Trio Ablation in Progenitors Leads to a Smaller and Twisted DG

To determine the function of Trio in hippocampal development, we first detected the protein level of Trio in the hippocampus at postnatal developmental stages (Fig. S1A) and generated Triofl/fl;Emx1-cre mice to specifically disrupt Trio in progenitors in the forebrain (Fig. 1A, B). We found that the brain size of conditional knockout mice was smaller than that of control mice (Fig. 1C). Interestingly, the brain sections showed smaller hippocampus and DG, especially severely twisted suprapyramidal/infrapyramidal blades of DG (Fig. 1D) and zigzagged arrangement of GCs in cKO mice at P2, P7, and P21 (Fig. 1E), with several piles of GCs detected out of the GC layer (arrowhead). However, neither the arrangement of GCs nor the size of DG was affected at P0 (Fig. S1B), and the pattern of DG anlage was not significantly changed in embryonic and perinatal DG of Triofl/fl;Emx1-Cre mice (Fig. S1C, D). Thus, our data indicated that ablation of Trio in progenitors disrupted the morphogenesis of the DG GC layer at the postnatal stage.

Fig. 1

Deletion of Trio in forebrain neural progenitors leads to DG hypoplasia at postnatal stages. A Trio mRNA expression in DG from Triofl/fl mice and Triofl/fl;Emx1-Cre mice at P4. Scale bars, 100 µm. B Trio protein expression in HIP from Triofl/fl mice and Triofl/fl;Emx1-Cre mice at P21. C Brain size of Triofl/fl;Emx1-Cre mice was smaller at P21. Scale bars, 2 mm. D P21 coronal sections revealed abnormal morphology of DG in Triofl/fl;Emx1-Cre mice. Scale bars, 500 µm. E Morphological changes in Trio-deleted DGs at postnatal developing stages (left) and the area of the DG decreased in different levels in Triofl/fl;Emx1-Cre DGs (right) (n = 5 WT; n = 5 cKO). Arrowheads indicated the ectopic granule cells in Triofl/fl;Emx1-Cre DG at P21. Data were shown as means ± SEM.*P <0.05, **P <0.01, ***P <0.001; n.s., no significance, two-tailed Student’s t-test. Scale bars, 100 µm.

Trio Deletion Leads to a Dynamic Reduction and Disturbed Distribution of Progenitors and GCs in Postnatal DG

To further clarify the cellular mechanism of DG malformation induced by Trio deletion, we labeled and quantified different types of cells at postnatal stages. We found that the total number of Pax6+ dentate progenitors (Fig. 2A) and Tbr2+ intermediate progenitors (IPCs; Fig. 2B) were significantly reduced at P2 and P7 but not P0 in Triofl/fl;Emx1-Cre mice. To further characterize the number of postmitotic GCs that were generated from progenitors, we performed Prox1 and NeuN immunostaining for immature and mature neurons, respectively. We found that the numbers of both Prox1+ and NeuN+ neurons were decreased in Triofl/fl;Emx1-Cre DG (Fig. 2C, D). In addition, the distribution patterns of progenitors and GCs were also disturbed in Trio deficiency DG. Progenitors settled in SGZ declined while many of them stayed in the molecular layer (ML) at P7 in Triofl/fl;Emx1-Cre mice (Fig. 2B, arrowhead), and GCs were loosely arranged with more GCs were scattered out of the granule cell layer (GL) in Triofl/fl;Emx1-Cre mice compared to control mice (Fig. 2C, star and arrowhead), suggesting that the localization of neural cells might be affected upon Trio deletion. These results indicated that Trio deletion was involved in the disruption of both the number and the location of neural cells in developmental DG.

Fig. 2

Dynamic reductions and disturbed distribution of progenitors and GCs in postnatal DG following Trio deletion. A The absolute number of progenitors in DG at P0 (n = 3 WT; n = 3 cKO), P2 (n = 3; 3), and P7 (n = 3; 3). Scale bars, 50 µm. B The absolute number of IPCs in DG at P0 (n = 3; 4), P2 (n = 4; 4) and P7 (n = 3; 3). The arrowhead indicated the missing IPCs in SGZ of Triofl/fl;Emx1-Cre mice. Scale bars, 50 µm. C The absolute number of postmitotic neurons in DG at P0 (n = 3; 4), P2 (n = 4; 4), and P7 (n = 3; 3). Dashed lines illustrated the potential borderlines of densely packed cell bands at P2 and P7. Stars indicated the ectopically distributed cells in the hilus region of Triofl/fl;Emx1-Cre mice at P2 and P7. Arrowheads indicated the loosely packed cells in ML of Triofl/fl;Emx1-Cre mice at P7. Scale bars, 100 µm. D The absolute number of GCs in DG at P14 (n = 3; 3) and P21 (n = 3; 3). Scale bars, 100 µm. Data were shown as means ± SEM. *P< 0.05, **P <0.01, ***P <0.001; n.s., no significance, two-tailed Student’s t-test.

Trio Deletion Results in Reduced Cell Proliferation and Ectopically Dividing Progenitors in DG

The male morphogenesis of DG might be due to the abnormalities of several highly relative processes including cell proliferation, apoptosis, differentiation, and migration. The number of proliferating cells was found significantly reduced at P2, P4, P7, and P14 but not at P0 (Figs. 3A–C, S2A), confirming a postnatal hypoplasia. Moreover, the alteration was only found in the suprapyramidal but not the infrapyramidal blade (Fig. S2C–D). Notably, cells that proliferated in GL and ML, considered ectopic dividing cells, were increased in Triofl/fl;Emx1-Cre mice at P14, the end of proliferation peak of postnatal DG (Fig. 3A arrowhead, Fig. S2B), suggesting that the ectopic proliferation might be involved during early postnatal stages.

Fig. 3

Trio deletion involves reduced and especially ectopic proliferation of IPCs in postnatal DG. A The absolute number of Ki67+ cells in DG at P2 (n = 4 WT; n = 4 cKO) and P14 (n = 4; 4). The arrowhead indicated the ectopic distributed Ki67+ cell in Triofl/fl;Emx1-Cre DG. Scale bars, 100 µm. B The absolute number of Ki67+ cells in DG, and their relative percentages in each part of the superior DG based on the total number of Ki67+ cells in the superior DG at P4 (n = 5; 5). Dashed lines illustrated the borderlines of the SPZ, GL, and superior parts of hilus. Scale bars, 100 µm. C The absolute number of Ki67+ cells in DG, and their relative percentages in each part of the superior DG based on the total number of Ki67+ cells in the superior DG at P7 (n = 3; 4). Dashed lines illustrated the borderlines of the ML, GL, and superior parts of hilus. Scale bars, 100 µm. D The relative percentages of proliferating IPCs in each part of the upper blade are based on the total number of Tbr2+EdU+ cells in the superior DG at P4 (n = 3; 4). Dashed lines illustrated the borderlines of the SPZ, GL, and superior parts of hilus. Mice were administrated EdU at P4 and sacrificed after 4 hours. Scale bars, 100 µm. E The relative number of cells existing cell-cycles in each part of the superior DG at P4 (n = 3; 3). Dashed lines illustrated the borderlines of the SPZ, GL, and superior parts of hilus. Mice were administrated EdU at P4 and sacrificed after 24 hours. Scale bars, 100 µm. F Relative percentages of the differentiating IPCs based on the total number of Tbr2+ cells in DG at P7 (n = 3; 3). Scale bars, 100 µm. G The absolute number of cleaved caspase-3+ cells in DG at P7 (n = 4; 4). Scale bars, 100 µm. Data were shown as means ± SEM. *P< 0.05, **P <0.01, ***P <0.001; n.s., no significance, two-tailed Student’s t-test.

Since progenitors gathered in SPZ and then transferred into SGZ from P4 to P7, we then examined the distribution pattern of Ki67+ cells at P4 and P7. The results showed that the distribution proportion of Ki67+ cells was increased in the SPZ and ML at P4 and P7, respectively, but decreased in the GL of Triofl/fl;Emx1-Cre mice at both P4 and P7 (Fig. 3B, C). Considering that Tbr2+ IPCs were the largest population of proliferating cells in SPZ at P4, we then further confirmed whether the ectopic IPCs underwent abnormal mitosis in ectopic positions. Brain sections were stained with Tbr2 following a 4 h EdU injection at P4. We found that the proportion of proliferative IPCs increased in SPZ while decreased in GL in the superior DG of Triofl/fl;Emx1-Cre mice (Fig. 3D). Together with the results of pulse EdU labeling the number of proliferating cells was decreased in the superior mutant DG, but the proportion of EdU+ cells was increased in SPZ (Fig. S2E), it indicated that more IPCs would keep multiplying in SPZ rather than migrated to form SGZ in Triofl/fl; Emx1-Cre mice. Results found in the cell cycle exit assay performed in specific regions of DG at P4 showed that the ectopic IPCs could exit the cell cycle in mutant DG as those in control ones(Fig. 3E). Taken together, our data indicated that though the total number of proliferating cells was decreased in DG, especially in the superior part of DG, due to Trio ablation, IPCs accumulated and multiplied in the SPZ instead of settling into SGZ at early postnatal stages, which might contribute to the ectopic distribution of GCs.

Furthermore, we performed double immunostaining for Tbr2 and Prox1 at P7 to analyze the differentiation from IPCs to postmitotic neurons by calculating the proportion of Tbr2+Prox1+ double-labeled cells in total Tbr2+ cells. We found that the differentiation was not disturbed in Triofl/fl;Emx1-Cre mice (Fig. 3F). Also, the effect of apoptosis was precluded by the Cleaved Caspase-3 staining (Fig. 3G).

Trio is Crucial for Neuron Migration in Postnatal DG Development

To assess the tangential migration of IPCs, we performed Tbr2 immunostaining and EdU labeling. We found that the percentage of Tbr2+ progenitors that arrived at the DG anlage was not dramatically changed in Triofl/fl;Emx1-Cre mice at P0 (Fig. 4A), which was further confirmed by the statistically equal percentage of Edu+ cells in the 3ry in both genotypes(Fig. S3A). However, when the SPZ was divided into three equal parts medial, middle, and lateral part, we found that the percentage of Tbr2+ progenitors in the medial part was increased at both P0 and P2 (Fig. 4B, C). Furthermore, the total number of IPCs in SPZ was detected reduced at P2 and P4 (Fig. S3B), indicating that the formation of SPZ was damaged as a consequence of the reduction of tangential migration velocity of progenitors in Triofl/fl;Emx1-Cre mice.

Fig. 4

Trio is crucial for neuron migration in postnatal DG development. A–C Tangential migration of IPCs was evaluated by calculating the relative percentage of Tbr2+ cells in the DG anlage based on the total number of IPCs that migrated from neuroepithelium at P0 (A), and the proportions of IPCs in three equal parts of SPZ at P0 (B) and P2 (C) (n = 3 WT; n = 3 cKO). Scale bars, 100 µm. D Distribution pattern of IPCs and postmitotic neurons in SPZ was evaluated at P4 by dividing SPZ into five equal parts and calculating the proportions of Tbr2+ and Prox1+ cells in each part respectively (n = 3; 3). Arrowheads indicated the ectopic distributed cells in Triofl/fl;Emx1-Cre mice. Scale bars, 50 µm (left), 10 µm (right). E IPC migration in SPZ was evaluated by calculating the proportions of Tbr2+ cells in ten equal parts of SPZ, and the number of IPCs accumulated in the twisted upper blade at P4 in Triofl/fl mice and Triofl/fl;Emx1-Cre mice (n = 3; 3). Scale bars, 100 µm (left), 10 µm (right). F The SPZ-SGZ transition of IPCs was evaluated by Tbr2 labeling at P7 and their relative percentages in each part of the superior DG were calculated (n = 3; 3). Scale bars, 100 µm. Data were shown as means ± SEM.*P <0.05, **P <0.01, ***P <0.001; n.s., no significance, two-tailed Student’s t-test.

To further characterize the migrating pattern in SPZ, double immunostaining for Tbr2 and Prox1 was performed. Unlike cells that were characterized with sharp boundaries in control mice, in Triofl/fl;Emx1-Cre mice, migratory streams of Tbr2+ and Prox1+ cells converged, so that more postmitotic cells located ectopically in SPZ, and more IPCs were crowded distributed beneath the pial surface, resulting Tbr2+ cells and Prox1+ cells were mixed up in local areas of the suprapyramidal blade (Fig. 4D, arrowhead). Thus, Tbr2+ IPCs accumulated focally in twisted parts in SPZ of Triofl/fl;Emx1-Cre mice (Fig. 4E). Taken together, these data suggested that the distribution patterns of neurons in SPZ were disrupted both tangentially and radially after the deletion of Trio.

To further detect the SPZ-SGZ transition of progenitors, the distribution proportion of Tbr2+ cells in each part of the suprapyramidal blade of DG was calculated at P7. The result showed that a greater percentage of IPCs were more likely to stay in ML, but not transit into SGZ in in Triofl/fl;Emx1-Cre mice, compared with control (Fig. 4F), while the IPCs in SGZ were distributed irregularly (Fig. S3C). To investigate further, we injected mice with EdU at P3 and harvested the brains at P7, that EdU+ cells migrating from SPZ gathered to form a symmetric cell band which would be the SGZ region. But in mutant SGZ, cells were distributed extremely asymmetrically, cells were even absent in some positions (Fig. S3D, arrowhead). These data suggested that the radial migration of IPCs was disrupted which resulted in a disturbance in the formation of SGZ in Trio ablated mice.

As neurons migrated along the scaffolds consisting of radial glia cells, we then examined the primary and secondary glia scaffolds by immunostaining the markers of radial glia cells during different developmental stages. BLBP and Vimentin staining showed normal density and extended long processes across the fimbria and radially to the hippocampal fissure at E16.5 and P0 (Fig. S4A, B) in Triofl/fl;Emx1-Cre mice, while the distribution and orientation of the radial glia scaffold were slightly changed in the DG anlage at P0 (Fig. S4B) in Triofl/fl;Emx1-Cre mice when comparing to controls. However, the secondary glia scaffold labeled with GFAP was missing above the twisted part of the GC band, and the orientation of the radial glia scaffold was also disrupted in Triofl/fl;Emx1-Cre mice at P7 and P14 (Fig. S4C, D).

In summary, our data showed that Trio deletion in DG led to the malformation of both SPZ and SGZ, indicating that Trio plays a critical role in neuron migration in postnatal DG development.

Trio Regulates the Migration of Postmitotic Cells That Provides the Framework of Postnatal Reorganization of Dentate Neurons

Since the localization of IPCs and postmitotic cells were both affected in Triofl/fl;Emx1-Cre mice, to further identify the type of neurons that play a dominant role in the postnatal morphogenesis of the suprapyramidal blade, we generated Triofl/fl;Nex-Cre mice, in which Trio was specifically knocked out in postmitotic excitatory neurons. In these mice, smaller brain size and reduced DG area were observed (Fig. 5A, D), with a remarkably twisted upper blade of DG (Fig. 5B). The GCs in Triofl/fl;Nex-Cre mice were also loosely and zigzagged packed (Fig. 5C). Meanwhile, converged migration streams of Tbr2+ and Prox1+ cells were detected in Triofl/fl;Nex-Cre mice at P2, suggesting a deficiency in neuron distribution and migration (Fig. 5E). However, when evaluating the glia scaffold of Triofl/fl;Nex-Cre DG by staining GFAP and Vimentin, no remarkable impairment was detected at any time point of postnatal development (Figs. 5F, S5A). In addition, the result of labeling proliferative cells by Ki67 at P14 showed that Ki67+ cells were reduced in the SGZ and ectopically distributed in the GCL and the ML of Triofl/fl;Nex-Cre mice (Fig. S5B), these results indicated a malformation of SGZ, which was similar to the deficiency in Triofl/fl;Emx1-Cre mice. According to the above results, similar impairments were observed in Triofl/fl;Nex-Cre mice, although the phenotype of twisted DG blades was milder than that in Triofl/fl;Emx1-Cre mice. Thus, we can infer that the damage in postmitotic cell migration was most associated with the malformation of postnatal DG.

Fig. 5

Similar but milder impairments were observed in postnatal DG of Triofl/fl;Nex-Cre mice. A Brain size of Triofl/fl;Nex1-Cre mice was smaller at P21. Scale bar, 2mm. B P21 coronal sections revealed abnormal morphology of DG in Triofl/fl;Nex-Cre mice. Scale bars, 100 µm. C-D Morphological changes in Trio-deleted DGs at postnatal developing stages (C and D left) and the area of the DG decreased in different levels in Triofl/fl;Nex-Cre DGs (D right) (n = 5 WT; n = 5 cKO). Data were shown as means ± SEM. ***P <0.001; n.s., no significance, two-tailed Student’s t-test. E Migration streams of Tbr2+ and Prox1+ cells were converged in Triofl/fl;Nex-Cre mice at P2, but not at P0. Scale bars, 100 µm. F The secondary radial glia scaffold was detected by GFAP staining at P14, and no impairments were found in Triofl/fl;Nex-Cre mice. Scale bars, 100 µm (left), 10 µm (right).

Spatial-Transcriptome Sequencing Revealed Different Functions of Trio in Various DG Excitatory Neurons

To investigate the underlying molecular mechanisms in abnormal postnatal development of DG, we generated and analyzed the spatial-transcriptomic data using direct approaches to define and select cell populations. We isolated 6 subclasses of DG neural cells at P0, based on the well-known markers of NSCs, progenitor cells, and immature and mature GCs, which were BLBP+, GFAP+, Tbr2+, Prox1+, CTIP2+, and NeuN+ cells. The expression level of Trio showed cell-specific diversity (Fig. 6A). At the cluster level, only 4.84% Tbr2+ cells expressed Trio, while 7.3% Prox1+ cells and 15.38% CTIP2+ cells expressed Trio (Fig. 6B). We performed additional confirmation by RNAscope co-staining of Trio with Tbr2 and CTIP2 respectively, and found that a higher Trio RNA expression level was detected in GCs rather than in progenitors (Fig. S6A).

Fig. 6

Spatial-transcriptome sequencing revealed different functions of Trio in various DG excitatory neurons. A Violin plot of the normalized expression level of Trio in different cell types of DG at P0 in spatial-transcriptome seq data with bin size of 30. B Relative percentages of cells expressing Trio based on the total detected numbers of each type of cell. C Spatial feature plots of prox1-, Tbr2- and CTIP2-expressing and prox1-Tbr2- and CTIP2-prox1-co-expressing cells in Triofl/fl and Triofl/fl;Emx1-Cre mice tissue sections. D GSEA analysis of Tbr2+ and prox1+ cells in neural migrating pathways. E Volcano plot of differentially expressed genes in spatial-transcriptome seq data form prox1+ cells in Triofl/fl and Triofl/fl;Emx1-Cre DGs (left). The bubble plot showed the top 20 terms of converged pathways that were down-regulated in Triofl/fl;Emx1-Cre DG by GO analysis (right). F Volcano plot of differentially expressed genes in spatial-transcriptome seq data form Tbr2+ cells in Triofl/fl and Triofl/fl;Emx1-Cre DGs (left). The bubble plot showed the top 20 terms of converged pathways that were down-regulated in Triofl/fl;Emx1-Cre DG by GO analysis (right).

Spatial-transcriptomic data was generated from a pair of Triofl/fl and Triofl/fl;Emx1-Cre mice brain at P0, which validated the ablation of Trio in DG by RNAscope (Fig. S6B). Alignment with single-cell bins data confirmed the spatial localization of excitatory neurons in the DG regions, and cells were isolated based on the expression of marker genes Tbr2, Prox1, and CTIP2 (Fig 6C). GSEA analysis was performed In Tbr2+ and Prox1+ cells in both Triofl/fl;Emx1-Cre mice and the control mice, showing that Prox1+ cells were highly involved in neuron migration and cytoskeleton organization and regulation, rather than Tbr2+ progenitor cells (Fig 6D). These results indicated that expression levels of Trio in these developing neuronal cells were variable, which led us to speculate on the variable roles that different neuronal cells play in postnatal DG development.

We then conducted differential gene expression analysis in progenitors, postmitotic cells, and radial glial cells respectively between Triofl/fl;Emx1-Cre mice and control. We found that Trio deletion in GFAP+ cells did not induce numerous genes down-regulating or up-regulating (Fig. S6C). However, Gene ontology (GO) analysis revealed that Trio was associated with neuron migration, regulation of reorganization of cytoskeleton components, and synapse organization among others in Prox1+ postmitotic neurons (Figs. 6E, S6D). In Tbr2+ progenitors, alteration of Trio was associated with the site of polarized growth, axonal growth, and GTP-protein activity (Figs. 6F, S6E). Thus, we can infer that Trio contributes to the postnatal reorganization of dentate GC blades by involving in diverse biological pathways in different cell populations, among which the migration of postmitotic neurons was directly regulated by Trio.

Furthermore, we accessed the expression level of reported ASD-related genes in the 6 subclasses of DG neural cells at P0 and found that the expression patterns of these genes were quite disparate in different developing neuronal cells of DG (Fig. S6F), suggesting a probable reason of the complicated etiology of ASD.

Trio-Deleted Mice Exhibit Autism-like Behaviors

To investigate the potential association of Trio deletion-induced DG malformation with autism-related abnormal behaviors, we conducted a battery of behavioral tests. In the three-chamber social interaction test, we found the deficiency of social novelty recognition in Triofl/fl;Emx1-Cre mice, while the social ability was not affected (Fig. 7A). Stereotyped behaviors were evaluated by hole-board test and marble burying test, showing that stereotyped number in Triofl/fl;Emx1-Cre mice were increased (Fig. 7B, C). However, the anxiety-like behaviors in Triofl/fl;Emx1-Cre mice tested by elevated plus-maze, light-dark box, and open field assay showed no difference compared to the control mice (Fig. 7D).

Fig. 7

Trio-deletion-induced DG hypoplasia is related to autism-like behaviors. A Three-chamber social interaction (SI) test. In the sociability trial (S1 vs EM), mice of both genotypes spent more time in close interaction with strangers1 (S1) than in the empty chamber (EM). In the social novelty preference trial (S1 vs S2), Triofl/fl;Emx1-Cre mice did not display a preference for the novel social partner, stranger2 (S2) (n = 15 WT; n = 13 cKO). B-C Stereotype behaviors were detected by the hole-board test and marble burying test. Triofl/fl;Emx1-Cre mice performed more stereotyped digs (B, n = 18; 16) and buried more marbles (C, n = 16; 16). D Anxiety-like behaviors were determined by calculating the time spent in the open arm in the elevated plus-maze (n = 14; 14), the time spent in the light box in the light-dark box test (n = 17; 15), and the time spent in the center region during the first 10 min in the open field test (n = 15; 15). Data were shown as means ± SEM. *P <0.05, **P <0.01, ***P <0.001; n.s., no significance, two-tailed Student’s t-test.

These results revealed that Triofl/fl;Emx1-Cre mice displayed social deficits and increased stereotyped behaviors, which were core abnormal behaviors related to ASDs.