Identifying SHROOM4 as a novel X-linked susceptibility gene for cerebral palsy in Chinese males

Abstract

Background

Cerebral palsy (CP) is a leading cause of childhood motor disability with a notable male predominance, suggesting that X-linked genetic factors may contribute to CP susceptibility. Although SHROOM4 has been implicated in several neurodevelopmental disorders, its role in CP remains unclear. This study aimed to investigate the contribution of SHROOM4 variants to male CP susceptibility.

Methods

Whole-exome sequencing was performed in 1,010 Chinese male patients with sporadic CP and 1,014 male controls. Association analysis focused on common variants and haplotypes within SHROOM4. Rare SHROOM4 variants identified in CP patients were further validated and characterized using qPCR, immunofluorescence, western blotting and CRISPR/Cas9-mediated knockout cell lines.

Results

A common T-A-G haplotype comprising rs2873098, rs2295544 and rs2295543 in SHROOM4 was significantly associated with male CP susceptibility (OR = 6.091, Pc = 3.26E-07) and was enriched in CP patients with intrauterine growth restriction. Additionally, a rare nonsense variant, c.C2050T (p.Arg684*), was identified in a patient presenting with spastic CP and intellectual disability. Functional analyses showed that p.Arg684* and a population-derived frameshift variant (p.Glu1140fs*42) were associated with reduced SHROOM4 transcript abundance, consistent with NMD-mediated transcript reduction, while residual mutant transcripts produced detectable truncated proteins with variant-specific effects on protein stability, subcellular localization, and actin cytoskeletal organization.

Conclusions

Our findings support SHROOM4 as an X-linked susceptibility gene associated with male CP, suggesting that both common haplotypes and rare SHROOM4 variants may contribute to the genetic susceptibility to CP. These results expand our current understanding of the genetic architecture of CP and highlight cytoskeletal regulation as a potentially relevant mechanism associated with SHROOM4 variants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-026-07486-y.

Introduction

Cerebral palsy (CP) is a clinically heterogeneous group of disorders of movement and posture characterized by non-progressive impairments arising from disturbances in the developing fetal or infant brain [1, 2], and is frequently accompanied by comorbidities such as intellectual disability (ID), epilepsy and autism spectrum disorders (ASDs) [3–5]. Despite advancements in obstetric practice and perinatal care, CP remains the most prevalent cause of childhood physical disability, affecting about 2 per 1,000 live births worldwide [6–8] and imposing a substantial lifelong medical and socioeconomic burden.

Extensive research has established associations between CP and various adverse perinatal events or risk factors, including prematurity, intrauterine growth restriction (IUGR) and birth hypoxia [9–12]. In addition, increasing evidence suggests that genetic factors also contribute to CP susceptibility. Recent advances in next-generation sequencing have identified causative genetic variants in familial CP cases [13–18], and it is estimated that genetic causes account for 24–31% of CP cases, highlighting the role of genetic factors in its pathogenesis [19, 20]. Nevertheless, compared with other neurodevelopmental disorders (NDDs), the genetic architecture of CP remains less well characterized, highlighting the importance of identifying novel CP-associated genes and biological pathways [21].

Male patients with CP have a higher predisposition, with a male-to-female ratio of approximately 1.3–1.5:1 [19, 22]. Given that males have only one X chromosome and are therefore hemizygous for X-linked variants, and given that our previous study demonstrated an enrichment of X-linked genetic causes in CP compared with the expected proportion of X-linked genes in the human genome, we speculate that X-linked variants may contribute to CP in males [19]. Emerging evidence suggests that the X chromosome harbors a number of genes important for neural development, synaptic organization and cytoskeletal regulation [23, 24], supporting a potential role of X-linked neurodevelopmental genes in male CP susceptibility.

SHROOM4, located at Xp11.22, encodes a member of the Shroom protein family containing a conserved motif Apx/Shrm domain 2 (ASD2) and an N-terminal PDZ domain [25]. Its ASD2 domain participates in cytoskeletal organization and stress fiber assembly through interactions with actin-associated regulatory proteins [26]. Cytoskeletal regulation is essential for multiple neurodevelopmental processes, including neuronal migration, neurite outgrowth, and brain structural maturation, suggesting that SHROOM4 may be relevant to neurodevelopmental vulnerability. Consistent with this concept, rare damaging variants in SHROOM family genes have been reported to be enriched in neural tube defect cases, and rare deleterious variants in SHROOM2 were shown to disrupt its interaction with ROCK1, supporting a broader role of Shroom-mediated cytoskeletal regulation in neurodevelopmental susceptibility [27]. SHROOM4 is highly expressed in the developing brain and vascular endothelial cells and exhibits strong intolerance to loss-of-function (LoF) variants (pLI = 1.0), suggesting potential roles in neurodevelopment and neuronal function. Consistent with this possibility, SHROOM4 have been implicated in neurodevelopmental disorders including epilepsy and ASD in previous studies [28–41]. However, the contribution of SHROOM4 to CP susceptibility remains largely unexplored.

In this study, we investigated the contribution of SHROOM4 to male CP in a Chinese cohort through an integrated genetic and functional approach. We evaluated the association between SHROOM4 variants and CP susceptibility. Our findings suggest that SHROOM4 as a candidate X-linked susceptibility gene associated with male CP and highlight cytoskeletal regulation as a potentially relevant mechanism underlying male CP susceptibility.

Materials and methods

Study patients

A total of 2,024 unrelated Chinese individuals were enrolled in this study, comprising 1,010 male patients with CP and 1,014 healthy male controls recruited through routine pediatric health examinations at collaborating hospitals. The control individuals had no known neurological abnormalities. All participants were recruited from Zhengzhou Children’s Hospital or the third affiliated hospital of Zhengzhou University. The diagnosis of CP was established by pediatric neurologists in accordance with guidelines from the Surveillance of Cerebral Palsy in Europe network [42]. Clinical data, including gestational age, and birth weight, risk factors, were collected from medical records.

Sequencing, variant filtering, and association analysis of SHROOM4

DNA extraction and Whole-Exome Sequencing (WES)

Blood samples were obtained from patients with vacuum EDTA anticoagulant tubes and genomic DNA was extracted using QIAamp DNA Blood Mini Kit (QIAGEN, Germany) following standard protocol. DNA samples concentration and purity were quantified by Qubit fluorometer. We performed WES on Illumina HiSeq platform with paired-end sequencing with the mean sequencing depth over 100×. Filtered reads were aligned to the GRCh38/hg38 reference genome. Variants were called using Genome Analysis Toolkit v4.2.3 following the best practice workflows and filtered by variant quality score recalibration.

Quality control and population stratification

Samples were filtered by individual relatedness, sex check and genotype completion. Cryptic relatedness was assessed by identity-by-descent analysis, and one individual from each potentially related pair with PI_HAT > 0.125 was removed to ensure sample independence for downstream association analyses. Variants with a minor allele frequency (MAF) ≥ 0.01 were classified as common variants and retained for association analysis. SNVs were removed if they met one or more of the following conditions: (1) multiple alternative alleles identified on one locus; (2) a calling rate < 95% in the combined case and control samples; (3) the MAF < 0.01 in both cases and controls.

To assess population stratification, principal component analysis (PCA) was performed on the X-chromosome SNVs using PLINK v1.9. Briefly, independent SNVs were obtained by linkage disequilibrium (LD) pruning (--indep-pairwise 1000 50 0.05), and the first two principal components were used as covariates in downstream association analyses. Genomic inflation factor (λ) was calculated using chi-square statistics from X-chromosome SNV association tests, and Q-Q plots were generated to assess deviation from expected distributions (Fig.S1). The genomic inflation factor (λ = 1.097) indicated limited inflation of association statistics.

Variants selection and SHROOM4-focused association analysis

The genomic sequence of the human SHROOM4 gene was determined according to the NCBI GenBank reference sequence NM_020717.5. Association analysis focused on the SHROOM4 locus (chrX: 50586796–50814194). Common SNVs (MAF ≥ 0.01) within SHROOM4 were retained for SHROOM4-focused association analysis. SNV association analysis was performed using a logistic regression model, with adjustment for the first two principal components to account for population stratification. Haplotype blocks were defined using LDBlockShow with D’ >0.5 as the LD cutoff. Haplotype frequencies were estimated, and their associations with CP susceptibility were assessed using logistic regression model implemented in R software (v4.4.1), adjusting for the first two principal components. Only haplotypes with frequencies ≥ 0.01 were analyzed to ensure statistical reliability, resulting in 9 common haplotypes.

Sensitivity analysis

To assess the robustness of the association results, sensitivity analyses were performed after excluding individuals who had received genetic diagnoses (n = 774 remaining CP patients). Logistic regression analyses were repeated using the same model and covariates as in the primary analysis.

Rare SHROOM4 variant prioritization for functional characterization

Variants with the allele frequency < 0.001 in gnomAD and absent from the control cohort were considered as rare candidate variants. Polyphen2, SIFT, MutationTaster, phyloP, CADD and the guidelines of American College of Medical Genetics and Genomics (ACMG) [43] were utilized to prioritize variants with potential functional relevance. Variants predicted to be deleterious by in silico tools and classified as variants of uncertain significance or potentially damaging according to ACMG guideline were selected for downstream functional characterization.

Plasmid construction

The full-length coding sequence of human SHROOM4 was cloned into the pcDNA3.1-FLAG vector. The SHROOM4 nonsense variant (p.R684*) identified in the study cohort and a frameshift variant (p.E1140fs*42) retrieved from gnomAD were generated using a Fast Site-Directed Mutagenesis Kit (Tiangen, China). A SHROOM4 overexpression plasmid was constructed by cloning the SHROOM4 coding sequence together with ~ 200 bp flanking intronic regions surrounding exon-intron junctions (exons 3–4, 4–5, 5–6, and 6–7) and a 1.5 kb 3’UTR segment downstream of exon 9 into the pCMV6-N-GFP vector. A mCherry-expressing plasmid (a gift from Lei Ji [44]) was co-transfected with SHROOM4-GFP constructs to serve as an internal reference for transfection efficiency and normalization of immunofluorescence intensity.

Cell culture and transfection

HEK293T, HeLa and HepG2 cells were cultured in DMEM medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37℃ and 5% CO₂. Cells were transfected with 1–2.5 µg of DNA using PolyJet™ In Vitro DNA Transfection Reagent (SignaGen, USA) according to manufacturer’s recommendations.

Generation of SHROOM4 knockout cell line

To simulate premature termination variants of SHROOM4, CRISPR-Cas9 system was used to introduce indel mutations in HepG2 cells. The pSpCas9(BB)-2 A-Puro (PX459) V2.0 (Addgene plasmid 62988) was used for sgRNA expression [45]. Two sgRNA targeting exon 4 (sg2191) and exon 6 (sg3556) of SHROOM4 were designed using CRISPick webtool (https://portals.broadinstitute.org/gppx/crispick/public) and evaluated for specificity with Cas-OFFinder (http://www.rgenome.net/cas-offinder/). Oligonucleotides encoding sgRNAs were cloned into PX459 and verified by Sanger sequencing.

T7 Endonuclease I (T7EI) assay

The T7EI assay was applied as a preliminary screen to detect potential CRISPR/Cas9-mediated edits. After 48 h post-transfection, cells were treated with puromycin to eliminate untransfected cells. Genomic DNA was extracted from the surviving bulk population. The target region was PCR-amplified using specific T7E1 verification primers (Additional file: Table S1). PCR products were denatured and re-annealed to form heteroduplexes, then digested with T7EI (New England Biolabs, USA) at 37 °C for 30 min. Digested products were separated on a 2% agarose gel in 1× TAE buffer alongside a 100 bp DNA ladder (Tiangen, China).

Following validation of positive T7EI signals, cells were seeded into 96-well plates for single-cell clonal expansion. Individual clones were subsequently genotyped by PCR amplification and indel mutations were confirmed by Sanger sequencing.

RNA extraction and qPCR analysis

To assess mRNA degradation, HEK293T cells were treated with 100 µg/ml cycloheximide (CHX) for 0 and 6 h, followed by RNA extraction and qPCR analysis. RNA extraction, reverse transcription and real-time quantitative PCR (qPCR) were performed according to the manufacturer’s instructions (Yeasen, China). Amplification was performed on QuantStudio™ 5 Real-Time PCR System (ThermoFisher, USA). Gene expression levels were normalized to the GAPDH. All the primers used in this study are listed in Additional file: Table S1.

Immunocytochemistry and confocal imaging

HeLa cells transfected with 5’-FLAG-tagged SHROOM4 vectors were fixed 48 h post-transfection with 4% paraformaldehyde in PBS for 15 min. For cytochalasin D treatment, cells were exposed to either 2 µM cytochalasin D (CD, MedChemExpress, USA) or DMSO (Sigma, USA) for 20 min prior to collection. Cells were incubated with primary antibodies FLAG (Gnipharma; 1:350) at 4℃ overnight, followed by incubation with Alexa Fluor 488-conjugated anti-mouse IgG and Alexa Fluor 647-conjugated anti-rabbit IgG (ThermoFisher; 1:1000). F-actin was visualized using Alexa-555 conjugated Phalloidin (CST; 1:200). For GFP-tagged SHROOM4 experiments, the co-transfected mCherry signal was used to normalize for transfection efficiency. Immunofluorescence intensity of SHROOM4-GFP was quantified as the GFP/mCherry ratio for comparison across conditions. Fluorescent images were acquired using a Leica SP8 confocal microscope with the 20× and 63× objectives.

Western blot

HEK293T cells were harvested 36 h post-transfection and lysed in ice-cold RIPA buffer (Epizyme, China) supplemented with 1% Triton X-100, 1% protease inhibitor cocktail (Meilunbio, China) and 1% phosphatase inhibitor cocktail (AGbio, China). Lysates were mixed with 5× SDS loading buffer, heated at 100℃ for 10 min, and subjected to Western blotting (WB).

For CHX chase assays, HEK293T cells were transfected with FLAG-tagged SHROOM4-WT or mutants (R684* and E1140fs*42) together with a GFP-expressing plasmid as internal control for transfection efficiency. Cells were treated with 50 µg/ml CHX for 0, 12 and 24 h. Cell lysates were collected at each time point, and protein levels were analyzed by WB using anti-FLAG, anti-GFP and anti-GAPDH antibodies.

Statistics analysis

SNV association analysis

Association between individual SHROOM4 SNVs and male CP was performed using logistic regression in PLINK v1.9, adjusting for the first two principal components to account for population stratification. Odds ratios (ORs) and 95% confidence intervals (Cls) were estimated from the logistic regression model. Two-tailed P-values < 0.05 were considered statistically significant.

Haplotype association analysis

Haplotype-specific associations were assessed using logistic regression in R (v4.4.1), adjusting for the first two principal components. ORs and 95% CIs were calculated for each haplotype. SNVs in moderate LD (D’ > 0.5) were grouped into haplotype blocks, and Bonferroni correction was applied across all 9 common haplotypes analyzed. A two-tailed P-value < 0.05 after correction was considered statistically significant.

Comparison of haplotype carriers and non-carriers

Clinical factors characteristics between carriers and non-carriers of each haplotype were compared using Fisher’s exact test for categorical variables (e.g., IUGR). Due to incomplete clinical information in controls, these analyses were restricted to the case group only.

Functional analyses of rare SHROOM4 variants

To investigate the effects of rare predicted LoF variants, we performed in vitro studies comparing wild-type (WT) and mutant (MU1: p.R684*, MU2: p.E1140fs*42) SHROOM4 constructs in HEK293T and HepG2 cells. Quantitative mRNA expression of SHROOM4-WT and mutant (p.R684*, p.E1140fs*42) constructs in HEK293T and HepG2 cells was assessed by qPCR and compared using one-way ANOVA followed by Dunnett’s post hoc test with WT as the control. Immunofluorescence intensity of GFP-tagged SHROOM4 was quantified using the co-transfected mCherry signal to normalize for transfection efficiency, and immunofluorescence intensity was quantified as the GFP/mCherry ratio using Fiji. Statistical analyses were performed in GraphPad Prism 9 using one-way ANOVA followed by Dunnett’s post hoc test. A two-tailed P-value < 0.05 was considered statistically significant.

Cytoskeletal assessment

For HeLa overexpression cells, actin fibers were scored 0–2: 0 = no long thick fibers, 1 = short or disorganized fibers, 2 = long thick fibers crossing the cell. At least 90 cells per condition were scored from at least three independent experiments. For HepG2 cells, actin fibers were scored 0–1: 0 = no thick fibers crossing the cell, 1 = thick fibers crossing the cell. At least 90 cells per condition were scored from at least three independent experiments. Statistical significance was assessed using Kruskal-Wallis with Dunn’s post hoc test. A two-tailed P-value < 0.05 was considered statistically significant.

Results

Association study of SHROOM4 with male CP

A total of 1,010 male patients with CP were enrolled in this study. The cohort was predominantly composed of Han Chinese individuals (91.6%). Notably, perinatal risk factors were prevalent, including birth asphyxia (25.4%), neonatal jaundice (24.1%), intracranial hemorrhage (3.3%) and IUGR (5.8%).

To investigate the correlation between SHROOM4 genetic variations and CP in males, we performed a case–control association analysis focusing on common variants (MAF ≥ 0.01) within the SHROOM4 gene in 1,010 CP boys and 1,014 healthy male controls. To assess potential population stratification, we evaluated the genomic inflation factor using X-chromosome association statistics after quality control. The genomic inflation factor was λ = 1.097, and the corresponding Q-Q plot is shown in Fig. S1. Within the SHROOM4 gene body, seven common variants were identified: rs2873098, rs2295544, rs2295543, rs12689863, rs41306882, rs3761506 and rs74774698 (Table 1; genotype data for selected SNVs were listed in Additional file: Table S2). Significant differences in allele frequencies were observed for one variant: rs2295543 (P = 0.005, OR = 1.244, 95% CI: 1.069–1.448).

Table 1.

Allele frequencies of SHROOM4 in 1,010 male CP male individuals and 1,014 control individuals

SNP	P value	OR [95% CIs]	Allele frequency
				T	C
rs2873098	0.724	0.982 [0.887–1.087]	Case	597(0.591)	413(0.408)
			Control	554(0.546)	460(0.453)
				G	A
rs2295544	0.517	0.953 [0.823–1.103]	Case	213(0.21)	797(0.789)
			Control	136(0.134)	878(0.865)
				G	C
rs2295543	0.005	1.244 [1.069–1.448]	Case	363(0.359)	647(0.64)
			Control	151(0.148)	863(0.851)
				G	A
rs12689863	0.411	0.950 [0.840–1.074]	Case	824(0.815)	186(0.184)
			Control	807(0.795)	207(0.204)
				A	T
rs41306882	0.180	0.911 [0.795–1.044]	Case	901(0.892)	109(0.107)
			Control	838(0.826)	176(0.173)
				C	T
rs3761506	0.042	0.642 [0.420–0.983]	Case	1000(0.99)	10(0.009)
			Control	991(0.977)	23(0.022)
				C	T
rs74774698	0.307	1.127 [0.896–1.418]	Case	975(0.965)	35(0.034)
			Control	968(0.954)	46(0.045)

Haplotype analysis is a powerful strategy for investigating the combined effects of multiple SNVs within a gene. Two haplotype blocks were formed based on LD analysis: Block 1 including SNVs rs2873098, rs2295544 and rs2295543, and Block 2 including SNVs rs12689863, rs41306882, rs3761506 and rs74774698, which were in at least moderate LD with each other (D’>0.5) (Fig. 1). The SHROOM4 T-A-G haplotype within Block 1 was significantly associated with CP after Bonferroni correction (Pc = 3.26E-07) and conferred a higher risk compared with non-carriers (OR = 6.091, 95% CI: 3.216–11.538; Table 2).

Fig. 1.

LD plot of selected SNVs. Two haplotype blocks (enclosed by black lines) were identified: Block 1 (rs2873098, rs2295544, rs2295543) and Block 2 (rs12689863, rs41306882, rs3761506, rs74774698), showing moderate LD (D’ > 0.5)

Table 2.

Haplotype association of SHROOM4 SNVs between 1,010 male CP patients and 1,014 controls

Haplotype	CP	Control	OR	95% CI	P value
C-A-C	365(0.361)	458(0.452)	0.954	0.778–1.170	6.50E-01
T-A-C	279(0.276)	401(0.395)	0.876	0.715–1.074	2.03E-01
T-A-G	108(0.107)	18(0.018)	6.091	3.216–11.538	2.96E-08
T-G-G	207(0.205)	132(0.130)	0.919	0.684–1.235	5.76E-01
A-A-C-C	91(0.090)	59(0.058)	0.936	0.620–1.414	7.53E-01
A-T-C-C	94(0.093)	148(0.146)	0.883	0.662–1.178	3.98E-01
G-A-C-C	766(0.758)	710(0.700)	1.176	0.946–1.462	1.44E-01
G-A-T-C	34(0.034)	46(0.045)	1.225	0.772–1.946	3.89E-01
G-T-C-C	14(0.014)	28(0.028)	0.566	0.277–1.156	1.18E-01

Block 1:rs2873098, rs2295544, rs2295543 and Block 2: rs12689863, rs41306882, rs3761506 and rs74774698. Haplotypes with a frequency < 0.01 in either controls or cases have been dropped

OR odds ratio, CI confidence interval

Given that patients with CP who have different histories of risk factor exposure may vary in their susceptibility to genetic variants, we further analyzed the relationship between the T-A-G haplotype and CP risk factors (Table 3). Notably, a highly significant association was identified with IUGR. The frequency of the T-A-G haplotype in CP patients with IUGR was significantly higher than in those without IUGR (18.52% vs. 4.32%; OR = 5.029, 95% CI: 2.817–9.020). Conversely, the occurrence of birth asphyxia was lower in haplotype carriers compared to non-carriers (11.11% vs. 27.16%; OR = 0.335, 95% CI: 0.181–0.608). No significant associations were found for family history, intracranial hemorrhage or neonatal jaundice.

Table 3.

Association of the T-A-G haplotype of SHROOM4 with risk factors in male CP individuals

Clinical Phenotype	Haplotype Carriers (n = 108)	Non-Carrier (n = 902)	P Value (Fisher’s exact)	OR	95% CIs
Family History
Present, n (%)	15 (13.89%)	99 (10.98%)	0.338	1.310	0.728–2.323
Absent, n (%)	93 (86.11%)	804 (89.13%)
Intrauterine Growth Restriction (IUGR)
Present, n (%)	20 (18.52%)	39 (4.32%)	6E-06	5.029	2.817–9.020
Absent, n (%)	88 (81.48%)	863 (95.68%)
Birth Asphyxia
Present, n (%)	12 (11.11%)	245 (27.16%)	2E-04	0.335	0.177–0.614
Absent, n (%)	96 (88.89%)	657 (72.84%)
Intracranial Hemorrhage
Present, n (%)	4 (3.70%)	29 (3.21%)	0.773	1.158	0.399–3.359
Absent, n (%)	104 (96.30%)	873 (96.79%)
Neonatal Jaundice
Present, n (%)	25 (23.15%)	218 (24.16%)	0.905	0.945	0.595–1.519
Absent, n (%)	83 (76.85%)	684 (75.83%)

To assess the stability of the observed association, we performed a sensitivity analysis excluding the individuals who had previously received genetic diagnoses (excluded: n = 236; remaining: n = 774). The results of this analysis reaffirmed for both SNVs (OR = 1.213, 95%Cls: 1.028–1.431, P = 0.022, Additional file: Table S3) and haplotypes (OR = 6.887, 95%Cls: 3.416–13.884, Pc = 6.192E-07, Additional file: Table S4). In addition, the SHROOM4 T-A-G haplotype demonstrated a persistent and significant association with IUGR remained significant (OR = 5.542, 95%Cls: 2.863–10.731, P = 3.23E-06, Additional file: Table S5). Collectively, these sensitivity results are consistent with the primary analysis and support the stability of the haplotype association between SHROOM4 and CP.

Identification of a rare truncating SHROOM4 variant for functional characterization

In addition to common variant analyses, a rare nonsense variant (c.2050 C > T; p.Arg684*, Fig. S2a, 2b), was identified in a 2-year-old boy with spastic CP and ID. The variant was prioritized for further functional characterization because it introduces a premature stop codon and was predicted to have deleterious effects by in silico tools (CADD: 35, Mutationtaster: deleterious) and was classified as a variant of uncertain significance according to ACMG guidelines.

Transcripts carrying truncating variants of SHROOM4 may partially escape NMD

The high intolerance of SHROOM4 to LoF variants underscores its biological importance. However, this apparent constraint seems inconsistent with the presence of multiple predicted LoF variants in the gnomAD database at non-negligible frequencies (e.g.: NM_020717.5: c.3413_3414insGG, p.Glu1140fs*42 and NM_020717.5:c.3415dupC, p.Glu1139fs*39). This discrepancy underscores the uncertainty surrounding the functional interpretation of SHROOM4 truncating variants.

Theoretically, both c.C2050T (p.R684*) located in exon 4 and c.3413_3414insGG (p.E1140fs*42) located in exon 6, are predicted to undergo NMD. To verify this prediction, we constructed WT, p.R684* and p.E1140fs*42 GFP-N-terminal expression plasmids and transfected them into HEK293T cells (Fig. 2a). Quantitative PCR (qPCR) results from the overexpression system showed markedly decreased mRNA levels for both variants relative to WT (Fig. 2b), indicating reduced abundance of mutant transcripts. Following CHX treatment, WT-transfected cells showed a subtle yet statistically significant upregulation of mRNA levels (mean ± SEM: 0.19 ± 0.05; P = 0.028), whereas cells transfected with p.R684* and p.E1140fs*42 exhibited more pronounced upregulation (mean ± SEM: 0.99 ± 0.19, P = 0.002; 1.08 ± 0.21, P = 0.015, respectively, Fig. 2c). These results are consistent with NMD-mediated reduction of mutant transcripts carrying these truncating variants. However, the mutant transcripts remained readily detectable, suggesting incomplete degradation and partially NMD escape.

Fig. 2.

Truncating variants in SHROOM4 exhibited reduced mRNA expression, consistent with partial NMD escape, and showed differential protein expression and stability. a Schematic presentation of the WT and mutant (p. R684*, p.E1140fs*42) GFP-tagged N-terminal SHROOM4 expression constructs. Red asterisks indicate the variants loci on the exons. b qPCR analysis of SHROOM4 mRNA levels in transfected cells, normalized to GAPDH. Data are presented as mean ± SEM from three independent biological replicates. ** P < 0.01. c qPCR analysis of SHROOM4 mRNA levels following treatment with CHX (100 µg/mL for 6 h). Each data point represents the mean value calculated from four independent biological replicates. * P < 0.05; ** P < 0.01; *** P < 0.001. d Representative immunofluorescence images of HEK293T cells transfected with the constructs in (a), showing GFP (green), mCherry (red, transfection control), and DAPI (blue). Scale bar: 200 μm. e Quantification of the GFP-SHROOM4 fluorescence intensity normalized to mCherry signal in (d). Data are presented as mean ± SEM; each data point represents the mean value calculated from 3–5 randomly selected fields of view. Five independent biological replicates were performed. *** P < 0.001. f WT and mutant SHROOM4 proteins expression at 0, 12, and 24 h following treatment with CHX (50 µg/mL) in HEK293T cells. g Quantification of CHX chase assays is presented as mean ± SEM from three independent experiments. During the 0–24 h chase period, WT, p.R684*, and p.E1140fs*42 proteins showed no statistically significant differences in protein stability changes

Consistent with incomplete transcript degradation, immunofluorescence analysis revealed reduced but detectable fluorescence signals for both c.C2050T (p.R684*) and the c.3413_3414insGG (p.E1140fs*42) compared with WT (Fig. 2d and e), suggesting that residual mutant transcripts may still give rise to detectable truncated proteins.

We therefore performed CHX chase followed by Western blotting (WB) to evaluate the stability of WT and truncated SHROOM4 proteins. Protein abundance was monitored over a 24 h chase period, and no statistically significant differences in stability changes were observed among WT, p.R684*, and p.E1140fs*42 proteins (Fig. 2f and g).

To better approximate the endogenous transcript context of SHROOM4 truncating variants, we further generated SHROOM4 knockout (KO) HepG2 cell lines using the CRISPR/Cas9 system. Guide RNAs (sgRNAs) were designed to target exon 4 (sg2191) and exon 6 (sg3556), corresponding to the approximate positions of p.R684* and p.E1140fs*42, respectively (Fig. 3a and b). qPCR analysis showed significantly reduced endogenous SHROOM4 expression in both KO cell lines compared with controls (Fig. 3c), supporting transcript reduction after disruption of these exons. The sg2191 cells exhibited a more pronounced reduction (0.1773 ± 0.0989) in mRNA levels than the sg3556 cells (0.2607 ± 0.0989), suggesting a potential position-dependent difference in NMD efficiency.

Fig. 3.

Generation of SHROOM4 knockout cells and functional validation of SHROOM4 variants in cytoskeletal organization. a Schematic of the sgRNA target sites in exons 4 (sg2191) and 6 (sg3556) in the SHROOM4 gene. b T7 endonuclease I (T7E1) assay confirming indel mutations in HepG2 cells. Representative T7EI assay results for preliminary screening. Transfected cells were enriched by puromycin selection, and PCR products of the target region were denatured, re-annealed, and digested with T7EI to detect potential indels. c qPCR analysis of SHROOM4 mRNA expression in monoclonal knockout (KO) cell lines. Relative expression normalized to controls is plotted as mean ± SEM (n = 3–7). Statistical significance was determined by one-way ANOVA, followed by Dunnett’s post hoc test with the WT group as the control; *** P < 0.001. d Phalloidin staining (red) of F-actin in HepG2 WT and KO cells (sg2191 or sg3556). Scale bar: 10 μm. Magnified insets highlight differences in actin fiber organization. e Representative immunofluorescence images of HeLa cells expressing FLAG-tagged SHROOM4-overexpression constructs, showing SHROOM4 (green), F-actin (red, stained with Phalloidin), and DAPI (blue). Scale bar: 10 μm. f Semi-quantitative analysis of actin fiber organization in HeLa cells. Cells were scored as 0, 1, or 2 based on the presence and morphology of actin stress fibers: score 0, absence of long actin fibers spanning the cell; score 1, presence of short or disorganized actin fibers; score 2, presence of thick and continuous actin fibers traversing the cell. Statistical significance was assessed using Kruskal-Wallis with Dunn’s post hoc test. Quantification was performed from at least three independent experiments, with more than 90 cells analyzed per group. * P < 0.05, *** P < 0.001. g Semi-quantitative analysis of actin fibers in HepG2 knockout cells. Cells were classified as score 0 (absence of actin fibers spanning the cell) or score 1 (presence of visible actin fibers traversing the cell). Statistical significance was assessed using Kruskal-Wallis with Dunn’s post hoc test. Quantification was performed from at least three independent experiments, with more than 90 cells analyzed per group. *** P < 0.001

Together, the overexpression and endogenous disruption systems suggest that SHROOM4 truncating variants are associated with NMD-mediated transcript reduction but may partially escape NMD. Residual mutant transcripts may allow production of detectable truncated proteins, with variant-specific differences in transcript abundance and protein stability potentially contributing to differential functional consequences.

Truncated SHROOM4 lead to functional impairment of cytoskeletal regulation

To assess the functional consequences of these variants, we examined their subcellular localization. The p. R684* and p.E1140fs*42 yield truncated proteins lacking the ASD2 domain while retaining the N-terminal PDZ domain. Immunofluorescence staining in HeLa cells showed that SHROOM4-WT localized to the cytoplasm, adopting a characteristic wave-like morphology and co-localizing with central stress fibers (Fig. 3e), with thick actin bundles. In contrast, SHROOM4⁻R684* was diffused distributed throughout the cytoplasm with absent of prominent actin fibers, whereas SHROOM4-E1140fs*42 exhibited a punctate, aggregated pattern with short or disorganized actin fibers. Semi-quantitative scoring confirmed that WT had the highest actin fiber scores, followed by p.E1140fs*42, and p.R684* showing the lowest scores (Fig. 3f). These data suggests that truncation position differentially affects SHROOM4-mediated actin organization.

Similar phenotypes were observed in the KO cell lines (Fig. 3d). Control HepG2 cells showed obvious actin fibers distributed throughout the cytoplasm. sg2191 cells showed fewer contiguous, red-stained actin fibers, whereas sg3556 cells exhibited a reduction, albeit less pronounced than sg2191, consistent with a differential effect of N-terminal versus C-terminal disruption on cytoskeletal integrity in overexpression cells (Fig. 3g).

To further investigate SHROOM4-actin interactions, HeLa cells expressing WT or mutant SHROOM4 were treated with cytochalasin D (CD) to disrupt actin filaments [26]. CD treatment completely abolished normal actin organization in all cells (Fig. S3). SHROOM4-WT remained fully co-localized with F-actin. In contrast, SHROOM4-E1140fs*42 showed partial co-localization with larger cytoplasmic aggregates, and SHROOM4-R684* was largely diffuse with minimal co-localization with actin, indicating a disruption of actin-dependent localization in R684*.

Discussion

CP arises from a complex interplay of environmental and genetic factors; however, the genetic architecture of CP remains less well characterized than that of other NDDs [21]. Male patients with CP have a higher predisposition. Given that males are hemizygous for X-linked variants and our previous study revealed an enrichment of X-linked genetic causes relative to the proportion of the X chromosome in the human genome in the CP cohort, we speculate that X-linked variants may also contribute to CP susceptibility in males [19]. Here, we identify SHROOM4, an X-linked cytoskeletal regulator implicated in various NDDs, as a gene contributing to male CP susceptibility.

In our case-control study of 1,010 male CP patients and 1,014 healthy male controls, a common T-A-G haplotype in SHROOM4 (rs2873098-rs2295544-rs2295543) was significantly enriched in male CP patients (OR = 6.091, Pc = 3.26E-07), suggesting that a combined effect of multiple linked variants within SHROOM4 may contribute to CP susceptibility. We also observed an enrichment of the SHROOM4 haplotype in CP patients with IUGR (OR = 5.029, P = 6.00E-6), suggesting a statistical association between the T-A-G haplotype and IUGR. Additionally, we identified a rare nonsense variant c.C2050T (p. R684*) exclusive to a male CP patient, and performed functional characterization. In vitro experiments showed that both c.C2050T (p.Arg684*) and the C-terminal frameshift variant c.3413_3414insGG (p.E1140fs*42) may partially escape NMD, producing detectable truncated proteins with variant-specific effects on protein stability, subcellular localization, and cytoskeletal regulation. Our findings support SHROOM4 as an X-linked susceptibility gene associated with male CP.

IUGR, a condition in which the fetus fails to achieve its genetically determined growth potential due to genetic or environmental factors, is a risk factor for CP with affected children showing an ~ 8-fold increased CP risk [46]. Chronic placental insufficiency or utero-placental dysfunction are the common causes of IUGR, leading to insufficient blood flow substrate delivery [47, 48]. Since SHROOM4 is highly expressed in various vascular endothelial cells, it may hypothetically be involved in placental and umbilical vascular development. The observed association between the SHROOM4 T-A-G haplotype and IUGR suggests a potential statistical link between the locus and fetal growth-related outcomes, offering a possible clinical exploratory angle in CP. However, this correlation remains hypothesis-generating and requires further investigation.

The high intolerance of SHROOM4 to LoF variation indicates strong selective pressure and supports its functional importance, consistent with prior reports implicating SHROOM4 LoF variants and copy number variants in NDDs. In our cohort, the male CP patient carrying the rare nonsense variant c.C2050T (p.R684*) presented with periventricular leukomalacia, ID, and spastic CP. This variant has previously been reported in a fetus with agenesis of the corpus callosum and Turner syndrome, suggesting that truncating variation at this position may have functional relevance [33]. In contrast, the c.3413_3414insGG (p.E1140fs*42) variant is more common in the population and retains a greater proportion of functional domains, consistent with a potentially milder functional consequence. Considering our functional findings, transcripts carrying p.R684* and p.E1140fs*42 may partial escape NMD, although further experiments are required to directly confirm pathway involvement. Importantly, both variants still produced detectable truncated proteins with variant-specific effects. p.R684* was associated with more severe disruption of actin stress fibers, while p.E1140fs*42 showed milder cytoskeletal effects and a different subcellular localization. This suggests that the truncation position in SHROOM4 may influence the severity of pathophysiological effects, pointing to variant-position-dependent functional differences.

The identification of SHROOM4-associated T-A-G haplotype and rare functional variants may provide useful information for future genetic risk assessment studies in CP, particularly in individuals with neurodevelopmental comorbidities or IUGR. Given the complex and multifactorial nature of CP, these findings should be interpreted as contributing factors to CP susceptibility. In addition, our findings suggest that altered cytoskeletal regulation and protein stability may represent potential biological processes underlying SHROOM4-associated phenotypes, although further functional studies are required to define the roles.

Our study has several limitations. First, the sample size is relatively modest, which may limit the detection of weaker genetic effects or the performance of more detailed subtype analyses. Second, all participants were recruited from a single Chinese center and consisted predominantly of Han Chinese individuals; therefore, replication in independent cohorts and diverse ethnic populations is necessary to evaluate the generalizability of our findings. Third, information on gestational age, birth weight, and other clinical covariates was not fully available for the control cohort, limiting the capacity to adjust for these factors in the association analyses. Fourth, the observed association between the SHROOM4 T-A-G haplotype and CP/IUGR should be considered hypothesis-generating, and the underlying biological mechanisms remain to be elucidated through functional studies. Fifth, functional experiments were performed in non-neuronal cell lines (HEK293T, HeLa and HepG2). While these models are suitable for dissecting basic molecular mechanisms, they may not fully recapitulate neuronal development; future studies in neuronal or brain organoid models are warranted to better understand the SHROOM4’s role in neurodevelopment. Finally, longitudinal follow-up of cognitive and motor outcomes in variant carriers is needed to fully characterize the phenotypic spectrum and long-term impact of SHROOM4 variants.

Conclusion

In summary, our study provides evidence that SHROOM4 genetic variations are associated with male CP susceptibility, and that altered cytoskeletal regulation may represent a potential mechanism underlying this association. Our findings offer new insights into the potential contribution of X-linked genetic factors to male-biased NDDs; however, further studies are needed to clarify the biological mechanisms linking SHROOM4 variants, IUGR, and CP risk.

Abbreviations

ACMG

American College of Medical Genetics and Genomics

ASD

Autism spectrum disorder

CD

Cytochalasin D

CIs

Confidence intervals

CP

Cerebral palsy

ID

Intellectual disability

IUGR

Intrauterine growth restriction

LD

Linkage disequilibrium

LoF

Loss-of-function

MAF

Minor allele frequency

NDDs

Neurodevelopmental disorders

NMD

Nonsense-mediated decay

ORs

Odds ratios

PCA

Principal component analysis

qPCR

Quantitative real-time PCR

SNVs

Single nucleotide variants

WES

Whole-exome sequencing

WT

Wild-type

Authors’ contributions

Yu Su: conceptualization, investigation, methodology, formal analysis, data curation, investigation and writing - original draft, review and editing; Yiran Xu: conceptualization, data curation, resources; Ye Cheng, and Jingzhou Li: formal analysis, methodology, investigation; Zheng Qi: methodology, resources; Jin Zhang: methodology and formal analysis; Ting Wang, methodology, investigation; Junjie Zhang: data curation and formal analysis; Yunqian Li: investigation and formal analysis; Xiaoyang Wang: resources; Qinghe Xing and Changlian Zhu: conceptualization, supervision, resources, funding acquisition and writing-review & editing.

Funding

This work was supported by the National Natural Science Foundation of China (31972880).

Data availability

All data generated or analyzed during this study are included in this published article. Variant call format (VCF) files for the 1,010 CP cases are available in the Genome Sequence Archive for Human (GSA-Human) under accession number BF2023081113944 (BioProject PRJCA023830). Access to controlled data can be requested via the GSA-Human portal (https://ngdc.cncb.ac.cn/gsa-human/document) and is subject to approval by the Data Access Committee. Correspondence regarding data access may be directed to the corresponding author (qhxing@fudan.edu.cn).

Declarations

Ethics approval and consent to participate

Written informed consent was obtained from parents or the legal guardians of all participants prior to their participation. This study was approved by the Ethics Committee of Zhengzhou University in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.