methods). Notably, the difference in FRAP of GFP/SP clusters in control and denervated cultures was still seen when clusters of equal size were compared (Fig. S4A), suggesting that SP cluster size does not determine GFP/SP cluster FRAP. Taken together, these data showed that the turnover of GFP/SP clusters is slowed following entorhinal denervation. Because partial deafferentation did not change the FRAP of GFP/SP clusters in the nondenervated IML (Fig. 2D), we concluded that SP cluster stability is mainly affected in the layer of denervation (i.e., the layer in which synaptic strength is scaled up at 3–4 dpl) (13).

To further strengthen these findings, SP cluster properties were evaluated in the OML during the early phase after denervation (between 0 and 6 h, 6 and 12 h, and 12 and 24 h following deafferentation) (Fig. S4 B–D). Indeed, a close interrelation between SP cluster remodeling, that is, an increase in SP cluster size/stability, and the induction of synaptic scaling was evident in these experiments. Of note, SP-mRNA levels were not significantly increased at 1, 2, and 4 dpl in the granule cell layer as determined by laser capture microdissection (LMD)-quantitative PCR analysis (Fig. S5), indicating that the observed changes are not regulated via changes of mRNA levels as seen following the induction of long-term potentiation (LTP) (42, 43). In summary, the results of these experiments disclosed that denervation-induced homeostatic synaptic strengthening is accompanied by an increase in SP cluster size and stability.

Denervation-Induced Ultrastructural Changes of SA Organelles.

The SP cluster analysis led us to the hypothesis that ultrastructural hallmarks of the SA could change after denervation. Because SP seems to be enriched in the so-called dense plates (22), which are localized between the stacked cisternae of smooth endoplasmic reticulum (22–24), we reasoned that changes in SP immunohistochemistry and FRAP, that is, reduced turnover and increased size of SP clusters after denervation, might reflect changes in the number of SA stacks, (i.e., changes in mean dense plate number). Hence, we subjected control and denervated cultures to EM analysis and cross-sections of SAs were assessed in the OML and IML. Whereas in control cultures 46% of identified SA cross-sections revealed single dense plates (i.e., two endoplasmic cisternae) and only 5% contained four or more dense plates, in the OML of denervated cultures 32% of SA cross-sections disclosed single dense plates and now 16% demonstrated four or more dense plates (Fig. 2 E and F). In line with this shift toward higher stack numbers of the SA, our analysis also revealed a significant increase in the mean stack number of SAs in the OML of denervated cultures (Fig. 2F). In the IML a slight increase in mean dense plate number was also seen (control IML, 2.60 ± 0.11 vs. 3–4 dpl, 3.0 ± 0.14; P = 0.033). These results reflected, at least in part, our immunohistochemistry and FRAP data (Fig. 2 A–D) and showed that SAs can change their stack number in response to denervation. We concluded that a compensatory increase in SA stacking accompanies the induction of denervation-induced homeostatic synaptic plasticity.

SP/SA Remodeling Is Regulated Inversely by Neuronal Activity.

The above findings raised the possibility that SP/SA remodeling could be part of an activity-dependent negative-feedback mechanism that aims at adjusting synaptic strength in a homeostatic manner (1, 3). Based on this theory we hypothesized that a reduction in neuronal activity, reflected by a decrease in intracellular [Ca^{], could induce the compensatory increase in SP cluster sizes and stability, which could in turn promote the accumulation of AMPA-R at excitatory postsynapses (}28).

To learn more about the mechanisms that control SP cluster properties in the context of homeostatic synaptic scaling, pharmacological inhibition of network activity of nondenervated cultures with TTX (2 µM, 3 d) was used first. Indeed, the effects of entorhinal denervation on SP cluster sizes (Fig. 3A), GFP/SP cluster FRAP (Fig. 3B), and SA stacking (Fig. S6), as well as changes in mEPSC amplitudes (Fig. 3C), were mimicked by the TTX treatment. To test for the role of Ca ^{entry sites in this context, nondenervated cultures were treated for 3 d with the NMDA receptor (NMDA-R) inhibitor R-2-amino-5-phosphonopentanoate (APV, 50 µM) and Nifedipin (20 µM), which blocks L-type voltage-gated calcium channels (L-VGCC). Indeed, combined APV+Nifedipin treatment revealed results comparable to those seen following TTX treatment and denervation (}Fig. 3 A–C).

Fig. 3.

The role of neuronal activity and ryanodine receptors (RyRs) in SP cluster remodeling and denervation-induced homeostatic synaptic strengthening. (A and B) In nondenervated wild-type cultures treatment with TTX (2 µM), APV (50 µM) + Nifedipin (20 µM), and APV or Nifedipin alone for 3 d resulted an increase in (A) Sp cluster sizes (n = 5–13 cultures per group; three to six visual fields per cultures) and (B) GFP/Sp cluster FRAP (n = 24–86 clusters per group; three to four cultures per group; statistical comparison performed with mean FRAP values between 65–80 min after bleaching). (C) Whereas TTX and APV+Nifedipin treatment increased mEPSC amplitudes as seen following denervation, APV or Nifedipin alone had no significant effect on mean mEPSC amplitude (n = 11–21 neurons per group; from four to eight cultures, each). (D) Inhibition of RyRs with a high dose of ryanodine (80 µM, 3 d) impairs denervation- and APV+Nifedipin-induced homeostatic synaptic strengthening (n = 13–15 neurons from four to six cultures each). (E) In the presence of ryanodine (80 µM) a reduction in GFP/SP cluster FRAP was observed in both nondenervated and denervated cultures (n = 34–38 clusters from five cultures each; six to nine clusters bleached per culture). The effects of ryanodine treatment and denervation were not additive. (F) Model of SP-dependent homeostatic synaptic plasticity. A reduction in neuronal activity leads to a compensatory increase in SP cluster sizes and SA stacking. SP remodeling may be part of a calcium-dependent negative feedback mechanism that promotes the ability of synapses to efficiently accumulate AMPA-R at excitatory synapses via a not-yet-fully-understood SP/SA-dependent mechanism. Earlier work has linked SP to spines with a bigger spine head volume and revealed that SP clusters associate with an intracellular source of calcium (28, 29) which regulates the accumulation of AMPA-R at excitatory postsynapses in a RyR-dependent manner (28). *P < 0.05; **P < 0.01; ***P < 0.001; NS, not significant.

We then determined whether inhibition of NMDA-R or L-VGCC underlies the observed changes. Interestingly, APV or Nifedipin treatment increased SP cluster size and stability while leaving mEPSC amplitudes unchanged (Fig. 3 A–C). Taken together, these results suggested that a reduction in neuronal activity (i.e., intracellular calcium levels) could trigger a compensatory increase in SP cluster size and stability. However, these changes seem to be insufficient to induce homeostatic synaptic scaling. Only when both NMDA-R and L-VGCC are blocked for 3 d is an increase in excitatory synaptic strength induced in cultured dentate granule cells.

Denervation-Induced Homeostatic Synaptic Strengthening Depends on Ryanodine Receptor Activity.

Previous work has revealed that SP clusters associate with ryanodine receptors (RyRs), which in turn promote the accumulation of AMPA-R at excitatory postsynapses of SP-containing dendritic spines (28). We therefore tested whether RyRs could also be involved in denervation-induced homeostatic synaptic strengthening. Nondenervated control cultures and denervated cultures were treated for 3 d with a high dose of ryanodine (80 µM) to block RyRs (44, 45) and mEPSCs were subsequently recorded. In the case of denervation RyRs were blocked immediately after lesioning. Because a compensatory increase in excitatory synaptic strength was not observed under these conditions (Fig. 3D), our results suggested that RyR activity is required for the homeostatic synaptic strengthening response of cultured dentate granule cells at 3–4 dpl. Ryanodine treatment also inhibited the APV+Nifedipin-induced increase in mEPSC amplitudes of nondenervated dentate granule cells (Fig. 3D).

Finally, we tested for the effects of RyR blockade on SP cluster properties. Control and denervated cultures were treated with ryanodine and SP cluster sizes and GFP/SP cluster FRAP were assessed in the OML. In these experiments an increase in SP cluster size (untreated, 26.9 ± 1.5; ryanodine, 36.1 ± 1.4; P < 0.05) and stability (Fig. 3E) was observed in nondenervated cultures. Because denervation and RyR blockade were not additive in their effects on GFP/SP cluster FRAP (Fig. 3E), we concluded that similar mechanisms (i.e., a reduction in RyR activity) may account, at least in part, for the denervation-induced stabilization of SP clusters.

Animals.

The following mouse strains were used in the present study: wild-type C57BL/6 mice, SP-deficient mice [internal strain designation SP-KO; official strain designation C57BL/6-Synpo ⁽21)], and a newly generated transgenic mouse line expressing GFP-tagged SP under the control of the Thy1.2 promoter crossed to SP-deficient mice (Thy1-GFP/SP × SP-KO).

Generation of Thy1-GFP/SP Transgenic Mice.

Details are given in SI Materials and Methods.

Preparation of Slice Cultures.

Entorhino-hippocampal slice cultures were prepared as described previously (31, 33).

Whole-Cell Patch-Clamp Recordings.

Whole-cell voltage recordings were carried out as described in ref. 31.

Fluorescent Recovery After Photobleaching Analysis.

FRAP experiments were performed and analyzed as previously described (50). Information is also given in SI Materials and Methods.

Immunostaining.

Cultures were fixed and stained as previously described (34) with antibodies against synaptopodin (SE-19; Sigma; 1:1,000).

Electron Microscopy.

Details are provided in SI Materials and Methods.

Laser Capture Microdissection and Quantitative PCR.

The granule cell layer of denervated cultures and age-matched nondenervated control cultures was isolated from 10-µm-thick resliced cultures using LMD as previously described (51). The probes were subjected to quantitative PCR analysis. Further details are provided in SI Materials and Methods.

Quantification and Statistics.

Electrophysiological data were analyzed as described in ref. 31. The Image-J software package (http://rsb.info.nih.gov/ij) was used to analyze immunostainings and FRAP (50); details are provided in SI Materials and Methods. Immunostainings were analyzed as described before (e.g., ref. 39). For EM, identified cross-sections of SAs were assessed in the respective regions and the number of dense plates and endoplasmic cisternae was noted. SA stacks were defined as x dense plates and x + 1 cisternae. Data obtained by quantitative PCR were analyzed as described by Pfaffl (52). Statistical comparisons were made using the Wilcoxon–Mann–Whitney test or Kruskal–Wallis test followed by Dunn’s post hoc analysis to correct for multiple comparisons. P values of less than 0.05 were considered a significant difference. All values represent means ± SEM.

Supplementary Material