Variants in the Kallikrein Gene Family and Hypermobile Ehlers-Danlos Syndrome

Hypermobile Ehlers-Danlos syndrome (hEDS) is a common heritable connective tissue disorder that lacks a known genetic etiology. To identify genetic contributions to hEDS, whole exome sequencing was performed on families and a cohort of sporadic hEDS patients. A missense variant in Kallikrein-15 (KLK15 p. Gly226Asp), segregated with disease in two families and genetic burden analyses of 197 sporadic hEDS patients revealed enrichment of variants within the Kallikrein gene family. To validate pathogenicity, the variant identified in familial studies was used to generate knock-in mice. Consistent with our clinical cohort, Klk15G224D/+ mice displayed structural and functional connective tissue defects within multiple organ systems. These findings support Kallikrein gene variants in the pathogenesis of hEDS and represent an important step towards earlier diagnosis and better clinical outcomes.


Main Text
2][3] Despite evidence of familial inheritance, the most common form of EDS, hypermobile EDS (hEDS), lacks a genetically validated cause.The impacts of hEDS can include frequent joint dislocations and subluxations, tendon and ligament laxity, skin hyperextensibility, connective tissue fragility, and chronic pain.Additionally, patients experience various systemic manifestations including comorbidities affecting the gastrointestinal tract, cardiovascular system, and immune system. 2,3 he combination of variable symptom presentation, limited clinical awareness, and the absence of genetic markers can result in delays of several decades before a diagnosis is made for hEDS patients, leading to worsened health outcomes. 4 identify genetic causes for hEDS, a genetic registry was developed at the Medical University of South Carolina (MUSC).Within this registry, a fourgeneration family was identi ed that presented with autosomal dominant hEDS.Eleven family members were enrolled for genetic analysis, of whom ve met the clinical diagnostic criteria for hEDS and three were coded as probable due to age and clinical history at time of analyses (Figure 1A, Figure S1).Whole exome sequencing (WES) of the proband (IV-1) and a second cousin (IV-4) was performed.Following variant ltering (see methods), four rare and potentially damaging variants were shared between the affected individuals (IV-1 and IV-4).PCR ampli cation and targeted sequencing of all enrolled family members identi ed only one of these variants with a perfect phenotype-genotype segregation throughout the pedigree (Figure 1A, C).This variant, located in the Kallikrein serine-protease gene KLK15 (chr19:50825890-C-T), was also found in affected members of a second family (Figure 1B, Figure S1).The single nucleotide polymorphism (SNP) results in a missense change (KLK15 p. Gly226Asp), is rare in the population with a minor allele frequency (MAF) of 0.002 (gnomAD v2.1.1),and is predicted to be damaging with CADD and DANN scores of 24 and 0.998, respectively.To determine relevance to connective tissue biology, RT-PCR was performed and con rmed KLK15 mRNA expression in glandular and connective tissues isolated from human and mouse biopsies (Figure 1D, E).
KLK15 is part of a contiguous cluster with 14 other members of the Kallikrein gene family on chromosome 19q13.33(Figure 1F).Given the known involvement of Kallikreins in regulating one another through activation cascades, and their shared expression patterns in connective tissues (Figure S2, S3), we evaluated the genetic burden of the entire KLK family of genes in a larger hEDS cohort.WES was performed on 197 clinically diagnosed, unrelated hEDS patients and ltered for KLK variants with MAFs less than 0.01 (<1%) in gnomAD.A total of 76 variants were identi ed, with 48 being unique in the cohort and 65 patients having at least one rare variant in a KLK gene (32.8%)(Figure S4).A gene-based burden test was used to evaluate enrichment of rare variants in individual KLK genes and the entire contiguous gene cluster in hEDS patients (see methods). 5Signi cant enrichment for qualifying variants was observed in 11 of the 15 KLK genes with p-value <0.05 as well for the entire KLK gene cluster (considered as whole, p = 2.28×10 −14 ) (Figure 1F, Figure S5); thus supporting a broad role for Kallikrein genes in hEDS.To provide functional support for the role of Kallikrein in hEDS pathogenesis, CRISPR-Cas9 was used to create knock-in mice with the corresponding familial KLK15 variant (Figure S6).The Achilles tendon, which shows high KLK15 expression in both mice (Figure 1D) and humans (Figure S2, S3), was chosen for analysis.This selection was further justi ed by reports of Achilles tendon ruptures in members of family 1, making it a relevant tissue to assess for structural and functional de cits in the mice.Freshly isolated Achilles tendons from Klk15 G224D/+ (N=8) and control mice (N=9) were subjected to mechanical testing.Microcomputed tomography (mCT) analyses demonstrated Klk15 G224D/+ tendons were similar in overall anatomical dimensions compared to controls (Figure S7, S8).Stress-strain curves revealed a larger displacement, higher strain, and lower toe modulus in Klk15 G224D/+ mice, consistent with a more elastic tissue (Figure 2A-E).This extended toe region occurs without signi cant changes in deformation at endpoints and is consistent with previous studies on ligament injuries. 6,7 s these mechanical changes are indicative of a structural collagen de cit, ultrastructural analyses were performed on Achilles tendons by transmission electron microscopy (TEM) (Figure 2F-H).Image analyses and quanti cation of collagen brils was performed on a total of 20 independent regions throughout tendons from ve Klk15 +/+ (N=3,240 brils) and 24 independent regions from six Klk15 G224D/+ (N=4,191 brils) mice.An overall 20% reduction in bril diameter was observed in mutant mice compared to controls with p<0.0001 (Figure 2G).Parsing data into 10nm increments revealed smaller collagen brils in Klk15 G224D/+ tendons compared to controls (Figure 2H), which correlated with increased elasticity and reduced mechanical strength in Klk15 G224D/+ tendons. 8 cardiovascular defects are a relatively common nding within the hEDS population, hearts from Klk15 G224D/+ (N=6) and Klk15 +/+ (N=5) mice were analyzed using echocardiography and histopathology.Although overall cardiac function was normal (Figure S9), valve dysfunction was observed in 83% (5/6) of the mutant mice, with 80% (4/5) having demonstrable prolapse of the mitral lea ets (Figure 2I).None of the control animals had detectable defects in valve function.Myxomatous degeneration of mitral and aortic valves, as assayed by Movat's Pentachrome stain, was evident in 75% (3/4) of mutant mice compared to 0% (4/4) of controls (Figure 2J).Given previous reports on hEDS patients having an increased risk for changes in aortic dimensions 9 , we examined whether this was apparent in the Klk15 G224D/+ hEDS mouse model.Consistent with these prior reports, Klk15 G224D/+ had slightly enlarged aortic dimensions, trending toward signi cance (p=0.0532) (Figure 2K, S10).
This study provides the rst genetic and biological evidence for the involvement of Kallikrein gene variants in hEDS.Through WES of two families, we identi ed the KLK15 p.G226D variant.Supporting this genetic discovery, in vivo data demonstrated that the KLK15 variant causes structural and functional effects across multiple organ systems in a murine model, consistent with an hEDS phenotype.In a cohort of 197 hEDS patients, we observed enrichment of Kallikrein variants, with 32.8% of patients harboring at least one KLK variant.While most Kallikrein variants identi ed are exceedingly rare, the familial KLK15 variant has a slightly higher MAF.1][12][13] Factors such as genetic background, environmental exposures, potential underdiagnosis, incomplete penetrance, and reduced expression likely contribute to the genotype-phenotype correlation in hEDS.
Exploring the molecular consequences of Kallikrein variants will not only uncover mechanisms of normal connective tissue development and disease but also shed light on the comorbidities commonly associated with hEDS.Notably, Kallikreins are known to interact with substrates in the extracellular matrix, in uencing the connective tissue environment in both homeostasis and disease. 14This class of genes also plays roles in blood pressure regulation and immune cell function, potentially contributing to various comorbidities such as postural orthostatic tachycardia syndrome (POTS) and mast cell activation syndrome, which are frequently observed in hEDS patients. 15Relatedly, it is curious that there is a tight interaction between Kallikrein's and in the innate immune system, speci cally the complement system.Kallikreins can cleave complement 3 (C3) and 5 (C5) directly, leading to the generation of C3a and C5a 16 , which are potent anaphylatoxins that enhance in ammation upstream of mast cell activation.Additionally, in certain pathological conditions such as hereditary angioedema (HAE), dysregulation of the interlinked kallikrein system and the complement pathway can lead to excessive in ammation and tissue swelling 17 , features observed in hEDS patients.
Given the autosomal dominant mode of inheritance and the variability of phenotypes associated with hEDS, it is likely that KLK gene variants, such as KLK15 G226D , function primarily in a dominant-negative manner.However, given the known synergistic hierarchy of Kallikreins, where they auto-activate and catalyze the activation of downstream Kallikrein enzymes, even subtle changes in expression levels due to loss-of-function alleles may have damaging effects.While we implicate KLK variants in hEDS, they represent just one aspect of the genetic landscape.The absence of damaging Kallikrein variants should not preclude patients from receiving a clinical diagnosis of hEDS.Although our study indicates the likelihood of additional families and sporadic individuals harboring rare monogenic causes, genome-wide approaches hold promise for revealing the complex genetic architecture of hEDS.
Hypermobile Ehlers-Danlos Syndrome is a connective tissue disorder that impacts many tissues and organ systems.Management of symptoms is challenging and often requires an interdisciplinary team of well-informed clinicians who understand the complexities of hEDS.Studies focusing on the pathophysiology of hEDS are in their infancy, underscoring the need for better clinical understanding.Elucidation of genetic causes and biological pathways involved in hEDS is critical for earlier diagnoses, which can reduce the healthcare burden and improve quality of life.This report provides a rst critical step toward mechanistic understanding of disease pathogenesis, paving the way for improved diagnostic tools and enhanced prognoses for patients with hypermobile Ehlers-Danlos Syndrome.

Study Design
Familial genetic studies and exon analyses were performed to determine gene variant enrichment in patients with hEDS.A genetically accurate model of hEDS based on the human KLK15 variant was generated.This model was used for in vivo assays to test the effect of the KLK15 variant on connective tissue mechanobiology, ligament laxity, collagen bril diameter, and functional heart diseases.Power analyses were conducted to determine sample size assuming α = 0.05 with a power of 0.80.A series of tests were conducted to analyze connective tissue defects in two independent study groups, a control group and an experimental group of genetically modi ed animals.The two animal/sample groups were differentiated by genotype.Animal genotype was de-identi ed, and researchers were blinded to the animals' genotypes for purpose of the analyses.Genotyping code was held by one individual not associated with measurement calculations.Analyses were conducted by at least two independent investigators who were blinded to genotype.After procurement of all measurements, the code was broken, and genotype/phenotype correlations were presented.The total numbers of replicates are represented in the gure legends.

Animal studies
All animal experiments were performed under protocols approved by the Institutional Animal Care and Use Committees at the Medical University of South Carolina.Prior to tissue biopsy, mice were euthanized by iso urane (Piramal) induction, followed by cervical dislocation in line with the Guide for the Care and Use of Laboratory Animals (NIH publication no.85-23, revised 1996).Combined data for both sexes are shown.All animal experiments were performed in accordance with IACUC procedures and approved protocol number IACUC-2020-00956.

Human subjects
Studies involving human research were approved by the Institutional Review Board of Partners Healthcare, (Boston, MA) and MUSC (Charleston, SC) and participants provided written informed consent.

Whole Exome Sequencing
Whole exome sequencing for the proband and her second cousin in Family 1 was performed through the Broad Institute.An aliquot of genomic DNA (100-150ng in 50µL) was used as the input for DNA fragmentation.Shearing was performed acoustically using a Covaris focused-ultrasonicator, targeting 385bp fragments.Following fragmentation, additional size selection was performed using a SPRI cleanup.Library preparation was performed using a commercially available kit provided by KAPA Biosystems (KAPA Hyper Prep with Library Ampli cation Primer Mix, product KK8504), and with palindromic forked adapters using unique 8-base index sequences embedded within the adapter (purchased from IDT).The libraries were then ampli ed by 10 cycles of PCR.Enzymatic clean-ups were performed using Beckman Coulter AMPure XP beads with elution volumes reduced to 30µL to maximize library concentration.Following library construction, library quanti cation was performed using the Invitrogen Quant-It broad range dsDNA quanti cation assay kit (Thermo Scienti c Catalog: Q33130) with a 1:200 PicoGreen dilution.Following quanti cation, each library was normalized to a concentration of 25 ng/µL, using a 10 mM Tris HCl pH 8.0 solution.All steps performed during the library construction process and library quanti cation process were performed on the Agilent Bravo liquid handling system.
After library construction, hybridization and capture were performed using the relevant components of IDT's XGen hybridization and wash kit and following the manufacturer's suggested protocol, with several exceptions.A single pre-hybridization pool was created.The pre-hybridization pool comprised of 96 unique libraries was created by equivolume pooling of the normalized libraries as well as 5 µL of Human Cot-1 and 2 µl of IDT XGen blocking oligos.The pre-hybridization pool underwent lyophilization using the Biotage SPE-DRY.Post lyophilization, 4 µL of custom exome bait (TWIST Biosciences) along with 13 µL of hybridization mastermix was added to the lyophilized pool prior to resuspension.An initial incubation was performed at 95ºC for 30 seconds, after which time the incubation temperature was lowered to 65ºC at which it remained overnight.Library normalization and hybridization setup were performed on a Hamilton Starlet liquid handling platform, while target capture was performed on the Agilent Bravo automated platform.
After post-capture enrichment, library pools were quanti ed using qPCR (automated assay on the Agilent Bravo), using a kit from KAPA Biosystems with probes speci c to the ends of the adapters.Based on qPCR quanti cation, pools were normalized using a Hamilton Starlet to 2nM, and sequenced using Illumina sequencing technology.The ow cells were then analyzed using RTA v.2.7.3 or later.Each pool of whole exome libraries was sequenced on paired 150 cycle runs with two 8 cycle index reads across the number of lanes needed to meet coverage for all libraries in the pool.
CRAM les aligned to hg38 were transformed to BAM les using SAMtools.BAM les were ltered for duplicates by RmDup and mate pairs xed using FixMateInformation.These processed BAM les were then input into three independent SNV calling tools: VarScan, LoFreq, and FreeBayes.The intersection of both probands with all three calling tools was used in further analysis (N = 48,720 common variants).To narrow likely causal variants, four ltering criteria were used: (1) Population-level minor allele fractions (MAF) from the 1000 Genomes project, with a cutoff of MAF < 0.01, (2) sequencing allelic fraction in the heterozygous range for both probands (average fraction of between 0.4 and 0.6 aligned reads), (3) SnpEff 4.3 to categorize variant type in relation to genes, and (4) PolyPhen-2 to annotate missense protein-coding variants.Variants in other known EDS genes, such as TNXB, were ruled out and 4 total shared variants were identi ed that t these criteria.These variants were: POTEJ V1002M, FRG2C D9N, FRG2C L210M and KLK15 G226D.Sanger sequencing for each of these variants throughout the family was performed using the following primer sets: POTEJ V1002M: 5' GCTATGTTGCCCTGGACTTC 3 and 5' ATTTGCGGTGGACAATGGAG 3'; FRG2C D9N: 5' AGGGACGTATAAAAGGCAGGTC 3' and 5' ACTAAGCCATTTCCCATCCCC 3'; FRG2C L210M: 5' CTCTGGTGAGTCTCTCACATGC 3' and 5' TCAGGGTGCTCCCAGCTTAG 3'; KLK15 G226D: 5' TTCTGTTCCATGTCAGGCGG 3' and 5' TTCACTCAACCTGAGACCCC 3' and BioRad iProof High Fidelity PCR Reagents.PCR cleanup or gel extraction of fragments were generated and sent for sequencing of both strands using the same PCR oligomers.Sequence data was aligned using BLAST and analyzed for phenotype/genotype segregation.One variant: KLK15 G226D had a perfect phenotype/genotype transmission through the family and was functionally evaluated further as detailed below.

KLK Variant Annotation and TRAPD analyses
Whole exome sequencing was performed on 205 patients with a clinical diagnosis of hEDS.Library preparations and sequencing reactions were conducted at Azenta US (South Plain eld, NJ, USA) as follows: Genomic DNA sample were quanti ed using Qubit 2.0 Fluorometer (ThermoFisher Scienti c, Waltham, MA, USA).Enrichment probes were designed against the region of interest and synthesized through Twist Biosciences -Twist Human Comprehensive Panel (South San Francisco, CA, USA).Library preparation was performed according to the manufacturer's guidelines.Brie y, the genomic DNA was fragmented by acoustic shearing with a Covaris S220 instrument.Fragmented DNAs were cleaned up and end repaired, as well as adenylated at the 3'ends.Adapters were ligated to the DNA fragments, and adapter-ligated DNA fragments were enriched with limited cycle PCR.Adapter-ligated DNA fragments were validated using Agilent TapeStation (Agilent Technologies, Palo Alto, CA, USA), and quanti ed using Qubit 2.0 Fluorometer.Adapter-ligated DNA fragments were hybridized with biotinylated baits.The hybrid DNAs were captured by streptavidin-coated binding beads.After extensive wash, the captured DNAs were ampli ed and indexed with Illumina indexing primers.Post-captured DNA libraries were validated using Agilent TapeStation (Agilent, Santa Clara, CA, USA) and quanti ed using Qubit 2.0 Fluorometer and Real-Time PCR (KAPA Biosystems, Wilmington, MA, USA).
The bioinformatics pipeline from FASTQ les to the variant analysis was performed suing the Franklin analysis platform (Genoox, Tel Aviv, Israel).The analysis was performed as follows: FASTQ les aligned to hg19 were transformed to BAM les using BWA.Duplicate reads were marked and ltered in the process.These processed BAM les were then put into two independent SNV calling tools: GATK haplotype caller, and FreeBayes in addition to a Franklin proprietary CNV caller.After removing 1 sample that failed QC and 6 samples that had pathogenic or likely pathogenic variants consistent with other connective tissue disease, 197 exomes were ltered for genes within the KLK locus.Variants were ltered for rare MAF (less than 0,01, < 1%) using gnomAD allele frequencies and assessed for conservation and pathogenicity through UCSC genome browser and dbNSFP, respectively, to generate the list of variants shown in Figure S4.
Previous reports have made claims for the involvement of the Methylenetetrahydrofolate reductase (MTHFR) gene. 18We performed genetic-follow up analyses in our WES cohort of 197 individuals and did not observe a signi cant enrichment of the common MTHFR polymorphisms.Chi-square test for trend revealed no association between hEDS patient genotypes for the C677T (p = 0.9864) or A1298C (p = 0.3156) SNPs when compared to Gnomad v2.1.1 non-Finnish European population.Thus, the high population frequency of these variants and lack of replication in our cohort render these common MTHFR variants as unlikely to cause hEDS.For an accurate synopsis of the impact of these speci c MTHFR variants on human physiology, we direct the audience to statements released by the Centers for Disease Control (CDC: https://www.cdc.gov/ncbddd/folicacid/mthfrgene-and-folic-acid.html) and American College of Medical Genetics (ACMG) recommendations. 19sociations between relevant variants located in the Kallikrein-related peptidase (KLK) genes and hEDS (N = 197) were tested using a gene-based burden test implemented in TRAPD (Testing Rare vAriants using Public Data) package. 5For the dominant test, for each gene, we calculated the number of individuals carrying at least one qualifying variant.We used non-Finnish European populations (N = 64603) summary statistics from gnomAD v2.1.1 as a control group to evaluate the enrichment of rare KLK genes alleles in patients.Also, as sensitivity analysis, we restricted control population to only samples from individuals who were not selected as a case in a case/control study of common disease (N = 24146).Results were consistent between the two analyses (Figure S5).Coverage for the variants in gnomAD v2.1.1 were taken into account and coverage summary for all the variants used in our analyses is shown below.For every variant, the mean and median depth of coverage, as well as the proportion of subjects with speci c depth coverage values (over 1, 5, 10, 15, 20, 25, 30, 50, 100)

Bgee Expression
To determine if KLK genes are expressed in tissues affected in hEDS patients, data mining in the Bgee portal (https://bgee.org/gene/ENSG00000174562/) was initially performed. 20Single cell and bulk RNA sequence analyses of connective tissues are poorly represented in most publicly available datasets.Bgee is a database compiled of RNA-Sqa, Affymetrix, in situ hybridization and EST data integrated for comparison of normal gene expression in various wild-type tissues.Bgee curates and integrates these large datasets, including GTEx, with many smaller ones.mRNA expression revealed broad overlap of the KLK genes in various organ systems including connective tissue (Figure S2).A deeper analysis of KLK15 revealed widespread expression with highest levels of mRNA observed in human tissues such as tendons, skin, veins, skeletal muscle, and the gastrointestinal tract (Figure S3).Expression scores for tissues with reported expression were retrieved for the entire KLK gene family in humans.

RT-PCR
Messenger RNA was puri ed from freshly dissected mouse tissues, frozen human ACL (Articular Engineering CDD-H-6800-F) and human dermal broblasts according to RNeasy Kit (Qiagen, 74104) and cDNA synthesis was performed using qScript XLT One-Step RT-PCR Kit (QuantaBiom 76047-074).Primers for mouse Klk15 were as follows: 5'-TGGCGACAAGGTGCTAGAAG-3', 5'-CGGGCAGGTTTGAAAAGTCG-3'.Primers for human KLK15 were as follows: 5'-AGT TGC TGG AAG GTG ACG AG-3', 5'-TGG TTT CCC TGA TCC ACT CC-3'.Negative controls included the RT-PCR reactions in the absence of reverse transcriptase.Results were con rmed by Sanger sequencing.Generation of Klk15 G224D mouse using CRISPR-Cas9 model the polymorphism ed from our genetic studies, CRISPR-mediated genome editing technology in mouse embryos was utilized to generate a p. Gly224Asp (G224D) substitution in the homologous region of mouse Klk15 (NP_777354.1).The mouse KLK15 protein has two fewer amino acids than the human ortholog, making the human G226D variant equivalent to the G224D variant in the mouse.Substitution was achieved through a c.671G > A single nucleotide exchange in Klk15 (NM_174865.1) of zygotes from C57BL/6J mice.Concomitantly, this nucleotide substitution resulted in the creation of a novel Tth111I restriction site suitable for PCR-based genotyping.Synthetic guide RNA (sgRNA; atcacaggggacatcgccccngg) and single-stranded oligonucleotide (ssODN;5'caagtagctgcagacttttgtgtagacgccaggcttggtggtagtatcacaggggacatcgtcccaggagacaatgccctgcagggcacccccacagaccaagggtcctccggagtcaccctg; opposite strand showing single nucleotide exchange in bold and underlined) were procured from Integrated DNA Technologies, Inc. (IDT), which were designed and validated at the Genome Engineering and iPSC Center (GEiC), Washington University in St. Louis, St. Louis, MO; for design strategy and sequences, see Figure S6.CRISPR reagents were delivered by electroporation (EP) to single-cell embryos using 1 mm cuvettes and a Gene Pulser Xcell Eucaryotic system (Bio-Rad Laboratories, Inc.); the EP cocktails contained the following concentrations of reagents in Opti-MEM medium (Gibco/Fisher ThermoScienti c): 4 µM Klk15 sgRNA, 4 µM Alt-R HiFi Cas9 Nuclease V3 (IDT),10 µM Klk15 ssODN.Electroporation conditions were as follows: square wave with 2 pulses of 30 V for 3 ms separated by 100 ms.Sequence analysis of 44 pups revealed four mice heterozygous for the G224D variant with a deletion or frameshift on the other allele, nine additional frameshift mice, and eight mice with deletions.The 4 founder Klk15 G224D/fs mice were bred to a second generation for true heterozygous (Klk15 G224D/+ ) pups that were used for breeding and the studies detailed below.Genotypes of the CRISPR-Cas9 Klk15 knock in mice were con rmed using the following primers and Tth111I restriction enzyme (New England

Figures
Figures

Figure 1 Kallikrein
Figure 1 5re available from the gnomAD database.These statistics have allowed us to ensure that we only consider variants with adequate coverage for reliable detection.In our dataset, the mean coverage depth for the variants of interest from exome sequencing ranges from approximately 17X to 98X, and for genomes sequencing, ranges from 27.7X to 33.6X.The median coverage depth varies from 14X to 100X in exomes and from 27X to 33X in genomes sequencing.In terms of subjects' coverage, for exomes sequencing, over 58.8% of subjects have coverage over 10X for all variants, and this value reaches 100% for many of the variants.The same is true for coverage over 15X and 20X, with the minimum proportion of subjects with such coverage being 49.9% and 35.7% respectively, and in many cases, this value reaches 100%.For genomes sequencing, over 99.1% of subjects have coverage over 10X for all variants, and this value reaches 100% for many of the variants.The same is true for coverage over 15X and 20X, with the minimum proportion of subjects with such coverage being 96.0% and 82.9% respectively, and in many cases, this value reaches 99.3% and 95.1%.These coverage statistics, both from exomes and genomes sequencing, ensure a high con dence in variant calling, and underline the robustness of our analysis.While it is true that different variants may have different coverage in gnomAD, our analysis takes this into account by considering only those variants that meet our stringent coverage criteria.This ensures that our ndings are based on robust variant calling and accurate estimation of the frequencies of these rare alleles.We used two-sided Fisher's exact test to estimate the enrichment P-values, consistent with previous reports.5