Hippocampal overexpression of Nos1ap increases the nNOS/PSD-95 interaction
For the experiments described in this study, we stereotaxically delivered rAAVs encoding murine Nos1ap tagged with 3xFLAG and mCherry [14] or the Nos1ap carboxyterminus (Nos1ap396 − 503), required for nNOS interaction [21] to the dorsal hippocampus of wild-type C57BL/6JRj mice. An rAAV expressing mCherry and 3xFLAG was used as a control (Fig. 1A,B).
Analysis of mCherry fluorescence (Fig. 1C) confirmed high expression levels in all subregions of the dorsal hippocampus (Figure S1). Congruent with previous findings [14] the Nos1ap rAAV had the lowest expression of the viruses. Ectopic expression was limited to the deep cortical layers above the injection site, comparable to previous observations [49]. Immunoblot analysis indicated that expression of the virally-encoded proteins was substantially higher than endogenous Nos1ap levels (Fig. 1D), though these blots were not readily quantifiable due to signal saturation. Expression analysis using qPCR (Fig. 1E, Table 1) showed ~ 294-fold increase of Nos1ap in Nos1ap rAAV injected mice (P < 0.001). As the primers targeted the aminoterminal region of Nos1ap, no expression changes were detected in Nos1ap396 − 503 mice (P = 0.593).
Table 1
Gene expression levels of Nos1ap and associated genes (see Figure S3 for graphical representation)
Gene
|
mCherry
|
Nos1ap
|
Nos1ap396 − 503
|
Statistics1
|
Cpe
|
1.0e -10 ± 0.545
|
0.021 ± 0.33
|
0.772 ± 0.235
|
F2,25=1.045, P = 0.366
|
Dlg1
|
-5.6e -18 ± 0.183
|
-0.412 ± 0.196
|
0.218 ± 0.154
|
F2,25=3.009, P = 0.067
|
Dlg3
|
-3.1e -18 ± 0.187
|
0.051 ± 0.187
|
0.048 ± 0.311
|
F2,25=0.017, P = 0.983
|
Dlg4
|
-1.9e -17 ± 0.214
|
0.23 ± 0.25
|
0.445 ± 0.264
|
F2,25=0.811, P = 0.456
|
Gria1
|
1.0e -10 ± 0.261
|
0.051 ± 0.247
|
0.695 ± 0.289
|
F2,25=1.978, P = 0.159
|
Gria2
|
4.163e -18 ± 0.251
|
-0.01 ± 0.157
|
0.506 ± 0.228
|
F2,25=1.725, P = 0.199
|
Grin2a
|
-5.573e -18 ± 0.259
|
-0.183 ± 0.233
|
0.474 ± 0.196
|
F2,25=1.903, P = 0.17
|
Grin2b
|
0.0 ± 0.491
|
0.739 ± 0.177
|
1.087 ± 0.211
|
F2,25=2.628, P = 0.092
|
Gucy1a1
|
1.0e -10 ± 0.393
|
0.317 ± 0.144
|
0.63 ± 0.205
|
F2,25=1.219, P = 0.312
|
Gucy1a2
|
-2.0e -10 ± 0.17
|
-0.463 ± 0.341
|
-0.204 ± 0.146
|
F2,25=0.539, P = 0.402
|
Gucy1b1
|
-1.0e -10 ± 0.469
|
0.425 ± 0.179
|
0.947 ± 0.128
|
F2,25=2.115, P = 0.142
|
Map2k3
|
1.0e -10 ± 0.39
|
0.56 ± 0.444
|
1.54 ± 0.241
|
F2,25=3.835, P = 0.035
|
Mapk14
|
-2.2e -17 ± 0.501
|
0.311 ± 0.15
|
0.886 ± 0.182
|
F2,25=1.668, P = 0.209
|
Nos1
|
-3.0e -10 ± 0.407
|
-0.368 ± 0.563
|
0.477 ± 0.358
|
F2,25=0.785, P = 0.467
|
Nos1ap
|
-1.0e -10 ± 0.191
|
8.199 ± 0.273
|
-0.168 ± 0.117
|
F2,25=512.807, P < 0.001
|
Rasd1
|
2.0e -10 ± 0.305
|
-0.253 ± 0.381
|
-0.645 ± 0.225
|
F2,25=0.956, P = 0.398
|
Scrib
|
1.0e -10 ± 0.627
|
0.884 ± 0.213
|
1.574 ± 0.189
|
F2,25=3.384, P = 0.05
|
Syn1
|
1.0e -10 ± 0.441
|
1.02 ± 0.295
|
1.044 ± 0.532
|
F2,25=2.076, P = 0.147
|
1 Nominal (i.e. uncorrected) P-values are shown. Bonferroni-corrected significance threshold: PBonf=0.00278
|
We previously showed that virally-overexpressed Nos1ap and a short isoform of Nos1ap, similar to Nos1ap396 − 503, interact with endogenous nNOS [14]. Given their overabundance, endogenous nNOS appeared to be predominantly bound by these virally-expressed proteins, strongly reducing the interaction with endogenous Nos1ap, as shown by co-immunoprecipitation (Fig. 1F,G; P < 0.001 for Nos1ap and Nos1ap396 − 503). Importantly, levels of endogenous Nos1ap and nNOS were not significantly affected by viral overexpression, as suggested by semi-quantitative immunoblot analysis (Figure S2A,B).
In keeping with recent findings suggesting that NOS1AP at least transiently mediates the function of the nNOS/PSD-95 NMDA receptor complex (discussed in [34]), using co-immunoprecipitation, we found that viral overexpression of Nos1ap and Nos1ap396 − 503 strongly increased the nNOS/PSD-95 interaction (Fig. 1F,G; P = 0.002 for both) while overall PSD-95 levels were not significantly affected (Figure S2C).
In addition, we wanted to investigate if Nos1ap or Nos1ap396 − 503 overexpression affects the expression of Nos1ap-associated genes. We found that the expression of Map2k3 and Scrib was nominally affected (P = 0.035 and P = 0.05 respectively), but not after Bonferroni correction (PBonf=0.00278). The expression of all other tested genes was not significantly affected (P > 0.05; Table 1 and Figure S3), suggesting that the treatment only had a limited effect on the regulation of these genes.
Hippocampal overexpression of Nos1ap does not affect synaptic electrophysiology but changes dendritic spine morphology
An increased nNOS/PSD-95 interaction may influence synaptic signaling dynamics. To understand the influence of the nNOS/Nos1ap interaction on synaptic properties, we injected mice with the mCherry, Nos1ap, or Nos1ap396 − 503 expressing rAAVs and, following 4–5 week of expression, performed patch clamp recordings on CA1 pyramidal neurons.
Paired-pulse ratios in CA1 Schaffer collateral synapses at inter-stimulus intervals of 30 or 100 ms were not altered (P = 0.13 and P = 0.719 respectively), indicating that overexpression of Nos1ap or Nos1ap396503 does not influence their vesicle release probability (Fig. 2A). Equally, mEPSC frequency (P = 0.713), amplitude (P = 0.793), 20–80% rise time (P = 0.158), and τdecay (P = 0.128) were not affected (Fig. 2B-E), suggesting that the number of functional synapses and the number of AMPA receptors per synapse were not influenced. AMPA and NMDA receptor-mediated currents in CA1 neurons, evoked by Schaffer collateral/commissural fiber stimulation showed that AMPA/NMDA ratios were comparable in all groups (P = 0.082), suggesting marginal effects on the relative number of AMPA and NMDA receptors (Fig. 2F).
We previously showed that overexpression of murine Nos1ap in cultured primary neurons resulted in an increase in filopodia and a reduction in dendritic spines [14]. Thus, we quantified dendritic spines in Golgi impregnated dorsal hippocampal CA1 neurons of mice with dorsal hippocampus injections (Fig. 2G-J). The total number of mature spines (i.e. excluding filopodia) was significantly reduced in brains overexpressing Nos1ap (P = 0.004), but not in those expressing Nos1ap396 − 503 (P = 0.196) when compared to control brains (Fig. 2I).
Functional properties of dendritic spines correlate strongly with their morphology [19]. Thus, we further analyzed different spine classes (Fig. 2J). We found a significant reduction in thin spines in brains overexpressing Nos1ap (P = 0.003) or Nos1ap396 − 503 (P = 0.029), but no changes in the number of stubby (P = 0.852) or mushroom (P = 0.107) spines, or filopodia (P = 0.615).
Overexpression of Nos1ap in dorsal hippocampus disrupts selective behaviors related to mental disorders
Given the substantial effects of viral Nos1ap and Nos1ap396 − 503 expression on nNOS/PSD-95 interaction and dendritic spines, we further investigated their effect on behavior.
Overexpression of Nos1ap or Nos1ap396 − 503 did not affect basic behavioral phenotypes (Fig. 3). Specifically, horizontal (distance traveled: Fig. 3A, P = 0.191) and vertical (number of rearings: Fig. 3B, P = 0.202) activity in the open field were comparable across all groups. Previous studies in mice of the ICR strain have suggested that Nos1ap overexpression in the dentate gyrus, targeting a region more posterior to the one used here, had anxiogenic effects [13]. However, in our study, anxiety-related behaviors in the open field (time in the center: Fig. 3C, P = 0.227), the light dark-box (light time: Fig. 3D, P = 0.164; number of transitions: Fig. 3E, P = 0.459), and the elevated zero maze (open arm time: Fig. 3F, P = 0.858) were not affected by Nos1ap/Nos1ap396 − 503 overexpression.
Sensorimotor gating, operationalized by PPI of the ASR is impaired in schizophrenia and other mental disorders (reviewed in [52]) and NOS1AP variants affecting PPI and startle have been identified [8]. Startle reactivity (Fig. 4A) was comparable in all groups (P = 0.611) and startle habituation (Fig. 4B) was intact in all groups (P < 0.001), with no effect of treatment (P = 0.683) or the interaction (P = 0.787). PPI significantly increased with prepulse intensity (Fig. 4C, P < 0.001), but was unaffected by the treatment (P = 0.599) or the interaction (P = 0.288) indicating that hippocampal Nos1ap/Nos1ap396 − 503 overexpression does not influence sensorimotor gating.
NOS1AP was previously linked to depression and depression phenotypes, but not specifically within the hippocampus [6, 10]. Therefore, we tested our mice for anhedonia, a common negative symptom caused by impairments in reward-related pathways including the hippocampus [53]. However, no differences in anhedonia, measured by sucrose preference, were observed between groups (P = 0.508; Fig. 4D).
Deficits in sociability and social cognition are commonly found in autism spectrum disorders [54], schizophrenia [55] and other mental disorders including depression [56]. When exposed to a juvenile conspecific, Nos1ap overexpressing mice showed a trend for reduced social interaction (Fig. 4E; P = 0.109) and social interaction in Nos1ap396 − 503 expressing mice was significantly reduced (P = 0.043). No effect on the number of social contacts was observed (Fig. 4F). Thirty min later mice were tested for social memory by exposure to the same (‘familiar’) and an unfamiliar (‘novel’) juvenile conspecific (Fig. 4G). We observed significantly longer interaction with the novel mouse in controls (P = 0.004), while Nos1ap and Nos1ap396 − 503 overexpressing mice showed no preference for the novel mouse (P = 0.32 and P = 0.144). These findings show that viral overexpression of Nos1ap and Nos1ap396 − 503 in hippocampus leads to deficits in sociability and social memory.
Deficits in working memory have been proposed as a transdiagnostic endophenotype for mental disorders [57]. Performance in the novel arm paradigm, a test requiring limited SWM capacity [49] was not affected (Fig. 4H, P = 0.6), suggesting that Nos1ap/Nos1ap396 − 503 overexpression does not influence basic SWM. Confirming this observation, mice from all groups performed well above chance level in the rewarded alternation paradigm on the T-maze (Fig. 4I). Control mice improved SWM performance (P = 0.006) across training. In contrast, neither Nos1ap (P = 0.833) nor Nos1ap396 − 503 (P = 0.363) expressing mice showed a significant increase in SWM performance indicating limited SWM capacity. This deficit cannot be attributed to an overall spatial memory deficit, as spatial reference memory was intact (Fig. 4J, P < 0.001) with no effect for the viral vector (P = 0.759) or the interaction (P = 0.552).