A sex difference in the response of the rodent postsynaptic density to synGAP haploinsufficiency

SynGAP is a postsynaptic density (PSD) protein that binds to PDZ domains of the scaffold protein PSD-95. We previously reported that heterozygous deletion of Syngap1 in mice is correlated with increased steady-state levels of other key PSD proteins that bind PSD-95, although the level of PSD-95 remains constant (Walkup et al., 2016). For example, the ratio to PSD-95 of Transmembrane AMPA-Receptor-associated Proteins (TARPs), which mediate binding of AMPA-type glutamate receptors to PSD-95, was increased in young Syngap1+/-mice. Here we show that only females and not males show a highly significant correlation between an increase in TARP and a decrease in synGAP in the PSDs of Syngap1+/-rodents. The data reveal a sex difference in the adaptation of the PSD scaffold to synGAP haploinsufficiency.

PSDs from the HET mice compared to WT. As we had predicted, the mean ratio of TARPs to PSD-95 48 showed a small (12%) but significant increase in PSDs from the HET animals compared to WT. We also 49 found a small but significant increase in the mean ratio of LRRTM2 (14%) and neuroligin-2 (9%) to 50 PSD-95. The mean ratio of neuroligin-1 to PSD-95 was unchanged.

51
Because the number of pooled brains in this previous study was small and WT and HET pools were not 52 perfectly balanced by developmental age or sex, we set out to expand these findings with a larger data set 53 gathered from PSDs isolated from individuals rather than from pooled animals. Data from individuals 54 allowed us to use a more rigorous statistical measure of correlation, the well-established Spearman's rank 55 correlation coefficient r. Comparison of mean levels of two proteins in pooled samples is not a perfect 56 measure of the correlation between the two levels in individuals. It is possible to have a correlation 57 between protein levels in individuals that is not reflected as a difference between mean levels. Spearman's 58 r tests whether a monotonic correlation exists between the rank order of magnitudes of two variables in a 59 data set. We used it to examine the correlation of levels of synGAP with levels of four other proteins in 60 individual PSD fractions. If the rank orders of two variables correlate perfectly, Spearman's r is 1; if 61 there is no correlation, it is zero; and if the ranks are perfectly anti-correlated, it is -1.

63
When the data were averaged for WT and HET animals in this large data set, we were surprised to find 64 that the mean TARP/PSD-95 ratio in PSDs was not different between WT and HET animals, in contrast 65 to our earlier finding. However, when we calculated Spearman's r for individual data sets, we made the 66 unexpected discovery that a strong inverse correlation between the levels of TARP and synGAP is present 67 only in females and not in males. The large and highly significant inverse correlation in HET females 68 drives a significant inverse correlation in data from all HET animals and from all female animals. The 69 inverse correlation is not found in any subset of animals that contains only males. We also established 70 that a similar sex difference is present in rats, as well as mice. We repeated the finding of the earlier 71 study that there is no difference in the level of neuroligin 1 between WT and HET rodents; but, there is a 72 small increase in the amount of neuroligin 2 in HETs. Finally, we made the additional discovery that the

97
The total sample included 81 females ( 39 WT, 42 HET), and 84 males (43 WT, 41 HET). In each 98 category, approximately half of the animals were rats and half were mice; approximately half were 7.5 99 weeks old and half were 12.5 weeks old. The ratio of synGAP/PSD-95 and TARP/PSD-95, averaged 100 over all of the rodents, are summarized in the two bars labeled "All" (Figure 2A

111
We examined the correlation between levels of TARP and synGAP among individual rodents in each 112 sample using the more sensitive measure, Spearman's r. We used Spearman's r rather than Pearson's r 113 for these measurements because many of the data sets showed a non-normal distribution. Pearson's r is 114 valid only for normally distributed data.

156
The average intensities of staining for proteins differed significantly between the cohorts, presumably 6 Figure 3 contains scatter plots of the TARP/PSD-95 ratio against the synGAP/PSD-95 ratio in 161 individual PSDs from the indicated data sets. Figure 3A contains the plot for all 165 animals. Rows two 162 (B., E., and H.) and three (C., F., and I.) show separate plots for WT and HET animals, respectively; 163 columns two (D., E., and F.) and three (G., H., and I.) show separate plots for females and males. Data 164 from HET animals are indicated in orange to highlight that they have a lower average synGAP/PSD-95 165 ratio than WT animals. In each plot, Spearman's r is given, along with the p-value indicating the 166 probability that Spearman's r differs from zero. P-values that indicate statistical significance are in red.

167
For reference, red lines through the points in each graph show the best linear fit determined by regression.

168
There is not a statistically significant inverse correlation (negative Spearman's r) between the two ratios 169 for the group of all animals ( Figure 3A).

184
The data from all females ( Figure 3D) show an inverse correlation between the two ratios that just 185 reaches statistical significance. In contrast, the data from all males ( Figure 3G

203
We also compared data sets from mice and rats at 7.5 weeks and 12.5 weeks (Figure 3.-figure 204 supplements 2 and 3). These data sets were small (9 or 10 animals). Nevertheless, they show a statistically 205 significant inverse correlation between TARP/PSD-95 and synGAP/PSD-95 in HET female mice at both 206 7.5 and 12.5 weeks. In data from HET rats at 7.5 weeks, the inverse correlation is very close to 207 significance; at 12.5 weeks, it is less significant, but still shows a trend. In the corresponding males, none 208 of the data sets shows a statistically significant inverse correlation. More data would be required to make 209 a definitive conclusion, but the results suggest that competition between synGAP and TARP for binding 210 to PSD-95 in females is more prominent at 7 weeks than at 12 weeks, and more prominent in mice than in 211 rats.

212
Effect of synGAP haploinsufficiency on the relative levels of other PSD proteins. In our previous paper, 213 we examined the levels of neuroligins 1 and 2 (NLG-1, -2), and of the surface protein LRRTM2. In this 214 study, we re-examined the effect of reduction of synGAP on the levels of NLG-1 and 2 in the PSD and 215 looked at the effect on levels of GluN2B, a subunit of the NMDA-type glutamate receptor that binds most 216 avidly to PDZ2 of PSD-95. We predicted that the level of GluN2B would be less affected than TARPs or 217 NLGs by reduction of synGAP because synGAP has lower affinity for PDZ2 than for PDZ1 and PDZ3 218 (Walkup et al., 2016).

219
The ratios of the three proteins to PSD-95 in HET and WT rodents, averaged over the same PSD 220 fractions shown in Figures 2A

377
There are many possible mechanistic explanations for the sex difference in sensitivity of TARPs to 378 the concentration of synGAP. One simple one is that additional protein(s) are present in females that 379 compete with synGAP for binding to PDZ domains of PSD-95. The resulting "crowding" could make 380 binding of TARP to PSD-95 more sensitive to reduction of synGAP in females.

381
Another is that an additional protein which can compete with TARPs more strongly than synGAP for 382 binding to PSD-95 is present in male PSDs, but not in female PSDs. In this case, reduction of synGAP in 383 PSDs of males would be expected to have little effect on the concentration of TARPs. Interestingly, 384 despite any differences between males and females in mechanism, the mean steady-state ratio of TARPs 385 to PSD-95 in the two sexes is not significantly different among WTs or HETs.

411
The simplest explanation is that NLG's have a higher affinity for PDZ3 than synGAP such that synGAP

532
One percent were larger than 2.75 and four percent were less than 0.5. After normalization within 533 cohorts, the ratios were used for the analyses in figures 2 and 4. This procedure allowed us to correct for 534 technical variation within cohorts while preserving true differences in ratios based on species, age, 535 genotype, or sex. Analyses for significance were carried out and graphs were created with Prism 8 536 (GraphPad Software, San Diego). The D'Agostino-Pearson omnibus test was used to determine the 537 normality of the data sets. The means of groups of data were compared for significant differences with 538 one-or two-tailed Mann-Whitney test (when non-normal), or one-or two-tailed paired T-Test, as 539 indicated. If the variances of the two groups were found to be different using an F test, a one-or two-540 tailed T-Test with a Welch's correction was applied.

541
To test rigorously among individual animals for correlations between the ratio of synGAP to PSD-95 542 and the ratio of the other target proteins to PSD-95 (Figures 3, 5, and 6), We corrected for differences in 16 intensity of signals between the four cohorts (i.e. 7.5 week old mice, 7.5 week old rats, 12.5 week old 544 mice, and 12.5 week old rats) which had been analyzed separately. (The previous normalization only 545 corrected for technical variation within each cohort.) Data for each cohort was divided into ratios from 546 WT males, WT females, HET males, and HET females. Averages of the ratios for each protein within 547 each of these groups, were calculated. The averaged ratios for each group were then further averaged 548 across the cohorts. A normalization factor was calculated for each protein in each cohort by dividing the 549 overall average for all the cohorts by the average for each cohort. Then the appropriate normalization 550 factor was applied to individual data points in each cohort. This sequence corrected for variation in the 551 average intensities of signals between cohorts and groups (for example, the overall lower expression of 552 TARPS in 7.5 week old HET females [ Figure 1B]) and allowed us to look for correlations between ratios distribution. Therefore, Pearson's r, the parametric equivalent of the non-parametric Spearman's 697 coefficient, was calculated for each set. A) A significant inverse correlation was found for 7 and 12 week 698 old female mice, and a strong trend was present for 7 and 12 week old female rats.