2. Rho Family Activity Is Regulated by GTPase-Activating Proteins (GAPs) and Guanine Nucleotide Exchange Factors (GEFs)
RhoA belongs to the RAS homolog (Rho) family, along with cell division control protein 42 (Cdc42) and RAS-related C3 botulinum toxin substrate 1 (Rac1)
[32]. Rho family proteins contain a conserved GDP/GTP binding domain and switch their activity by cycling between GDP-bound (inactive) and GTP-bound (active) states
[32][33]. This cycling is regulated by GAPs, GEFs, and guanine nucleotide dissociation inhibitors (GDIs)
[32][34]. GAPs consist of more than 70 members, and conversion from a GTP-bound form to a GDP-bound form suppresses their activity
[32][35]. In contrast, GEFs (>74 members) accelerate the exchange of tightly bound GDP for GTP, resulting in the activation of Rho family proteins
[32][35]. GAPs and GEFs exhibit high selectivity for RhoA, Cdc42, and Rac1
[35][36]. GDIs, of which there are only three members in the human genome, form soluble complexes with GDP-bound Rho protein and control its cycling between the cytosol and membrane
[32][34].
3. Rho Family Protein Effectors and Their Physiological Roles in the Brain
Rho family proteins are associated with over 70 potential effector proteins
[37]. Rho-kinase, a serine/threonine kinase, is a representative downstream effector of RhoA
[38]. In vascular smooth muscle, for example, Rho-kinase phosphorylates myosin phosphatase-targeting subunit 1 (MYPT1) at Thr696 and Thr853. This converts MYPT1 to an inactivated state, increases the phosphorylation of myosin light chain, and promotes actomyosin contractility
[39][40][41]. P21-activated kinase (PAK) acts as a downstream effector for Cdc42 and Rac1 and affects actin dynamics by regulating the LIM kinase–cofilin pathway
[42][43]. PAK also inhibits myosin light chain kinase, resulting in decreased myosin light chain phosphorylation and, thus, decreased actomyosin contractility
[44].
Rho GTPases regulate cell morphology. For instance, RhoA promotes stress fiber formation and focal adhesions in cells
[45]. Rho GTPases also modulate neuronal development. For instance, RhoA inhibits growth of dendrites and axons, while Rac1 and Cdc42 promote axonal elongation
[42][46]. In addition, RhoA/Rho-kinase signaling promotes spine shrinkage and destabilization, while Rac1 and Cdc42/PAK signaling promotes spine stabilization and maintenance
[23][47]. Accordingly, Rho GTPase signaling is involved in neuronal maturation through the regulation of actin dynamics.
4. Schizophrenia-Associated Genes Involved in Small GTPase RhoA Signaling
Recently, several variants of RhoA-associated GAPs, including
ARHGAP10 [48],
ARHGAP18 [49][50][51], and
p250GAP [52], and GEFs such as
KALRN [53][54][55][56] and
ARHGEF11 [57], were reported to be significantly associated with the development of schizophrenia. In addition, variants in genes that activate RhoA via GEF, such as
RTN4R [58][59][60], or in those that degrade RhoA, such as
KCTD13 [61], were also identified in schizophrenia (
Figure 1).
Figure 1. Schizophrenia-associated genes involved in small GTPase RhoA signaling. Genes shown in red are schizophrenia-associated genes involved in small GTPase RhoA signaling. RhoA contains a conserved GDP/GTP binding domain, and its activity cycles between GDP-bound (inactive) and GTP-bound (active) states. ARHGAP10, ARHGAP18, and p250GAP are GTPase-activating proteins (GAPs) that convert RhoA from the GTP- to GDP-bound form, thereby suppressing its activity. In contrast, Kalirin and ARHGEF11 are guanine nucleotide exchange factors (GEFs) that accelerate the exchange of tightly bound GDP for GTP, resulting in RhoA activation. GDIs form soluble complexes with GDP-bound RhoA and control its cycling between the cytosol and membrane. Reticulon 4 receptor (RTN4R) (also called Nogo-66 receptor, NgR1), a RTN4 receptor subunit, is activated by RTN4 and binds leucine-rich repeat and immunoglobulin domain-containing protein (Lingo-1) and either the p75 neurotrophin receptor or tumor necrosis factor (TNF) receptor orphan Y (TROY), resulting in RhoA activation by GEF. Potassium channel tetramerization domain-containing 13 (KCTD13) is the only identified signaling protein capable of inducing the microcephaly phenotype associated with 16p11.2 duplication, which is associated with schizophrenia
[61][62]. KCTD13 is functionally related to cullin 3 (Cul3), a core component of E3 ubiquitin-protein ligase complexes that mediates the ubiquitination and subsequent proteasomal degradation of target proteins such as RhoA.
4.1. GAPs
4.1.1. ARHGAP10
The
ARHGAP10 gene, which encodes Rho GTPase-activating protein 10 (ARHGAP10), is located on chromosome 4q31.23 and exhibits GAP activity for RhoA and Cdc42
[32][63]. ARHGAP10 is expressed in the brain
[63][64], and its mRNA levels rise in the cerebellum, striatum, and frontal cortex from E4 to P56 in mice
[64].
CNVs in
ARHGAP10 were identified in seven patients with schizophrenia (six with deletions and one with duplication) but not in controls, and there was a significant association of
ARHGAP10 CNVs with schizophrenia in Japanese patients (OR = 12.3,
p = 0.015)
[48]. Most
ARHGAP10 CNVs were exonic deletions at the Bin1/amphiphysin/Rvs167 domain, the RhoGAP domain, or both. The relative expression levels of
ARHGAP10 mRNA in lymphoblastoid cell lines established from the peripheral blood of patients with exonic
ARHGAP10 CNVs were significantly decreased compared to those in patients with schizophrenia without
ARHGAP10 CNVs and in a control group
[48]. One of the patients (case #5) with
ARHGAP10 CNVs had a missense variant (p.S490P) in exon 17 that overlapped with the exonic deletion on the other allele
[48]. ARHGAP10 p.S490P showed weaker binding to active RhoA compared to wild-type ARHGAP10, suggesting that this single-nucleotide variation (SNV) exhibits the loss of function of ARHGAP10. Of note, clinical data of these seven patients with
ARHGAP10 variants showed that treatment response was poor in most individuals, including case #5
[48].
4.1.2. ARHGAP18
ARHGAP18 is ubiquitously expressed throughout the body, including the brain, and shows GAP activity for RhoA but not for Rac1 or Cdc42
[65][66][67]. Through RhoA/Rho-kinase signaling, ARHGAP18 regulates cell spreading and migration and also the formation of stress fibers and focal adhesions
[65]. ARHGAP18 knockdown, which leads to RhoA activation, causes significantly increased formation of stress fibers and focal adhesions in HeLa cells, while these changes are abolished by a Rho-kinase inhibitor or by dominant-negative RhoA transfection
[65]. In contrast, the overexpression of wild-type ARHGAP18, but not GAP-defective ARHGAP18, suppresses the formation of stress fibers and focal adhesions in HeLa cells
[65]. ARHGAP18 also contributes to cell migration
[65]. The knockdown or knockout of
ARHGAP18 impairs migration and cellular polarity in breast cancer cells (MDA-MB-231 or SUN-159 cells)
[65][67].
Single-nucleotide polymorphisms (SNPs) in
ARHGAP18 are associated with schizophrenia
[49][50]. The genotypes and allelic frequencies of two SNPs, rs7758025 and rs9483050, were significantly different between patients and controls in a Chinese-Han population (genotype: rs7758025,
p = 0.0002, and rs9483050,
p = 7.54 × 10
−6; allelic frequencies: rs7758025,
p = 4.36 × 10
−5, and rs9483050,
p = 5.98 × 10
−7). In addition, the AG haplotype in rs7758025-rs9385502 and the CG haplotype in rs11753915-9483050 were associated with schizophrenia (AG haplotype in rs7758025-rs9385502:
p = 0.0012, 95% confidence interval [CI] = 0.48–0.93; CG haplotype in rs11753915-9483050:
p = 9.6 × 10
−6, 95% CI = 0.44–0.78)
[49]. Another group reported an association between SNPs in
ARHGAP18 and schizophrenia in Caucasian people
[50][51]. They also performed a combined analysis with GWAS and functional magnetic resonance imaging scanning and demonstrated that these SNPs were significantly correlated with neuronal activity in the dorsolateral prefrontal cortex during a working memory task
[50].
4.1.3. p250GAP (ARHGAP32)
P250GAP (ARHGAP32) is expressed in the brain
[68] and exerts GAP activity for RhoA but not Cdc42 and Rac1 in mouse primary hippocampus neurons
[69]. P250GAP regulates spine morphology in primary hippocampus neurons and neurogenesis in Neuro-2A cells through the regulation of RhoA activity
[69][70]. In addition, p250GAP interacts with the NR2B subunit of N-methyl-D-aspartate (NMDA) receptors and is involved in NMDA receptor-mediated RhoA activation
[69][70].
An SNP in
p250GAP (rs2298599) was shown to be associated with schizophrenia in a Japanese cohort (
p = 0.00015)
[52]. The minor genotype frequency was higher in patients with schizophrenia (18%) than in healthy controls (9%) (
p = 0.000083)
[52]. rs2298599 is located 2.9 kb downstream of p250GAP and showed no significant association with p250GAP expression levels in immortalized lymphoblasts in in silico analysis (
p = 0.28)
[52]. Thus, further study is needed to clarify the mechanism of the association between p250GAP and schizophrenia.
4.2. GEFs
4.2.1. KALRN
Kalirin has two GEF domains, with activity targeting Rac1 (GEF1) and RhoA (GEF2), respectively
[27][71]. The alternative splicing of
KALRN gives rise to several isoforms, including Kalirin-4, Kalirin-5, Kalirin-7, Kalirin-8, Kalirin-9 (Kal9), and Kalirin-12 (Kal12)
[27][71]. Kal9 and Kal12 contain both the GEF1 and GEF2 domains, while other isoforms contain only the GEF1 domain
[27][71]. Kalirin is involved in neurite and dendritic outgrowth and in dendritic arborization in the brain
[71]. Expression levels of Kal9 and Kal12 in the brain were found to be higher during early postnatal development than in adulthood
[72]. Knockout of Kal9 and Kal12 by shRNA decreased the complexity of rat hippocampal neurons on days in vitro (DIV) 4 and 7 (immature neurons)
[73]. In rat cortical neurons, Kal9 overexpression on DIV 2 (immature neurons) resulted in neurite elongation
[72], while that on DIV 28 (mature neurons) reduced dendritic length and complexity
[74]. These reports indicate that the role of Kal9 in neurons might change depending on the developmental stage.
A transcriptome-wide association study in patients with schizophrenia revealed an increase in exon skipping immediately prior to the GEF2 domain in
KALRN transcripts
[53]. In addition, some missense variants in
KALRN, such as P2255T and T1207M, showed a higher frequency in schizophrenia cases compared to control cases
[54][56]. In particular, P2255T in
KALRN was significantly associated with schizophrenia (OR = 2.09,
p = 0.048) in a Japanese population
[54]. Of note, a P2255 residue exists near the RhoA-GEF2 domain
[55]. The P2255T variant in Kal9 (Kal9-P2255T) leads to highly stable Kal9 mRNA, resulting in increased protein levels of Kal9
[55], which, for instance, were detected in the auditory cortex of patients with schizophrenia by post-mortem analysis
[74]. Furthermore, overexpression of Kal9-P2255T in rat primary neurons and HEK 293 cells increased RhoA activity but had no effect on Rac1 activity
[55][75]. These data indicate that Kal9-P2255T increases the expression levels of Kal9, leading to the activation of RhoA. From the viewpoint of neuronal morphology, Kal9-P2255T overexpression in cortical primary neurons led to a significant reduction in proximal dendritic complexity and dendritic spine size compared to wild-type Kal9 (Kal9-WT)
[55]. On the other hand, it is known that reticulon 4 receptor (RTN4R) activates Kal9 and subsequently leads to the activation of RhoA
[76][77]. The RTN4R/Kal9/RhoA pathway is known to modulate neurite outgrowth
[77]. Myelin-associated inhibitors such as oligodendrocyte-myelin glycoprotein (OMGp) have been identified as additional RTN4R (NGR1) ligands and these also suppress neurite outgrowth
[75][78]. Pharmacologic inhibition of RhoA with the RhoA inhibitor CT04 prevented the OMGp-induced decrease in neuronal complexity
[75]. The overexpression of Kal9-P2255T in cortical primary neurons made them more sensitive to OMGp and decreased both the length and complexity of dendritic arbors
[75]. These results suggest that Kal9-P2255T-induced RhoA activation causes morphological changes in neurons.
4.2.2. ARHGEF11
ARHGEF11, also referred to as KIAA0380 or GTRAP48, shows GEF activity for RhoA but not Rac1 or Cdc42
[79]. ARHGEF11 is expressed in the brain
[80][81] and in cortical neurons, including dendrites and spines
[82]. ARHGEF11 regulates glutamate transport activity by direct binding of excitatory amino acid transporter 4
[81]. The overexpression of ARHGEF11 was shown to decrease spine density in rat cortical neurons
[82][83].
ARHGEF11 haplotypes such as C-C of rs6427340-rs6427339 and A-C-C of rs822585-rs6427340-rs6427339 were shown to be associated with schizophrenia (
p = 0.0010 and 0.0018, respectively), but the
ARHGEF11 SNPs were not
[57]. The functions of
ARHGEF11 haplotypes associated with schizophrenia have not been clarified. On the other hand, in situ hybridization analysis indicated that
ARHGEF11 mRNA levels in the thalamus of patients with schizophrenia were higher than those in healthy controls
[84]. These findings raise the possibility that
ARHGEF11 activation is associated with schizophrenia pathology.
4.3. Others
4.3.1. RTN4R
RTN4R (also called Nogo-66 receptor, NgR1) is a RTN4 receptor subunit located at chr22q11.2, and it has been shown that deletion of chr22q11.2 is associated with a high risk of developing schizophrenia
[85]. RTN4R binds leucine-rich repeat and immunoglobulin domain-containing protein (Lingo-1) and either the p75 neurotrophin receptor or tumor necrosis factor (TNF) receptor orphan Y (TROY) and activates RhoA through Kal9; this results in the collapse of growth cones, which prevents further axonal growth and inhibits myelination
[76][86].
Several SNPs in RTN4R are associated with schizophrenia
[58][59]. In samples from individuals of Afrikaner origin, significant associations with schizophrenia were seen for SNP rs696880 in women (OR = 0.73,
p = 0.046) and for rs701427 (OR = 1.21,
p = 0.019), rs696880 (OR = 1.18,
p = 0.029), and rs854971 (OR = 1.20,
p = 0.021) in men
[59]. Diffusion tensor imaging revealed that the SNP rs701428 was associated with white matter abnormalities in 22q11.2 deletion syndrome
[87]. Other groups identified several rare missense variants in patients with schizophrenia, specifically p.R68H (rs145773589), p.R119W (rs74315508), p.R196H (rs74315509), p.D259N (rs3747073), p.R292H (rs1432033565), and p.V363M (rs149231717), and p.R292H was significantly associated with schizophrenia (OR = 3.9,
p = 0.048)
[58][60]. RTN4R-R292H is located in the ligand binding site, and its overexpression in E5.5 chick retinal neurons significantly decreased growth cone collapse induced by treatment with RTN4, a ligand of RTN4R, compared to that resulting from treatment with RTN4R-WT
[58]. In addition, a glutathione S-transferase binding assay showed that RTN4R-R292H exhibited reduced interaction with LINGO1 compared to RTN4R-WT
[58]. Although these data suggest that RTN4R-R292H has impaired function, the effect of RTN4R-R292H on RhoA signaling remains obscure. In a post-mortem brain analysis, the expression levels of RTN4R were decreased in the dorsolateral prefrontal cortex but increased in the hippocampal CA3 region of patients with schizophrenia compared to healthy controls
[88].
In addition to the above, genetic variations in components of the RTN4R signaling pathway, such as RTN4 (
p = 0.047 and 0.037 for rs11894868 and rs2968804, respectively) and myelin-associated glycoprotein (
p = 0.034 and 0.029 for rs7249617 and rs16970218, respectively), were shown to be associated with schizophrenia
[89].
4.3.2. 16p11.2 CNVs and the KCTD13-Cul3-RhoA Pathway
16p11.2 microduplication was associated with schizophrenia in two large cohorts (OR = 25.8,
p = 1.2 × 10
−5; and OR = 8.3,
p = 0.022)
[61]. In a zebrafish model, potassium channel tetramerization domain-containing 13 (KCTD13) was identified as the sole signaling protein capable of inducing the microcephaly phenotype associated with 16p11.2 duplication
[62]. A spatiotemporal protein–protein interaction network analysis showed that KCTD13 is functionally related to cullin 3 (Cul3)
[90]. Cul3 is a core component of E3 ubiquitin–protein ligase complexes and mediates the ubiquitination and subsequent proteasomal degradation of target proteins such as RhoA, but not Rac1 and Cdc42
[91][92]. Although it is estimated that 16p11.2 microduplication is associated with decreased RhoA protein levels, organoids derived from patients with autism spectrum disorder who had 16p11.2 duplications showed RhoA activation and slightly increased KCTD13 expression
[93]. Therefore, further research should analyze 16p11.2 microduplication in patients with schizophrenia.