TRPC4 regulates limbic behavior and neuronal development by stabilizing dendrite branches through actomyosin-driven integrin activation

Significance

Neurons continue to grow after birth, developing intricate dendritic arbors that are essential for assembling the neural circuits underlying all brain functions. The process of dendritic arborization involves both the outgrowth and stabilization of new dendritic branches and is fundamental to proper neurodevelopment. This study reveals a previously unrecognized, force-sensitive mechanism by which the calcium-permeable ion channel TRPC4 connects metabotropic glutamate receptor signaling to integrin activation in the dendrites of mouse hippocampal neurons. This pathway plays a critical role in stabilizing dendrite branches, a key step in supporting early postnatal brain development that underlies everyday functioning and social behavior. These findings offer insights into the cellular mechanisms that may contribute to neurodevelopmental disorders such as autism spectrum disorder.

Abstract

Transient Receptor Potential Canonical 4 (TRPC4) channels have been implicated in multiple neurological functions, including anxiety and sociability. TRPC4 variants were also found in patients with autism. However, the contributions of TRPC4 to neurodevelopment remain undefined. Here, we show that neurobehavioral deficits appear early in young TRPC4 knockout (Trpc4^−/−) mice immediately after weaning, manifesting as alterations in multiple, limbic-related behaviors, such as nesting, marble burying, burrowing, self-grooming, and social interactions. Hippocampal neurons of Trpc4^−/− mice exhibit reduced dendritic arborization both in vivo and in vitro. Mechanistically, we found that TRPC4 expression in dendrites surged at the same time when glutamate exerted its stimulatory effect on dendritic branching. In live-cell imaging assays, glutamate induced the formation of new dendrite branches in both wild-type and Trpc4^−/− neurons. However, many of the new and preexisting branches retracted in the mutant neurons. Furthermore, TRPC4 mediated Ca²⁺ entry in dendrites downstream from metabotropic glutamate receptors, leading to phosphorylation of nonmuscle myosin light chain (MLC) by the Ca²⁺-dependent myosin light chain kinase (MLCK), which in turn supported integrin activation in dendrite branches. Our findings underscore the essential role of TRPC4 channels in both dendrite morphogenesis and modulation of motivation-related behaviors. The mechanistic underpinning elucidated for the impaired dendritic development and abnormal behaviors in juvenile Trpc4^−/− mice provides insights into the understanding of neurodevelopmental disorders, particularly autism spectrum disorder, opening potential avenues for targeted therapeutic interventions against TRPC4.

TRPC4 is a member of the Transient Receptor Potential Canonical (TRPC) family of Ca²⁺-permeable nonselective cation channels. In the lateral septum and a few other brain regions, TRPC4 regulates afterdepolarization and action potential firing of neurons (1, 2). In the lateral amygdala, TRPC4 regulates innate fear responses (3). TRPC4 also participates in hippocampus-related functions, including sociability in adult mice (4), and as a part of the heteromeric TRPC1/4/5 channel complex, working memory formation, and regulation of presynaptic signaling in the CA3-CA1 synapses (5). However, little is known about the role of TRPC4 in neurodevelopment despite the implication of this channel in sociability or even autism (6).

Two related channels, TRPC5 and TRPC6, have also been linked to autism spectrum disorder (ASD) and/or intellectual disability (7, 8). Consistent with the general view that impairments in dendritogenesis and synaptogenesis are strongly implicated in neurodevelopmental disorders (9, 10), both TRPC5 and TRPC6 play roles in dendrite growth, with TRPC5 having a negative whereas TRPC6 having a positive effect (11, 12). These channels primarily affect dendritic arborization through transcriptional regulation, mediated by CaMKIIβ (TRPC5) and CaMKIV (TRPC6) (11, 13). TRPC4 shares ~78% sequence similarity with TRPC5 (14, 15), but its role in dendrite development remains unclear.

In the mouse hippocampus, TRPC4 mRNA is detected from postnatal day 4 (P4) to P28, peaking at P14 (16), a time corresponding to early adolescence in rodents and coinciding with structural and functional transformations in limbic regions, which are associated with learning, memory formation, and behavioral changes (9, 17). During this postnatal phase, dendrites undergo dynamic changes such as growth, branching, pruning, and stabilization, with significant changes occurring through P21. Thus, in addition to regulating synapse maturation in adult mice as recently suggested (4), TRPC4 likely also contributes to the early development of neurons before synapse formation, particularly in the hippocampus. As such, deregulation of neurodevelopment may also account, at least partially, for autism and developmental delays found in patients carrying TRPC4 mutations (6) and in rodents with Trpc4 gene ablation (3–5, 18).

Here, we examined neurological behaviors and neuronal morphogenesis in young (P21–P32) TRPC4 knockout (Trpc4^−/−) mice. We found that in addition to sociability deficits, the mutant juvenile mice also display defects in other behaviors, including nesting, marble burying, burrowing, and self-grooming, indicative of a general delay in neurodevelopment. Given the importance of the limbic system in regulating these behaviors and the pronounced hippocampal expression of TRPC4 during early development, we interrogated the contribution of TRPC4 in dendrite morphogenesis. We found that TRPC4 deficiency results in marked reductions of dendritic arborization of hippocampal neurons both in vivo and in vitro. Mechanistically, TRPC4 acts by mediating Ca²⁺ entry into dendrites downstream from metabotropic glutamate receptors (mGluRs), which in turn activates integrins in a manner dependent on phosphorylation of nonmuscle myosin light chain (MLC) through myosin light chain kinase (MLCK). This pathway is pivotal in stabilizing nascent dendrite branches during dendritogenesis. These findings enhance our understanding of glutamate and Ca²⁺-dependent pathways in dendritic development, opening broad perspectives on TRPC channel regulation of neurological behaviors in early development.

Results

Trpc4^−/− Mice Display Alterations in Sociability and Other Neurobehaviors During Early Postnatal Days.

Adult Trpc4^−/− mice were reported to exhibit reduced anxiety and impaired sociability (3, 4). To explore whether TRPC4 deficiency affects animals at early developmental stages, we conducted the three-chamber social interaction test on young mice (P26). Trpc4^−/− mice displayed reduced sociability, showing less preference to check on an unfamiliar mouse than wild-type (WT) mice (Fig. 1 A and B). This suggests an early emergence of social interaction deficits in Trpc4^−/− mice.

Fig. 1.

Behavioral alterations in juvenile Trpc4^−/− mice. (A and B) Social interactions of Trpc4^+/+ and Trpc4^−/− mice (P26, males) during three-chamber test, showing time spent sniffing Stranger (S) or Empty (E) cage (A) and preference index of social approach (B) calculated using the formula (S−E)/(S + E), where S and E stand for values from (A). Shown are data points of individual animals for n = 11 Trpc4^+/+ and n = 15 Trpc4^−/− mice. Statistics: two-way ANOVA with Tukey’s multiple comparisons test (A); unpaired two-tailed t test (B). (C) Experimental timeline for assessing nest building ability of group-housed mice. (D) Daily nesting scores of group-housed mice. The test started on P21 with photos taken for scoring 24 h later on P22. First cage change occurred on P24 and then every day afterward. Statistics: P22-P24, two-way ANOVA; P25-P27, two-way ANOVA with Šídák’s multiple comparisons test. (E) Summary of nestlet shredded for the same groups of mice in (D) based on weight of shredded portions. n = 6 Trpc4^+/+, n = 5 Trpc4^−/−. Statistics: two-way ANOVA with Šídák’s multiple comparisons test. (F) Daily nesting scores over 10 d after weaning for individually housed mice. A one-time cage change was made on P26. n = 10 Trpc4^+/+, n = 8 Trpc4^−/−. Statistics: two-way ANOVA with Šídák’s multiple comparisons test, performed separately for P22-P26 and P27-P31. (G) Experimental timeline for assessing nest building ability of individually housed mice and their marble burying, burrowing, and self-grooming behaviors. (H) Representative photos of nestlets and surrounding areas on indicated postnatal days of individually housed Trpc4^+/+ and Trpc4^−/− mice with daily cage changes. (I) Summary data of nestlet shredded based on weight for individually housed mice with daily cage change. Statistics: two-way ANOVA with Šídák’s multiple comparisons test. (J) Summary of the marble burying test on P28 (number of marbles buried in 30 min). (K) Summary of the burrowing test on P30 (food pellets burrowed over a 12-h period). (L) Summary of the self-grooming test on P32 (time spent on self-grooming for 10-min period). Statistics: J unpaired, two-tailed Mann–Whitney U test; K and L, unpaired, two-tailed t test. For J–L, n = 15 Trpc4^+/+, n = 14 Trpc4^−/−. All summary data represent means ± SEM. ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. Trpc4^−/− of the corresponding date or indicated by brackets.

We next examined nesting behaviors by comparing nesting scores (quality of constructed nests) and nestlet shredding (the amount of shredded material) between WT and Trpc4^−/− mice. In group-housed animals (3 to 4 per group, male), we first scored nest building for 3 d after weaning (P22–P24) without a cage change and then evaluated nesting scores and nestlet shredding for 3 consecutive days (P25–P27) with daily cage changes (Fig. 1C). We noticed a continued improvement of nesting scores in both WT and Trpc4^−/− mice during the first three days, although the mutant mice performed poorly compared to WT controls (Fig. 1D and SI Appendix, Fig. S1A; two-way ANOVA for the first 3 d: P < 0.0001 for postnatal days; P < 0.0063 for genotype). After the cage change, however, WT mice maintained high nesting scores and shredded nestlet extensively throughout the three days with daily cage changes, while Trpc4^−/− mice showed persistently low nesting scores and minimal nestlet shredding (Fig. 1 D and E and SI Appendix, Fig. S1A).

To assess whether the observed deficits were influenced by social dynamics, nesting behaviors were also examined under individual housing conditions. During the first 5 undisturbed days (P22–P26), Trpc4^−/− mice exhibited a delay in nest building compared to WT controls (Fig. 1F and SI Appendix, Fig. S1B; two-way ANOVA: P < 0.0001 for postnatal days; P < 0.0001 for genotype). Following the cage change on P26, the nesting score of WT mice displayed a small, insignificant decrease, while that of Trpc4^−/− dropped to the baseline level and took >2 d to return to the precage change level (Fig. 1F; P26 vs. P27, P = 0.0557 for Trpc4^+/+, P < 0.0093 for Trpc4^−/−, paired t test). These indicate that Trpc4^−/− mice are unable to maintain their acquired skills in nest building and they may also be more sensitive to environmental disturbance than WT mice.

Comparing group-housed and individually housed animals, we detected no difference in nesting scores for WT mice during the first three days after weaning. However, for Trpc4^−/− mice, group housing improved the overall nesting scores (SI Appendix, Fig. S1C; two-way ANOVA: P < 0.0369 for postnatal days; P < 0.0001 for housing style). This suggests that nesting impairment may be partially alleviated through social reinforcement, a communal aspect of nesting behavior (19). Alternatively, by providing food pellets in bedding material, we also observed markedly improved nesting behaviors in Trpc4^−/− mice (SI Appendix, Fig. S1 D and E; two-way ANOVA for the first 5 d: P < 0.0001 for postnatal days, P = 0.0001 for food vs. no food), suggesting that the nesting deficit may result from a lack of motivation and can be partially ameliorated by introducing a nonsocial stimulus (20). Nonetheless, after the cage change, the nesting scores of Trpc4^−/− mice dropped to the initial level again despite the presence of food pellets. Thus, neither social nor nonsocial reinforcement can fully correct the motivational and behavioral deficits caused by TRPC4 deficiency.

To further evaluate neurobehavioral abnormalities in Trpc4^−/− mice, we made daily cage changes with a new nestlet provided each time in individually housed animals from P21 to P29. On P28, P30, and P32, we also tested marble burying, burrowing, and self-grooming behaviors, respectively (Fig. 1 G–L). Whereas WT mice exhibited a gradual increase in their ability to shred nestlets and make nests from P21 to P29, Trpc4^−/− mice not only performed poorly on P21 but also showed no improvement over the next eight days (Fig. 1 H and I; two-way ANOVA: P < 0.0001 for postnatal days; P < 0.0001 for genotype).

Marble-burying tests showed that Trpc4^−/− mice buried fewer marbles than their WT counterparts (Fig. 1J and SI Appendix, Fig. S1F; WT: 7.93 ± 0.97 vs. Trpc4^−/−: 5.14 ± 1.48, P = 0.035), in line with reduced anxiety in mutant animals (3). Noticeably, two of the 14 Trpc4^−/− mice tested displayed higher marble-burying than WT mice (Fig. 1J), which may reflect behavioral variability associated with potential obsessive-compulsive or repetitive behaviors (21). In the burrowing test, Trpc4^−/− mice displaced much less food pellets from the burrowing tube than WT mice (22) (Fig. 1K and SI Appendix, Fig. S1G), reflecting deficits in behaviors associated with tunnel construction and maintenance, potentially influenced by hippocampal functions (23). Furthermore, Trpc4^−/− mice engaged in self-grooming for a longer time than WT mice (Fig. 1L), implying a possible increase in repetitive behaviors (24). Together, these findings reveal substantial behavioral alterations in young Trpc4^−/− mice, highlighting the role of TRPC4 in the development of daily living, social interaction, and survival-related behaviors. The deficits observed likely arise from a combination of motivational impairments, heightened sensitivity to environmental changes, and delays in task execution or learning and behavioral adaptation during the early adolescence period.

Trpc4^−/− Hippocampal Neurons Exhibit Reduced Dendritic Arborization.

Numerous studies have established that integration and processing of electrical information within neural circuits depend on the proper development of dendritic arbors, which starts from prenatal and continues during postnatal period (9, 10, 25). TRPC4 expression is readily detected in the mouse hippocampus during early development (SI Appendix, Fig. S2 A–D). Given that in ASD and other neurodevelopmental disorders, behavioral abnormalities are commonly associated with dendritogenesis deficits (10), we examined how Trpc4 loss affects dendritic arborization in hippocampal neurons.

In Golgi-Cox stained brain samples, hippocampal CA1 pyramidal neurons from Trpc4^−/− mice exhibited fewer apical dendrite branches compared to that from WT mice on P14, P21, and P26 (Fig. 2A). Quantification of number of dendritic branch points and total length showed significant reductions in Trpc4^−/− compared to WT mice (Fig. 2 B and C), suggesting an important role of TRPC4 in dendritic arbor development or its maintenance during early postnatal period in vivo.

Fig. 2.

Impaired dendritic arborization of developing hippocampal neurons in Trpc4^−/− mice. (A) Representative 3D reconstructions (Left) and confocal images (Right) of Golgi-stained CA1 pyramidal neurons of Trpc4^+/+ and Trpc4^−/− mice killed on P14, P21, and P26. In reconstructed images: blue, soma; red, apical dendrites; green, branch points. (Scale bar, 50 µm.) (B and C) Quantification of number of branch points (B) and total apical dendrite length (C). n = 11-15 neurons pooled from 3 – 4 Trpc4^+/+ and Trpc4^−/− mice in each group. Data points of individual neurons and means ± SEM are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by the unpaired t test. (D and E) Representative confocal images (Top) and 3D reconstructions (Bottom) of Map2-stained primary cultured hippocampal neurons from Trpc4^+/+ and Trpc4^−/− mice on days in vitro (DIV) as indicated. Reconstructed images: blue, soma; red, dendrites; green, branch points. (Scale bar, 50 µm.) (F and G) Quantification of number of branch points (F) and total dendrite length (G). Shown are data points of individual neurons (n = 40 to 66) pooled from 3 independent experiments for each DIV. *P < 0.05, ***P < 0.001, ****P < 0.0001, by two-way ANOVA with Šídák’s multiple comparisons test.

Next, we used primary hippocampal neuron cultures, in which dendritic arborization predominantly occurs from days in vitro (DIV) 5 to DIV10, with rapid development stages being around DIV7 and DIV9 (26). In low-density cultures, we visualized dendrites using Map2 staining and confocal microscopy (Fig. 2D), followed by dendritic reconstruction using NeuronStudio and Shutu software (Fig. 2E). Quantification of number of dendritic branch points and total dendritic length (27) revealed significant decreases in Trpc4^−/− compared to WT neurons on all DIVs (DIV7, 10, 14, 16) tested (Fig. 2 F and G). Thus, when cultured in vitro with glial cells largely removed (28), TRPC4 deficiency also impairs dendrite development, supporting a crucial role of TRPC4 in dendritic arborization of mouse hippocampal neurons during the early postnatal period.

The difference in dendritic arborization between WT and Trpc4^−/− neurons was readily detected by using Sholl analysis, which measures the number of intersections concentric circles drawn from the soma at increasing radii with the dendrite branches (SI Appendix, Fig. S2 E and G). However, despite that TRPC1 and TRPC4 often form heteromeric channels, cultured Trpc1^−/− hippocampal neurons displayed similar levels of dendritic arborization as the corresponding WT controls (SI Appendix, Fig. S2F). Moreover, while WT neurons showed increased arborization in response to brain-derived neurotrophic factor (BDNF, 100 ng/ml, 48 h), this effect was missing in Trpc4^−/− neurons and rescued by expression of TRPC4β-GFP in Trpc4^−/− neurons (SI Appendix, Fig. S2 G–J). The latter effect indicates that TRPC4 regulates dendritic arborization cell autonomously.

TRPC4 Responds to Glutamate to Stabilize Nascent Dendrite Branches.

Growing neurons secrete both BDNF and glutamate (29). While BDNF is well established to promote dendritic growth, glutamate exhibits both stimulatory and inhibitory effects on dendritic growth depending on its concentrations and culture conditions (30–32). Most studies on the role of glutamate in dendritic growth have focused on AMPA and NMDA receptors (30). However, glutamate also activates metabotropic glutamate receptors (mGluRs). In this context, several studies have demonstrated the link between glutamatergic signal and TRPC4 channel activation through mGluRs in neurons (2, 3). Thus, we tested whether glutamate serves as an upstream signal for TRPC4 regulation of dendritic arborization.

We applied a low concentration of glutamate (Glu, 5 µM, Na⁺ salt) to cultured hippocampal neurons on either DIV5 or DIV6. The neurons were fixed after 24 h and stained with anti-Map2. Interestingly, Sholl analysis revealed that Glu application on DIV5 WT neurons did not alter dendritic arborization compared to the basal increase observed from DIV5 to DIV6 (Fig. 3A and SI Appendix, Fig. S3A; two-way ANOVA: P = 0.0053 for DIV5 vs. DIV6, P = 0.3091 for DIV6 vs. DIV6+Glu). However, Glu application on DIV6 resulted in a significant increase in dendritic arborization when quantified on DIV7 (Fig. 3B and SI Appendix, Fig. S3B; two-way ANOVA: P = 0.1063 for DIV6 vs. DIV7; P < 0.0001 for DIV7 vs. DIV7+Glu). Glu failed to alter dendritic arborization of Trpc4^−/− neurons no matter if it was applied on DIV5 or DIV6 (Fig. 3 C and D and SI Appendix, Fig. S3 C and D; two-way ANOVA: P = 0.8801 for DIV6 vs. DIV6+Glu; P = 0.3003 for DIV7 vs. DIV7+Glu). These results indicate a pivotal role of TRPC4 in Glu-induced dendritic arborization, which is dependent on the stage of dendrite development.

Fig. 3.

Contribution of TRPC4 to dendritic arborization in response to glutamate during a specific time period of in vitro culture. (A and B) Sholl analysis of Map2-stained dendrite branches of primary cultured low-density Trpc4^+/+ neurons on different DIVs as shown. Na⁺-glutamate (Glu, 5 µM) was added 24 h before fixation. Note while adding Glu on DIV5 did not affect dendritic arborization measured on DIV6 (A), adding it on DIV6 increased dendritic arborization measured on DIV7 (B). Neurons for DIV5 (n = 37), DIV6 (n = 39), DIV6 with Glu treatment (n = 30) and DIV6 (n = 64), DIV7 (n = 66), DIV7 with Glu treatment (n = 47) were pooled from 3 independent experiments. Data represent means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, for DIV5 vs. DIV6 (A) and for DIV7 vs. DIV7 with Glu-treatment (B) of the corresponding distance, by two-way ANOVA with Tukey’s multiple comparisons test. (C and D) Similar experiments as in (A and B) for Trpc4^−/− neurons. Neurons for DIV5 (n = 60), DIV6 (n = 63), DIV6 with Glu treatment (n = 50) and DIV6 (n = 56), DIV7 (n = 49), DIV7 with Glu treatment (n = 42) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, for DIV5 vs. DIV6 (C) and DIV7 vs. DIV7+Glu (D) of the corresponding distance, by two-way ANOVA with Tukey’s multiple comparisons test. (E) Representative N-SIM images of Trpc4^+/+ (Top) and Trpc4^−/− (Bottom) neurons on DIV7. Neurons were fixed on DIV7 and stained for TRPC4 (green), Map2 (red), and F-actin (blue, with phalloidin). (Scale bar, 5 µm.) (F) Summary of TRPC4 density in Map2-positive dendritic regions for neurons on DIV5, DIV6, and DIV7. The spot analysis function of Imaris software was used to identify TRPC4-positive spots within the contours of dendrites drawn according to Map2 staining. Phalloidin-positive but Map2-negative areas were considered axons and not counted. Values for individual neurons and means ± SEM (n = 22 to 25) are shown. ns, not significant, ****P < 0.0001, by two-way ANOVA with Tukey’s multiple comparisons test. (G) Schematic diagram of EGFP-CAAX construct with the 0.4 kb CaMKII promoter. (H) Representative confocal images of live cell recording of Trpc4^+/+ and Trpc4^−/− neurons that expressed EGFP-CAAX and treated with Glu (5 µM). Boxed areas are shown below the main images for time-lapse images at the indicated time points after Glu addition. (Scale bar, 20 µm.) White and red arrowheads indicate extending and retracting dendrite branches, respectively. (I) Time courses of mean branch length extended over 10 h after Glu addition for Trpc4^+/+ and Trpc4^−/− neurons, as well as for Trpc4^−/− neurons that had overexpression of mouse TRPC4β (C4OE). Note the rescue effect of C4OE. Data represent means ± SEM of n = 9 to 10 neurons pooled from 3 independent experiments. ****P < 0.0001, for comparison between curves indicated by brackets, by two-way ANOVA. (J) Representative growth patterns of dendrite branches, representing the four major categories seen in live cell recording after Glu addition. (K) Pie charts showing distributions of the four major growth patterns in Trpc4^+/+ (Top) and Trpc4^−/− (Bottom) neurons in response to Glu, for n = 37 branches pooled from Trpc4^+/+ neurons (n = 10) and n = 41 branches pooled from Trpc4^−/− neurons (n = 9). Note the predominance of retracted, including both newly formed and preexisting, dendrite branches in Trpc4^−/− neurons.

Intriguingly, immunofluorescent staining revealed that TRPC4 protein levels in WT neuron dendrites markedly increased between DIV6 and DIV7 (Fig. 3 E and F and SI Appendix, Fig. S3 E, F; two-way ANOVA: P < 0.0001 for DIV, P < 0.0001 for genotype). This suggests that TRPC4 expression in dendrites is augmented at a specific stage of dendrite development (i.e., DIV6-7), which coincides with the onset of Glu-induced dendritic arborization.

During dendrite morphogenesis, new or nascent dendritic branches are formed by interstitial sprouting from existing shaft (primary) or side branches. These newly formed branches undergo continuous remodeling, including extension and retraction, until a subset of them become stabilized (33). To investigate how TRPC4 affects dendrite branching, we monitored dendritic growth in live hippocampal neurons that expressed farnesylated GFP (GFP-CAAX) driven by the CaMKII promoter (0.4 kb CaMKIIα) (Fig. 3G). This probe allows for the observation of detailed membrane structures, including growth cones (34) (Fig. 3H). Using time-lapse imaging with the neuron culture placed in an environmental chamber (37 °C, 5% CO₂) mounted on the stage of a confocal microscope, we monitored dynamic dendritic growth patterns over a period of 10 h. After adding Glu on DIV6, many new branches formed in both WT and Trpc4^−/− neurons (Fig. 3H and Movies S1 and S2); however, this growth period was short-lived in Trpc4^−/− neurons as compared to WT. On average, while the growth of WT neuron branches did not begin to level off until 6 h after Glu application, that of Trpc4^−/− branches stopped extension after 2 h (Fig. 3I). Importantly, expression of TRPC4β-GFP (SI Appendix, Fig. S3G) in Trpc4^−/− neurons restored the growth to the levels seen in WT neurons (Fig. 3I and SI Appendix, Fig. S3H).

Previously, the growth behaviors of nascent and preexisting dendritic branches have been categorized into at least four general patterns: 1) extended growth, 2) maintained growth after extension (stop growing after a certain period of extension), 3) repetitive growth and retraction (oscillation), and 4) retraction (35) (Fig. 3J). Applying this classification method, we found that Trpc4^−/− neurons had a greater proportion of retracting branches than WT neurons (Fig. 3K), indicating that TRPC4 may play a role in stabilizing newly formed dendrite branches.

TRPC4 Mediates Ca²⁺ Influx in Hippocampal Dendrites via mGluRs and G_i/o Signaling.

To evaluate TRPC4 function in dendrites, we expressed membrane-targeted Ca²⁺ indicator, GCaMP6s-CAAX in hippocampal neurons. To prevent cytosolic Ca²⁺ rise evoked by action potentials and ionotropic glutamate receptors (iGluRs), we included D-AP5 (50 μM), CNQX (25 μM), and tetrodotoxin (TTX, 1.5 μM) in all solutions (36, 37). Application of Glu (30 μM) caused robust fluorescence increases in dendrites in both WT and Trpc4^−/− neurons (DIV6.5), which reached the peak within 5 s and then declined (SI Appendix, Fig. S4 A and B). Similar patterns were observed in the shaft and side dendrite branches (SI Appendix, Fig. S4 B and C).

Quantification of fluorescence changes at the peak Ca²⁺ response revealed no difference between WT and Trpc4^−/− neurons at shaft dendrites, while a smaller peak in Trpc4^−/− than WT neurons at side branches (SI Appendix, Fig. S4D; two-way ANOVA: p = 0.0161 for genotype, p < 0.0001 for branch type). At 60 s after the peak, Trpc4^−/− neurons showed significantly smaller Ca²⁺ responses in both shaft and side branches than WT controls (SI Appendix, Fig. S4E; two-way ANOVA: p < 0.0001 for genotype, P = 0.0229 for branch type). These findings highlight the critical role of TRPC4 in maintaining sustained near membrane Ca²⁺ elevation in response to Glu in dendrites.

Furthermore, the sustained Ca²⁺ response (at 60 s) was inhibited by the TRPC4/5 antagonist ML204 in WT neurons (SI Appendix, Fig. S4 B, C, F, and H; one-way ANOVA, P = 0.0004 for F, P = 0.0007 for H, Tukey’s post hoc test, ML204 vs. Veh, P = 0.0202 for F, P = 0.0134 for H). Additionally, the Glu-induced sustained Ca²⁺ response was attenuated by group I mGluR blockers (YM298198 plus MPEP) and pertussis toxin (PTX) (SI Appendix, Fig. S4 F, H; Tukey’s post hoc test, YM298198+MPEP vs. Veh, P = 0.0051 for F, P = 0.0081 for H; PTX vs. Veh, P = 0.0075 for F, P = 0.0345 for H). Given that group I mGluRs couple to G_q/11 and PTX suppresses G_i/o, this finding is consistent with the notion that TRPC4 is dependent on coincident activation of both G_q/11 and G_i/o signaling (2, 38). By contrast, none of these drugs significantly altered the sustained phase of Glu-induced Ca²⁺ rise in Trpc4^−/− neuron dendrites (SI Appendix, Fig. S4 G and I; one-way ANOVA, P = 0.1983 for G, P = 0.2701 for I).

TRPC4 Promotes Integrin Activation and Myosin Phosphorylation in Dendrites.

To link TRPC4-mediated Ca²⁺ influx to its role in stabilizing dendrite branches, we examined integrins. Integrins are essential for a range of cellular functions, including regulation of neural morphology and stabilization (39), and their activation is dependent on coupling to forces generated by actin–myosin interaction (40–43). While the contribution of TRPC4 through actomyosin force to smooth muscle contraction and endothelial permeabilization has been proposed (44), the functional significance of this regulation in neurons has not been explored. Particularly, the role of TRPC4 in activating integrins in neurons remains uncertain.

To determine whether Glu-induced dendritic arborization is dependent on integrin activation, we expressed the head domain of Talin1, which retains its ability to bind the cytoplasmic domain of integrin β1 but lacks the tail domain required for actin binding and force transduction needed for integrin activation (45, 46). In WT neurons, expression of the Talin head domain suppressed Glu-induced dendritic arborization (SI Appendix, Fig. S5 A and B), highlighting the critical role of integrin signaling and regulation by actomyosin force in Glu-induced dendritic growth.

Next, we tested the integrin peptide inhibitor RGDS. Addition of RGDS (5 µg/ml) 1 h prior to Glu (5 µM) strongly suppressed Glu-induced dendritic arborization in WT neurons, while also having a moderate effect in the absence of exogenously added Glu (Fig. 4 A–C; two-way ANOVA, P = 0.0109 for basal vs. RGDS; P < 0.0001 for Glu vs. Glu+RGDS). Interestingly, RGDS affected neither the basal nor Glu-induced dendritic arborization in Trpc4^−/− neurons (Fig. 4 A, D, and E), suggesting that integrin functions are compromised in the absence of TRPC4. Thus, integrin activation is required for Glu-induced dendritic arborization, and it is dependent on TRPC4.

Fig. 4.

TRPC4-dependent integrin activation and its role in Glu-induced dendritic arborization of primary hippocampal neurons. (A) Representative confocal images of Map2-stained WT neurons untreated or treated with 5 µM Glu on DIV6 and fixed on DIV7. RGDS (5 µg/ml) was added at 1 h prior to Glu application. (Scale bar, 50 µm.) (B) Sholl analysis of dendritic arborization of Trpc4^+/+ neurons under basal conditions in the absence and presence of RGDS. Neurons of vehicle- (Veh, n = 67) and RGDS-treated (n = 81) were pooled from 3 independent experiments. Data represent means ± SEM. **P < 0.01, ****P < 0.0001, vs. RGDS of the corresponding distance, by two-way ANOVA with Šídák’s multiple comparisons test. (C) Sholl analysis of dendritic arborization of Trpc4^+/+ neurons under Glu-stimulated conditions in the absence and presence of RGDS. Neurons of Glu-treated (n = 43) and RGDS+Glu-treated (n = 54) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. RGDS+Glu of the corresponding distance, by two-way ANOVA with Šídák’s multiple comparisons test. (D and E) Sholl analysis of dendritic arborization of Trpc4^−/− neurons under basal (D) and Glu-stimulated (E) conditions in the absence and presence of RGDS. Neurons of vehicle-treated (n = 43), RGDS-treated (n = 40), Glu-treated (n = 40), and RGDS+Glu-treated (n = 40) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant between the two conditions, by two-way ANOVA. (F and G) Representative N-SIM images of DIV7 Trpc4^+/+ (F) and Trpc4^−/− (G) neurons untreated (basal) or treated with 30 µM Glu for 2 min and stained for active β1 integrins with HUTS-4 under nonpermeabilized conditions (green). After permeabilization, the neurons were stained for Map2 (red) and F-actin (blue, by phalloidin). (Scale bar, 5 µm.) (H and I) Quantification of active integrins, based on the density of HUTS-4-positive spots in Map2-positive dendritic regions, in shaft (H) and side (I) dendrite branches of Trpc4^+/+ and Trpc4^−/− neurons. The spot analysis function of Imaris software was used to identify HUTS-4-positive spots within the contours of dendrites drawn according to Map2 staining. Trpc4^+/+ shaft branches for basal (n = 25) and Glu-treated (n = 23), Trpc4^−/− shaft branches for basal (n = 23) and Glu-treated (n = 20), Trpc4^+/+ side branches for basal (n = 212) and Glu-treated (n = 177), and Trpc4^−/− side branches for basal (n = 169) and Glu-treated (n = 138) were pooled from Trpc4^+/+ (n = 23-25) and Trpc4^−/− (n = 20) neurons of 3 independent experiments. For shaft branches, values for individual branches and means ± SEM are shown. For side branches, violin plots are used due to the large number of values. **P < 0.01, ****P < 0.0001, by two-way ANOVA with Tukey’s multiple comparisons test. (J and K) Schematics of integrin conformations showing bent-closed, extended-closed, and extended-open conformations. RGDS binds the extended-closed conformation, while the HUTS-4 antibody recognizes active β1 integrins with extended-open conformation (J), which are increased upon glutamate stimulation (K). The red arrow in (J) indicates the involvement of force from the cytoplasmic side that stabilizes the extended open conformation.

To directly examine the effect of Glu on integrin activation in dendrites, we used the HUTS-4 antibody, which detects the active, ligand-bound conformation of β1 integrins (47–49) (Fig. 4J). Within 2 min of Glu (30 µM) treatment, we observed significant increases in the density of active β1 integrin puncta in both shaft and side dendrite branches of WT neurons (Fig. 4 F, H, and I), which were not detected in Trpc4^−/− neurons (Fig. 4 G–I). In WT neurons, the increases were also blocked by RGDS (SI Appendix, Fig. S5 C–F). Noticeably, both the Trpc4 deletion and RGDS treatment also significantly reduced integrin activation in side, but not shaft, branches under basal conditions (Fig. 4 H and I and SI Appendix, Fig. S5 E and F). This is consistent with the view that endogenously released Glu tends to induce the formation of higher order, i.e., side, branches. These results suggest that Glu activates integrins in dendrite branches in a TRPC4-dependent manner.

Given that the HUTS-4 antibody can detect high-affinity integrin activation, which is stabilized by actomyosin force and ligand binding (43, 48, 50) (Fig. 4J), we suspected that myosin ATPase, crucial for cytoskeleton contractility, also plays a role in Glu-induced dendritic arborization. As expected, the myosin ATPase inhibitor, blebbistatin (Blebb, 30 µM) (51), prevented the increase in dendritic arborization induced by 5 µM Glu in WT, but not Trpc4^−/−, neurons (Fig. 5 A–E; WT, P = 0.9561 for basal vs. Blebb; WT, P = 0.0069 for Glu vs. Glu+Blebb; Trpc4^−/−, P = 0.0974 for basal vs. Blebb; Trpc4^−/−, P = 0.1737 for Glu vs. Glu+Blebb).

Fig. 5.

Involvement of myosin ATPase in Glu-induced dendritic arborization and TRPC4-dependent MLC phosphorylation in response to Glu in hippocampal neuron dendrites. (A) Representative confocal images of DIV7 Trpc4^+/+ (Top panels) and Trpc4^−/− (Bottom panels) neurons treated with vehicle or 5 µM Glu for 24 h without and with Blebbistatin (Blebb, 30 µM), added at 1 h prior to Glu application. (Scale bar, 50 µm.) (B and C) Sholl analysis of dendritic arborization of Trpc4^+/+ neurons under basal (B) and Glu-stimulated (C) conditions. Neurons for basal (n = 35), Blebb-treated (n = 37), Glu-treated (n = 35), and Glu+Blebb-treated (n = 44) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant, *P < 0.05, **P < 0.01, vs. Blebb (B) or Glu+Blebb (C) of the corresponding distance, by two-way ANOVA with Šídák’s multiple comparisons test. (D and E) Sholl analysis of dendritic arborization of Trpc4^−/− neurons under basal (D) and Glu-stimulated (E) conditions. Neurons for basal (n = 34), Blebb-treated (n = 41), Glu-treated (n = 54), and Glu+Blebb-treated (n = 50) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant between the two conditions, by two-way ANOVA. (F and G) Representative N-SIM images of DIV7 Trpc4^+/+ (F) and Trpc4^−/− (G) neurons untreated (Basal) or treated with 30 µM Glu for 1 and 2 min. After the treatment, neurons were stained for phospho-myosin light-chain 2 (Ser19, pMLC) under permeabilized conditions (green), followed by staining for Map2 (red) and F-actin (blue, by phalloidin). (Scale bar, 5 µm.) (H) Quantification of pMLC spot densities, based on pMLC-positive spots in Map2-stained dendritic regions, in shaft (Left) and side (Right) dendrite branches of Trpc4^+/+ and Trpc4^−/− neurons without (0 time point) or with Glu treatment for 1 and 2 min. Trpc4^+/+ shaft branches for control (n = 25), 1 min Glu-treated (n = 20), and 2 min Glu-treated (n = 20); Trpc4^−/− shaft branches for control (n = 20), 1 min Glu-treated (n = 22), and 2 min Glu-treated (n = 23); Trpc4^+/+side branches for control (n = 153), 1 min Glu-treated (n = 131), and 2 min Glu-treated (n = 130); Trpc4^−/− side branches for control (n = 142), 1 min Glu-treated (n = 148), and 2 min Glu-treated (n = 165) were pooled from 20 to 25 Trpc4^+/+ and Trpc4^−/− neurons of 3 independent experiments. Data represent means ± SEM. ns, not significant, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. Trpc4^−/− at the corresponding time point, or ^#P < 0.05, ^####P < 0.0001 indicated by brackets for Trpc4^+/+ branches, by two-way ANOVA with Šídák’s multiple comparisons test. (I) Schematic of myosin activation by MLC phosphorylation.

As TRPC4 is implicated in generating force or contraction in nonneuronal cells (52), we next tested the effect of Glu on nonmuscle myosin II using an antibody for phospho-(Ser-19) myosin-light chain II (pMLC). In WT neuron dendrites, the density of pMLC puncta increased within 1 min of Glu (30 µM) treatment, whereas in Trpc4^−/− dendrites, no increases were observed at 1 and 2 min following Glu addition (Fig. 5 F and G). Quantification of pMLC spot density revealed significant increases in both shaft and side dendrite branches of WT, but not Trpc4^−/−, neurons (Fig. 5H; two-way ANOVA, P < 0.0001 for treatment times, P < 0.0001 for genotypes between shaft branches; P < 0.0001 for treatment times, P < 0.0001 for genotypes between side branches). Thus, TRPC4 is critical for Glu-induced nonmuscle myosin II activation (Fig. 5I), which, through MLC phosphorylation, drives the mechanical force required for high-affinity integrin activation and subsequent dendritic stabilization/maturation.

MLCK Mediates Glu-Stimulated MLC Phosphorylation and Integrin Activation.

Nonmuscle myosin II is regulated by at least two kinases: Ca²⁺-dependent MLCK and Rho-associated protein kinase (ROCK) (53) (Fig. 6A). To determine the kinase involved, we first investigated the impact of ML7 (an MLCK inhibitor) and Y27632 (a ROCK inhibitor) (53) on Glu-induced dendritic arborization. While the basal dendritic arborization of WT neurons was affected by neither ML7 (10 µM) nor Y27232 (30 µM) (Fig. 6 B and C; two-way ANOVA: P = 0.3021 for treatment conditions), the Glu-induced arborization was strongly suppressed by ML7, but not Y27632 (Fig. 6 B and D; two-way ANOVA: P < 0.0001 for ML7, P = 0.1518 for Y27232). Therefore, MLCK, but not ROCK, activity is crucial for Glu-induced dendritic branching. This finding is surprising as ROCK is known to phosphorylate MLC to induce force in nonneuronal cells, and it also inhibits the phosphatase that dephosphorylates MLC (53); however, it appears that ROCK plays a limited role in dendritic arborization.

Fig. 6.

Involvement of MLCK in Glu-induced dendritic arborization and MLC phosphorylation in hippocampal neuron dendrites. (A) Schematic of kinases that phosphorylate MLC. (B) Representative confocal images of Map2-stained Trpc4^+/+ and Trpc4^−/− mouse hippocampal neurons untreated (Basal) or treated with 5 µM Glu on DIV6 and fixed 24 h later on DIV7. ML7 (10 µM) or Y27632 (30 µM) was added at 1 h prior to Glu application. (Scale bar, 50 µm.) (C and D) Sholl analysis of dendritic arborization of Trpc4^+/+ neurons under basal (C) and Glu-stimulated (D) conditions. Neurons for basal (n = 41), ML7-treated (n = 55), Y27632-treated (n = 46), Glu-treated (n = 31), Glu+ML7-treated (n = 31), and Glu+Y27632-treated (n = 34) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant, **P < 0.01, ***P < 0.001, ****P < 0.0001 for basal vs. ML7 (C) or Glu vs. Glu+ML7 (D) of the corresponding distance, by two-way ANOVA with Šídák’s multiple comparisons test. No significance was noted for basal vs. Y27632 (C) or Glu vs. Glu+Y27632 (D). (E and F) Sholl analysis of dendritic arborization of Trpc4^−/− neurons under basal (E) and Glu-stimulated (F) conditions. Neurons for basal (n = 31), ML7-treated (n = 31), Glu-treated (n = 40), and Glu+ML7-treated (n = 40) were pooled from 3 independent experiments. Data represent means ± SEM. ns, not significant between the two conditions, by two-way ANOVA. (G) Representative N-SIM images of DIV7 Trpc4^+/+ hippocampal neurons untreated or treated with 30 µM Glu for 1 and 2 min without or with ML7 (10 µM) added at 1 h prior to Glu application. After the treatment, neurons were stained for pMLC under permeabilized conditions (green), followed by staining for Map2 (red) and F-actin (blue, by phalloidin). (Scale bar, 5 µm.) (H) Quantification of p-MLC spot densities in shaft (Left) and side (Right) dendrite branches of Trpc4^+/+ hippocampal neurons untreated (0 time point) or treated with Glu. Shaft branches for basal (0 time, n = 18), 1 min Glu-treated (n = 14), 2 min Glu-treated (n = 21), ML7-treated (0 time, n = 15), 1 min Glu+ML7-treated (n = 17), and 2 min Glu+ML7-treated (n = 17); Side branches for basal (0 time, n = 124), 1 min Glu-treated (n = 81), 2 min Glu-treated (n = 87), ML7-treated (0 time, n = 123), 1 min Glu+ML7-treated (n = 113), and 2 min Glu+ML7-treated (n = 111) were pooled from n = 14-21 Trpc4^+/+ neurons from 3 independent experiments. Data represent means ± SEM. ns, not significant, *P < 0.05, ***P < 0.001, ****P < 0.0001, vs. ML7 at the corresponding time point, or ^#P < 0.05, ^##P < 0.01, ^####P < 0.0001 indicated by brackets for Veh-treated, by two-way ANOVA with Šídák’s multiple comparisons test.

Consistent with the involvement of TRPC4 in dendritic arborization, ML7 affected neither the basal nor Glu-induced dendritic arborization in Trpc4^−/− neurons (Fig. 6 B, E, and F). By immunofluorescence staining, we detected comparable levels of MLCK in hippocampal neuron dendrites of WT and Trpc4^−/− mice (SI Appendix, Fig. S6 A and B). Knocking down MLCK expression with shRNA, which reduced dendritic MLCK spot density by ~74.2% (SI Appendix, Fig. S6 C and D), effectively inhibited Glu-induced dendritic arborization of WT neurons (SI Appendix, Fig. S6 E–G). These findings indicate MLCK as the primary kinase that supports Glu-induced TRPC4-dependent dendritic arborization.

To directly evaluate whether MLCK mediates MLC phosphorylation, neurons were pretreated with ML7 (10 µM) for 1 h before Glu (30 µM) application, and pMLC densities were measured after 1 and 2 min. While ML7 blocked Glu-induced increases in pMLC density in both shaft and side branches of WT neurons (Fig. 6 G and H), it had no effect on pMLC density in Trpc4^−/− neurons (SI Appendix, Fig. S7 A and B), further confirming that TRPC4 is required for MLCK activation and subsequent MLC phosphorylation.

Next, we asked whether actomyosin activation, as indicated by the increased pMLC density, is critical for Glu-induced integrin activation in dendrites. By staining with HUTS-4, we found that both ML7 and Blebb abolished the Glu-induced increase in the density of activated β1 integrins in WT neuron dendrites (SI Appendix, Fig. S8 A and C). This indicates that MLCK, through activating actomyosin to generate force, plays a pivotal role in integrin activation in the developing dendrites. In contrast, neither ML7 nor Blebb affected activated β1 integrin density in Trpc4^−/− neurons (SI Appendix, Fig. S8 B and D). Moreover, shRNA knockdown of MLCK in WT neurons also prevented β1 integrin activation induced by Glu (SI Appendix, Fig. S9 A and B).

Together, these findings strongly argue for a TRPC4-dependent activation of MLCK in hippocampal neuron dendrites that promotes actomyosin activity to support integrin activation (SI Appendix, Fig. S9C), which in turn facilitates dendritic arbor formation in response to Glu. This pathway provides a plausible mechanistic link between TRPC4 and stabilization of dendrite branches, in which integrin-mediated adhesion serves as a key step.

Discussion

TRPC4 Supports Dendritic Arborization by Activating Integrins to Stabilize Dendrite Branches.

This study reveals a function of TRPC4 in stabilizing dendrite branches in developing neurons. Mechanistically, TRPC4 exerts this action primarily by promoting integrin activation. Not only did blocking integrins with RGDS abolish glutamate-induced dendritic arborization of mouse hippocampal neurons, but the density of active β1 integrins in dendrites markedly increased within 2 min of glutamate exposure (Fig. 4). More importantly, TRPC4 is required for both dendritic arborization and integrin activation induced by glutamate (Figs. 3 A–D and 4 F–I), establishing a pathway by which glutamate regulates dendritic arborization through TRPC4 and then integrins. Given the general role of integrins in cell adhesion and signaling, specifically β1 integrins in dendrite growth and stability (54, 55), this TRPC4-dependent integrin signaling pathway provides the most plausible explanation for the higher incidence of retracting dendrite branches seen in TRPC4 deficient neurons in live-cell imaging (Fig. 3 H–K).

During dendrite morphogenesis, nascent branches undergo extension and stabilization, and a failure to stabilize results in retraction (35). Previously, dendrite retraction was observed following β1-integrin inactivation in chick retinal ganglion cells (54). The ability of integrins to stabilize morphological changes has also been implicated in long-term potentiation of mature neurons, where they play critical roles in stabilizing new synaptic structures (56). Thus, a failure to stabilize likely accounts for the decreased overall dendritic lengths and branch points in hippocampal neurons of Trpc4^−/− mice both in vivo and in vitro (Fig. 2), which most likely also underlies the behavioral alterations of mutant animals (Fig. 1).

Integrin activation involves transitions from low-affinity bent-closed to intermediate-affinity extended-closed, and then high-affinity extended-open conformations (42, 43). Intracellular signaling shifts integrins between bent-closed and extended-open high-affinity conformations, where the latter is stabilized through coupling of the β-subunit cytoplasmic domain to actomyosin force (40–42). As a Ca²⁺-permeable channel, TRPC4 most likely exerts its action through conducting Ca²⁺ influx, which in turn activates MLCK, a well-known target of Ca²⁺-calmodulin, leading to phosphorylation of MLC of the nonmuscle myosin II complex. The resultant contractility then provides force to support integrin activation. Indeed, we demonstrate that in cultured mouse hippocampal neurons, β1 integrins are activated within 2 min of glutamate exposure in all dendrite branches (Fig. 4). This timing coincides with actomyosin activation in these structures (Fig. 5). Importantly, the activations of both integrins and actomyosin were inhibited by not only the MLCK blocker ML7, but also TRPC4 deficiency, demonstrating a tight mechanistic link between TRPC4 and MLCK function in developing dendrites.

Previously, TRPC-mediated Ca²⁺ signals have been linked to MLC phosphorylation mainly in endothelial cells, where these channels are thought to contribute to endothelium permeation, barrier disruption, or migration, and both MLCK and RhoA might be involved in mediating the effects (57). However, despite the known effect of actomyosin force on integrin activation, only a few studies have linked TRPC-induced MLC phosphorylation to integrin function. Sparse evidence exists for TRPC4 involvement in β1-integrin dependent alveolar epithelial repair (58). In platelets, TRPC6 may play a role in the activation of integrins αIIbβ3 for clot retraction or MLC phosphorylation (59). However, these studies did not address the causal relationship between TRPC-dependent actomyosin activation and integrin function. In a different context, cross-linking of β1-containing integrins with gangliosides activated TRPC5 in neurons (60). Whether reciprocal regulation between TRPCs and integrins can form a feed-forward loop to strengthen this pathway is an interesting question that warrants future investigation. Moreover, it was previously shown that TRPC4 overexpression in cultured WT neurons reduced dendritic complexity (61). This suggests that TRPC4 function in dendritic morphogenesis is finely tuned, with both excessive and deficient TRPC4 activities leading to aberrant dendritogenesis.

Glutamate Acts through mGluRs and then TRPC4 to Promote Dendritic Arborization.

Glutamate is commonly released in developing neurons, primarily from axonal boutons (62), which can be enhanced by BDNF (63). This may explain the loss of both BDNF- and Glu-induced dendritic arborization in TRPC4-deficient neurons. Intriguingly, glutamate exerts both stimulatory and inhibitory effects on dendritic outgrowth depending on the neuron’s age, time, and concentration of glutamate used, and stiffness of the culture substrate (31, 32), and NMDARs are implicated in these effects (30). More significantly, during dendrite morphogenesis, spontaneous glutamate release from axonal boutons provides a long-range signal (up to 20 µm) to shape dendritic arbor formation through activation of NMDARs (62). Our observations are not in conflict with these studies but rather enrich the knowledge on glutamate regulation by showing a previously unrecognized role of TRPC4 in stabilizing either nascent or preexisting dendrite branches through mGluRs.

Plausibly, this mechanism is important not just at the time of dendritogenesis prior to synapse formation, but during synaptogenesis as well, as evidence exists for mGluRs to support both dendritic arborization and synapse maturation (64). Moreover, impaired spine maturation represents the most obvious morphological alteration of dentate gyrus granule neurons in adult Trpc4^−/− mice (4). Considering the role of integrins in stabilizing new synaptic structures of mature neurons during long-term potentiation (56), the molecular mechanism illustrated here may be more general, representing a pivotal step of neural morphological changes during development, remodeling, and plasticity, where new structures need stabilization in order to function. Thus, the glutamate-mGluR-TRPC4-integrin pathway joins the frequently studied semaphorins-integrin systems (55) for integrin activation and function in neurons, in line with integrin signaling being subject to regulation by multiple mechanisms, allowing enhanced dynamics and versatility.

The exogenously applied glutamate enhanced integrin activation in both shaft and side dendrite branches (Fig. 4 F, H, and I), indicating that the mGluR-TRPC4-integrin machinery is present throughout all dendritic areas. Under basal conditions, RGDS only inhibited integrin activation in side branches (SI Appendix, Fig. S5 C–F), suggesting that the spatiotemporal control of this machinery may rely on endogenous cues. Consistent with the notion that spontaneously released glutamate can act as a long-range local branching or guidance cue in developing dendrites (62), TRPC4 deficiency primarily impaired arborization in distal dendritic regions under basal conditions (SI Appendix, Fig. S2E).

While the initiation and extension of the new branch activated by glutamate may require NMDARs, its stabilization likely involves the mGluR-TRPC4-integrin pathway. This local signaling mechanism differs from the global one reported in adult neurons, in which the control of synaptic maturation involved binding of miR-138-2 by matrin-3 in the nucleus (4). Moreover, although TRPC5 and TRPC6 have also been shown to regulate dendritic arborization, the suggested mechanisms were also global: While TRPC5 acted through CaMKIIβ to free the E3 ubiquitin ligase CDC20-APC from the centrosome (13), TRPC6 signaled via CREB to regulate transcription of genes involved in dendritogenesis (8). Given that local Ca²⁺ signaling critically determines detailed branching patterns of dendritic arbors (65), it would be interesting to know whether, in addition to acting in the soma to produce global effects, these channels also exert local actions like TRPC4. Presumably, the ability to activate integrins can be shared by many Ca²⁺-permeable channels, but the specificity lies in signaling mechanisms that lead to their activation and their coexistence in time and space with integrin complexes. In this respect, it is intriguing that the glutamate-induced increase in arborization was not detected in cultured hippocampal neurons until the dendritic distribution of TRPC4 protein underwent a sharp increase between DIV6 and DIV7 (Fig. 3F).

TRPC4 Deficiency Causes Neurodevelopmental Disorders.

TRPC4 has recently been reported as an autism gene based on exome sequencing of patient families (6). Consistently, Trpc4 gene deletion caused sociability deficits in adult rats and adult mice (4, 18). Here, we extend these findings by demonstrating multiple behavioral alterations in juvenile Trpc4^−/− mice (Fig. 1), indicative of early onset of neurodevelopmental disorders. Since alterations in dendritogenesis are strongly implicated in neurodevelopmental disorders, including ASD (9, 10), the observed impairments in dendritic arborization likely underlie the behavioral deficits exhibited by young Trpc4^−/− mice. Yet not all behavioral changes can be accounted for by dendritogenesis deficits of hippocampal neurons. Future studies should examine whether similar deficits exist in other brain areas with significant TRPC4 expression and upon targeting TRPC4 in a specific brain region, which behavioral changes emerge.

Gene disruptions of TRPC5 and TRPC6, respectively, were also found in patients with X-linked intellectual disability with ASD and ASD (7, 8), suggesting that the three TRPC subtypes may share functions in neurodevelopment. Indeed, both TRPC5 and TRPC6 have previously been implicated in neurite and dendritic growth and patterning (8, 12). More importantly, induced pluripotent stem cell–derived neurons from the ASD patient bearing TRPC6 mutation displayed reduced dendritic arborization and spine density (8). It would be interesting to know whether the dendritic arborization defect was also due to an impairment to stabilize nascent branches, as has been reported for other ASD-linked mutations, such as Ube3A/E6AP and GluN2B (66, 67). If so, then a failure to stabilize newly formed dendrite branches could be a main underlying cause of ASD.

Our data suggest that TRPC4 functions in the limbic system, including the hippocampus, during the early postnatal period to influence behaviors crucial for survival and adaptability. Notably, TRPC4 deficiency led to a decrease in motivation-related behaviors, such as nesting and burrowing, emphasizing the importance of TRPC4-mediated pathways in normal behavioral responses. Environmental and social reinforcements, such as food pellets and group housing (20), did not fully alleviate the diminished nesting behavior in the mutant mice, indicating a defect in adaptability. Moreover, the decreased sociability and increased repetitive behaviors in Trpc4^−/− mice offer insights into the potential role of TRPC4 in social interactions and the potential link to ASD. Notably, the early onset of these behavioral phenotypes in Trpc4^−/− mice argues for the importance of TRPC4 in neurodevelopment from an early postnatal stage. The crucial role of TRPC4 in both dendritic and behavioral modulations opens broad avenues for therapeutic interventions. Targeting the TRPC4-mediated pathways may be a promising strategy for ameliorating symptoms associated with ASD and other neurological diseases.

In conclusion, this study establishes behavioral manifestations of TRPC4 during the postnatal stage and the regulatory role of TRPC4 in dendritogenesis through modulation of integrin activation and stabilization of dendrite branches. Mechanistically, this involves TRPC4 activation downstream from mGluRs to generate Ca²⁺ signals that lead to MLC phosphorylation by MLCK, supporting integrin activation (SI Appendix, Fig. S9C). These findings advance our understanding of the role of the TRPC4-Ca²⁺-mediated regulatory mechanism underlying dendritic arborization and will shed light on approaches to treat neurodevelopmental disorders.

Materials and Methods

WT and Trpc4^−/− mice in C57BL/6 background were obtained and housed as described (2). Male juvenile mice (P21–P32) were used for behavioral tests. The FD Rapid GolgiStain kit (NeuroTechnologies) was used for Golgi-Cox staining of whole brains. Primary hippocampal neurons were cultured from newborn (P0) mouse pups and transfected with desired cDNA or shRNA on DIV4-DIV6 using Lipofectamine 2000. Immunofluorescence staining was performed under either nonpermeabilized or permeabilized conditions based on experimental needs. Fluorescence imaging was performed using either confocal microscopy or superresolution microscopy (N-SIM). GraphPad Prism 10 Software was used for statistical analyses and graph preparation. Details on materials and experimental procedures are included in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Representative time-lapse video for confocal live cell recording of Trpc4^+/+ hippocampal neurons that expressed EGFP-CAAX and treated with Glu (5 μM). The video, beginning immediately after the addition of Glu, shows progressive dendritic branch growth, including extension, stabilization, and retraction for 10 hours. (Scale bar, 20 μm.) Experimental conditions were identical to those described in Fig. 3 H–K.

Movie S2.

Representative time-lapse video for confocal live cell recording of Trpc4^-/- hippocampal neurons that expressed EGFP-CAAX and treated with Glu (5 μM). The video, beginning immediately after the addition of Glu, shows dendritic branch growth and retraction for 10 hours, with a noticeably shorter growth period than Trpc4^+/+ neurons. (Scale bar, 20 μm.) Experimental conditions were identical to those described in Fig. 3 H–K.

Data, Materials, and Software Availability

All data necessary for the conclusions of the study are available in Figs. 1–6 and SI Appendix, Figs. S1–S9. All study data are included in the article and/or supporting information.