Eye movement defects in KO zebrafish reveals SRPK3 as a causative gene for an X-linked intellectual disability

Intellectual disability (ID) is a common neurodevelopmental disorder characterized by significantly impaired intellectual and adaptive functioning. X-linked ID (XLID) disorders, caused by defects in genes on the X chromosome, affect 1.7 out of 1,000 males. Employing exome sequencing, we identified three missense mutations (c.475C>G; p.H159D, c.1373C>A; p.T458N, and c.1585G>A; p.E529K) in the SRPK3 gene in seven XLID patients from three independent families. Clinical features common to the patients are intellectual disability, agenesis of the corpus callosum, abnormal smooth pursuit eye movement, and ataxia. SRPK proteins are known to be involved in mRNA processing and, recently, synaptic vesicle and neurotransmitter release. In order to validate SRPK3 as a novel XLID gene, we established a knockout (KO) model of the SRPK3 orthologue in zebrafish. In day 5 of larval stage, KO zebrafish showed significant defects in spontaneous eye movement and swim bladder inflation. In adult KO zebrafish, we found agenesis of cerebellar structures and impairments in social interaction. These results suggest an important role of SRPK3 in eye movements, which might reflect learning problems, intellectual disability, and other psychiatric disorders.


INTRODUCTION
Intellectual disability (ID) affects approximately 1-3% of the general population [1]. Males exceed 52 females in the ID population by 20-30%, likely due to an enrichment of genes on the X-  Recent estimates suggest that the number of genes linked to ID has risen to over 1,500 genes 63 [4]. The availability of a complete sequence for the X chromosome, coupled with the capability of 64 next-generation sequencing and functional analysis, presents an exciting opportunity to identify 65 the genetic basis for at least one-third of this group of disorders for which the causative gene has 66 not yet been identified. 67 Through exome sequencing, we initially identified a missense mutation in the SRPK3 gene 68 (serine/arginine-rich protein-specific kinase 3) in an XLID family. The mutation was maternally 69 inherited from a Manitoba mother of Anglo-Saxon heritage. Subsequently, we became aware of 70 a second mutation identified by the Undiagnosed Disease Program (UDP) at NIH, USA. Given 71 the presence of two missense mutations in SRPK3, we conducted a search of a database 72 generated by our previous Sanger X-chromosome resequencing project [5], which revealed a 73 third missense mutation. The combination of these three missense mutations made SRPK3 a 74 compelling candidate gene for in-depth analysis as a novel XLID gene. SRPK3 is known to 75 specifically phosphorylate serine-arginine (SR) proteins which act as splicing factors [6]. 76 Phosphorylation is required for SR proteins to enter the nucleus and play a role in alternative 77 splicing of pre-mRNA, mRNA export, and other processing events. Recently, new functions of 78 SRPK2 were reported in synaptic vesicle and neurotransmitter release [7]. Although our 79 bioinformatic analysis predicted that the three variants were likely pathogenic and related to XLID 80 3 phenotypes in the families, further investigation was required to confirm these predictions and 81 elucidate the molecular mechanisms underlying the observed XLID phenotype. 82 In order to investigate SRPK3 as a novel XLID gene, a knockout zebrafish model of srpk3 was 83 generated and evaluated at different developmental stages to understand how SRPK3 deficiency 84 adversely affected its function in vivo. 85

87
Further details about the procedures are described in the Supplementary Methods. 88 The human subject research protocol for the study was approved by the Institutional Review 89 Boards (IRBs). Informed consent was obtained from each study patient and/or their parents or 90 legal guardians. Patients enrolled in the Greenwood Genetic Center XLID study were evaluated 91 by clinical geneticists and other specialists and underwent comprehensive laboratory studies for 92 ID. All patients were found to have a normal karyotype and negative molecular testing for Fragile 93 X syndrome. Given that each center independently identified SRPK3 as the novel candidate gene 94 for ID, the enrichment kits used and sequencing protocols varied between centers.

95
Family 1 (K8765, GGC) 96 Patients with XLID and normal control males were recruited at Greenwood Genetic Center (GGC), 97 USA. X-chromosome exome sequencing on probands from XLID families was conducted using a    The gene sequences for srpk3 were obtained from the NCBI database, and primers were 142 designed for in vitro transcription of sgRNAs targeting srpk3. Cas9 expression vector pT3TS-143 nCas9n was used to synthesize Cas9 mRNA, which was injected along with sgRNAs into one-144 cell stage zebrafish embryos. Mutations were validated using T7E1 assay, and founder (F0) fish 145 were raised to adulthood and out-crossed with wild-type zebrafish to generate germ line mutations.

146
The F1 generation was raised to adulthood, and individuals with the heterozygous genotype were 147 in-crossed to produce stable srpk3 KO zebrafish lines (F2) [14].

149
Zebrafish larvae were maintained at a constant temperature of 28.5°C until they reached 5 dpf, 150 which is a stage commonly used for spontaneous eye movement analysis in zebrafish. Zebrafish 151 larvae at 5 dpf were mounted on a petri dish using 2% low melting agarose and a 3-minute video 152 of their eye movements was recorded using a brightfield microscope. The recorded video was 153 then analyzed using MATLAB software to track and analyze the eye movements of the zebrafish   156 In the social interaction assay [16], behavior of zebrafish was observed in a tank divided into two 157 sections. One section was designated for the cue fish and the other for the tester fish. The tank 158 was divided into four equal chambers (zone 1, 2, 3, and 4); the zone nearest to the social cue 159 was designated zone "1", the second nearest zone "2", the third zone "3", and the last zone "4". 160 During the experiment, three adult zebrafish were added as cue fish and a single srpk3 KO sibling 161 5 was added as the tester fish. A video was taken for 15 minutes and the behavior was analyzed 162 using Ethovision XT software.

163
Statistical analysis 164 Statistical analysis was performed using Graphpad Prism 8. Data were expressed as mean ± 165 standard error of the mean (SEM) with a p value of < 0.05 being considered significant. For 166 parametric measures of two groups, a two-tailed T-test was performed and for more than two 167 groups, One-Way ANOVA with Tukey's post-test was performed. convergence insufficiency, lazy eye, and/or poor attention span, were also clinically described in 181 the four patients (Table S1). Bioinformatic analysis predicted that the p.H159D, p.T458N, and     carboxyl group decreased to 39 contacts due to H159D substitution (Fig. 3B). The T458N 246 mutation appeared to cause steric hindrance with neighboring bulky residues such as Phe454, 247 Tyr528, and Trp530, presumably resulting in disturbance of SRPK3 protein folding (Fig. 3C). In 248 contrast, the side chain of Glu529 was exposed to the outside of the protein, and thus its 249 substitution to lysine seems not to have affected protein stability. Instead, we noticed that SRPK1 250 and SRPK3 share a high structural similarity with each other with a root-mean-square deviation 251 value of 0.73 Å over 343 aligned Cα atoms, and Glu529 was found to be located adjacent to 252 SRPK1-bound protein fragment in the superposed model (Fig. 3D). Therefore, we hypothesize 253 that E529K substitution led to alteration of protein shape and surface charge which ultimately 254 affected protein-protein interaction-mediated target protein recognition of SRPK3 (Fig. 3E).

255
Generation of srpk3 KO zebrafish lines using CRISPR-Cas9 system 256 The human homologue of SRPK3 is located on Xq28, while in zebrafish, srpk3 homolog is located 257 on chromosome 8. We performed a ClustalX 2.1 alignment of the amino acid sequences of both 258 homologues, which revealed a high degree of conservation (Fig. S3). Specifically, the human 259 homologue contains 567 amino acids, while the zebrafish homologue contains 701 amino acids. including body length, pigment patterns, and eye size were analyzed. Although we did not find 271 any significant defects in these aspects, we observed that the majority of KO zebrafish failed in 272 swim bladder inflation (Fig. 3G, H). Swim bladder inflation is an early marker for survival in larval 273 zebrafish, as it is a vital organ which enables fish to swim. In total, 2,367 larvae from 10 274 independent clutches were examined. The incidence of swim bladder defects was decreased from 275 23.4% to 4.3% in subsequent generations (Table S2). 6 out of 73 (8%) adult srpk3 KO zebrafish 276 survived into adulthood (Fig. S4). Therefore, the lack of swim bladder inflation in srpk3 KO 277 zebrafish was considered a significant feature which prompted molecular marker analysis. zebrafish compared to the wild type. We also analyzed the motoneuron marker islet 1 and the 283 dopaminergic neuronal marker tyrosine hydroxylase (th) in KO zebrafish, but again we did not 284 observe significant changes in homozygous KO zebrafish compared to its wild type siblings ( Fig.   285 3I-L). Additionally, we examined the expression of other neuronal markers (phox2b, ascl1, dlx2), 286 cell cycle marker (ccdn1), and muscle markers (bin1b, ttnb) but did not detect significant 287 differences in expression levels at early developmental stages between KO and wild type siblings 288 (Fig. S6). Thus, the introduction of genome-scale analysis is needed to identify underlying in eye movement frequency ( Fig. 4; Movie S1), indicating its involvement in maintaining the 298 pattern of movement. Although most wild types displayed robust synchronized movements in both 299 eyes, KO zebrafish showed a reduction in ocular angle (Fig. 4C). To further investigate this, we 300 analyzed one spontaneous movement pattern in WT and KO zebrafish and found a significant 301 reduction in reset time (Fig. 4D). Next, we examined whether KO zebrafish lost their visual acuity.

302
In a color preference test, which we had previously developed [20], KO zebrafish showed normal 303 visual activity of color preference at 5 dpf, compared to WT siblings (Fig. S7). Taken together, we  (Table S1) with poor attention span in 317 two patients (F1:III-4 and F1:III-5) and eye movement defects in two patients (F2 and F3:III-1), 318 we challenged srpk3 KO zebrafish to social interaction testing (Fig. 5). As a result, we observed 319 that srpk3 KO adult zebrafish displayed defects in social interaction, especially at the 10-15-320 minute time point (Fig. 5A). We analyzed the behavior of fish at two time points: early phase (4-5 321 min) and late phase (10-15 min) of the session. During early phase, both srpk3 KO and control 322 WT fish showed social interaction behavior by staying in zone 1, close to the social cue fish.

323
However, at late phase srpk3 KO fish started to lose this socially interactive behavior, which might 324 reflect the "poor attention span" exhibited by some of the patients. To investigate social interaction 325 in zebrafish, we divided the tank into four zones (1, 2, 3, and 4). One section of the tank contained 326 the test fish, including srpk3 wild type or KO sibling, while the other section contained social cue 327 fish. Three wild type fish were used as social cue. A transparent acrylic plate separated the two 328 sections to allow for fish interaction. Our results showed that WT siblings tended to stay in zone 329 1 and interacted with the cue fish frequently. In contrast, srpk3 KO zebrafish showed significantly 330 less interaction with the cue fish and instead explored the whole tank, including zones 2, 3, and 331 4 ( Fig. 5C-E). Based on these observations, we hypothesized that impaired social interaction may 332 be due to some neurological defects in the brain, although we could not detect significant changes 333 9 in the expression of neuronal makers at early developmental stages. Therefore, we proceeded 334 with brain sectioning to analyze anatomical defects in the zebrafish adult brain. 336 Although we observed impaired social interaction in adult srpk3 KO zebrafish, we could not find 337 significant difference in growth and body size between WT and srpk3 KO siblings. We then tried 338 to examine whether brain size was affected in srpk3 KO zebrafish, given that we had previously 339 observed reduced brain size in a zebrafish autism model [16]. We wanted to identify how srpk3 340 was involved in maintaining anatomical structures of the brain. In dissected adult whole brains, 341 we did not observe any significant difference in the overall size of dissected brains between WT 342 and srpk3 KO siblings (Fig. 6A,B). Next, we performed brain sections of 7 µm in wild type and KO 343 zebrafish, performed H & E staining, and examined the brain structures in detail. To our surprise, 344 we noticed a significant reduction in the specific brain region, so called the valvular cerebelli in 345 fish, which is suggested to be functionally equivalent to the pontine nuclei in mammals (Fig. 6C-346 D"'). ImageJ analysis showed that the relative size of valvular cerebelli in the whole brain of srpk3 347 KO zebrafish was reduced to 50%, 59%, and 41% from that of WT zebrafish in three different  In this study, we provided multiple lines of evidence for the role of SRPK3 in XLID. We identified 352 novel SRPK3 mutations segregating with the phenotype in three independent families. Using AI-353 based structural protein modelling, we described the effects of the missense mutations on SRPK3 354 structure and how they could disrupt normal SRPK3 functioning. We described for the first time 355 early human embryonic and fetal brain expression patterns of SRPK3 and its correlation to the 356 brain phenotypes observed in the neuroimaging of our patients. Finally, using histological analysis, 357 social interaction assays, and modelling oculomotor behavior, we recapitulated the human 358 phenotype in a novel srpk3 zebrafish knockout model.

359
One critical cellular activity is processing of pre-mRNAs which is essential for gene expression  future studies that selectively probe valvular cerebelli activity will provide further insights into the       of srpk3 in developing zebra sh embryo showing transcripts of srpk3 being present at 24 hpf in the brain, heart and muscles, and in the brain and retinal layers at 72 hpf. Scale bars: 1 mm for A-C, 100 µm for D.

Figure 3
Molecular modeling of SRPK3 and generation of srpk3 zebra sh KO line. A-E Molecular modeling of SRPK3 variants. A Side chains of three amino acids involved in XLID are indicated as spheres on the SRPK3 structure modeled using AlphaFold. Unstructured loop regions of SRKP3 are omitted for clarity, including residues 1-46 and 310-385. B, C Interior misfolding due to H159D and T458N variations. SRPK3 wild-type (left) and the H159D (B, right) or T458N (C, right) mutant form modeled based the AlphaFold structure prediction are shown together. DStructural superposition of the AlphaFold-modeled SRPK3 (green) onto the crystal structure of ICP27 (red)-bound SRPK1 (Navy; PDB code 6FAD). Side chain atoms of SRPK3 Glu529 are presented as spheres. E Alteration of protein shape and surface charge due to E529K variation. SRPK3 wild-type (left) and the E529K mutant form (right) are shown as sticks and loops (top) or electrostatic surface representation (bottom). F-L Characterization of srpk3 zebra sh KO line. F Disruption of protein domains in KO zebra sh while wild type zebra sh show intact srpk3 protein. G, H KO zebra sh fail to develop swim bladder in ation at 5 dpf when compared to wild type siblings (red arrow). I-L Expression analysis of neuronal markers (neurogenin 1, her 4, islet 1 and th) in developing zebra sh embryos. Scale bars are 100 µm.

Figure 4
Analysis of spontaneous eye movements in KO zebra sh. A Individual representation of spontaneous eye movement in wild type (WT) and srpk3 KO siblings. Red lines represent right eye and blue lines represent left eye of 5 dpf zebra sh larvae. B Eye movement frequency in the left (L) and right (R) eye of srpk3 KO zebra sh when compared to its WT siblings. C Ocular angle signi cantly decreases in both left and right eye in KO zebra sh, compared to WT siblings. D srpk3KO zebra sh take signi cant longer time to reset in slow phase when compared to their WT siblings during spontaneous eye movement. Video recording of WT (n = 6) and KO (n = 9) zebra sh for 3 minutes. All data represented as mean ± SEM. *p < 0.05, ** p < 0.01, *** p < 0.001 by t test.