Cn deficiency induces immature dentate gyrus phenotype as assessed by typical neuronal maturation markers
We first examined the expression patterns of typical markers of mature and immature GCs in the GC layer. Expression of calbindin (Fig. 1a), GluR1 (Fig. 1b), and GluR2 (Fig. 1c) [63], markers of mature GC, was dramatically decreased in Cn mutants. The number of Hoechst-stained nuclei in the GC layer was not significantly different between genotypes (mutants: 112.64 ± 3.59 cells/mm2, controls: 113.43 ± 5.10 cells/mm2; P = 0.90), indicating that the decreased expression of those mature GC markers in Cn mutants was not due to the loss of GCs. Expression levels of mRNA for other mature GC markers, Dsp and Tdo2 [64], were also significantly lower in the DG of Cn mutants (Fig. 1i). In contrast, the expression of doublecortin (Fig. 1d), PSA-NCAM (Fig. 1e), and calretinin (Fig. 1f), markers for progenitors or immature GCs, was higher in the Cn mutants. Doublecortin-positive cells were observed within the GC layer in Cn mutants, whereas in controls, such cells were mostly located in the subgranular zone, where GCs proliferate and differentiate into new neurons (Fig. 3d). We also found that the expression of phospho-cAMP-dependent response element binding protein (CREB), which is predominantly expressed in immature GCs and located in the inner GC layer in the normal adult DG [65–67], was increased throughout the GC layer in Cn mutants (Fig. 1g), while the expression levels of total CREB were almost normal (Fig. 1h). We also found that the expression of Drd1a mRNA was increased in the DG of Cn mutants compared to that in controls (Fig. 1i).
An additional feature common to iDG mouse models is the activation of astrocytes in the DG [21]. Studies in human post-mortem brains have suggested activation of astrocytes in the brain of patients with neuropsychiatric disorders, including ASD [68,69], ID [70], epilepsy [71,72], schizophrenia [73,74], and AD [75], which are considered inflammatory conditions of the brain [73]. In Cn mutants, the expression of glial fibrillary acidic protein (GFAP), an astrocytic marker, increased significantly in the molecular layer of the DG, while the number of GFAP-positive cells was not different from that in controls (Fig. 2a). We found that neither expression of the microglial marker Iba1 nor the number of Iba1-positive cells was changed in Cn mutants (Fig. 2b). These results suggest astrogliosis with no apparent microgliosis in the DG of Cn mutants. Thus, Cn mutants showed the iDG phenotype as assessed by typical marker expressions that were common to previously identified mouse models with iDG [21].
Transcriptomic evidence for the immaturity of the DG in Cn mutant mice
Next, we evaluated the iDG phenotype in Cn mutants at a genome-wide gene expression level. Microarray analysis revealed that of 45,037 transcripts tested, 353 were differentially expressed in the DG of Cn mutants compared to controls (absolute fold change > 1.2, P < 0.05, without correction for multiple tests; Fig. 3a). Cnb1 (Ppp3r1) displayed the lowest value among the differentially expressed genes (DEGs) (Additional file 2: Table S3). Pathway analysis showed that terms related to cell proliferation and development were enriched in the DEGs (Fig. 3b; Additional file 2: Table S4), supporting the idea of maturation abnormalities in the DG of Cn mutants.
To further define the iDG phenotype in Cn mutants, we used publicly available microarray datasets to conduct a comparative transcriptomic analysis. This analysis examined whether, or to what extent, overall gene expression patterns were similar between adult Cn mutants and typically developing infant mice (Fig. 3c). Transcriptome datasets were compared using the Running Fisher test, a non-parametric rank-based statistical method, which calculates overlap P-values between two given gene sets in consideration of fold-change-based rank and the direction of changes [57]. In this analysis, we identified a highly significant overlap P-value with a positive correlation between the DG of adult Cn mutants (mutants vs. controls) and the DG of infant mice (2-week-old vs. 1-month-old) (P = 2.7 × 10-13; Figs. 3d and 3e, Additional file 2: Table S5), indicating a significant similarity in the gene expression patterns between them. Furthermore, the patterns of gene expression changes in Cn mutants were similar to those in Hivep2 KO mice [3], Camk2a+/- mice [2], and fluoxetine-treated mice [45,46], and other mouse models with iDG (Fig. 3f). These results confirmed the iDG phenotype in Cn mutants from a transcriptomic standpoint.
Chronic rolipram treatment ameliorates iDG phenotype and nest-building behavior
Increased expression of Drd1a, which stimulates intracellular cAMP signaling, is a common feature found in previously identified mouse models with iDG [20], and an increase was also found in Cn mutants (Fig. 1i). We also found an increase in CREB phosphorylation (Figs. 1g and 1h), which are known to be increased by cAMP/protein kinase A (PKA) [76], while CREB is suggested to be not a direct target of Cn [77]. These results suggest that cAMP signaling is elevated in the DG of Cn mutants. We previously found that chronic treatment with rolipram, a cAMP-specific phosphodiesterase inhibitor that elevates intracellular cAMP levels (in combination with ibuprofen), rescued the iDG phenotype in Hivep2 KO mice [3], raising the possibility that the increased cAMP signaling in Cn mutants is due to a compensatory mechanism. To determine whether the increased cAMP signaling is related to compensatory or pathological mechanisms underlying the iDG and behavioral phenotypes, we investigated the effect of rolipram on the phenotypes of Cn mutants.
The rolipram treatment significantly increased the expression levels of the mature GC marker calbindin (P = 0.0057; Figs. 4a and 4b) and decreased the expression levels of the immature GC marker phospho-CREB (P = 0.033; Figs. 4a and 4c) in Cn mutants. Expression levels of GluR1 were not affected by rolipram treatment (Additional file 1: Fig. S2a), and the number of doublecortin-positive (Fig. 4d) and calretinin-positive cells (Fig. 4e) tended to be lower following rolipram treatment in Cn mutants. These results suggest that chronic treatment with rolipram partially rescued the iDG phenotype in Cn mutants. Moreover, chronic rolipram treatment decreased the expression of the astrocytic marker, GFAP, in mutants (P = 0.010; Fig. 4f), which may be due to the anti-inflammatory effects of the drug [78].
A series of behavioral tests revealed that chronic rolipram treatment improved the impaired nest-building activity in Cn mutants at a nominal significance level (raw P = 0.025; Fig. 4g). Chronic treatment with rolipram did not significantly affect locomotor activity or anxiety-like behavior in the open field test (Additional file 1: Figs. S3a–S3d), prepulse inhibition (Additional file 1: Figs S3e and S3f), or working memory assessed using the T-maze spontaneous alternation task (Additional file 1: Fig. S3g). Thus, rolipram treatment selectively rescued impaired nesting behavior, which is considered to be associated with negative symptoms of psychiatric disorders [43].
Increased Drd1a/PKA signaling activity in the DG of Cn mutant mice
Considering the increased expression of Drd1a in Cn mutants and the cAMP-modulating effect of rolipram, we investigated cAMP-dependent PKA signaling activity downstream of Drd1a to gain mechanistic insights into the rescue effects of rolipram. After the 3-week treatment of mice with rolipram or vehicle, we prepared DG slices and examined the phosphorylation levels of known substrates of PKA (P-Ser845 GluR1, P-Thr34 dopamine-and cAMP-regulated phosphoprotein of 32 kDa [DARPP-32], and P-Ser133 phosphodiesterase 4 [PDE4]) or a downstream substrate of PKA (P-Thr202/Tyr204 extracellular signal-regulated kinase 2 [ERK2]) with or without incubation with SKF81297, a Drd1a agonist.
In the vehicle-treated group (Figs. 4h–4k and S4), the effect of genotype on phosphorylation levels of all four substrates for PKA signaling was examined under both basal and SKF81297-stimulated conditions, suggesting that Drd1a-mediated PKA signaling is activated in Cn mutants. Interestingly, in the rolipram-treated group (Figs. 4l–4o), phosphorylation levels of P-Ser845 GluR1 and P-Thr202/Tyr204 ERK2 were not significantly different between Cn mutants and controls (Figs. 4l and 4m), whereas levels of P-Thr34 DARPP-32 and P-Ser133 PDE4 remained increased in Cn mutants (Figs. 4n and 4o). In Cn mutants, chronic rolipram treatment shifted P-Ser845 GluR1 and P-Thr202/Tyr204 ERK2 levels downward, and there was a tendency of decrease in P-Ser845 GluR1 levels (P = 0.090 for P-Ser845 GluR1 and P = 0.13 for P-Thr202/Tyr204 ERK2) (Additional file 1: Fig. S5). In contrast, in control mice, P-Ser845 GluR1 and P-Thr202/Tyr204 ERK2 levels were shifted upward following chronic rolipram treatment, and the effect of rolipram treatment on P-Thr202/Tyr204 ERK2 levels was significant (P = 0.0070) (Additional file 1: Fig. S5). These results indicate that chronic rolipram treatment modulates PKA activity downstream of Drd1a in a substrate-selective manner with the opposite effect between Cn mutant and control mice.