Molecular origin of somatostatin-positive neuron vulnerability

Abstract

Reduced somatostatin (SST) and dysfunction of SST-positive (SST⁺) neurons are hallmarks of neurological disorders and associated with mood disturbances, but the molecular origin of SST⁺ neuron vulnerability is unknown. Using chronic psychosocial stress as a paradigm to induce elevated behavioral emotionality in rodents, we report a selective vulnerability of SST⁺ neurons through exacerbated unfolded protein response (UPR) of the endoplasmic reticulum (ER), or ER stress, in the prefrontal cortex. We next show that genetically suppressing ER stress in SST⁺ neurons, but not in pyramidal neurons, normalized behavioral emotionality induced by psychosocial stress. In search for intrinsic factors mediating SST⁺ neuron vulnerability, we found that the forced expression of the SST precursor protein (preproSST) in SST⁺ neurons, mimicking psychosocial stress-induced early proteomic changes, induces ER stress, whereas mature SST or processing-incompetent preproSST does not. Biochemical analyses further show that psychosocial stress induces SST protein aggregation under elevated ER stress conditions. These results demonstrate that SST processing in the ER is a SST⁺ neuron-intrinsic vulnerability factor under conditions of sustained or over-activated UPR, hence negatively impacting SST⁺ neuron functions. Combined with observations in major medical illness, such as diabetes, where excess ER processing of preproinsulin similarly causes ER stress and β cell dysfunction, this suggests a universal mechanism for proteinopathy that is induced by excess processing of native endogenous proteins, playing critical pathophysiological roles that extend to neuropsychiatric disorders.

Figure 1.. Gene set enrichment analysis of ER stress/UPR-related pathways expressed in neurons of cortical circuitry in UCMS mice.

a. Gene set enrichment analysis (GSEA) summary of ER stress/UPR-related pathways significantly enriched in neurons of cortical circuitry in UCMS mice. PYR: pyramidal neurons; SST: SST⁺ neurons; NES: normalized enrichment score. b. Enrichment plot for each pathway shown in a. c. GSEA summary of PERK, IRE1α and ATF6 pathways expressed in cortical SST⁺ neurons of UCMS mice as compared with control mice. d. Enrichment plot for each pathway shown in c. e. Heatmap analysis of the effects of UCMS on selected representative genes in each pathway. Gene name and normalized enrichment score are shown in each box.

Figure 2.. ER stress in SST⁺ neurons induced by UCMS.

a. Phospho-eIF2α (Ser-51) levels (normalized by total eIF2α levels) in SST⁺ neurons in PFC of male control mice (Perk^+/+;Sst^IRES-Cre) as compared with mice with reduced Perk gene level (Perk^flox/+;Sst^IRES-Cre) before and after 5 weeks of UCMS. Arrow points to SST⁺ neurons labeled green (via Ai6lsl-ZsGreen/+;SstIRES-Cre/+). Scale bar: 20 μm. b. Time-course of p-eIF2α levels in SST⁺ neurons during UCMS (1, 3 and 5 weeks). SST⁺ neurons in PFC of mice with reduced Perk levels show lower ER stress levels than those in control mice. (N=4/genotype/treatment/sex, 3–4 months old; Kruskal-Wallis test with Dunn’s multiple comparisons, *P<0.05, **P<0.01)

Figure 3.. Genetic suppression of ER stress in SST⁺ neurons ameliorates UCMS-induced behavioral emotionality.

a. Behavioral assay scheme: Mice (6♂+6♀/genotype, 3–4 months old) were subjected to 5 weeks of UCMS or kept under no stress, and tested for behavioral emotionality via 8 independent assays (weekly coat state evaluation and PhenoTyper (PT) followed by OFT, EPM, NSF, NIH, sucrose consumption (SC) and FST at the end of 5 weeks of UCMS). Results of independent tests are depicted in Fig. S1. b. Z-scoring across all 8 assays shows consistent upregulated behavioral emotionality after UCMS in male and female mice. Blocking ER stress abolished UCMS-induced behaviors and lowered behavioral emotionality at baseline in female mice. Kruskal-Wallis test with Dunn’s multiple comparisons; *P<0.05, **P<0.01, ***P<0.001. c. Blocking ER stress in CaMKII⁺ neurons failed to rescue UCMS-induced behaviors. Z-scoring of emotionality assays in mice of the indicated genotypes shows elevated behavioral emotionality after UCMS in both male and females. Results of independent tests are depicted in Fig. S2. Kruskal-Wallis test with Dunn’s multiple comparisons; **P<0.01.

Figure 4.. SST⁺ neuron-specific ER stress induced by preproSST.

a. AAV-vector design: A human synapsin promoter (hSyn) drives the transgene expression when the host-derived Cre recombinase excises the stop cassette (lox_SV40-polyA_lox). Each transgene is fused in frame with eGFP via T2A self-cleavage signal. Transgene-expressing cells are labeled by eGFP, which is released from the expressed transgene product. Full-length preproSST and two truncation mutants (processing incompetent and mature SST) used are shown. b. Fractionation of cytosolic and membrane fractions for proteins expressed from AAV vectors in HEK293T heterologous expression system. Proteins were detected by Western blot or dot blot using the indicated antibodies. c. GFP-tagged full-length preproSST transgene or GFP alone were introduced via AAV into SST⁺ neurons in PFC of male mice (N=4/group, 3–4 months old). The p-eIF2α (Ser-51) levels (normalized by total eIF2α levels) in GFP⁺ cells (pointed by arrow) were evaluated in IF assays. Scale bar: 50 μm. d. AAV encoding each transgene was delivered into a specific cell type in PFC (“cell type” > “transgene”) and the p-eIF2α levels in each infected cell were plotted (mean±SEM in red). Kruskal-Wallis test with Dunn’s multiple comparisons; *P<0.05. e. Overexpressing preproSST in SST⁺ neurons is sufficient to induce elevated behavioral emotionality. Z-scoring of emotionality assays in mice (Sst^IRES-Cre/+) injected into PFC with the indicated AAV (N=7/group/sex). Results of independent tests are depicted in Fig. S4. Kruskal-Wallis test with Dunn’s multiple comparisons; *P<0.05, #P=0.086.

Figure 5.. ER stress-dependent increase in insoluble preproSST peptides in PFC during UCMS.

a. Proteins extracts (50, 100, 150 and 200 μg) prepared from PFC of mice (N=4/genotype/treatment, 3–4 months old, male) kept under no stress or 5 weeks of UCMS conditions were subjected to filter-trap assays to capture insoluble protein fractions and blotted with anti-SST antibody. The blot confirmed linearity of detection within the range of proteins used in this assay (50~200 μg). The graph to the right shows the densitometry of insoluble SST proteins included in 100 μg of protein extracts from each sample. Kruskal-Wallis test with Dunn’s multiple comparisons; *P<0.05 (as compared with no stress Perk^+/+;Sst^IRES-Cre control mice). b. Total SST peptides solubilized in RIPA buffer were detected by Western blot. Each lane contains 25 μg protein extracts. α-Tubulin was used as loading control. # indicates non-specific immunoreactivity. Control mice (Perk^+/+;Sst^IRES-Cre) treated for 5 weeks of UCMS show a decreased trend of total soluble SST proteins (P=0.066, as compared with no stress Perk^+/+;Sst^IRES-Cre control mice). c. Sarkosyl-insolubility assays: Protein extracts from PFC of mice (N=4/genotype/treatment, 3–4 months old, female) kept under no stress or 5 weeks of UCMS conditions were solubilized in 0.2% sarkosyl buffer. Following ultracentrifugation, the insoluble proteins were analyzed by dot blot using an anti-SST antibody. The graph below shows the densitometry of insoluble SST proteins included in 200 μg of protein extracts for each condition. Kruskal-Wallis test with Dunn’s multiple comparisons; *P<0.05 (as compared with no stress Perk^+/+;Sst^IRES-Cre control mice). d. Sarkosyl-insoluble protein fractions were further analyzed by Western blot using an ubiquitin antibody. Mice exposed to UCMS (5 weeks) show elevated ubiquitin⁺ signals compared to no stress mice, regardless of genotypes, consistent with elevated ER stress/UPR.