ALPK1 missense pathogenic variant in five families leads to ROSAH syndrome, an ocular multisystem autosomal dominant disorder

Purpose To identify the molecular cause in five unrelated families with a distinct autosomal dominant ocular systemic disorder we called ROSAH syndrome due to clinical features of retinal dystrophy, optic nerve edema, splenomegaly, anhidrosis, and migraine headache. Methods Independent discovery exome and genome sequencing in families 1, 2, and 3, and confirmation in families 4 and 5. Expression of wild-type messenger RNA and protein in human and mouse tissues and cell lines. Ciliary assays in fibroblasts from affected and unaffected family members. Results We found the heterozygous missense variant in the ɑ-kinase gene, ALPK1, (c.710C>T, [p.Thr237Met]), segregated with disease in all five families. All patients shared the ROSAH phenotype with additional low-grade ocular inflammation, pancytopenia, recurrent infections, and mild renal impairment in some. ALPK1 was notably expressed in retina, retinal pigment epithelium, and optic nerve, with immunofluorescence indicating localization to the basal body of the connecting cilium of the photoreceptors, and presence in the sweat glands. Immunocytofluorescence revealed expression at the centrioles and spindle poles during metaphase, and at the base of the primary cilium. Affected family member fibroblasts demonstrated defective ciliogenesis. Conclusion Heterozygosity for ALPK1, p.Thr237Met leads to ROSAH syndrome, an autosomal dominant ocular systemic disorder.


INTRODUCTION
Retinal dystrophy describes progressive degeneration or abnormality of photoreceptors (rods and cones) leading to vision loss, with a worldwide prevalence of approximately 1 in 3500 (ref. 1 ). Cases may be isolated or syndromic with a number of additional organs involved. Syndromal forms include metabolic conditions such as Refsum disease (MIM 266500), storage disorders such as neuronal ceroid lipofuscinosis 3 (MIM 204200), and conditions where there are primary ciliary abnormalities including Bardet-Biedl (MIM 209900), Senior-Løken (MIM 266900), and Joubert syndromes (MIM 213300) (ref. 2 ). Each of these disorders has specific clinical and/ or laboratory features that facilitate diagnosis, and are usually associated with autosomal recessive inheritance.
We previously described, in Tantravahi et al., 3 a family with an autosomal dominant mode of inheritance of a unique constellation of ocular and systemic features that presented to the ophthalmology and hematology clinics of the University of Utah. The features common to all three affected individuals in this family included a severe retinal dystrophy of the cones and rods, optic nerve edema, mild pancytopenia, and idiopathic massive splenomegaly. Two of the affected individuals had anhidrosis and migraine headaches, with one also experiencing episodic urticaria. Despite extensive literature search and discussion with colleagues, no other similarly affected individuals were identified at that time.
Here, we describe the clinical features of four additional unrelated affected families, allowing us to phenotypically define this new ocular systemic disorder, which we call ROSAH syndrome. ROSAH syndrome is named due to the predominant features in affected individuals including retinal dystrophy, optic nerve edema, splenomegaly, anhidrosis, and migraine headache. Utilizing different genomic technologies and analysis pipelines, we link the disease-causing variant in ALPK1 (ɑ-kinase 1 [MIM 607347], c.710C>T, [p.Thr237Met]) to ROSAH syndrome in all five families. ALPK1 is a member of a class of atypical protein kinases and our cellular, expression, and functional studies suggest a critical role in centrosome and cilia biology. This study defines the genetic cause of ROSAH syndrome.

MATERIALS AND METHODS
The study protocol at the University of Utah for phenotyping and analysis of patients in the Utah, Virginia, and Delaware cohorts was approved by the Institutional Review Boards of the University of Utah. The Australian patients were enrolled in a research project approved by the ethics committee of The Children's Hospital at Westmead, Sydney Children's Hospital Network, Sydney, Australia. The patient from the Netherlands was examined in a diagnostic setting. These projects conform to the Declaration of Helsinki, and written, informed consent was obtained from all participants and from the parents of minors.
All study subjects underwent complete ophthalmic and systemic examinations as dictated by the clinical needs of each. Unaffected subjects also received a complete exam for the purpose of phenotyping, and examination locations are detailed in Supplementary Methods: Clinical Examination Locations. For all families, whole blood was obtained and genomic DNA was extracted from peripheral lymphocytes according to previously described protocols. [4][5][6] Exome sequence analysis: family 1 (Utah) Exome sequencing was performed at the University of Utah on four members of the Utah cohort: the proband (III-4, in family 1, Fig. 1a), her unaffected sister (III-5), her affected mother (II-3), and her unaffected father (II-4).
After exome sequence alignment and processing (Supplementary Methods) the exome data of the family quartet was analyzed using VAAST (v2.0) (ref. 7 ) by comparing the exomes of the affected mother and daughter with a control data set consisting of 54 genomes from the Complete Genomics Diversity Panel 8 and 1057 nonduplicative exomes from the phase I release of the 1000 Genomes Project. 9 The analysis was restricted to the subset of variants that were present in affected mother and daughter and absent in the father and unaffected daughter. Given the extreme nature of the phenotype and rarity of the disease, all variants in our control data set and in the National Heart, Lung, and Blood Institute (NHLBI) Exome Sequencing Project (ESP) were excluded. The initial analysis resulted in a set of 44 candidate genes and none of the identified variants were located in genes known to be associated with ocular diseases such as retinitis pigmentosa or cone-rod dystrophy. Variants that were predicted to cause severe changes in protein function, were located in genes known to be expressed in retina and spleen, or were located in genes located in pathways not likely to be lethal were given highest priority. After this analysis, of 23 potential disease-causing variants, 5 were sequencing artifacts and of the others, only the ALPK1 variant segregated appropriately with disease. The ALPK1 variant was confirmed as a de novo pathogenic variant in the proband's mother (family I, II-3, Fig. 1a), using the following primers: forward primer 5′-GGCAAA-TAGTTCATATGGAGGAA-3′ and reverse primer 5′-CCTCCCAATCTCCTGAGAAA-3′.
Genome sequence analysis: family 2 (Australia) Independent genome sequencing was performed for all nine members of the Australian cohort (family 2, Fig. 1a). After genome sequence alignment and processing (Supplementary Methods) a variant call list was generated. Phen-Gen 10 was used for post-variant call analysis. Phen-Gen incorporates disease symptoms to use the phenotype to genotype link to improve the likelihood of novel disease gene identification. Phen-Gen discards common variants (minor allele frequency [MAF] >1% in dbSNP, 1000 Genomes, or ESP) and estimates the deleterious impact of pathogenic variants, allowing for a direct comparison of putative damaging variants across different classes (nonsynonymous, splice site, and indels). It further prioritizes the candidate gene list based on patientspecific symptoms by integrating available disease knowledge of phenotypically similar disorders, the genes implicated for them, and other functionally similar genes. Symptoms listed in Table 1 were used in the analysis. There were two candidate variants suggested by the analysis, but one was of low genotype quality and did not validate with Sanger sequencing. Mutation validation of the ALPK1 pathogenic variant and segregation with disease was shown by Sanger sequencing in all family members (Macrogen Inc., Seoul, Korea) with the following primers for ALPK1 (NM_001102406): forward 5′-TGATCCTTCAAGCTCTGTGGC-3′, and reverse 5′-CCACC AAGGACACGATTGACA-3′.

Exome sequence analysis: family 3 (Netherlands)
Diagnostic exome sequencing was performed in the proband and his parents in the Netherlands cohort. After exome sequence alignment and processing (Supplementary Methods) variants were filtered with the Cartagenia software package on quality (read depth ≥10), MAF (≥0.1% in 200 alleles in dbSNP, ESP6500, 1000 Genomes Project, GoNL, or ExAC), location (within an exon or first/last 3 bp of introns). Subsequently variants were further selected based on three inheritance models (de novo autosomal dominant, autosomal recessive, and X-linked recessive). Only variants predicted to change the protein sequence (nonsynonymous single-nucleotide variants and short insertions and deletions) or in the recognition of consensus RNA splice sites were considered. Ninety-six percent of the RefSeq sequences were covered at least 10×. Confirmation of the ALPK1 variant was performed by Sanger sequencing as described for the Utah family.   Table S1). Cells were incubated with appropriate secondary antibodies (Supplementary  Table S1) for 2 hours at room temperature. Images were captured using a Zeiss LSM 880 laser scanning microscope with Airyscan. Super-resolution images were captured using 63×/1.4 Oil Plan-Apochromat objective and processed using Zeiss ZEN Black software. For ciliation and multinucleation studies, data were obtained from three independent replicates where 100 cells were examined for each sample in each experiment. Statistical analysis of the results was performed using one-way analysis of variance (ANOVA) (GraphPad Prism software) and p values were calculated by the two-tailed t test.

Immunohistofluorescence
Adult female ARC(S) mouse eyes were collected, and the corneas were removed. The eyes were embedded in optimal cutting temperature compound (Tissue-Tek, cat #4583) and  snap frozen in liquid nitrogen. The paws were collected and fixed in 4% paraformaldehyde for 1 hour. Samples were then placed in 20% sucrose followed by 30% sucrose and embedded in optimal cutting temperature compound. Tissue samples were sectioned at a thickness of 14 µm. The tissue sections were blocked in a buffer consisting of 0.1% fish gelatin (Sigma Aldrich, cat #G7765), and 0.02% Triton X-100 in phosphate-buffered saline for 1 hour. Tissue sections were incubated with primary antibodies (Supplementary Table S1) overnight at 4 ºC, washed, and incubated with appropriate secondary antibodies for 2 hours. Images were captured using a Zeiss LSM 880 confocal microscope.

Clinical phenotype of ROSAH syndrome
There is marked similarity in the clinical features we characterized in affected individuals from all five of these unrelated families (Fig. 1a, Table 1, Fig. 2). We propose the name ROSAH syndrome (retinopathy, optic nerve edema, splenomegaly, anhidrosis, and migraine headache), due to the consistency of the ophthalmic, hematologic, and systemic signs, and detailed clinical case reports are provided in the Supplementary Clinical Information Appendix. The most frequent earliest feature is decreased vision associated with presence of optic nerve edema on ophthalmic examination, evident as early as 4 years of age and present by age 12 years in all of the cases identified so far (Tantravahi et al., 3 Table 1, Fig. 2a, l). This is not associated with elevated intracranial pressure. Low-grade ocular inflammation is common in affected individuals, and does not resolve with steroid or other forms of immunosuppression. Later in childhood or the second decade of life, patients have increasing visual impairment and most have abnormal cone function and later loss of rod function (Fig. 2f). By the third decade of life, visual dysfunction is severe with visual acuity ranging from counting fingers to no light perception. In addition to the ocular findings, patients generally have splenomegaly that may begin in childhood or the second decade, and may eventually require splenectomy. Hematological assessment shows chronic mild pancytopenia, which may be exacerbated during periods of viral or other infection. Peripheral smears frequently show leukopenia with reactive lymphocytosis. Bone marrow examination is normocellular with trilineage hematopoiesis and relative erythroid hyperplasia. Affected individuals also have anhidrosis. Headaches and migraine are a frequent feature, often associated with episodic fever and back pain. Mild renal impairment was present in some members of family 2 (Fig. 1a), but is not a consistent feature in all patients.
Heterozygous ALPK1 pathogenic variant identified in individuals with ROSAH syndrome Independent exome sequencing in four members of the Utah cohort (family 1, Fig. 1a), three members of the Netherlands family (family 3, Fig. 1a), and genome sequencing in all affected and unaffected family members of the Australian cohort (family 2, Fig. 1a) was conducted. Comparison of variants, also conducted independently, between affected and unaffected individuals led to the identification of a missense change in ALPK1, (Chr. 4q25, c.710C>T, [p.Thr237Met]) ( Fig. 1b-d) as the likely disease-causing variant in each of the families. This variant arose de novo in the Utah (family 1) and Netherlands (family 3) families, and comparison of available haplotype data between the Utah and Australian (family 2) families was not consistent with a recent shared common ancestor for these two families. Two additional unrelated families with ROSAH syndrome were identified (family 4 [Virginia] and family 5 [Delaware]), and affected members were found to be heterozygous for the ALPK1, (c.710C>T, [p. Thr237Met]) variant, which arose de novo in the Delaware patient. All families were of non-Hispanic, Caucasian ancestry. This same C to T transition in all five unrelated families, affected the bp 710-711 CG dimer (Fig. 1b) and may represent a hotspot for mutation due to uncorrected spontaneous deamination of a methylcytosine to thymine, which is a known source of mutation in humans. 12 Bioinformatic analysis with PolyPhen-2 (ref. 13 ) (HumVar score 0.953) predicted the ALPK1, (c.710C>T, [p. Thr237Met]) variant to be probably damaging, and SIFT 14 predicted it to be damaging. The variant was highly conserved across species, with a PhastCons 15 score of 0.986 (Fig. 1e), and was absent from the population database gnomAD. The Thr237 amino acid is in a region of poorly defined function in the protein, N terminal to the ɑ-kinase domain (Fig. 1d).

ALPK1 expression in the RPE, photoreceptors, and other tissues
To further evaluate the possible role of the ALPK1 variant in this disease, expression levels of ALPK1 were assessed via quantitative real-time PCR in human donor eye tissue as well as in FFPE spleen tissue from two affected individuals (family 1, II-3 and III-2) along with two age-matched controls. All donor eye tissue was processed at 6 hours or less postmortem time as previously described. 16 In the eye, ALPK1 expression was evaluated in the optic nerve, macula, RPE, and extramacular retina. Pooled RNA from two normal eyes was reverse transcribed and real-time PCR was performed. The resulting data demonstrates increased expression of ALPK1 in the RPE and optic nerve compared with retina (Fig. 3a). Realtime PCR performed on spleen tissue samples confirmed the expression of APLK1 in the normal spleen (data not shown). However, the quality of the RNA obtained from the FFPE tissue was too low to effectively compare diseased tissue samples with normal samples.
In the mouse retina, confocal microscopy showed that Alpk1 localized to the region of the photoreceptors where the basal bodies of connecting cilia are located and to the RPE, with some additional distribution across the inner and outer plexiform layers (Fig. 3b, c). In the connecting cilium region, costaining with a pancentrin antibody 17 indicated localization of Alpk1 to the basal body of the photoreceptor cilium and adjacent centriole region (Fig. 3c-f). In mouse skin, Alpk1 was present in the sweat glands, especially in the myoepithelial cells where there was colocalization with the myoepithelial cell marker α-SMA (Fig. 3g-r). There was also broad expression in other mouse tissues (Supplementary Figure S1).

ALPK1 subcellular localization in centrosomes, spindle poles, and primary cilia
Our immunofluorescence studies on ARPE19 cells revealed the localization of ALPK1 in the spindle poles at metaphase (Fig. 4a-f), and in centrosomes of cells during interphase (Fig. 4g-i). Furthermore, cells were serum starved for 24 hours  and then stained with ALPK1 antibody and acetylated αtubulin, which is a marker for primary cilia. Super-resolution confocal imaging showed the presence of ALPK1 in the base of primary cilia (Fig. 4j-l). This suggested a possible role for ALPK1 in centrosome biology, which is critical for mitotic spindle pole assembly and primary ciliary assembly. 18 Abnormal cilia function due to ALPK1 variant, p.Thr237Met Given the centrosomal localization of ALPK1 and its presence in the basal regions of the primary cilium and the photoreceptor cilium, we hypothesized that the ALPK1 pathogenic variant, p.Thr237Met, may affect ciliogenesis. The evaluation of ciliogenesis in the Australian family revealed, on average, 40% of the cells were ciliated in an unaffected control, while only 15% and 13% of the cells could assemble the primary cilium in two affected individuals (p<0.05) (Fig. 4m, n). These findings suggest that phenotypic features in this family may relate to abnormalities of cilia-centrosomal function. 19 Increased proportion of multinucleated cells due to ALPK1 variant, p.Thr237Met In view of the presence of ALPK1 in the spindle poles of mitotic cells, we considered that the variant in this gene may affect cell division and cytokinesis. Transfection of HeLa cells with p.Thr237Met ALPK1 compared with wild type, empty vector, and untransfected revealed that cells transfected with the mutant protein had a significantly higher number of multinucleated cells (16%), compared with the cells transfected with wild-type ALPK1 (3.7%) or empty vector (5.7%) (p<0.01) (Supplementary Figure S2).

DISCUSSION
We define a phenotypically distinct autosomal dominant retinal dystrophy syndrome, ROSAH, with features of retinopathy, optic nerve edema, splenomegaly, anhidrosis, and headache in five unrelated families, and show heterozygous (c.710C>T, [p.Thr237Met]) ALPK1 as the causative variant in all cases. Some patients also exhibit low-grade ocular inflammation and recurrent infections. Three independent exome and genome sequencing family studies led to the identification of this ALPK1 variant, which was confirmed in the two additional families. The occurrence of the same c.710C>T transition in a CG dimer in all five families indicates the existence of a mutational hotspot. ALPK1 is an ɑ-protein kinase of poorly defined function. Consistent with the phenotypic features in our patients, we showed that ALPK1 was highly expressed in retinal tissues, notably the retinal pigment epithelium and the connecting cilium region of photoreceptors, localized to myoepithelial cells of sweat glands, and was expressed across most other tissues. Our cellular investigations suggest a role for ALPK1 in centrosomal cilia function. Our findings show that ROSAH syndrome is a distinctive autosomal dominant retinal dystrophy syndrome, caused by a recurrent heterozygous variant in ALPK1, and indicate a role for ALPK1 across several different organ systems. ALPK1 was first identified as a member of a family of atypical protein kinases, the ɑ-kinases, through homology to the catalytic domain of myosin heavy chain kinase. 20 These kinases phosphorylate sites that are surrounded by peptide regions with ɑ-helical secondary structure. 21 The ALPK1 catalytic domain lies between amino acids 1027 and the carboxy-terminus at 1244 (Fig. 1d). 22 It has homology to the myosin heavy chain kinase of Dictyostelium. 20 In addition, ALPK1 has been shown to phosphorylate myosin 1a in MDCK cells. 23 In MDCK cells, ALPK1 has been shown to play a role in trafficking sucrose-isomaltase lipid rafts from the Golgi network through phosphorylation of myosin 1a. 23  When disrupted, ALPK1 regulation of apical transport may affect cellular establishment of polarity. 24 In all patients with ROSAH syndrome there is evidence of an early-onset retinal dystrophy. We noted presence of Alpk1 in the basal region of the connecting cilium of the photoreceptors (Fig. 3b-f), as well as the centrosomes and base of the primary cilium ( Fig. 4a-l). We identified reduced ciliation of fibroblast cells from patients affected with ROSAH syndrome compared with controls. Pathogenic variants in centrosome-cilia proteins have been extensively described in syndromic and nonsyndromic forms of retinal dystrophy disease, such as pathogenic variants in TOPORS, RPGR, RP2, CC2D2A, and CEP290 (refs. [25][26][27][28][29][30]. These pathogenic variants indicate that ciliary-centrosomal abnormality is one of the known pathways that contributes to the pathogenesis of retinal dystrophy through defects in ciliary formation and regulation of cargo trafficking in the photoreceptors. 19 In addition, cilia formation is linked with cell cycle progression 31 and localization of ALPK1 at the spindle poles during mitosis (Fig. 4a-f) raises the possibility that abnormality of ALPK1 may impact cell cycle progression, another factor that may contribute to a retinal degeneration phenotype. 32 Noting that the population database gnomAD did not reveal likely loss-offunction intolerance of ALPK1, we investigated the impact of overexpression of ALPK1 (c.710C>T,[p.Thr237Met]) in HeLa cells and showed increased multinucleated cells (Supplementary Figure S2), suggesting a possible gain-of-function disease mechanism. ALPK1 was also expressed in the RPE where its role in maintenance of epithelial cell polarity may also be important for normal function of the RPE and prevention of retinal dysfunction. Patients with ROSAH syndrome also exhibited anhidrosis, so the role of ALPK1 in maintenance of epithelial cell polarity may also be relevant for normal function of the sweat glands.
There are several reports of proposed association of singlenucleotide polymorphisms (SNPs) in ALPK1 with increased susceptibility to inflammatory conditions, including gout and chronic kidney disease. 33,34 In the mouse, Alpk1 was found to mediate innate immune responses to bacterial infection 35  susceptibility to inflammatory bowel disease. 36 The role of ALPK1 in inflammation may be mediated through regulation of NF-κB signaling. 37 Interestingly, the N terminal region of ALPK1, including the Thr237 amino acid affected in our patients, has been found to be critical for stimulation of the ALPK1 kinase domain and phosphorylation and activation of downstream targets in the NF-κB signaling pathway. 35 Patients with ROSAH syndrome also show signs of inflammation, exhibiting recurrent or chronic low-grade uveitis, although the uveitis does not appear to be clinically alleviated by immunosuppression. Abnormalities of the innate immune pathway may also contribute to retinal degeneration, raising a further avenue for investigation in ROSAH syndrome patients. Some patients with ROSAH syndrome also exhibit episodic infections, fevers, and pancytopenia, further indicating an inflammatory component to this condition.
Our report is the first where a pathogenic variant in ALPK1 is found in affected individuals with Mendelian disease. Motor coordination deficits of unknown mechanism have been found in mice with the catalytic domain of ALPK1 disrupted by a piggyBAC transposon. 38 None of the individuals in the families we describe have motor coordination defects. These mice and the Alpk1 loss-of-function mice with inflammatory abnormalities were not tested for blindness or retinal dystrophy. 35,36 The etiology of the optic nerve edema, splenomegaly, and headache is unclear in ROSAH syndrome patients. Two patients had lumbar punctures performed due to concerns about raised intraocular pressure because of the appearance of the optic nerve head, but no increased pressure was detected. ALPK1 may be important in the development of the red blood cell (RBC) cytoskeleton, which if impaired may lead to the propensity to form elliptocytes noted in one case, and may be filtered at the level of the spleen resulting in splenomegaly. In addition, our patients exhibit headache, which has been associated with alterations in kinase function affecting calcitonin gene related peptide and trigeminal nerve pain. 39,40 In summary, we name this autosomal dominant condition ROSAH syndrome to reflect the most common features across affected individuals: retinopathy, optic nerve edema, splenomegaly, anhidrosis, and headache, and identify p.Thr237Met ALPK1 as the causative variant. Future studies are warranted to examine the precise pathophysiological mechanism of this new autosomal dominant retinal dystrophy syndrome.