Integrated Genomics Analysis Identies Recessive Ciliopathy Mutations in Primary Endocardial Fibroelastosis: a Rare Neonatal Cardiomyopathy

Among neonatal cardiomyopathies, primary endocardial broelastosis (pEFE) remains a mysterious disease of the endomyocardium, affecting 1/5000 live births and accounting for 25% of the entire pediatric dilated cardiomyopathy with a devastating course and grave prognosis. To investigate the potential genetic cause that underlies pEFE, we performed integrative genomic analysis, using whole exome sequencing (WES) and RNA-seq in a female infant with conrmed pathological diagnosis of pEFE. Within regions of homozygosity in the proband genome, WES analysis revealed novel parent-transmitted homozygous mutations affecting three genes with known roles in cilia assembly or function. Among them, a novel homozygous variant [c.1938delA] of uncertain signicance in ALMS1 was prioritized for functional genomic and mechanistic analysis. Loss of function mutations of ALMS1 have been implicated in Alstrom syndrome (AS) [OMIM 203800], a rare recessive ciliopathy that has been associated with cardiomyopathy. The variant of interest results in a frameshift introducing a premature stop codon. RNA-seq of the proband’s dermal broblasts conrmed the impact of novel ALMS1 variant on RNA-seq reads and revealed dysregulated cellular signaling and function, including the induction of epithelial mesenchymal transition (EMT) and activation of TGFβ signaling. ALMS1 loss enhanced cellular migration in patient broblasts as well as neonatal cardiac broblasts, while ALMS1-depleted cardiomyocytes exhibited enhanced proliferation activity. Herein we present the unique pathological features of pEFE compared to DCM and utilize integrated genomic analysis to elucidate the molecular impact of a novel mutation in ALMS1 gene in an AS case. Our report provides insights into pEFE etiology and suggests, for the rst time to our knowledge, ciliopathy as a potential underlying mechanism for this poorly understood and incurable form of neonatal cardiomyopathy. with primary antibodies overnight and then appropriate AlexaFluor-conjugated secondary antibodies (Invitrogen) for 1 hour. Cell nuclei were eventually counterstained by DAPI in the mounting medium (Invitrogen). Images were recorded and analyzed on a Revolve microscope (ECHO). Primary antibodies used to detect the expression of the proteins of interest are ALMS1 (Abcam), Histone H3 (Cell Signaling Technology), Phospho-Histone H3 (Millipore), Ac-alpha-Tubulin (Cell Signaling Technology), Collagen IV (Cell signaling Technology) and Pericentrin (Abcam). Cell Culture and Small Interfering RNA Studies: Neonatal rat ventricular myocytes (NRVMs) and neonatal rat cardiac broblasts (NRCFs) were isolated from 2-day/old rat pups and cultured in DMEM/10% (vol/vol) FBS with antibiotics at 37°C and 5% CO2. pEFE patient-derived dermal broblasts and human neonatal dermal broblasts (hNDFs) were cultured in broblast growth medium (ATCC). Transient transfection of cells with siRNA (Bioland Scientic) was performed in 6-well plates using RNAiMAX (Invitrogen). Rat Alms1-siRNA (30 pmol/well) was transfected into NRVM or NRCF cultured in DMEM. For hNDFs, human ALMS1-siRNA (30pmol/well) was used. After 48 h or 72 h of incubation, the cells were harvested. two-tailed t-test

Aiming to elucidate the genetic cause of pEFE, we carried out an integrative genomic analysis, using trio whole-exome sequencing (WES) and RNA-seq, on a female proband with con rmed diagnosis of pEFE and her unaffected parents. We identi ed a novel homozygous single nucleotide deletion, which results in a frameshift [p.Val649fs] and introduction of a premature stop codon in ALMS1, suggesting the genetic diagnosis of AS [12,13]. Clinical follow-up and integrative genomic analysis of WES and RNA-seq of the proband-derived broblasts con rmed the causality of the novel ALMS1 variant and elucidated the molecular signature of pEFE learned from our index AS case. To our knowledge, our report provides the rst evidence linking ciliopathy to pEFE etiology.

Results
Human Studies: All human studies were conducted in accordance with regulation of the University of California Los Angeles (UCLA) Institutional Review Board (IRB). Subjects provided written informed consent to participate in this study. Electronic medical record, family pedigree and specimen collection were acquired through the UCLA Congenital Heart Defect (CHD) BioCore [21] following the UCLA-IRB approved protocols. Specimens were de-identi ed and coded following acquisition. Potential Contribution of Recessive Ciliopathy Genes to pEFE Etiology:To investigate the genetic contribution to pEFE etiology, we performed WES on our pEFE proband and her parents combined with RNA-seq analysis of pEFE proband dermal broblasts [Supplemental Figure 1]. Parents received pre-test genetic counseling and were then asked to provide informed consent to make anonymous clinical and genomic data available for research and publication. Using a custom-made primary gene list of 44 known cardiomyopathy genes, including TAZ [Supplenmental Table 1], no pathogeneic variants were identi ed. Based on the sporadic nature of pEFE presentation in the proband cohort, we expected that de novo germline mutations might contribute to the genetic basis of pEFE. However, no denovo mutation was detected. The family pedigree did not reveal parental consanguinity or another affected family member.
On the contrary, homozygosity analysis revealed 8 large runs of homozygosity (ROH) greater than 5 MBs, totaling 51.8 MBs, which correspond to 1.62% of the proband's genome [Supplemental Table 2], suggesting that both parents descend from a common, genetically isolated community. Importantly, within these ROHs, three novel homozygous variants in cilia-related genes, DNAH8, DNAH17, and ALMS1, were detected, while both of the unaffected parents were heterozygous carriers for each of these variants [ Figure 2. A]. Integrative Genomics Viewer (IGV) analysis con rmed the zygosity of the variants [ Figure 2. B]. The detected variants have not been previously reported in gnomAD, ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), the 1000 Genomes database  [22], and have never been reported in association with cardiomyopathy. Therefore, these two genes are unlikely to contribute to pEFE phenotype.
Finally, ALMS1 encodes the centrosome and basal body associated protein 1, which functions in the formation and maintenance of cilia and in microtubule organization [12,13]. Prioritizing the Candidate Genes for Functional Assessment: To further prioriterize the candidate novel variants in the three cilia-related genes, we set out rst to con rm the impact of these variants on RNA expression. We isolated total RNAs from pEFE proband dermal broblasts and from hNDFs (as a control) and performed paired-end RNA-seq as we previously described [23,24]. Out of the three candidate genes, ALMS1 was the only gene expressed in dermal broblasts with good read coverage as illustrated by IGV viewer, while neither DNAH8 nor DNAH17 was detected. IGV con rmed the impact of the novel ALMS1 c.1938delA variant at the RNA level in pEFE broblast compared to control hNDF [ Figure 2. D]. Next, we examined the expression of these genes in our previously reported RNA-seq data of congenital heart defects samples obtained from infants with structural heart defects [24]. Among the three genes, only ALMS1 was expressed in heart tissue samples [ Figure 2. E].
Given that the cardiac dysfunction of pEFE manifests clinically during the early postnatal period [5], we next examined the expression of the three candidate genes in our previously reported RNA-seq data derived from neonatal mouse heart left and right ventricles at postnatal day P0, P3 and P7 [24], available in the NCBI's Gene Expression Omnibus repository under the Neonatal Heart Maturation SuperSeries [http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE85728]. Again, among the three candidate genes, ALMS1 was the only gene expressed in both ventricular chambers, exhibiting dynamic regulation in neonatal heart during perinatal stages [ Figure 2. F]. To further ascertain the expression pattern of the candidate genes in neonatal heart at the cellular level, we examined their expression in primary cultured neonatal rat ventricular myocytes (NRVMs) and neonatal rat cardiac broblasts (NRCFs) isolated from neonatal rat hearts. Indeed, out of the three candidate genes, only ALMS1 was expressed in both cardiac cell types [ Figure 2. G]. Altogether, our ndings led to prioritizing ALMS1, which is expressed in neonatal heart and has been associated with cardiomyopathy, as the potential causal candidate gene contributing to pEFE. Therefore, we focused the following studies on determining the molecular and cellular impact of the novel ALMS1 variant in pEFE pathogenesis.
ALMS1 Protein Expression is Absent in pEFE Proband Dermal Fibroblasts: ALMS1 gene is located on chromosome 2p13.1 spanning 23 exons. It encodes a protein of 4,169 amino acids (461.2 kDa), which lacks known catalytic domains [25,26], but has several sequence features of unknown function, including a large tandem repeat domain (TRD), three short coiled-coil domains and a stretch of ~130 residues at the C-terminus termed as the ALMS motif [aa: 4032-4164 (Ensembl)] [ Figure 3. A]. The ALMS1 protein is widely expressed and co-localizes with the centrosomes and basal bodies of ciliated cells in different tissues and organs, where ALMS1 has been shown to play important roles in ciliary function and intracellular tra cking [25]. Deletion analysis of ALMS1 suggests that the ALMS motif may play a key role in centrosome-targeting [26]. Importantly, the variant of interest [c.1938delA] is predicted to result in a severely truncated ALMS1 protein [p.Val649fs], missing most of the tandem repeats and the highly conserved ALMS motif at the C-terminus. Based on these previously reported observations, we predicted the ALMS1 protein expression and/or stability in the centrioles to be affected by the novel ALMS1 variant in pEFE cells. Indeed, immunocytochemistry (ICC) assay using anti-ALMS1 antibodies con rmed ALMS1 protein localization to the centrioles in control hNDFs, while ALMS1 protein was completely absent in the proband pEFE-derived dermal broblasts, indicating a null effect [ Figure 3. B].
Novel ALMS1 Variant Alters the Functional Phenotype of pEFE Fibroblasts: To examine the functional impact of the novel ALMS1 variant at the cellular level, we rst examined cilia morphology in pEFE dermal broblasts compared to control hNDFs using electron microscopy (EM). We observed a unipolar organization of thin, over-branched, microvilli, nonmotile nger-like protrusions from the surface of epithelial cells that function to increase the adhesion, the cell surface area, the e ciency of absorption, that was more prominent in pEFE cells compared to control hNDFs [ Figure 3. C]. Next we examined the impact on cellular proliferation over time using an automated proliferation assay performed on an xCELLigence RTCA SP instrument. We observed decreased proliferation index in pEFE broblasts, compared to hNDFs [ Figure 3. Dand E]. FACS analysis revealed no change in G2/M phase in pEFE broblast compared to control [Supplemental Figure 2]. Then, we examined the impact on cellular migration over time using the xCELLigence RTCA DP instrument and observed increased migration activity in pEFE broblasts compared to the control hNDFs [ Figure 3. F and G]. Together, the novel ALMS1 variant altered the functional phenotype and cellular physiology of pEFE broblasts. To further con rm the impact of ALMS1 loss, we treated primary cultured hNDF1 with ALMS1-siRNA. Indeed, ALMS1depleted hNDFs exhibited enhanced migration replicating the changes seen in pEFE broblasts [Supplemental Figure 3].
Novel ALMS1 Variant Impact on pEFE Proband Heart: As previously stated, few reports have implicated homozygous ALMS1 mutations in an extremely rare form of neonatal cardiomyopathy associated with AS and characterized by prolonged mitotic window of postnatal cardiomyocytes, hence termed "mitogenic cardiomyopathy" [16,17]. Furthermore, mice with Alms1 mutations exhibit delayed exit from active cell cycle progression [18], supporting a functional role for ALMS1 in regulating postnatal cardiomyocyte maturation. Based on these reported observations, we sought to determine the pathogenic impact of the mutant ALMS1 in pEFE heart. First, we determined that ALMS1 protein is also localized at the centrosomic poles (centrioles) in neonatal cardiomyocytes by using primary cultured neonatal rat ventricular myocytes (NRVMs) isolated from the neonatal rat heart [ Figure 4. A]. Next, we con rmed that ALMS1 loss is su cient to extend the proliferative activity of postnatal cardiomyocytes by using small interfering RNA (siRNA)-mediated inhibition of Alms1 in primary cultured NRVMs. We observed increased mitotic activity in Alms1-de cient myocytes, as indicated by increased number of phospho-Histone H3 (pH3) positive cardiomyocytes and upregulation of the mitotic marker genes, including Ki67 and Cdc25c, compared to control scramble [ Figure 4. B and C]. These results are consistent with previous studies [16][17][18], indicating that ALMS1 regulates neonatal cardiomyocyte proliferation. Then, we con rmed that ALMS1 expression at the centrosomes was absent in histopathological sections obtained from the proband pEFE heart, but appeared normal in the age matched DCM heart with a pathogenic mutation in TNNT2 [ Figure 4. D]. Finally, we evaluated cardiomyocyte proliferation activity in proband pEFE heart sections, using pH3 immunohistochemistry (IHC) analysis and revealed increased pH3 positive cardiomyocytes in pEFE heart compared to the age-matched DCM heart [ Figure 4. E and F]. Together, in agreement with other reported cases of AS-associated mitogenic cardiomyopathy, the novel ALMS1 variant abolished ALMS1 protein expression and localization at the centrioles and delayed cardiomyocyte proliferation arrest in pEFE heart. However, how novel ALMS1 vaiant leads to pEFE phenotype remains unclear.
Novel ALMS1 Variant Alters Global Transcriptome Signature in pEFE Fibroblasts: Having established the causal role of the novel ALMS1 variant [c.1938delA] in pEFE, we aimed to gain further insights into the mechanisms that underlie ALMS1 function in endocardial broelastosis process by evaluating the global impact of ALMS1 loss on transcriptome programming of proband pEFE cells that carry the mutation. We systematically analyzed pEFE dermal broblast-derived RNA-seq datasets using total RNA obtained from three independent biological replicates. In addition, three independent hNDF replicates were subjected to the same RNA-seq protocol as we previously described [23,24].
Principal component analyses of the top 1000 varied genes showed that transcripts from pEFE and control broblasts formed distinct clusters [ Figure 5. A]. Likewise, expression heatmap across the entire transcriptome revealed distinct molecular signatures for pEFE and control samples [ Figure 5. B], indicating that the variation pattern was consistent across the two methods. In total, 8970 protein coding genes were expressed at ≥3 RPKM in at least three sample (3 biological replicates) with coe cient of variation (CV) exceeding 0.2. Of these expressed genes, 3910 genes exhibited signi cant differential gene expression (DGE) in pEFE versus control at Benjamini-Hochberg (B-H) adjusted P value less than 0.05 [ Figure 5. C]. Together, these ndings indicate signi cant impact of ALMS1 perturbation on global transcriptome signature in pEFE broblasts compared to the control hNDF cells.
To further annotate the enriched functional identity and signaling pathways in the differentially expressed genes, we used Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and identi ed cellular adhesion, exon guidance, TGFβ and RAP1 signaling as the top enriched functional pathways that were also interconnected by sharing several overlapping genes [ Figure 5. D and E]. For example, genes involved in EMT signi cantly overlapped with cell adhesion molecule and TGFβ pathways. Among these hub genes, we identi ed BMPR1B, a known member of the bone morphogenetic protein (BMP) family of transmembrane serine/threonine kinases that has been associate with left ventricular mass [29]. The ligands of this receptor are members of the BMPs and TGF-β superfamily. Consistently, upstream analysis using IPA (Ingenuity Pathway Analysis) of the connected networks predicted TGFβ as the top upstream regulator with remarkable induction of its downstream signaling mediators, including BMPR1B, ID1-4, SMAD9, INHBB, CHRD, and BMP6, many of which are key players of EMT during embryonic development and cardiogenesis [28] [Supplemental Tables 3-5]. Together, our ndings suggested TGFβ mediated activation of EMT program in the proband pEFE broblasts.
To validate these ndings, we examined the impact of the novel ALMS1 variant on selected EMT marker genes in primary cultured pEFE broblasts compared to control hNDFs using qRT-PCR. Indeed, TGFβ was induced along with the transcription factor SNAIL, the ECM marker ELN as well as the myo broblast marker gene aSMA. Consistently, CDH1, a key signaling mediator for the induction of EMT cascade, was signi cantly upregulated, while CDH2 was downregulated. Finally, cell cycle markers (Ki67 and Plk1) were suppressed [ Figure 6. A]. Together, the data support that ALMS1 loss leads to EMT induction, potentially via the activation of TGFβ signaling.
Novel ALMS1 Mutation Induces EMT in Neonatal Cardiac Fibroblasts: Current evidence suggests a role for EMT in EFE [30,31]. Our transcriptome analysis suggested that ALMS1 loss altered the physiological proprieties (enhanced migration) and induced EMT in pEFE broblasts potentially via activating TGFβ signaling. We sought next to determine whether ALMS1 regulates EMT in cardiac broblasts by performing siRNA-mediated knockdown of Alms1 in primary cultured NRCFs. Remarkably, the cells exhibited enhanced migration and induced EMT marker genes, including the Snail1, Tgfb, and Cdh1 [ Figure 6. B-D]. Importantly, unlike neonatal cardiomyocytes, NRCFs remained quiescent and did not exhibit increased proliferation activity as demonstrated by proliferation marker genes (Plk1 and Ki67) expression [ Figure 6. D]. Together, the data suggest EMT induction and Tgfb activation in ALMS-de cient neonatal cardiac broblasts. These ndings replicate the changes observed in the proband pEFE dermal broblasts and ALMS1-de cient hNDFs, and correspond to a known role of cilia-mediated signaling in driving EMT process in cardiac development and brosis in response to cardiac injury [32].
Further Surveillance for Alstrom Syndrome and Patient Care: According to the diagnostic criteria from Marshall et al (2007) [13], detecting the novel ALMS1 variant [c.1938delA] suggested the genetic diagnosis of AS. We provided detailed genetic consultation about AS to the family and monitored our pEFE proband closely for potential multisystem involvement, in addition to the regular monitoring and management by the primary cardiologist and heart transplantation clinic. Bilateral nystagmus, ptosis and photophobia were documented at 6-12 month of age. Visual impairment was diagnosed at one year of age. Standard electroretinography (ERG) at 7 years of age showed absence of scotopic, photopic, maximal-combined and icker responses, consistent with severe pan-retinal abnormalities of both rodand cone-mediated retinal functions [ Figure 7. A]. Retinal imaging revealed bilateral retinal dystrophy [ Figure 7. B]. Short stature, increased weight gain and early onset obesity were observed at 2-3 years of age associated with elevated triglycerids, acanthosis nigricans and increased hemoglobin A1C (HbA1C) suggesting insulin resistance [ Figure 7. C-E]. Hearing evaluation, thyroid function tests, blood sugar levels, serum lipid pro le, liver function tests, serum creatinine, abdominal ultrasound and blood chemistry remained normal. The patient maintained normal bilateral audiology and psychomotor functions.

Discussion
In this report, we describe the unique pathological nding of pEFE compared to DCM. We also elucidate novel genetic contribution to pEFE etiology involving a known ciliopathy gene. By complementing triobased WES with RNA-seq of patient-derived broblasts, our integrative genomic analysis allowed the identi cation of a novel pathogenic variant. Herein we report our novel discovery of ALMS1 as a causal gene underlying pEFE pathogenesis. To our knowledge, this is the rst case of pEFE associated with Alstrom syndrome (AS), a rare recessive ciliopathy.
Our report expands the phenotypic and genetic features of pEFE and AS. The patient reported here initially presented with a unique form of isolated neonatal cardiomyopathy that could lead to early death [5]. Subsequent WES analysis revealed the genetic diagnosis of AS. Having survived heart transplantation at one year of age, surveillance for multi-organ involvement and targeted interventions can help improve long-term outcomes for this patient [19]. Since ALMS1 is the only gene in which recessive mutations are known to cause AS mani satations, including cardiomyopathy [12][13][14][15][16][17][18][19][20], we ascertained the inheritance pattern, leading to appropriate counseling. Therefore, we propose utilizing WES as the rst diagnostic tool to establish an early diagnosis for neonatal cardiomyopathies, ALMS1 should be added to the cardiomyopathy gene panel, the cardiomyopathy in ALMS1, in this case at least, is pEFE, and that patients with cardiomyopathy/pEFE should be investigated for syndromic features of Alstrom syndrome The giant ALMS1 protein, predicted to comprise 4,169 amino acids, is widely expressed and localized to the centrosomes and basal bodies of ciliated cells of different organs [25,26]. ALMS1 loss-of-function has been linked to defects in primary cilium formation, positioning and maintenance, leading to AS being classi ed as a ciliopathy [12,13]. Reportedly, the ALMS motif is the only region that shares obvious sequence similarity with other human proteins and confers an evolutionally conserved function in targeting ALMS1 protein to the centrioles [26]. Consistent with previous reports, the ALMS1 protein was absent in the centrioles of the proband pEFE broblasts and heart tissue sections, con rming a deleterious effect of the novel ALMS1 variant on ALMS1 protein translation and localization. Although the exact functions of the ALMS1 protein are yet to be fully revealed, it has been demonstrated to play a role in cell cycle regulation [16][17][18]. Consistently, proliferation activity was enhanced in pEFE heart tissue compared to DCM. Likewise, Alms1-depleted cardiomyocytes exhibited a modest increase in their mitotic activity. On the contrary, Alms1-depleted NRCFs and dermal broblasts from pEFE proband exhibited reduced proliferative capacity. However, they demonstrated enhanced migration abilities when compared to their control counterparts. Together, these con icting phenotypic properties in ALMS1-depleted broblasts, in contrast to cardiomyocytes, may suggest cell type-speci c role for ALMS1 protein in heart and potentially other organs.
Alms1-de cient mice have been studied and the initial formation of primary cilia appeared normal in all three mouse models reported to date. However, age-dependent loss of primary cilia was observed in two of these mouse models [Alms1 L2131X/L2131X and Alms1 foz/foz ] [15]. Likewise, the primary cilia formation seems to be unaffected in reported AS patient broblasts in vitro [33]. Consistent with these reports, primary cilia formation appeared normal in our proband [pEFE] broblasts. The observed unipolar localization and con uent networking of branching cilia is potentially consistent, in part, with the observed enhanced cellular motility and migration properties. Whether these differences in the cilia-loss occurrence/timing, seen in previous murine studies, are mutation speci c or cell type-speci c remains to be interrogated.
Cilia play essential roles throughout cardiac development. During cardiogenesis, primary cilia are present on the majority of heart cell types acting as cellular sensors for signal transduction that control cardiac myocytes and broblast differentiation [34][35][36]. Deletions in cilia-related genes have been implicated in congenital heart defects such as heterotaxy, which may manifest clinically as an isolated disorder or associated with other features of situs inverses [34,35]. In the case of AS, dilated cardiomyopathy during childhood is a common feature [19,20]. Meanwhile, only few reports have described an extremely rare feature of AS called "mitogenic cardiomyopathy" that is characterized by persistent mitotic activity in neonatal cardiomyocytes [16][17][18]. Importantly, like pEFE, AS-associated mitogenic cardiomyopathy can be lethal during infancy, before other clinical features of AS normally manifest. However, in contrast to the unique pEFE phenotype, no abnormalities in the endocardium have been described in mitogenic cardiomyopathy cases that are reported thus far. Indeed, our report provides the rst link between ciliopathy and pEFE.
How ALMS1 contributes to pEFE remains to be fully revealed. The involvement of the endocardium is a distinct, yet mysterious, phenomenon in pEFE hearts. Unlike secondary EFE that associates with hypoplastic left heart or myocarditis, no structural defect is seen in pEFE cases, meriting further investigations. Hence, unusual human phenotypes offer the opportunity to understand disease pathogenesis, if the cause can be determined. Having identi ed a causal homozygous ALMS1 variant in pEFE4, the availability of dermal broblasts derived from pEFE4 allowed us to examine the broader impact at the molecular and cellular levels by performing unbiased transcriptome analysis of the patient cells that harbor the ALMS1 [c.1938delA] mutation, using hNDFs as control. EFE is a specialized form of cardiac brosis that involves excessive broblast-mediated deposition of elastin (ELN) and collagen bers, predominantly, in the subendocardium layer of the left ventricle [6][7][8][9]. Consistently, ALMS1-de cient broblasts from pEFE4 exhibited signi cant upregulation of several ECM constituents compared to hNDF [Supplemental Table 5].
Different cardiac lineages arise from one or more EMT waves during caridiogenesis. Errors of the EMT process, including ciliopathies, may lead to cardiac defects and brosis [28,29]. An epithelial cell must disassemble the cell-cell junctions, lose epithelial cell polarity, and acquire a motile phenotype to undergo EMT. The Snail proteins are key EMT-inducing transcription factors activated by EMT-inducing stimuli such as TGFβ. In turn, Snail suppresses adhesion protein genes such as E-cadherin and upregulates the cytoskeletal proteins leading to cytoskeletal reorganization. Furthermore, it stimulates the expression of matrix metalloproteases (MMPs) that break down the ECM and facilitate cellular migration.
It was previously suggested that the EFE broblasts arise from postnatal endocardial cells that undergo an aberrant endothelial-to-mesenchymal transition (End-MT) [30]. However, using an EFE-like mouse model with hemodynamically unloaded heart and multiple genetic lineages tracing approach, a more recent study showed that the epicardium-derived mesenchymal cells (Epi-MCs) serve as the major source for EFE broblasts [31]. In agreement, pEFE broblast transcriptome pro le uncovered clear activation of EMT including reprogramming of broblasts to migratory cells exhibiting markers of myo broblasts, along with the upregulation of MMPs and suppression of apical junction genes, likely triggered by TGFβ signaling activation. Further investigations to determine the cellular origins of EFE broblasts are essential to delineate EFE pathogenesis. Moreover, the role of cilia signaling pathways that govern Epi-MCs expansion needs to be fully addressed.
Finally, consistent with the previous transcriptomic analysis of ALMS1 mutants in other species [37, 38], RNA-seq data derived from pEFE broblasts revealed induced exon guidance and increased secretory activity, endosomal transport and cellular tra cking. Furthermore, retina, leptin and insulin-related pathways were dysregulated [Supplemental Table 6]. Since obesity, insulin resistance and related metabolic defects are cardinal characteristics of AS [39], it is quite interesting to observe that the insights gained from transcriptome analysis of pEFE broblasts recapitulate physiological defects of AS patients and mouse AS models. These conserved effects underscore the utility of integrated genomics analysis (Exome/transcriptome) in patient-derived cells to interpret human phenotypes.
Several limitations need to be acknowledged in our study. Our functional studies are limited to in vitro analysis. Further mechanistic studies of Alms1 mediated ciliary signaling in Alms1 mouse models are nessecary to fully elucidate the mechanisms that underlie pEFE brosis and neonatal cardiomyocyte proliferation in AS. It would be essential to characterize ALMS1 function and potential downstream mediators that might impact neonatal heart maturation and to identify disease-speci c therapeutic targets.

Methods
Human Studies: All human studies were conducted in accordance with regulation of the University of California Los Angeles (UCLA) Institutional Review Board (IRB). Subjects provided written informed consent to participate in this study. Electronic medical record, family pedigree and specimen collection were acquired through the UCLA Congenital Heart Defect (CHD) BioCore [21] following the UCLA-IRB approved protocols. Specimens were de-identi ed and coded following acquisition. pEFE diagnosis was determined based on clinical pathological examination of the explanted heart. Skin and Muscle Biopsy, Heart Tissue Specimens: Using betadine, then alcohol wipe, the skin was prepped and draped in a sterile fashion. For local anesthesia, approximately 1 ml Lidocaine 1% was injected intradermally and 3mm diameter biopsy punch was performed through the skin. The biopsy sample was placed in DMEM culture media and transported to the lab for adherent primary dermal broblast culture following standard protocol. The control human neonatal dermal broblasts (hNDFs) were purchased from ATCC. Heart tissue sections were obtained from UCLA clinical pathology laboratory.
All samples were de-identi ed and coded upon procurement.
Whole Exome Sequencing (WES): Genomic DNA was extracted using standard methods (Purelink Genomic DNA Mini Kit, Invitrogen) at the UCLA Congenital Heart Defect-BioCore. Library preparation, sequencing and data analysis were performed at the CCRD (California Center for Rare Disease) and the UCLA Clinical Genomics Center, using the CLIA (Clinical Laboratory Improvement Amendments) and CAP (College of American Pathologists) -validated protocols. Genomic DNA (3 µg) from the proband and parents was subjected to library preparation and exome capture following the Agilent SureSelect Human All Exon 50 Mb Illumina Paired-End Sequencing Library Prep Protocol. Sequencing was performed on an Illumina HiSeq2000 as a 50 bp paired-end run. For each sample, approximately 200 million independent paired reads were generated for an average coverage of 140X of RefSeq protein coding exons and anking introns (+2 bp) with at least 95% of these bases covered at ≥10X. The sequences were aligned to the hg19/b37 genome release using Novoalign. PCR duplicates were marked using Picard. Genome Analysis Toolkit (GATK) [40] was used for indel realignment and base quality recalibration. Both SNVs (single nucleotide variants) and small INDELs (insertions and deletions) were called using GATK uni ed genotyper. All variants were annotated using the customized VEP (variant effect predictor) engine from Ensembl. Regions of homozygosity by descent were determined using PLINK. Rare protein-altering variants were examined under different inheritance models. Rare variants with minor allele frequency <1% in public databases were retained for analysis. Variants were classi ed based on their pattern of inheritance, predicted consequence at the protein level, and their location within the gene. All variants were interpreted in the context of the patient's phenotype.
RNA Sequencing and Bioinformatics Analysis: RNA sequencing, data processing, and bioinformatics analysis steps follow exactly the procedures we previously described [23,24]. Brie y, total RNAs were isolated, ribosomal RNAs were depleted (RiboZero Gold, Illumina), and the resulting RNA transcripts were processed into short fragments (about 200~500nt). Strand-speci c cDNA libraries were constructed using random hexamer-primed reverse transcription and subsequently were used to generate secondstrand. Sequencing adaptors were ligated using the Illumina Paired-End Sample Prep Kit. Fragments of 200 bp were isolated by gel electrophoresis to select the suitable fragments for PCR ampli cation.
Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System were used for quanti cation and evaluation of the library quality. Lastly, the libraries were sequenced according to Illumina HiSeqTM 4000 sequencer protocol (BGI-Genomics) in paired-end sequencing mode (2 x 100 nt long reads). After adaptors removal and quality control procedures, reads containing ribosomal RNA (rRNA) were removed by SOAP (Short Oligonucleotide Alignment Program) alignment. Subsequently the raw reads were aligned to reference genome using Tophat. The background noise was reduced by employing fragments abundance threshold and coverage analysis considering only those of 100 bp or longer. HISAT2 (Hierarchical Indexing for Spliced Alignment of Transcripts) was used for mapping step and transcript assembly. Normalized gene expression values were presented using mean reads per kilobase of transcript per million mapped reads (RPKM).
Differential Gene Expression (DGE) Analysis: DGE analysis was performed with expression levels normalized for gene length, library size and GC contents. DEseq2 algorithms was used to assess differential expression in log2 [normalized RPKM] values for each gene. Signi cant results were reported at Benjamini_Hochberg FDR ≤ 0.05. The statistical signi cance for DGE was assessed by Fisher's exact test. Pearson's correlation coe cients (r) for gene expression were calculated in R. Principal Component Analysis (PCA) was conducted using R function prcomp. The top 1000 varied mRNAs based on alignment results were used to generate PCAs. The heat map function of R, which employs a hierarchical cluster algorithm, was used to draw heat map gure. The log2-transformed data were preprocessed by median centering of the data for each set and then hierarchically clustered using centered correlation as the similarity metric and average linkage as the clustering method. Volcano plot function of R was used to visualize upregulated and downregulated genes.
Other Bioinformatics and Computational Methods: Expression data analysis was supported by functional enrichment analysis, including gene ontology analysis using the Gene Set Enrichment Analysis (GSEA) suite, which calculates the global signi cance scores by summarizing the overall level of statistical signi cance of each gene in each pathway. Pathway analysis of the differentially expressed genes was done using Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways. For upstream analysis, QIAGEN's Ingenuity Pathway Analysis (IPA) was used.
Quantitative Real-Time PCR: Total RNA was isolated from cells or tissue using an RNeasy Mini Kit (QIAGEN). For reverse transcription, one microgram of total RNA was used to generate rst-strand cDNA with random primers (Applied Biosystems). Real-time PCR was performed using the SYBR Green Mix (Bio-Rad) on the CFX96 Real-time System (Bio-Rad). Data were presented as average ± SEM. Two-tailed Student's t-tests were used to calculate statistical signi cance.
Histology and Immuno uorescence Staining: Cells were xed in 4% (v/v) formaldehyde in PBS for 20 minutes. After rinsing with PBS, the cells were incubated with 0.1% Triton X-100 in PBS for 15 minutes at room temperature and rinsed. Heart tissue was xed in 4% (v/v) formaldehyde, embedded into para n, and cut into 5-μm-thick tissue sections. After depara nization, slides were subjected to antigen retrieval. After blocking in PBS containing 10% bovine serum albumin for 1 hour, the cells or tissue sections were incubated with primary antibodies overnight and then appropriate AlexaFluor-conjugated secondary antibodies (Invitrogen) for 1 hour. Cell nuclei were eventually counterstained by DAPI in the mounting medium (Invitrogen). Images were recorded and analyzed on a Revolve microscope (ECHO). Primary antibodies used to detect the expression of the proteins of interest are ALMS1 (Abcam), Histone H3 (Cell Signaling Technology), Phospho-Histone H3 (Millipore), Ac-alpha-Tubulin (Cell Signaling Technology), Collagen IV (Cell signaling Technology) and Pericentrin (Abcam).
Proliferation Assay: An xCELLigence Real Time Cell Analysis (RTCA) Single Plate (SP) instrument (Agilent) was used to monitor real-time cell proliferation. This technology works by measuring electron ow transmitted between gold microelectrodes fused to the bottom surface of a 96-well RTCA E-plate, in the presence of an electrically conductive solution such as tissue culture medium. Adhering cells disrupt the interaction between the electrodes and the solution and thus impede electron ow. The impedance of electron ow caused by adherent cells is calculated via computed mathematical algorithms and reported as an arbitrary parameter called Cell Index (CI), the magnitude of which re ects the rate and extent of cell proliferation. Brie y, patient-derived dermal broblasts and control hNDFs were cultured in broblast growth medium (ATCC) and seeded at a density of 2500 cells per well in a nal volume of 100 µl into a 96-well RTCA E-plate. The E-plate was placed in the RTCA SP station housed inside a standard CO2 cell culture incubator. Cellular impedance data (CI value) was continuously recorded over a 4-day period using the xCELLigence RTCA Software installed on the control unit outside the incubator. Subsequently CI curves were analyzed and the Slope parameters were plotted to determine cell proliferation activity over time. Data were presented as average +SEM. A two-tailed Student's t-test was used to calculate statistical signi cance.
Migration Assay: An xCELLigence RTCA Double Purpose (DP) instrument (Agilent) was used to monitor real-time cell migration. This instrument utilizes an electronically integrated Boyden chamber, the CIM plate, to make continuous measurements of electrical impedance generated by cell migration through the microporous membrane of the upper chamber to the gold microelectrodes-fused bottom surface of the lower chamber. Brie y, patient-derived dermal broblasts and control hNDFs, or Alms1 siRNA treated neonatal rat cardiac broblasts (NRCFs) and scramble treated NRCFs, or ALMS1 siRNA treated hNDFs and scramble treated hNDFs, were cultured in DMEM with 10% FBS and passaged the day prior to the experiment. Cells in the range of 60-80% con uence were trypsinized using 0.05%/Trypsin/EDTA solution, washed once with serum-free medium (SFM), and resuspended in SFM. Subsequently cells were seeded at a density of 10,000/20,000 cells per well in a nal volume of 100 µl into the top chamber of a CIM-Plate16 (10,000 cells per well for human dermal broblasts; 20,000 cells per well for NRCFs). The CIM plate containing the cells were placed at room temperature for 30 minutes to allow the cells to settle down to the bottom surface (the microporous membrane) of the upper chamber, and then loaded into the DP System inside the incubator. Impedance signal was measured with an interval of 15 minutes for 24 hours. Subsequently CI curves were analyzed and the Slope parameters were plotted to determine cell migration activity. Data were presented as average +SEM. A two-tailed Student's t-test was used to calculate statistical signi cance.

Figure 3
The broblasts exhibited enhanced migration activity as demonstrated by higher cellular index over time. The migration assay was performed using an xCELLigence RTCA DP instrument over a 25-hour period. 10000 cells per well were seeded into the upper chamber of CIM-Plate16. N= 3 biological replicates per group. G.
Quantitative analysis of broblast migration assay shown in F.

Figure 4
The Novel ALMS1 Variant Abolishes ALMS1 Protein Expression and Induced Cardiomyocytes Proliferation in pEFE Heart. A. Representative uorescence microscopy images of neonatal rat ventricular myocytes (NRVMs) show ALMS1 localization at the centromeric poles of the mitotic spindle. B.
Representative uorescence microscopy images of phospho Histone 3 (pH3) stained NRVMs depict enhanced proliferation activity in Alms1-depleted neonatal cardiomyocytes. C. qRT-PCR analysis of Alms1-depleted NRVMs. N=3 biological replicate per condition. * p value <0.05, *** p value <0.005, **** p value <0.001. D. Representative uorescence microscopy images of ALMS1 IHC in pEFE heart tissue (middle panel/pEFE) and an age matched proband with DCM (lower panel/DCM). E. Representative uorescence microscopy images of pH3-stained heart sections demonstrate increased proliferation activity in pEFE heart (Arrows) compared to age matched DCM heart. F. Quantitative analysis of pH3 positive cardiomyocytes in pEFE heart compared to DCM heart (n=3 sections per heart, 5 surface areas per section).  Alms1-depleted NRCFs exhibited enhanced migration activity as demonstrated by higher cellular index over time. The migration assay was performed using an xCELLigence RTCA DP instrument over a 25-hour period. 20000 cells per well were seeded into the upper chamber of CIM-Plate16. N= 3 biological replicates per condition. C. Quantitative assessment of NRCF migration assay shown in B. ** p value <0.01. D. qRT-PCR analysis of EMT and proliferation marker genes in Alms1-depleted NRCF compared to control scramble. Inhibition of Alms1 expression in Alms1 siRNA treated NRCFs was con rmed as well.