Mutations in PIKFYVE cause autosomal dominant congenital cataract

Congenital cataract, an ocular disease predominantly occurring within the first decade of life, is one of the leading causes of blindness in children. Through whole exome sequencing of a Chinese family with congenital cataract, we identified a disease-causing mutation (p.G1943E) in PIKFYVE, which affecting the PIP kinase domain of the PIKfyve protein. We demonstrated that heterozygous/homozygous disruption of PIKfyve kinase domain, instead of overexpression of PIKFYVEG1943E in zebrafish mimicked the cataract defect in human patients, suggesting that haploinsufficiency, rather than dominant-negative inhibition of PIKfyve activity caused the disease. Phenotypical analysis of pikfyve zebrafish mutants revealed that loss of Pikfyve caused aberrant vacuolation (accumulation of Rab7+Lc3+ amphisomes) in lens cells, which was significantly alleviated by treatment with the V-ATPase inhibitor bafilomycin A1 (Baf-A1). Collectively, we identified PIKFYVE as a novel causative gene for congenital cataract and demonstrated the potential application of Baf-A1 in treatment of congenital cataract.


Introduction
. To maximally mimic the 134 p.G1943E mutation in human patients, we designed single-guide RNA (sgRNA) in exon 40 to 135 target the C terminal PIPK domain and screened out the pikfyve Δ8 allele, which harbored an 8-bp 136 deletion and a 112-bp insertion in exon 40. As shown in Figure 3A, this pikfyve Δ8 mutation 137 introduced a premature stop codon immediately downstream of the insertion site, thereby 138 presumably generating a truncated Pikfyve protein that lacks the kinase activity. Phenotypical 139 analysis revealed that pikfyve Δ8 homozygous mutants developed normally before 5 days post-140 fertilization (dpf), but later showed severe developmental defects and all died between 7 dpf and 141 9 dpf (Figure 3-figure supplement 1 A and B). Under stereomicroscope, we noticed that 142 compared to siblings (sib), lens of pikfyve Δ8 mutants were less transparent at 5 dpf ( Figure 3B). 143 To characterize this phenotype in more details, we followed up the development of lens in all 144 genotypes at different stages under differential interference contrast (DIC) microscope. Our 145 results showed that bubble-like vacuoles first appeared in the developing lens of pikfyve Δ8 146 homozygous mutants at 3 dpf ( Figures 3C and 3D). Strikingly, the number and size of vacuoles 147 both increased drastically by 5 dpf and almost dominated the lens of mutants. Interestingly, 148 compared to the wild-type (WT) control, pikfyve Δ8 heterozygote mutants also manifested 149 significantly higher number of vacuoles in their lens ( Figures 3C and 3D), suggesting that 150 normal development of lens highly depends on the activity of Pikfyve and heterozygous 151 disruption of Pikfyve is sufficient to cause cataract phenotype. In addition to loss of function 152 study of Pikfyve, we also utilized a ubiquitously expressed ubiquitin (ubi) promoter to generate transgenic lines Tg(ubi:PIKfyve WT ) and Tg(ubi:PIKfyve G1943E ) to overexpress PIKfyve WT and 154 PIKfyve G1943E in zebrafish respectively. As expected, both of these two zebrafish lines showed 155 normal development of the lens during early stages (Figure 3-figure supplement 1 C). 156 Collectively, these findings strongly supported the hypothesis that haploinsufficiency, rather than  To delineate the details of cataract phenotype in pikfyve Δ8 mutants, we crossed pikfyve Δ8 mutant 168 with the reporter transgenic line Tg(cryaa:DsRed), in which expression of DsRed was controlled 169 by promoter of the crystalline gene cryaa, the major constitutive components of lens fibers.

170
Intriguingly, in comparison with the evenly distributed DsRed signals in the lens of siblings, 171 vacuoles in pikfyve Δ8 mutants almost occupied the surface of the lens and all of them were DsRed negative ( Figure 4A), indicating that these vacuoles did not contain lens fibers. Meanwhile, 173 hematoxylin-eosin (HE) staining, together with ZL-1 antibody and DAPI co-staining revealed 174 that while disruption of Pikfyve seemingly had no effects on the enucleation process of lens, lens 175 fibers in pikfyve Δ8 mutants were less organized than those in siblings ( Figures 4B and 4C). To 176 further characterize the vacuoles and lens structure in high resolution, we utilized transmission 177 electron microscope (TEM) to visualize the ultrastructure of lens in both siblings and pikfyve Δ8 178 mutants. We could detect large vacuoles in the mutant lens at 3 dpf and 5 dpf, while only several 179 tiny vacuoles in the sibling lens ( Figure 4D). Lens fiber cells of WT embryos were mildly 180 edema, and their nuclei were oval with uniform chromatin. The lens fibers (white arrows) were 181 arranged neatly and tightly. Furthermore, mitochondria (black triangles) of WT were slightly 182 swollen without vacuoles. In contrast, lens fiber cells of the pikfyve Δ8 mutants, were obviously 183 edematous; and the nuclei were irregularly shaped. Besides, the arrangement of lens fibers was 184 loose and deformed. Also, lipid droplets were formed in pikfyve Δ8 mutants. Compared to the WT, 185 mitochondria (black triangles) of pikfyve mutants were obviously swollen and enlarged.

186
Mitochondrial crests were reduced, and most of them were aberrantly vacuolated. In addition, 187 autophagic lysosomes appeared swollen and vacuolated.

Vacuoles in pikfyve Δ8 mutants were amphisomes
To define the nature of vacuoles in the lens of pikfyve Δ8 mutants, we conducted time-lapse 191 imaging to monitor their behaviors during zebrafish development. Our results showed that 192 vacuoles in pikfyve Δ8 mutants were highly dynamic and small vacuoles were frequently found to 193 be fused with each other to form larger vacuoles ( Figure 5A, white arrows). This feature, 194 together with previous findings showing that PIKfyve was an essential regulator of 195 endomembrane homeostasis (Hasegawa, Strunk, & Weisman, 2017), prompted us to further 196 investigate whether vacuoles in pikfyve Δ8 mutants were actually endocytic vesicles. To probe this 197 issue, we generated fusion mRNAs encoding GFP and the markers of endocytic vesicles (i.e., the 198 small GTPase Rab5c, Rab7 and Rab11a), which were injected into zebrafish embryos to 199 specifically label early, late and recycling endosomes in pikfyve Δ8 mutants. Our results showed 200 that in pikfyve Δ8 mutants, almost all vacuoles were positive for the late endosome maker Rab7-201 GFP ( Figure 5C, white arrows). By contrast, the early endosome marker Rab5c-GFP and 202 recycling marker Rab11a-GFP showed no co-localization with vacuoles ( Figures 5B-5D). As 203 autophagy was also closely-related to the organelle membrane system and had been shown to be 204 regulated by PIKfyve (Vicinanza et al., 2015), we therefore checked the status of 205 autophagosomes in pikfyve Δ8 mutants by injecting lc3b-mcherry fusion mRNA. As shown in 206 Figure 5E, a proportion of vacuoles in pikfyve Δ8 mutants were also positive for the 207 autophagosome marker Lc3b. Taken together, these data implied that vacuoles in pikfyve Δ8 208 mutants were amphisomes formed by the fusion of autophagosome and late endosome. Baf-A1 is a specific inhibitor of V-ATPase. Previous work has shown that the vacuole phenotype 213 induced by PIKfyve deficiency in COS-7 cells could be rescued by Baf-A1 (Compton,214 Ikonomov, Sbrissa, Garg, & Shisheva, 2016). We thus investigated whether the vacuolation 215 defect in the lens of pikfyve Δ8 mutant zebrafish could also be rescued by Baf-A1. Indeed, we 216 found that vacuole number in the lens of pikfyve Δ8 mutants treated with 1 µM Baf-A1 was 217 significantly lower than that in the control group treated with dimethyl sulfoxide (DMSO; 218 Figures 6A and 6B). To further validate that Baf-A1 could directly alleviate cataract defect, 219 rather than just delay the phenotype, we imaged the lens of the same mutant zebrafish before and 220 after Baf-A1 treatment. As shown in Figures 6C and 6D, while vacuole numbers of all the 221 mutant embryos showed a slight increase after DMSO treatment or without treatment, we 222 observed significant decrease in vacuole number in all mutants with mild or severe phenotype.

223
On the other hand, a previous study also showed overexpression of transient receptor potential

232
In this study, we identified a missense mutation (p.G1943E) in PIKFYVE responsible for 233 congenital cataract in a Chinese Korean family (Figures 1 and 2). This mutation is very rare in 234 the general population. In the gnomAD database, only 4 individuals (including 3 East Asians and 235 1 Latino/Admixed American) carry the p.G1943E mutation, representing an extremely low allele 236 frequency of 0.00002 ( Table 2). The degree and morphology of lens opacity in the cataract 237 family were phenotypically heterogeneous ( Table 1), which is consistent with previous findings 238 (Berry, Ionides, et al., 2020). Most of the patients in this family developed cataract and vision 239 loss in their childhood. Meanwhile, the majority of patients had nuclear pulverulent cataract, 240 while the others had nuclear Y-sutural cataract or peripheral cortical punctate cataract (Table 1). 241 The heterogeneity in clinical manifestations might be ascribed to interactions between genetic 242 and environmental factors during lens development (Berry, Ionides, et al., 2020). 243 Mutations in PIKFYVE have been reported to be associated with CFD (Gee et al., 2015;244 Kawasaki et al., 2012;Kotoulas et al., 2011;S. Li et al., 2005). However, none of the 245 PIKFYVE mutations in these studies caused congenital cataract in addition to CFD, except two  Accordingly, we did not observe any cornea defect in the zebrafish mutants either.

264
Inhibition of PIKfyve in COS-7 cells had been shown to induce large vacuoles through 265 promoting the enlargement of both early and late endosomes (Ikonomov, Sbrissa, & Shisheva, 266 2006;Rutherford et al., 2006). Further studies revealed that the Ca 2+ releasing channel, 267 endolysosome-localized mucolipin TRPML1 acts downstream of PIKfyve to trigger membrane 268 fusion/fission process (Dong et al., 2010) or promote lysosome/phagosome maturation 269 (Dayam, Saric, Shilliday, & Botelho, 2015;Kim, Dayam, Prashar, Terebiznik, & Botelho, 2014). Interestingly, while overexpression of TRPML1 partially alleviated the vacuole 271 phenotype in PIKfyve-deficient macrophages (Krishna et al., 2016), the rescue effect was not 272 observed in the lens of pikfyve Δ8 zebrafish mutants ( Figure 6figure supplement 1), 273 suggesting that PIKfyve might function in a context dependent manner. Consistent with this idea, 274 we also noted that vacuoles in macrophages and lens cells of pikfyve Δ8 mutants were differently 275 stained by lysosome marker (Figure 5figure supplement 1). On the other hand, vacuole 276 formation in PIKfyve-deficient COS-7 cells and macrophages could also be inhibited by a drug 277 called Baf-A1 (Compton et al., 2016;Isobe et al., 2019). In this study, we found that Baf-A1 278 could partially rescue the vacuole defect in the lens of pikfyve Δ8 mutant zebrafish ( Figures 6A-279   6D). Baf-A1 is a macrolide antibiotic that inhibits V-ATPase, the ATP-dependent proton pump 280 located on the membrane of organelles. The cellular acidification process mediated by V-ATPase 281 may affect many basic biological processes, including membrane trafficking (in particular 282 endosome maturation and fusion between autophagosomes and lysosomes (Hammond et al., 283 1998;Yamamoto et al., 1998), protein degradation and autophagy (Bowman, Siebers, & 284 Altendorf, 1988;Yamamoto et al., 1998;Yoshimori, Yamamoto, Moriyama, Futai, & 285 Tashiro, 1991). Interestingly, several works in yeast also identified mutations in V-ATPase that 286 did not affect proton pump function, but indeed caused defects in vacuole fusion (Strasser, 287 Iwaszkiewicz, Michielin, & Mayer, 2011). Moreover, another study in drosophila 288 demonstrated that inhibition of vesicle fusion by Baf-A1 did not depend on V-ATPases, but 289 relied on Ca 2+ sarco/endoplasmic reticulum Ca 2+ -ATPase (SERCA) pump, the secondary target 290 of Baf-A1 (Mauvezin, Nagy, Juhasz, & Neufeld, 2015). Therefore, the target of Baf-A1 and 291 underlying mechanisms still need to be further investigated. These mechanisms will finally 292 contribute to the potential clinical application of the drug in the treatment of congenital cataract. A 4-generation pedigree consisting of 31 family members ( Figure 1A)      Zebrafish embryos were fixed in 4% PFA at 5 dpf and then dehydrated in 30% sucrose. The 372 whole embryos were mounted in optimal cutting temperature compound and frozen in -80°C.

388
Coding sequences of rab5c, rab7 or rab11a were fused with GFP sequence and cloned into the       Figure 6-source data 1. Raw data for quantification in Figure 6B and 6D.
695 Figure 6 supplement 1-source data 1. Raw data for quantification in Figure 6B.