Effect of Immune Activation during Early Gestation or Late Gestation on Inhibitory Markers in Adult Male Rats

People with schizophrenia exhibit deficits in inhibitory neurons and cognition. The timing of maternal immune activation (MIA) may present distinct schizophrenia-like phenotypes in progeny. We investigated whether early gestation [gestational day (GD) 10] or late gestation (GD19) MIA, via viral mimetic polyI:C, produces deficits in inhibitory neuron indices (GAD1, PVALB, SST, SSTR2 mRNAs) within cortical, striatal, and hippocampal subregions of male adult rat offspring. In situ hybridisation revealed that polyI:C offspring had: (1) SST mRNA reductions in the cingulate cortex and nucleus accumbens shell, regardless of MIA timing; (2) SSTR2 mRNA reductions in the cortex and striatum of GD19, but not GD10, MIA; (3) no alterations in cortical or striatal GAD1 mRNA of polyI:C offspring, but an expected reduction of PVALB mRNA in the infralimbic cortex, and; (4) no alterations in inhibitory markers in hippocampus. Maternal IL-6 response negatively correlated with adult offspring SST mRNA in cortex and striatum, but not hippocampus. These results show lasting inhibitory-related deficits in cortex and striatum in adult offspring from MIA. SST downregulation in specific cortical and striatal subregions, with additional deficits in somatostatin-related signalling through SSTR2, may contribute to some of the adult behavioural changes resulting from MIA and its timing.

timing: F (1,24) = 0.032, treatment × timing: F (1,24) = 0.18, all p > 0.05). No more than two male offspring per litter were used in each experimental group to avoid litter-effect confounds and were randomly assigned to the study. Whole brains, snap-frozen in isopentane (−40 °C) for storage (−80 °C), were sectioned (coronal, 14 μm) using a cryostat (Leica, Wetzlar, Germany) and mounted onto gelatin-coated glass slides. Table 1) were generated with 35 S-UTP (CAT# NEG039H001MC Perkin Elmer, Waltham, Massachusetts, USA) using an in vitro transcription kit (CAT# P1121, Promega, Madison, Wisconsin, USA). In situ hybridisation was performed as previously described 57 , using 5 ng/ ml radiolabelled riboprobes in hybridisation buffer, and 35 S-UTP labelled sense strand riboprobes as a negative control ( Supplementary Figures 1 and 2). Slides were exposed to BioMax MR (Kodak, Rochester, NY, USA) autoradiographic film (Supplementary Table 1  Statistical analysis. Graphs were plotted using Graph Pad Prism (v6) and data analysed with IBM SPSS statistics (v23). All data passed Shapiro-Wilk normality tests. Three-way mixed analysis of variance (ANOVA) was conducted for each region (cortex, striatum, and hippocampus), separately; hence there were three 2 × 2 × 3 repeated-measure (RM) ANOVAs in total. For each of these analyses, the two between-subject factors were treatment (polyI:C or vehicle) and timing (GD10 or GD19), and the within-subject factor was subregions (cortical subregions: IL, Cg, and Aud cortex; striatal subregions: DS, AC, and AS; hippocampal subregions: CA1, CA3, and CA4). Bonferroni tests were used for pairwise comparisons when the overall ANOVA was significant. The Greenhouse Geisser correction was used if Mauchley's sphericity test was violated for within-subjects interaction effects. Deming regressions were conducted to query the relationship between each offsprings' regional gene expression of each marker and their respective maternal IL-6 protein exposure. Data are expressed as the mean ± standard error of the mean (SEM) and two-sided p < 0.05 was deemed statistically significant.

Results
GAD1 mRNA signals were punctate in the cerebral cortex and hippocampus, and darker and more homogeneous across the striatum (Fig. 1E). PVALB mRNA signals were punctate in the cerebral cortex, hippocampus, and striatum. PVALB mRNA signals in the cortex was stronger and denser in comparison to signals in the hippocampus and striatum (Fig. 1E). SST mRNA signals were punctate in all regions, and denser in the cerebral cortex in comparison to the hippocampus and striatum (Fig. 1E). SSTR2 mRNA signals were highest in the molecular layer of the hippocampus, followed by the deeper layers of the cerebral cortex, the superficial layers of the cerebral cortex, and then the striatum (Fig. 2). GAD1 mRNA 59,60 , PVALB mRNA 61 , and SST mRNA 62 signals correspond to previously described distributions. Although cortical SSTR2 mRNA signals were lighter in comparison to some studies 63 , the distribution corresponds to SSTR2 mRNA and SSTR2 binding results in multiple other studies (Supplementary Figure 1A) [64][65][66][67][68] . Control (sense) riboprobe hybridization, for all transcripts, was at background levels (Supplementary Figure 1B, 2).
There was a main effect of subregion on the expression pattern of GAD1 mRNA, PVALB mRNA, SST mRNA, and SSTR2 mRNA for cortical, striatal, and hippocampal comparisons in all offspring, except for SST mRNA in hippocampus (Supplementary Table 2). A summary of all treatment effects and treatment-related interactions are highlighted in Table 1. Briefly, there were significant treatment effects for SST mRNA in cortex (Fig. 3A) and striatum (Fig. 3B), treatment × timing interactions for SSTR2 mRNA in cortex (Fig. 3A) and striatum (Fig. 3B), and treatment × subregion interactions for SST mRNA in cortex (Fig. 4A) and striatum (Fig. 4B). There were no significant treatment × timing × subregion interactions for any of the markers (Supplementary Figure 3A-C).
Cortical PVALB mRNA and SST mRNA are reduced in polyI:C offspring in a subregion-specific manner, whereas cortical SSTR2 mRNA is reduced in polyI:C offspring in a timing-specific manner.  (Table 1). As expected, PVALB mRNA was reduced by 22% in the IL cortex of polyI:C offspring (a priori univariate ANOVA: F (1, 22) = 5.08, p = 0.04; Fig. 4A).

Discussion
Our findings demonstrate differential, long-term impacts of early gestation and late gestation MIA on inhibitory indices within the cortex and striatum of adult rat polyI:C offspring. As hypothesised, we found PVALB mRNA reductions in the infralimbic cortex of polyI:C offspring, which aligns with previous studies that report reductions in PVALB-positive cells in the medial prefrontal cortex in both early gestation and late gestation adult 49,52 Figure 4. Inhibitory neuron gene expression is altered in subregions of (A) cortex and (B) striatum, but not (C) hippocampus, in adult male polyI:C offspring. Pregnant dams were treated with vehicle (Control, grey) or 4 mg/kg (PolyI:C, dotted) during either early gestation or late gestation (data herein represent gestational timing combined, data is presented separately in Supplementary Fig. 3). Glutamate decarboxylase 1 (GAD1) parvalbumin (PVALB), somatostatin (SST) and somatostatin receptor 2 (SSTR2) mRNAs were quantified in vehicle (control, grey bar) and polyI:C (dotted bar) offspring. Treatment × subregion interaction effects were not present for GAD1 or SSTR2 mRNAs in any subregion. PolyI:C offspring had reductions in PVALB mRNA in the infralimbic cortex compared to controls (a priori univariate ANOVA: F(1, 22) = 5.08, p = 0.04). SST mRNA was significantly reduced in (A) cingulate cortex and (B) nucleus accumbens shell of the striatum in polyI:C offspring. Refer to Table 1 for treatment × subregion interaction for SST mRNA. Data are mean ± SEM (n = 12-16 rats per group). Infralimbic cortex (IL), cingulate cortex (CG), auditory cortex (AUD), dorsal striatum (DS), nucleus accumbens core (AC), nucleus accumbens shell (AS), cornu ammonis (CA). # p < 0.05, **p < 0.01.

Scientific RepoRtS |
(2020) 10:1982 | https://doi.org/10.1038/s41598-020-58449-x www.nature.com/scientificreports www.nature.com/scientificreports/ and juvenile 53 mouse polyI:C offspring. Somewhat surprisingly, we did not find significant changes in the inhibitory neuron marker, GAD1 mRNA, in cortical subregions, and no significant changes in either PVALB or GAD1 mRNAs in the hippocampus of polyI:C offspring. Although this lack of apparent change in GAD1 mRNA aligns with the majority of the reports in polyI:C offspring, methodological variations may contribute to inconsistent findings in the field (see Supplementary Table 3 for a brief overview). We also report no significant changes in either PVALB or GAD1 mRNA in the striatum. Although, to our knowledge, there have been no investigations of PVALB gene expression in the striatum of adult polyI:C offspring, our results align with studies that also find no significant change in GAD1 mRNA in the striatum of GD14 polyI:C adult rat offspring 38 . Our findings suggest that some inhibitory neuron deficits may not consistently result from MIA, or may arise due to MIA at a different gestational time point. Further, we acknowledge that our sample size may not detect subtle changes. Indeed, we find approximate trend reductions in PVALB mRNA in the cortex (p < 0.11), striatum (p < 0.09) and hippocampus (p < 0.07) of MIA offspring, and qualitative observation of our data suggests that this may arise from MIA at GD19. Our data suggests that future studies are warranted with larger sample sizes to investigate the effect of MIA timing on these inhibitory markers, particularly GAD1 and PVALB.
We present the first report of reduced SST mRNA in both cortex and striatum in the rat MIA model of schizophrenia. In the cortex of people with schizophrenia, SST mRNA reductions are typically of a larger magnitude and can be as robust (if not more robust) than the more often studied PVALB mRNA changes 8,20,21,24 , with widespread decreases in SST mRNA reported in dlPFC, orbital frontal cortex, anterior cingulate cortex, motor cortex, and visual cortex 8,20,21,24 . The anatomically specific MIA-induced reductions in SST mRNA in the cingulate cortex, but not in infralimbic and auditory cortex, suggest that MIA may impact specific circuitry differentially. If MIA alone is not sufficient to recapitulate the fairly pervasive cortical reductions in SST mRNA observed in people with schizophrenia 8,20,21,24,69 , a subsequent postnatal stressor, or 'second hit' , may be needed to elicit a widespread change in SST gene expression in the MIA model [see 70 for review]. The SST mRNA deficits detected in the cortex and striatum, but not hippocampus, may relate to the distinct origins of inhibitory neurons in these regions. In cortex, the majority of PVALB-and SST-containing inhibitory neurons are derived from the medial ganglionic eminence (MGE) ( 71 , see 72 for review). In hippocampus, although some PVALB-and SST-containing inhibitory neurons also arise from the MGE, approximately 40% of SST-containing inhibitory neurons are derived from the caudal ganglionic eminence (CGE) 73 . The dominant period of MGE neurogenesis (GD9.5-GD13.5) precedes that of CGE neurogenesis (GD12.5-16.5) 74,75 . It is possible that the time points we chose to elicit MIA may alter neurogenesis for each ganglionic eminence distinctly and somewhat spare the interneurons destined for the hippocampus.
We also present the first report of cortical and striatal SSTR2 mRNA reductions in a rodent MIA model of schizophrenia, which aligns with the SSTR2 mRNA reduction seen in the cortex of people with schizophrenia 26 . Reductions in both regions, however, are contrary to our hypothesis that postulated that the cortex is more susceptible to MIA than the striatum. Given that our previous behavioural study showed working memory impairments in GD19 polyI:C offspring, but not GD10 polyI:C offspring 50 , and multiple studies show a central role of somatostatinergic signalling in cognitive function [76][77][78][79] , our present findings suggest that the combined SSTR2 mRNA deficits in cortex and striatum (found only at GD19), in combination with the SST mRNA deficits in cortex and striatum, may contribute to the working memory deficits apparent in late gestation MIA offspring.
Qualitative observation of our data shows that SST mRNA deficits in cortex in polyI:C offspring seem to occur at both GD10 and GD19, whereas SST mRNA deficits in striatum in polyI:C offspring seem driven by MIA at GD10. This is possibly evidence of a timing-specific effect of MIA in striatum but not cortex; however, we did not detect a significant treatment × timing interaction effect or treatment × timing × subregion interaction effect that would permit us to investigate these further. Our data suggests that future studies are warranted with larger sample sizes may reveal significant effects of MIA timing on inhibitory markers in offspring.
Further, maternal IL-6 elevations are prolonged when polyI:C is administered in rodent dams during late gestation versus early gestation 80 , and maternal IL-6 levels in humans are associated with altered offspring neonatal functional connectivity and cognitive outcome [81][82][83][84] . Although we only measured maternal IL-6 at one time point, we also report a moderate negative relationship between maternal IL-6 and offspring SST gene expression in cortex and striatum. Overall, our data supports the notion that SST and SSTR2 gene expression deficits may contribute to the working memory deficits apparent in late gestation MIA offspring.
Increased SST-positive interstitial white matter neurons (IWMN) and concurrent decreases in SST mRNA in the grey matter is reported in post-mortem frontal cortex from schizophrenia cases 24,85 . Recent studies show that both early gestation and late gestation polyI:C treatment increased the number of SST-positive IWMNs in regions that extend underneath the cingulate cortex in adult rat offspring 86 . Our detection of reduced SST mRNA in the grey matter suggests MIA during either the initial development of the ganglionic eminence 54 , or the time of neuronal tangential migration 55,56 could alter the migration or survival of SST cortical interneurons. Interestingly, late gestation polyI:C offspring exhibit increased SST-positive IWMNs in more extensive regions of white matter compared to early gestation polyI:C offspring 86 . Given that SSTR2 is highly expressed on migrating neurons during early neurodevelopment in both rat and human 87 , our present findings may indicate a link between exaggerated IWMN pathology and SSTR2 deficits in late gestation polyI:C offspring. Indeed, reductions in SST mRNA and reductions in SSTR2 mRNA are correlated in dlPFC in schizophrenia 26 .
Overall, our current and prior findings are consistent with our hypothesis and earlier findings that gestational inflammation contributes to inhibitory neurotransmission deficits. This supports the growing evidence of altered inhibitory indices in the cortex and striatum of adult polyI:C offspring and recapitulates some aspects of neuropathology present in schizophrenia. We present the first rodent MIA study of cortical and striatal deficits in SST and SSTR2 gene expression that concurs with cortical SST and SSTR2 mRNA deficits found in post-mortem schizophrenia studies. Our novel finding of SST and SSTR2 mRNA reductions in striatum suggests that further examination of SST-related neurotransmission may be warranted in the striatum of people with schizophrenia.