Genetic studies identify known and novel variants for recessively inherited moderate to severe hearing loss in consanguineous families from Pakistan Authors: Memoona Ramzan, Hafiza Idrees, Hina Khan, Rabia Faridi, Zunaira Munir, Fariha Muzaffar, Kanwal Shabbir, Ayesha Imtiaz, Muhammad Noman, Alina Ahmed, Rasheeda Bashir, Niaz Muhammad Khan, Azra Maqsood, Ghulam Mujtaba, Midhat Salman, Ihtisham Bukhari, Hafiz Muhammad Waqas Munir, Huma Tariq, Muhammad Waqas, Muhammad N. Iqbal, Elizabeth Wohler, P. Dane Witmer, Nara Sobreira, Robert J. Morell, Go Hun Seo, Sayaka Inagaki, Thomas B. Friedman, Sadaf Naz Published: 2026 ABSTRACT Variants of at least twenty-seven human genes are associated with stable or progressive nonsyndromic moderate to severe hearing loss inherited as a recessive condition. We ascertained 54 consanguineous families, predominantly from the Punjab province in Pakistan, with multiple individuals affected by moderate to severe or progressively profound hearing loss. After Sanger sequencing identified 4 families with GJB2 variants or a non-coding pathogenic variant in HGF, exome sequencing was carried out for selected samples from the remaining 50 families. Analyses revealed a total of 24 novel and 33 reported variants in 26 different genes associated with hearing loss. Overall, there were 23 missense variants, 3 in-frame deletions and 1 intronic deletion, while 30 variants likely impacted splicing, introduced premature termination codons or caused frameshifts. Thus, over half of the alleles are predicted to severely impair gene function. Genetic heterogeneity was observed in members of 9 families. Variants of SLC26A4 were key contributors with a frequency of 30%, while those affecting CDH23, MYO15A, GJB2 and OTOF explained 28% of hearing loss. Our findings corroborate the contribution of many well-studied gene variants associated with hearing loss and also implicate LHFPL5 and PCDH15 in the etiology of moderate or progressive hearing loss.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-48515-1. Introduction The worldwide incidence of newborn and early childhood hearing loss is at least 2 in 1000 individuals1 and about 85% of hearing loss is inherited as an autosomal recessive disorder2. According to estimates, around 14.5 million people in Pakistan have some degree of hearing loss3. The high prevalence of endogamous unions in the Pakistani population increases the likelihood of recessively inherited disorders such as hearing loss4. Consequently, the Pakistani population is a valuable resource for studying autosomal recessive inherited disorders, which is evident by multiple studies aimed at estimating the genetic epidemiology of this population5. Studies of the Pakistani population have been instrumental for genetic mapping and identifying many recessive nonsyndromic hearing loss loci and genes that subsequently were found to be associated with hearing loss in communities worldwide (Hereditary Hearing Loss Homepage, https://hereditaryhearingloss.org/, accessed February 2026). Of the 156 genes with variants associated with nonsyndromic hearing loss, 88 are inherited in an autosomal recessive manner (Hereditary Hearing Loss Homepage) More than 50 of these genes have pathogenic variants described in individuals affected by hearing loss in Pakistan3. Among these, variants of 10 genes; SLC26A4, MYO7A, GJB2, CIB2, HGF, TMC1, MYO15A, CDH23, TMPRSS3 and OTOF account for approximately 75% of all hearing loss in Pakistan6. Of these, genetic variants in five genes SLC26A4, GJB2, MYO7A, HGF and TMC1 account for over 60% of profound hearing loss in Pakistan6. Similarly, variants of SLC26A4, GJB2, MYO15A, TMC1, TMPRSS3, OTOF, MYO7A and CLDN14 contribute to 54% of moderate to severe hearing loss in Pakistan5. Variants in many of the same genes and in some cases even identical variants, contribute to both moderate to severe hearing loss or even profound hearing loss6. Due to this overlapping genotypic and phenotypic spectrum of hearing loss, predicting an association of variants of a specific gene or even a particular allele as solely causing moderate to severe hearing loss is unrealistic. Variants in a few genes such as STRC7, TECTA8 and OTOG9 predominantly cause mild or moderate to severe hearing loss while profound hearing loss is either less frequently reported or may occur later in life10. In contrast, variants in genes such as OTOF11, POU3F412, and ESRRB13 usually cause severe to profound hearing loss; nevertheless some alleles of these three genes can result in a milder hearing loss14,15. Here, we present the results of molecular genetic characterization of 54 consanguineous Pakistani families predominantly segregating moderate to severe hearing loss in multiple individuals. Consistent with our previous studies, SLC26A4 emerged as a prevalent gene identified in our cohort5. Although no additional gene associated with hearing loss was discovered in our study, we identified previously unreported pathogenic variants in several well-established hearing loss genes including CDH23, MYO15A, OTOF and SLC26A46. Moreover, we observed variants in LHFPL5 and PCDH15 associated with a progressive or moderate hearing loss rather than the typically reported severe to profound hearing loss16,17. Our findings broaden the allelic spectrum of moderate to severe hearing loss. We posit that our analyses of hearing loss in Pakistan will continue to inform the molecular genetics of clinically heterogeneous hearing loss evaluated in communities worldwide. Results Members of most families had a bilateral, moderate to severe hearing loss All families enrolled in this study have consanguineous unions with multiple affected and unaffected participants (Fig. 1 and data not shown). However, only affected individuals were available to participate in families HLMS1819, HLMS2022, and HLSHI01. Hearing loss was bilateral in the affected members of all families which was moderate to severe or severe in degree for most participants. Six families had individuals with a moderate degree of hearing loss (Table 1). One family primarily affected with moderate to severe hearing loss, as well as 3 families with severe hearing loss, each had at least one participant who was profoundly deaf. In family HLMN3, hearing loss ranged from mild in one participant to severe in the other participant. Hearing loss was reported by the parents to be progressive in the affected members of families HLAI-03 and HLAI-18, but no audiometry had been performed prior to this study. In families HLGM07, HLGM23, HLGM26, HLMR12, HLZM06, HLHI03, and HLHI07 at least one affected participant had an asymmetric hearing loss (Table 1, Supplementary Table S1, Figs. 2 and 3). Fig. 1Pedigrees of families showing genetic heterogeneity. Families HLMR1, HLMR4, HLMR5, HLMN4, HLMN5, HLAI-18, HLAI-27, HLGM07 and HLGM26 exhibit intrafamilial genetic heterogeneity. Circles and squares represent females and males, respectively. Shaded shapes indicate affected individuals; asterisks denote participants for whom exome sequencing was performed. Genotypes of individuals are shown below their respective symbols. Two different gene variants were identified for the affected individuals of families HLMR4, HLMR5, HLAI-18 and HLAI-27. In families HLMN5, HLMR1, HLGM07 and HLGM26 the affected individuals IV:7, V:5, V:3 and V:6 respectively, were found to be heterozygous for the identified variants while in family HLMN4 and HLGM26 the affected participants IV:5 and V:10 were homozygous for the wild-type allele. Pedigrees of families showing genetic heterogeneity. Families HLMR1, HLMR4, HLMR5, HLMN4, HLMN5, HLAI-18, HLAI-27, HLGM07 and HLGM26 exhibit intrafamilial genetic heterogeneity. Circles and squares represent females and males, respectively. Shaded shapes indicate affected individuals; asterisks denote participants for whom exome sequencing was performed. Genotypes of individuals are shown below their respective symbols. Two different gene variants were identified for the affected individuals of families HLMR4, HLMR5, HLAI-18 and HLAI-27. In families HLMN5, HLMR1, HLGM07 and HLGM26 the affected individuals IV:7, V:5, V:3 and V:6 respectively, were found to be heterozygous for the identified variants while in family HLMN4 and HLGM26 the affected participants IV:5 and V:10 were homozygous for the wild-type allele. Table 1Molecular genetic findings in participating families.GeneFamilyNo. of affected individualsPTA0.5–4 kHz of affected individuals and type of HLVariantEffectPathogenicityFrequencyClinVar ID/PMIDACMG criteria and classificationCADDREVEL/MaxEnt (Δ%)PolyPhenSIFTgnomAD v2 BSND NM_057176.3HLHT01/025Audiograms not available, reported to be moderate to severec.35T > Cp.(Ile12Thr)MissensePPM2_sup, PP3_sup, PP1_strong, PM3_strong25.60.941, D0, D0.00001989VCV000004388.16/19,646,679 CDH23 NM_022124.6HLMR11765, 75, 82, 83Moderate to severec.1373 C > Tp.(Pro458Leu)MissenseVUSPM2_sup, PP1_strong310.681, D0, D0VCV000517256.4/ This studyHLHI123Audiograms not available, reported to be moderate to severec.1621G > Ap.(Glu541Lys)MissenseVUSPM2_sup, PP1_mod,PM3_mod23.60.320.44, UN/A0.0001308VCV000252736.8/ 35,020,051HLMN7258Moderatec.4568 A > T p.(Asn1523Ile)MissenseVUSPP1_sup, PP3_supPM2_sup,27.70.840.999, DN/A0This studyHLMR8484, 89Severec.4730G > Cp.(Arg1577Pro)MissenseVUSPP1_strong,PM2_sup26.10.511, DN/A0VCV001502044.5This study1HLAI-18 IV:24(1 Affected)73Severe; Progressive reportedc.7814 A > G p.(Asn2605Ser)MissenseLPPP1_sup, PM2_sup, PP3_sup, PM3_strong25.20.770.99, D0, D0.00003111VCV000862821.7 /27,573,290 EPS8 NM_004447.6HLHI09575, 80Severec.204 + 1G > ASplice donorPPVS1_strong, PM2_sup, PP1_strong34-127.4--0This studyESPN NM_031475.3HLMR9865, 74, 74, 74, 74 80, 81Moderate to severec.1720delGp.(Ala574ArgfsTer72)FrameshiftPPVS1_strong,PM2_sup, PP1_strong----0This studyHLMR6364, 75Moderate to severec.2254G > Tp.(Glu752Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_mod46---0.000009410This study ESRRB NM_001379180.1HLHI052Audiograms not available(Reported moderate to severe)c.322G > Tp.(Asp108Tyr)MissenseVUSPP3_sup,PM2_sup, PP1_sup, PM3_srong29.60.97N/A0, D031835641GIPC3 NM_133261.3HLZM02260Moderatec.356_357delAGp.(Glu119GlyfsTer36)FrameshiftPPVS1_strong,PM2_sup, PP1_sup----0This studyHLAM03576, 80, 85Severec.788-9G > AIntronic/SplicingVUSPP1_strong, PM2_sup23.9-82.16--0.00003257This study GJB2 NM_004004HLZM04244, 48Moderatec.358_360delp.(Glu120del)In-frame deletionLPPM3_strong, PS3_sup, PM2_sup, PP1_sup----0.00007101VCV000017006.36/10,544,226HLZM01553, 86, 94Moderate to severec.439G > Ap.(Glu147Lys)MissensePPP3_sup,PP1_strong,PM2_sup, PM3_strong,25.30.981, D0.001,U0.00001195VCV000371709.37/14,676,473HLNK22Moderate to severec.598G > Ap.(Gly200Arg)MissenseLPPM3_strong, PP1_sup, PP3_sup, PS3_sup, PM2_sup26.80.710.9620.0410VCV000225222.919567088 GRXCR1 NM_0010804762HLGM075(4 Affected)79, 84, 90 (106)Severec.412 C > Tp.Arg138CysMissenseLPPP1_strong, PM3_modPM2_sup29.50.521, D0, D0.00006022VCV000000196.3/ 20137778 HGF NM_000601.5HLMR12763, 64, 66, 68 (85)Moderate to severec.482+1986_1988delTGADeletionLPPM2_sup, PP1_strong, PM3_strong,----019576567 ILDR1 NM_001199799.22HLMN54(3 Affected)53, 55, 80Moderate to severec.1384 C > Tp.(Arg462Ter)NonsenseLPPM2_sup, PP1_mod, PM3_mod, PVS1_strong,35---0.00003622VCV000562073.5/30,303,587LHFPL5 NM_182548.42HLGM266(4 Affected)45 (61), 84, 99Moderate to severe; Profoundc.250delCp.(Leu84Ter)FrameshiftPPM2_sup, PP1_strong, PM3_mod, PVS1_strong----0.000007953VCV000001694.3/16,459,341 MARVELD2 NM_001038603.33HLMR43(2 Affected)76, 89, 96Severe; Profoundc.1331 + 2T > Cp.(Cys395fsTer403)Splice donorPPM3_mod, PVS1_strong, PM2_sup, PP1_sup34-110.7--0.00003980VCV000871776.24/17,186,462 MYO3A NM_017433.5HLMN3339, 73Mild; Severec.892 C > Tp.(Gln298Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_mod40---0.0001154VCV001687422.3/This study MYO6 NM_004999.4IHT02385, 90, 93.75Severec.261 + 7 A > GIntronic/SplicingVUSPM2_sup, BP4_sup16.67-Benign-0This study COL9A1 NM_001851.4c.2504-5T > CIntronic/SplicingVUSPM2_sup, BP4_sup14.311754Benign-0.0001157VCV001112686.5/This studyMYO7A NM_000260.4HLMR7359, 61Moderatec.3136dupCp.(Leu1046ProfsTer9)FrameshiftPPVS1_strong, PM2_sup, PM3_mod, PP1_mod----0VCV000996717.3/18,181,211MYO15A NM_016239.4HLMN6264, 85Moderate to severec.1657del (p.Arg553GlyfsTer76)FrameshiftLPPM3_strong, PVS1_strong, PM2_sup----0VCV002572402.1/27573290HLSHI012Severec.2199dupCp.(Asp734ArgfsTer246)FrameshiftLPPP1_sup, PVS1_strong, PM2_sup----0.00001935This studyHLHI08263, 70Moderate to severec.3756 + 1G > TSplice donorLPPVS1_strong, PM2_sup, PM3_sup29.9-141.6--0.00003185VCV000006954.1/ 117350291HLAI-27(V:4)4(1 affected)Reported moderate to severec. 8340G > A p.(Thr2780Thr)Synonymous/SplicingLPPVS1_strong, PM3_mod, PM2_sup22.8-96.8-0.000008027VCV000236038.22/ 30139988 OTOA NM_144672.4HLHKA01359, 63Moderate to severePlay audiometry performed for the third affected demonstrated moderate hearing lossc.635 + 2T > ASplice donorLPPVS1_strong, PM2_sup, PP1_mod, PM3_sup31-113.45--0.00002875VCV002990394.1/ 37561809 OTOF NM_194248.3HLGM20461, 61, 63, 66Moderatec.2965_2967delp.(Phe989del)In-frame deletionLPPM2_sup, PM3_mod, PP1_strong,----0.000004088VCV000634821.5/27,573,290HLHKA03288Severec.3262del p.(Leu1088CysfsTer58)FrameshiftLPPVS1_strong, PM2_sup, PP1_sup----0This studyHLGM11384, 85, 85Severec.4362 + 1G > CSplice donorLPPVS1_strong, PM2_sup, PP1_mod33-132.91--0VCV000634822.2/This study PCDH15 NM_001142764HLAI-03469, 94, 96Progressive reported, Moderate to profoundc.2368_2370delp.(Val790del)In-frame deletionLPPM2_sup, PM4, PM3_sup, PP1_strong----025930172HLMN2259, 76Moderate to severec.2858_2862dupTTTATp.(Ala955PhefsTer16)FrameshiftLPPVS1_strong, PM2_sup, PP1_sup----0VCV001526060.1/This study POU3F4 NM_000307.5HLHI103No audiometry due to noisec.747delG p.(Lys249AsnfsTer3)FrameshiftLPPVS1_strong,PM2_sup, PP1_mod----0This study SLC26A4 NM_000441.2HLGM09671, 89Severec.235 C > Tp.(Arg79Ter)NonsensePPM3_mod,PVS1_strong,PM2_sup, PP1_strong37---0VCV000188950.10/19,648,736HLZM07588, 90, 95, 103Severe; Profoundc.269 C > Tp.(Ser90Leu)MissensePPM3_strong, PS3_sup, PM2_sup, PP3_sup, PP1_strong26.70.890.96, U0.001, U0.00001988VCV000188842.14/12,676,8933HLMR54(3 Affected)59, 69Moderatec.293 C > A p.(Ala98Asp)MissenseVUSPM2_sup, PP1_strong22.10.610.911, D0.067, U0This studyHLGM23476, 78, 80 (98), 81 (105)Severec.302 A > Cp.(Gln101Pro)MissenseLPPP3_sup, PM2_sup, PP1_strong28.80.981, D0, D0.000007957This studyHLMR10369, 78Moderate to severec.416G > Tp.(Gly139Val)MissenseLPPM3_mod, PP3_sup, PM2_sup, PP1_mod340.991, D0, D0.00001988VCV000996638.12/ 19287372HLMS20222Severec.617T > Ap.(Leu206Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_sup---380VCV001526244.1/This studyc.2145G > Tp.(Lys715Asn)MissenseLPPM3_strong, PM2_sup, PP1_sup0.35, U0.280.252, B19.120.0001154VCV000043541.30/19,287,372HLMN1258, 75Moderate to severec.716T > Ap.(Val239Asp)MissenseLPPM3_mod, PP1_sup, PS3_sup, PM2_sup, PP3_sup26.60.940.86, U0.01, U0.0002028VCV000043566.42/12,676,893HLRB13369, 83Moderate to severec.1003T > Cp.(Phe335Leu)MissenseLPPM2_sup, PM3_mod, PP1_mod, PP3_sup27.10.861, D0.727, B0.001306VCV000004842.90/ 11317356HLMR13479(PTA0.25–2 kHz), 84, 88Severec.1172G > A p.(Ser391Asn)MissenseVUSPM2_sup, PM5, PP3_sup, PM3_mod26.60.78N/A0.032, U015355436c.2145G > T p.(Lys715Asn)MissenseLPPM3_mod, PS3_mod, PM2_sup, PP1_strong19.380.350.28, U0.252, B0.0001154VCV000043541.30/19287372HLGM272Moderate to severePediatric conditioned play audiometryc.1337 A > G p.(Gln446Arg)MissensePPM3_strong, PM2_sup, PP3_sup, PP1_strong27.10.911, D0.242, B0.00007578VCV000550505.11/107004802HLMR14`83, 83, 91, 99Severe; ProfoundHLHI07265 (76), 81Moderate to severeHLHI11279Severec.1341 + 3 A > CIntronic/SplicingVUSPM2_sup, PP1_sup, PM3_sup23.2-107.2--0VCV000550280.1 /19287372HLNK1273Severec.1544 + 1G > ASplice donorLPPVS1_strong, PM2_sup, PP1_sup35-124.28--0VCV000228395.7 /This studyHLHI03553, 63, 73, 73, 79 (99)Moderate to severec.1667 A > G p.(Tyr556Cys)MissensePPM3_strong, PM2_sup, PP3_sup, PP1_strong26.40.891, D0, D0.00001594VCV000446456.12/9,618,167HLMS18192Moderate to severec.2028delGp.(Arg677fs2)MissenseLPPVS1_strong, PM2_sup, PP1_sup----0VCV001333518.1/ This study TECTA NM_005422.24HLSNA1264, 73Moderate to severec.4248 C > T p.(Gly1416=)Synonymous/ SplicingVUSPP1_sup, PM2_sup15.38267.2--0This study2HLMN43(2 Affected )73Severec.6103G > Tp.(Glu2035Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_sup49---0VCV001687445.1/This study TMPRSS3 NM_001256317.3HLHI04476, 80, 89Severec.581G > Tp.(Cys194Phe)MissenseLPPM2_sup, PM3_sup, PM2_sup, PP1_strong240.551, D0, D0.000007077VCV000996722.1/11,424,922• TPRN NM_001128228.3HLZM06466 (82), 74, 79, 90Moderate to Severec.225_235del11p.(Gly76AlafsTer150)FrameshiftPPM3_sup, PVS1_strong, PM2_sup, PP1_strong----0.0001420VCV000000137.11/20,170,898Variants were classified according to ACMG guidelines as P (pathogenic), LP (likely pathogenic), VUS (variant of uncertain significance), U (uncertain) or LB (likely benign). The number of affected individuals refers to affected participants in each family who were sequenced either through exome sequencing or targeted Sanger sequencing. For families that showed genetic heterogeneity, values in parentheses represent the number of affected individuals which segregated the variant in the specified gene. PTA0.5–4 kHz values given in parenthesis represent the threshold for the ear showing the worse hearing loss for those having asymmetric losses. CADD (Combined Annotation Dependent Depletion) scores of above 20 indicate the variants predicted to be among the top 1% most deleterious variants in the human genome. Variants are homozygous except in families HLMS2022 and HLMR13 in which compound heterozygous variants were detected. 1Families for which results of some affected individuals were previously reported by us, but one or a few affected members remained undiagnosed. Exome sequencing performed as part of the current study identified a different causative gene variant, indicating genetic heterogeneity within the families. 2Variant not segregating in all affected participants of a family. Either a variant is not present in an affected individual, or he/she is a heterozygous carrier for the respective pathogenic variant. 3Genetic heterogeneity, the second gene identified in the family members, is reported in Table 2. The frequencies of all variants were also checked in 300 unrelated ethnically matched controls. Except for a few, none were detected in any control individual. The exceptions were one heterozygous individual each for the six SLC26A4 variants c.416G > T; p.(Gly139Val), c.716T > A; p.(Val239Asp), c.1003T > C; p.(Phe335Leu), c.1337A > G; p.(Gln446Arg), c.1667 A > G; p.(Tyr556Cys) and c.2145G > T; p.(Lys715Asn). Table 1Molecular genetic findings in participating families.GeneFamilyNo. of affected individualsPTA0.5–4 kHz of affected individuals and type of HLVariantEffectPathogenicityFrequencyClinVar ID/PMIDACMG criteria and classificationCADDREVEL/MaxEnt (Δ%)PolyPhenSIFTgnomAD v2 BSND NM_057176.3HLHT01/025Audiograms not available, reported to be moderate to severec.35T > Cp.(Ile12Thr)MissensePPM2_sup, PP3_sup, PP1_strong, PM3_strong25.60.941, D0, D0.00001989VCV000004388.16/19,646,679 CDH23 NM_022124.6HLMR11765, 75, 82, 83Moderate to severec.1373 C > Tp.(Pro458Leu)MissenseVUSPM2_sup, PP1_strong310.681, D0, D0VCV000517256.4/ This studyHLHI123Audiograms not available, reported to be moderate to severec.1621G > Ap.(Glu541Lys)MissenseVUSPM2_sup, PP1_mod,PM3_mod23.60.320.44, UN/A0.0001308VCV000252736.8/ 35,020,051HLMN7258Moderatec.4568 A > T p.(Asn1523Ile)MissenseVUSPP1_sup, PP3_supPM2_sup,27.70.840.999, DN/A0This studyHLMR8484, 89Severec.4730G > Cp.(Arg1577Pro)MissenseVUSPP1_strong,PM2_sup26.10.511, DN/A0VCV001502044.5This study1HLAI-18 IV:24(1 Affected)73Severe; Progressive reportedc.7814 A > G p.(Asn2605Ser)MissenseLPPP1_sup, PM2_sup, PP3_sup, PM3_strong25.20.770.99, D0, D0.00003111VCV000862821.7 /27,573,290 EPS8 NM_004447.6HLHI09575, 80Severec.204 + 1G > ASplice donorPPVS1_strong, PM2_sup, PP1_strong34-127.4--0This studyESPN NM_031475.3HLMR9865, 74, 74, 74, 74 80, 81Moderate to severec.1720delGp.(Ala574ArgfsTer72)FrameshiftPPVS1_strong,PM2_sup, PP1_strong----0This studyHLMR6364, 75Moderate to severec.2254G > Tp.(Glu752Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_mod46---0.000009410This study ESRRB NM_001379180.1HLHI052Audiograms not available(Reported moderate to severe)c.322G > Tp.(Asp108Tyr)MissenseVUSPP3_sup,PM2_sup, PP1_sup, PM3_srong29.60.97N/A0, D031835641GIPC3 NM_133261.3HLZM02260Moderatec.356_357delAGp.(Glu119GlyfsTer36)FrameshiftPPVS1_strong,PM2_sup, PP1_sup----0This studyHLAM03576, 80, 85Severec.788-9G > AIntronic/SplicingVUSPP1_strong, PM2_sup23.9-82.16--0.00003257This study GJB2 NM_004004HLZM04244, 48Moderatec.358_360delp.(Glu120del)In-frame deletionLPPM3_strong, PS3_sup, PM2_sup, PP1_sup----0.00007101VCV000017006.36/10,544,226HLZM01553, 86, 94Moderate to severec.439G > Ap.(Glu147Lys)MissensePPP3_sup,PP1_strong,PM2_sup, PM3_strong,25.30.981, D0.001,U0.00001195VCV000371709.37/14,676,473HLNK22Moderate to severec.598G > Ap.(Gly200Arg)MissenseLPPM3_strong, PP1_sup, PP3_sup, PS3_sup, PM2_sup26.80.710.9620.0410VCV000225222.919567088 GRXCR1 NM_0010804762HLGM075(4 Affected)79, 84, 90 (106)Severec.412 C > Tp.Arg138CysMissenseLPPP1_strong, PM3_modPM2_sup29.50.521, D0, D0.00006022VCV000000196.3/ 20137778 HGF NM_000601.5HLMR12763, 64, 66, 68 (85)Moderate to severec.482+1986_1988delTGADeletionLPPM2_sup, PP1_strong, PM3_strong,----019576567 ILDR1 NM_001199799.22HLMN54(3 Affected)53, 55, 80Moderate to severec.1384 C > Tp.(Arg462Ter)NonsenseLPPM2_sup, PP1_mod, PM3_mod, PVS1_strong,35---0.00003622VCV000562073.5/30,303,587LHFPL5 NM_182548.42HLGM266(4 Affected)45 (61), 84, 99Moderate to severe; Profoundc.250delCp.(Leu84Ter)FrameshiftPPM2_sup, PP1_strong, PM3_mod, PVS1_strong----0.000007953VCV000001694.3/16,459,341 MARVELD2 NM_001038603.33HLMR43(2 Affected)76, 89, 96Severe; Profoundc.1331 + 2T > Cp.(Cys395fsTer403)Splice donorPPM3_mod, PVS1_strong, PM2_sup, PP1_sup34-110.7--0.00003980VCV000871776.24/17,186,462 MYO3A NM_017433.5HLMN3339, 73Mild; Severec.892 C > Tp.(Gln298Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_mod40---0.0001154VCV001687422.3/This study MYO6 NM_004999.4IHT02385, 90, 93.75Severec.261 + 7 A > GIntronic/SplicingVUSPM2_sup, BP4_sup16.67-Benign-0This study COL9A1 NM_001851.4c.2504-5T > CIntronic/SplicingVUSPM2_sup, BP4_sup14.311754Benign-0.0001157VCV001112686.5/This studyMYO7A NM_000260.4HLMR7359, 61Moderatec.3136dupCp.(Leu1046ProfsTer9)FrameshiftPPVS1_strong, PM2_sup, PM3_mod, PP1_mod----0VCV000996717.3/18,181,211MYO15A NM_016239.4HLMN6264, 85Moderate to severec.1657del (p.Arg553GlyfsTer76)FrameshiftLPPM3_strong, PVS1_strong, PM2_sup----0VCV002572402.1/27573290HLSHI012Severec.2199dupCp.(Asp734ArgfsTer246)FrameshiftLPPP1_sup, PVS1_strong, PM2_sup----0.00001935This studyHLHI08263, 70Moderate to severec.3756 + 1G > TSplice donorLPPVS1_strong, PM2_sup, PM3_sup29.9-141.6--0.00003185VCV000006954.1/ 117350291HLAI-27(V:4)4(1 affected)Reported moderate to severec. 8340G > A p.(Thr2780Thr)Synonymous/SplicingLPPVS1_strong, PM3_mod, PM2_sup22.8-96.8-0.000008027VCV000236038.22/ 30139988 OTOA NM_144672.4HLHKA01359, 63Moderate to severePlay audiometry performed for the third affected demonstrated moderate hearing lossc.635 + 2T > ASplice donorLPPVS1_strong, PM2_sup, PP1_mod, PM3_sup31-113.45--0.00002875VCV002990394.1/ 37561809 OTOF NM_194248.3HLGM20461, 61, 63, 66Moderatec.2965_2967delp.(Phe989del)In-frame deletionLPPM2_sup, PM3_mod, PP1_strong,----0.000004088VCV000634821.5/27,573,290HLHKA03288Severec.3262del p.(Leu1088CysfsTer58)FrameshiftLPPVS1_strong, PM2_sup, PP1_sup----0This studyHLGM11384, 85, 85Severec.4362 + 1G > CSplice donorLPPVS1_strong, PM2_sup, PP1_mod33-132.91--0VCV000634822.2/This study PCDH15 NM_001142764HLAI-03469, 94, 96Progressive reported, Moderate to profoundc.2368_2370delp.(Val790del)In-frame deletionLPPM2_sup, PM4, PM3_sup, PP1_strong----025930172HLMN2259, 76Moderate to severec.2858_2862dupTTTATp.(Ala955PhefsTer16)FrameshiftLPPVS1_strong, PM2_sup, PP1_sup----0VCV001526060.1/This study POU3F4 NM_000307.5HLHI103No audiometry due to noisec.747delG p.(Lys249AsnfsTer3)FrameshiftLPPVS1_strong,PM2_sup, PP1_mod----0This study SLC26A4 NM_000441.2HLGM09671, 89Severec.235 C > Tp.(Arg79Ter)NonsensePPM3_mod,PVS1_strong,PM2_sup, PP1_strong37---0VCV000188950.10/19,648,736HLZM07588, 90, 95, 103Severe; Profoundc.269 C > Tp.(Ser90Leu)MissensePPM3_strong, PS3_sup, PM2_sup, PP3_sup, PP1_strong26.70.890.96, U0.001, U0.00001988VCV000188842.14/12,676,8933HLMR54(3 Affected)59, 69Moderatec.293 C > A p.(Ala98Asp)MissenseVUSPM2_sup, PP1_strong22.10.610.911, D0.067, U0This studyHLGM23476, 78, 80 (98), 81 (105)Severec.302 A > Cp.(Gln101Pro)MissenseLPPP3_sup, PM2_sup, PP1_strong28.80.981, D0, D0.000007957This studyHLMR10369, 78Moderate to severec.416G > Tp.(Gly139Val)MissenseLPPM3_mod, PP3_sup, PM2_sup, PP1_mod340.991, D0, D0.00001988VCV000996638.12/ 19287372HLMS20222Severec.617T > Ap.(Leu206Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_sup---380VCV001526244.1/This studyc.2145G > Tp.(Lys715Asn)MissenseLPPM3_strong, PM2_sup, PP1_sup0.35, U0.280.252, B19.120.0001154VCV000043541.30/19,287,372HLMN1258, 75Moderate to severec.716T > Ap.(Val239Asp)MissenseLPPM3_mod, PP1_sup, PS3_sup, PM2_sup, PP3_sup26.60.940.86, U0.01, U0.0002028VCV000043566.42/12,676,893HLRB13369, 83Moderate to severec.1003T > Cp.(Phe335Leu)MissenseLPPM2_sup, PM3_mod, PP1_mod, PP3_sup27.10.861, D0.727, B0.001306VCV000004842.90/ 11317356HLMR13479(PTA0.25–2 kHz), 84, 88Severec.1172G > A p.(Ser391Asn)MissenseVUSPM2_sup, PM5, PP3_sup, PM3_mod26.60.78N/A0.032, U015355436c.2145G > T p.(Lys715Asn)MissenseLPPM3_mod, PS3_mod, PM2_sup, PP1_strong19.380.350.28, U0.252, B0.0001154VCV000043541.30/19287372HLGM272Moderate to severePediatric conditioned play audiometryc.1337 A > G p.(Gln446Arg)MissensePPM3_strong, PM2_sup, PP3_sup, PP1_strong27.10.911, D0.242, B0.00007578VCV000550505.11/107004802HLMR14`83, 83, 91, 99Severe; ProfoundHLHI07265 (76), 81Moderate to severeHLHI11279Severec.1341 + 3 A > CIntronic/SplicingVUSPM2_sup, PP1_sup, PM3_sup23.2-107.2--0VCV000550280.1 /19287372HLNK1273Severec.1544 + 1G > ASplice donorLPPVS1_strong, PM2_sup, PP1_sup35-124.28--0VCV000228395.7 /This studyHLHI03553, 63, 73, 73, 79 (99)Moderate to severec.1667 A > G p.(Tyr556Cys)MissensePPM3_strong, PM2_sup, PP3_sup, PP1_strong26.40.891, D0, D0.00001594VCV000446456.12/9,618,167HLMS18192Moderate to severec.2028delGp.(Arg677fs2)MissenseLPPVS1_strong, PM2_sup, PP1_sup----0VCV001333518.1/ This study TECTA NM_005422.24HLSNA1264, 73Moderate to severec.4248 C > T p.(Gly1416=)Synonymous/ SplicingVUSPP1_sup, PM2_sup15.38267.2--0This study2HLMN43(2 Affected )73Severec.6103G > Tp.(Glu2035Ter)NonsenseLPPVS1_strong, PM2_sup, PP1_sup49---0VCV001687445.1/This study TMPRSS3 NM_001256317.3HLHI04476, 80, 89Severec.581G > Tp.(Cys194Phe)MissenseLPPM2_sup, PM3_sup, PM2_sup, PP1_strong240.551, D0, D0.000007077VCV000996722.1/11,424,922• TPRN NM_001128228.3HLZM06466 (82), 74, 79, 90Moderate to Severec.225_235del11p.(Gly76AlafsTer150)FrameshiftPPM3_sup, PVS1_strong, PM2_sup, PP1_strong----0.0001420VCV000000137.11/20,170,898Variants were classified according to ACMG guidelines as P (pathogenic), LP (likely pathogenic), VUS (variant of uncertain significance), U (uncertain) or LB (likely benign). The number of affected individuals refers to affected participants in each family who were sequenced either through exome sequencing or targeted Sanger sequencing. For families that showed genetic heterogeneity, values in parentheses represent the number of affected individuals which segregated the variant in the specified gene. PTA0.5–4 kHz values given in parenthesis represent the threshold for the ear showing the worse hearing loss for those having asymmetric losses. CADD (Combined Annotation Dependent Depletion) scores of above 20 indicate the variants predicted to be among the top 1% most deleterious variants in the human genome. Variants are homozygous except in families HLMS2022 and HLMR13 in which compound heterozygous variants were detected. 1Families for which results of some affected individuals were previously reported by us, but one or a few affected members remained undiagnosed. Exome sequencing performed as part of the current study identified a different causative gene variant, indicating genetic heterogeneity within the families. 2Variant not segregating in all affected participants of a family. Either a variant is not present in an affected individual, or he/she is a heterozygous carrier for the respective pathogenic variant. 3Genetic heterogeneity, the second gene identified in the family members, is reported in Table 2. The frequencies of all variants were also checked in 300 unrelated ethnically matched controls. Except for a few, none were detected in any control individual. The exceptions were one heterozygous individual each for the six SLC26A4 variants c.416G > T; p.(Gly139Val), c.716T > A; p.(Val239Asp), c.1003T > C; p.(Phe335Leu), c.1337A > G; p.(Gln446Arg), c.1667 A > G; p.(Tyr556Cys) and c.2145G > T; p.(Lys715Asn). Molecular genetic findings in participating families. BSND NM_057176.3 c.35T > C p.(Ile12Thr) P PM2_sup, PP3_sup, PP1_strong, PM3_strong VCV000004388.16/ 19,646,679 CDH23 NM_022124.6 65, 75, 82, 83 Moderate to severe c.1373 C > T p.(Pro458Leu) VUS PM2_sup, PP1_strong c.1621G > A p.(Glu541Lys) VUS PM2_sup, PP1_mod, PM3_mod 58 Moderate VUS PP1_sup, PP3_sup PM2_sup, 84, 89 Severe c.4730G > C p.(Arg1577Pro) VUS PP1_strong, PM2_sup VCV001502044.5 This study 4 (1 Affected) 73 Severe; Progressive reported LP PP1_sup, PM2_sup, PP3_sup, PM3_strong EPS8 NM_004447.6 75, 80 Severe P PVS1_strong, PM2_sup, PP1_strong 65, 74, 74, 74, 74 80, 81 Moderate to severe c.1720delG p.(Ala574ArgfsTer72) P PVS1_strong, PM2_sup, PP1_strong 64, 75 Moderate to severe c.2254G > T p.(Glu752Ter) LP PVS1_strong, PM2_sup, PP1_mod ESRRB NM_001379180.1 Audiograms not available (Reported moderate to severe) c.322G > T p.(Asp108Tyr) VUS PP3_sup, PM2_sup, PP1_sup, PM3_srong 60 Moderate c.356_357delAG p.(Glu119GlyfsTer36) P PVS1_strong, PM2_sup, PP1_sup 76, 80, 85 Severe VUS PP1_strong, PM2_sup GJB2 NM_004004 44, 48 Moderate c.358_360del p.(Glu120del) LP PM3_strong, PS3_sup, PM2_sup, PP1_sup 53, 86, 94 Moderate to severe c.439G > A p.(Glu147Lys) P PP3_sup, PP1_strong, PM2_sup, PM3_strong, c.598G > A p.(Gly200Arg) LP PM3_strong, PP1_sup, PP3_sup, PS3_sup, PM2_sup VCV000225222.9 19567088 GRXCR1 NM_001080476 5 (4 Affected) 79, 84, 90 (106) Severe c.412 C > T p.Arg138Cys LP PP1_strong, PM3_mod PM2_sup HGF NM_000601.5 63, 64, 66, 68 (85) Moderate to severe LP PM2_sup, PP1_strong, PM3_strong, ILDR1 NM_001199799.2 4 (3 Affected) 53, 55, 80 Moderate to severe c.1384 C > T p.(Arg462Ter) LP PM2_sup, PP1_mod, PM3_mod, PVS1_strong, 6 (4 Affected) 45 (61), 84, 99 Moderate to severe; Profound c.250delC p.(Leu84Ter) P PM2_sup, PP1_strong, PM3_mod, PVS1_strong MARVELD2 NM_001038603.3 3 (2 Affected) 76, 89, 96 Severe; Profound c.1331 + 2T > C p.(Cys395fsTer403) P PM3_mod, PVS1_strong, PM2_sup, PP1_sup MYO3A NM_017433.5 39, 73 Mild; Severe c.892 C > T p.(Gln298Ter) LP PVS1_strong, PM2_sup, PP1_mod VCV001687422.3/ This study MYO6 NM_004999.4 85, 90, 93.75 Severe VUS PM2_sup, BP4_sup COL9A1 NM_001851.4 VUS PM2_sup, BP4_sup VCV001112686.5/ This study 59, 61 Moderate c.3136dupC p.(Leu1046ProfsTer9) P PVS1_strong, PM2_sup, PM3_mod, PP1_mod 64, 85 Moderate to severe LP PM3_strong, PVS1_strong, PM2_sup VCV002572402.1/ 27573290 c.2199dupC p.(Asp734ArgfsTer246) LP PP1_sup, PVS1_strong, PM2_sup 63, 70 Moderate to severe LP PVS1_strong, PM2_sup, PM3_sup 1HLAI-27 (V:4) 4 (1 affected) LP PVS1_strong, PM3_mod, PM2_sup OTOA NM_144672.4 59, 63 Moderate to severe Play audiometry performed for the third affected demonstrated moderate hearing loss LP PVS1_strong, PM2_sup, PP1_mod, PM3_sup OTOF NM_194248.3 61, 61, 63, 66 Moderate c.2965_2967del p.(Phe989del) LP PM2_sup, PM3_mod, PP1_strong, 88 Severe LP PVS1_strong, PM2_sup, PP1_sup 84, 85, 85 Severe LP PVS1_strong, PM2_sup, PP1_mod PCDH15 NM_001142764 69, 94, 96 Progressive reported, Moderate to profound c.2368_2370del p.(Val790del) LP PM2_sup, PM4, PM3_sup, PP1_strong 59, 76 Moderate to severe c.2858_2862dupTTTAT p.(Ala955PhefsTer16) LP PVS1_strong, PM2_sup, PP1_sup POU3F4 NM_000307.5 LP PVS1_strong, PM2_sup, PP1_mod SLC26A4 NM_000441.2 71, 89 Severe c.235 C > T p.(Arg79Ter) P PM3_mod, PVS1_strong, PM2_sup, PP1_strong 88, 90, 95, 103 Severe; Profound c.269 C > T p.(Ser90Leu) P PM3_strong, PS3_sup, PM2_sup, PP3_sup, PP1_strong 4 (3 Affected) 59, 69 Moderate VUS PM2_sup, PP1_strong 76, 78, 80 (98), 81 (105) Severe c.302 A > C p.(Gln101Pro) LP PP3_sup, PM2_sup, PP1_strong 69, 78 Moderate to severe c.416G > T p.(Gly139Val) LP PM3_mod, PP3_sup, PM2_sup, PP1_mod c.617T > A p.(Leu206Ter) LP PVS1_strong, PM2_sup, PP1_sup VCV001526244.1/ This study c.2145G > T p.(Lys715Asn) LP PM3_strong, PM2_sup, PP1_sup 58, 75 Moderate to severe c.716T > A p.(Val239Asp) LP PM3_mod, PP1_sup, PS3_sup, PM2_sup, PP3_sup 69, 83 Moderate to severe c.1003T > C p.(Phe335Leu) LP PM2_sup, PM3_mod, PP1_mod, PP3_sup 79(PTA0.25–2 kHz), 84, 88 Severe VUS PM2_sup, PM5, PP3_sup, PM3_mod LP PM3_mod, PS3_mod, PM2_sup, PP1_strong VCV000043541.30/ 19287372 Moderate to severe Pediatric conditioned play audiometry P PM3_strong, PM2_sup, PP3_sup, PP1_strong VCV000550505.11/ 10700480 `83, 83, 91, 99 Severe; Profound 65 (76), 81 Moderate to severe 79 Severe VUS PM2_sup, PP1_sup, PM3_sup 73 Severe LP PVS1_strong, PM2_sup, PP1_sup 53, 63, 73, 73, 79 (99) Moderate to severe P PM3_strong, PM2_sup, PP3_sup, PP1_strong c.2028delG p.(Arg677fs2) LP PVS1_strong, PM2_sup, PP1_sup TECTA NM_005422.24 64, 73 Moderate to severe VUS PP1_sup, PM2_sup 3 (2 Affected ) 73 Severe c.6103G > T p.(Glu2035Ter) LP PVS1_strong, PM2_sup, PP1_sup TMPRSS3 NM_001256317.3 76, 80, 89 Severe c.581G > T p.(Cys194Phe) LP PM2_sup, PM3_sup, PM2_sup, PP1_strong TPRN NM_001128228.3 66 (82), 74, 79, 90 Moderate to Severe c.225_235del11 p.(Gly76AlafsTer150) P PM3_sup, PVS1_strong, PM2_sup, PP1_strong Variants were classified according to ACMG guidelines as P (pathogenic), LP (likely pathogenic), VUS (variant of uncertain significance), U (uncertain) or LB (likely benign). The number of affected individuals refers to affected participants in each family who were sequenced either through exome sequencing or targeted Sanger sequencing. For families that showed genetic heterogeneity, values in parentheses represent the number of affected individuals which segregated the variant in the specified gene. PTA0.5–4 kHz values given in parenthesis represent the threshold for the ear showing the worse hearing loss for those having asymmetric losses. CADD (Combined Annotation Dependent Depletion) scores of above 20 indicate the variants predicted to be among the top 1% most deleterious variants in the human genome. Variants are homozygous except in families HLMS2022 and HLMR13 in which compound heterozygous variants were detected. 1Families for which results of some affected individuals were previously reported by us, but one or a few affected members remained undiagnosed. Exome sequencing performed as part of the current study identified a different causative gene variant, indicating genetic heterogeneity within the families. 2Variant not segregating in all affected participants of a family. Either a variant is not present in an affected individual, or he/she is a heterozygous carrier for the respective pathogenic variant. 3Genetic heterogeneity, the second gene identified in the family members, is reported in Table 2. The frequencies of all variants were also checked in 300 unrelated ethnically matched controls. Except for a few, none were detected in any control individual. The exceptions were one heterozygous individual each for the six SLC26A4 variants c.416G > T; p.(Gly139Val), c.716T > A; p.(Val239Asp), c.1003T > C; p.(Phe335Leu), c.1337A > G; p.(Gln446Arg), c.1667 A > G; p.(Tyr556Cys) and c.2145G > T; p.(Lys715Asn). Fig. 2Audiograms of affected participants from selected families exhibiting bilateral symmetric hearing loss. Pure tone air conduction thresholds shown for sounds of different frequencies. A circle, cross and hexagon show threshold levels for available affected members in a family. Audiometry was performed in ambient noise conditions. The hearing thresholds are shown for better hearing ears. Audiograms of affected participants from selected families exhibiting bilateral symmetric hearing loss. Pure tone air conduction thresholds shown for sounds of different frequencies. A circle, cross and hexagon show threshold levels for available affected members in a family. Audiometry was performed in ambient noise conditions. The hearing thresholds are shown for better hearing ears. Fig. 3Audiograms of affected participants exhibiting asymmetric hearing loss. Pure-tones for both ears are presented for participants with asymmetric hearing loss. A circle represents the profile for the right ear and a cross represents hearing ability in the left ear. Audiograms of affected participants exhibiting asymmetric hearing loss. Pure-tones for both ears are presented for participants with asymmetric hearing loss. A circle represents the profile for the right ear and a cross represents hearing ability in the left ear. Molecular genetic characterization reveals variants in bona fide human hearing loss genes Genetic analyses revealed 57 variants in 26 genes previously associated with hearing loss (Tables 1 and 2). Among the identified likely pathogenic variants, 93% were homozygous, while in two families HLMS2022 and HLMR13, the variants were inherited in a compound heterozygous state. Of these, 24 variants (42%) were novel while 33 variants (58%) were previously reported. A few variants were identified in participants of more than one family. The SLC26A4 variant c.1337 A > G; p.(Gln446Arg) was detected in 3 families. Additionally, the SLC26A4 variant c.2145G > T; p.(Lys715Asn) was found in participants of 2 families, existing in a compound heterozygous state, along with a different variant in each family. Significantly, 46% of the pathogenic variants were detected in only three genes, including SLC26A4, which accounted for approximately 30% of hearing loss, while variants of CDH23 and MYO15A contributed to 9% and 7%, respectively, of individuals with hearing loss (Fig. 4). Among the deleterious variants identified in this study, 23 (40%) were missense, 13 (23%) were predicted to alter splice sites, 11 (19%) were frameshifts, 3 (5%) were in-frame deletions, 1 (2%) was an intronic deletion and 6 (11%) introduced premature stop codons (Table 1). Table 2Genetic heterogeneity observed in the cohort.Family IDFirst gene identified in the familySecond gene identified and corresponding affected individualsVariantin the second geneEffectPathogenicitygnomAD v2 frequencyPMID/ClinVar ID/First reportACMGREVELPolyPhen2SIFTCADDHLMR4 MARVELD2 NM_001038603.3c.1331+2T> Cp.(Cys395Ter)(V:2, VI:2) MYO7A NM_000260.4(V:2, V:5)c.4994_4997del4p.(Asp1665ValfsTer21)FrameshiftVUSPM2_sup, PVS1_strong----0This studyHLMR5 SLC26A4 NM_000441.2c.293 C> A p.(Ala98Asp)(IV:1, IV:6, IV:7) MSRB3 NM_001031679.3(II:8)c.412-1G> ASplice acceptorVUSPM2_sup, PM3_sup---340.000032VCV000562077.3 /30303587 Table 2Genetic heterogeneity observed in the cohort.Family IDFirst gene identified in the familySecond gene identified and corresponding affected individualsVariantin the second geneEffectPathogenicitygnomAD v2 frequencyPMID/ClinVar ID/First reportACMGREVELPolyPhen2SIFTCADDHLMR4 MARVELD2 NM_001038603.3c.1331+2T> Cp.(Cys395Ter)(V:2, VI:2) MYO7A NM_000260.4(V:2, V:5)c.4994_4997del4p.(Asp1665ValfsTer21)FrameshiftVUSPM2_sup, PVS1_strong----0This studyHLMR5 SLC26A4 NM_000441.2c.293 C> A p.(Ala98Asp)(IV:1, IV:6, IV:7) MSRB3 NM_001031679.3(II:8)c.412-1G> ASplice acceptorVUSPM2_sup, PM3_sup---340.000032VCV000562077.3 /30303587 Genetic heterogeneity observed in the cohort. MARVELD2 NM_001038603.3 c.1331+2T> C p.(Cys395Ter) (V:2, VI:2) MYO7A NM_000260.4 (V:2, V:5) c.4994_4997del4 p.(Asp1665ValfsTer21) VUS PM2_sup, PVS1_strong SLC26A4 NM_000441.2 c.293 C> A p.(Ala98Asp) (IV:1, IV:6, IV:7) MSRB3 NM_001031679.3 (II:8) VUS PM2_sup, PM3_sup VCV000562077 .3 /30303587 Fig. 4Contribution of moderate to severe hearing loss genes in our Pakistani cohort. Percentages were calculated for the families reported in this study and combining data from our previous reports5,23,60, including only families with clear genetic diagnoses after targeted capture and sequencing or those subjected to exome sequencing. Other genes include BSND, ESP8, ESRRB, GRXCR1, HGF, ILDR1, LHFPL5, MARVELD2, MYO3A, MYO6, MYO7A, OTOA, POU3F4, TMPRSS3 and TPRN in the cohort of 54 families. In families with genetic heterogeneity, the gene affecting most participants in the respective family is considered for calculations. For the percentage contribution of 124 families, other genes include, TRIOBP, PTPRQ, MET, GRXCR2, CDC14A, DFNB59 (PJVK), EPS8, ESRRB, GRXCR1, ILDR1, KCNQ4, LHFPL5, MARVELD2, MYO3A, OTOA, POU3F4, USH1C and USH1G. SLC26A4 is the top contributor to hearing loss, accounting for 24% of the families in our cohort which are recruited mainly from the Punjab Province of Pakistan. Contribution of moderate to severe hearing loss genes in our Pakistani cohort. Percentages were calculated for the families reported in this study and combining data from our previous reports5,23,60, including only families with clear genetic diagnoses after targeted capture and sequencing or those subjected to exome sequencing. Other genes include BSND, ESP8, ESRRB, GRXCR1, HGF, ILDR1, LHFPL5, MARVELD2, MYO3A, MYO6, MYO7A, OTOA, POU3F4, TMPRSS3 and TPRN in the cohort of 54 families. In families with genetic heterogeneity, the gene affecting most participants in the respective family is considered for calculations. For the percentage contribution of 124 families, other genes include, TRIOBP, PTPRQ, MET, GRXCR2, CDC14A, DFNB59 (PJVK), EPS8, ESRRB, GRXCR1, ILDR1, KCNQ4, LHFPL5, MARVELD2, MYO3A, OTOA, POU3F4, USH1C and USH1G. SLC26A4 is the top contributor to hearing loss, accounting for 24% of the families in our cohort which are recruited mainly from the Punjab Province of Pakistan. Members of nine families (17%) (Tables 1 and 2; Fig. 1), HLAI-18, HLA1-27, HLGM07, HLGM26, HLMN4, HLMN5, HLMR1, HLMR4, and HLMR5 have intrafamilial genetic heterogeneity5. In particular, individuals IV:2, V:4, V:5 and II:8, in four families HLAI-18, HLAI-27, HLMR4 and HLMR5 respectively, were homozygous for pathogenic variants in different genes as compared to those detected in other affected family members (Fig. 1). Notably, a 19 years old individual (V:2) in family HLMR4, who is profoundly deaf (PTA 96 dB HL), was found to be homozygous for pathogenic variants in two genes; MYO7A (c.4994_4997del; p.(Asp1665Ter) and MARVELD2 (c.1331 + 2T > C; p.(Cys395Ter). In contrast, the other two affected individuals in the same family (11 and 30 years old) have a less severe hearing loss (PTA 76 dB HL and 89 dB HL, respectively) and were each homozygous for variants in only one of these two genes. This scenario is reminiscent of a previous study in which co-occurrence of GJB2 and SLC26A4 variants in one individual resulted in profound hearing loss18. In such families, an individual with two sets of biallelic variants of two different hearing loss genes is an observation consistent with the assumption that genetic modifiers are a source of some intrafamilial hearing loss variation5,19,20. In our previous report, the HGF c.482+1986_1988delTGA validated pathogenic variant21,22 was identified in family HLAI-18 that explained the hearing loss of two of four affected siblings5. Here, we identified a homozygous CDH23 c.7814 A > G; p.(Asn2605Ser) variant associated with hearing loss in another affected individual (IV:2). The hearing loss in the fourth similarly affected sibling (IV:6) could not be attributed to any of the reported hearing loss genes after exome sequencing, though he is heterozygous for a pathogenic TMPRSS3 c.316 C > T; p.(Arg106Cys) variant and heterozygous for a well characterized pathogenic HGF c.482+1986_1988delTGA deletion21,22 (Fig. 1). For family HLAI-27, we previously reported a TMC1 c.596 A > T; p.(Asn199Ile) variant to cause hearing loss in three out of four affected individuals23. In the current study, exome sequencing identified a known pathogenic homozygous synonymous variant in MYO15A c.8340G > A; p.(Thr2780=) in their affected relative V:4 (Table 1), which causes intron retention by introducing a premature stop codon in the mRNA24. In family HLMR5, a homozygous splice site variant in MSRB3 c.412-1G > A was identified for individual II:8. Exome data analyses followed by Sanger sequencing revealed that three of the third-degree relatives (grand-niece and grand-nephews) IV:1, IV:6 and IV:7 in the same family were homozygous for a SLC26A4 c.293 C > A; p.(Ala98Asp) variant (Fig. 1). Ala98 is not conserved in many vertebrate species (Fig. 5). However, a partially evolutionarily conserved amino acid can also be disease-causing when mutated. For example, the SLC26A4 p.(Thr99Arg) variant is associated with hearing loss in humans25 but in zebrafish the corresponding residue is a cysteine. In human, a well-established pathogenic substitution of SLC26A4 p.(Val239Asp)26 affects a residue which in zebrafish is an isoleucine and a leucine in African lungfish. Fig. 5Amino acid sequence alignments of CDH23 and SLC26A4 from diverse vertebrates. Alignments of amino acids affected by novel homozygous missense variants show conservation among representative vertebrate species performed by Clustal Omega. In the case of CDH23 Pro458, Asn1523, Arg1577 and SLC26A4 Gln101, the amino acids altered by the corresponding variants were highly conserved across orthologous proteins. In the case of human SLC26A4 Ala98, Ala is not conserved at the corresponding position, but a similar small volume amino residue is present at this position. There is an Ala residue in M. mulatta, L. africana, G. gallus, A. caroliensis, a Gly residue in S. scrofa, M. musculus, X. laevis, and a Cys in D. rerio. The symbols “*”, “:”, “.” indicate an identical residue, conserved residue with strongly similar properties, and a weakly conserved residue with weakly similar properties, respectively. The accession numbers used for the amino acid alignments are from NCBI and UniProt databases. CDH23: H. sapiens (NCBI RefSeq: NP_001165401.1); M. mulatta (NCBI RefSeq: XP_015002911.2); S. scrofa (UniProt: A0A4 × 1T995); M. musculus (NCBI RefSeq: NP_075859.2); L. africana (UniProt: G3UNP5); G. gallus (UniProt: A0A8V0ZCC7); A. carolinensis (UniProt: R4GAX0); X. laevis (NCBI RefSeq: XP_041424827.1); D. rerio (NCBI RefSeq: NP_999974.1). SLC26A4: H. sapiens (NCBI RefSeq: NP_000432.1); M. mulatta; (NCBI RefSeq: XP_014990157.2); Sus scrofa (NCBI RefSeq: XP_003357559.1); M. musculus (NCBI RefSeq: NP_035997.1); L. africana (UniProt: G3T5S4); G. gallus (NCBI RefSeq: XP_425419.3); A. carolinensis (UniProt: G1KQS1); X. laevis (NCBI RefSeq: NP_001087123.2); D. rerio (NCBI RefSeq: NP_001159387.1). Amino acid sequence alignments of CDH23 and SLC26A4 from diverse vertebrates. Alignments of amino acids affected by novel homozygous missense variants show conservation among representative vertebrate species performed by Clustal Omega. In the case of CDH23 Pro458, Asn1523, Arg1577 and SLC26A4 Gln101, the amino acids altered by the corresponding variants were highly conserved across orthologous proteins. In the case of human SLC26A4 Ala98, Ala is not conserved at the corresponding position, but a similar small volume amino residue is present at this position. There is an Ala residue in M. mulatta, L. africana, G. gallus, A. caroliensis, a Gly residue in S. scrofa, M. musculus, X. laevis, and a Cys in D. rerio. The symbols “*”, “:”, “.” indicate an identical residue, conserved residue with strongly similar properties, and a weakly conserved residue with weakly similar properties, respectively. The accession numbers used for the amino acid alignments are from NCBI and UniProt databases. CDH23: H. sapiens (NCBI RefSeq: NP_001165401.1); M. mulatta (NCBI RefSeq: XP_015002911.2); S. scrofa (UniProt: A0A4 × 1T995); M. musculus (NCBI RefSeq: NP_075859.2); L. africana (UniProt: G3UNP5); G. gallus (UniProt: A0A8V0ZCC7); A. carolinensis (UniProt: R4GAX0); X. laevis (NCBI RefSeq: XP_041424827.1); D. rerio (NCBI RefSeq: NP_999974.1). SLC26A4: H. sapiens (NCBI RefSeq: NP_000432.1); M. mulatta; (NCBI RefSeq: XP_014990157.2); Sus scrofa (NCBI RefSeq: XP_003357559.1); M. musculus (NCBI RefSeq: NP_035997.1); L. africana (UniProt: G3T5S4); G. gallus (NCBI RefSeq: XP_425419.3); A. carolinensis (UniProt: G1KQS1); X. laevis (NCBI RefSeq: NP_001087123.2); D. rerio (NCBI RefSeq: NP_001159387.1). Among the other families with genetic heterogeneity, affected individuals IV:5 in family HLMN4, IV:2 in HLMR1 and V:10 in HLGM26, had only the wild-type alleles for the genes mutated in other affected members (Fig. 1). Furthermore, individuals V:5 in family HLMR1, IV:7 in HLMN5, V:3 in HLGM07 and V:6 in HLGM26 were heterozygous for the deleterious gene variants while other affected individuals in their families were homozygous. Of particular note, heterozygous affected individuals in families HLMR1 and HLGM07 were from different marriages in relation to the affected individuals in their families with homozygous variants (Fig. 1). It is therefore possible that they may have inherited a second pathogenic allele in trans from a parent which could account for their hearing loss. Sequencing analyses in future could elucidate this. Different variants affect the same exons of MYO15A and SLC26A4 Two pathogenic variants of MYO15A, (c.1657delC and c.2199dupC), were located within exon 2. The association of variants in this exon with less severe hearing loss has been previously noted27. Interestingly, for SLC26A4 we observed four different deleterious variants (c.235 C > T, c.269 C > T, c.293 C > A and c.302 A > C) within exon 3 and two variants (c.1337 A > G and c.1341 + 3 A > C) in exon 11 or its intron. The clustering of variants in SLC26A4 exon 3 warrants analysis to identify whether this occurs by chance or due to it being a mutational hotspot. Impact of substitutions of CDH23 and SLC26A4 on predicted protein structure and function The longest isoform of CDH23 is 3,354 residues with an N-terminal signal peptide and 27 extracellular ectodomains (EC) (NCBI RefSeq: NP_071407.4) (Supplementary Fig. S1). Three novel variants, p.(Pro458Leu), p.(Asn1523Ile), and p.(Arg1577Pro) of human cadherin 23 (CDH23) in families HLMR11, HLMN7 and HLMR8, respectively, are predicted by AlphaMissense28 (https://alphamissense.hegelab.org/) to be likely pathogenic (Supplementary Table S2). All three residues are evolutionarily conserved (Fig. 5). Pro458 is in a linker region between EC4 and EC5. Asn1523 is located at the end of EC14, and Arg1577 is in EC15 (Supplementary Fig. S1). Reported crystal structures of sections of CDH2328 do not cover the domains where these three novel substitutions are located. Therefore, we evaluated the impact of each variant using structure modeling and with help of the reported structures of EC domains with their functions. The amino acid sequence of human CDH23 EC repeats are not well-conserved28 (Supplementary Fig. S1). However, the wild type residues Pro458, Asn1523, and Arg1577 are 93%, 70% and 19% conserved at the corresponding position of each EC repeat. Pro458 is part of a conserved XPXF/L sequence next to a Ca2+-binding motif (452-NENDN-456). The Pro458Leu substitution may disrupt Ca2+ binding, protein folding and orientation of EC4 with EC5. In this regard, there are 18 reported examples of substitutions of prolines in EC repeats associated with hearing loss (Supplementary Table S3). Asn1523 is in a Ca2+-binding motif (1519-DINDN-1523), and substitutions of Asn may change the binding affinity of Ca2+. The modeled protein structures show that Pro458 resides in a flexible linker region between EC4 and EC5 while Asn1523 is in a flexible region at the end of EC14 (Supplementary Fig. S1). The stability predictions of Pro458Leu and Asn1523Ile in the flexible region are not definitive (Supplementary Methods and Results, Supplementary Fig. S2 and Supplementary Table S4). Arg1577 is in EC15 and is not conserved among EC repeats of CDH23 (Supplementary Fig. S2). A modeled protein structure shows that Arg1577 is in a β-strand in EC15, and the substitution with Pro is predicted to destroy its β-sheet structure, which would disrupt the function of CDH23 (Supplementary Fig. S2 and Supplementary Table S4). Two novel variants SLC26A4, p.(Ala98Asp) and p.(Gln101Pro) identified in families HLMR5 and HLGM23 respectively are predicted as “likely pathogenic” by AlphaMissense29 (Supplementary Table S2). We explored protein stability changes of these two amino acid substitutions using the AlphaFold predicted structure of human SLC26A4 apo-form since a structure of human SLC26A4 has not been reported. Both Ala98 and Gln101 are in transmembrane helix (TM) 1 of human SLC26A4 (Supplementary Fig. S2). Amino acid sequence alignment shows that Ala98 in human SLC26A4 is conserved except for wild boar where there is a glycine and in fish there is a cysteine at the corresponding position (Fig. 5), suggesting that small volume residues are acceptable at this position30. The human SLC26A4 Gln101 residue is conserved among species (Fig. 5). Both p.(Ala98Asp) and p.(Gln101Pro) substitutions of human SLC26A4 are predicted to destabilize the structure SLC26A4 (Supplementary Table S4). We also evaluated amino acid substitution effects at the corresponding positions of wild boar SLC26A4 using its two cryo-EM structures determined as apo-and Cl−-bound forms. The amino acid sequence comparison showed that human SLC26A4 (NCBI RefSeq: NP_000432.1) is 91% identical and 95% similar to wild boar SLC26A4 (NCBI RefSeq: XP_003357559.1). Wild boar Gly98 and Gln101 are located at the corresponding positions of human SLC26A4 Ala98 and Gln101. Wild boar SLC26A4 p.(Gly98Asp) and p.(Gln101Pro) are predicted to destabilize both the apo and Cl−-forms (Supplementary Table S4). Substitution p.(Ala98Asp) in TM1 is predicted to form an extra hydrogen bond to the neighboring Arg409 in TM10 altering TM local structure which may destabilize the protein structure (Supplementary Fig. S2). A proline residue is known to make a kink in α-helix and destabilize the structure of soluble proteins. However, there are also reports that proline residue supports folding of transmembrane α-helices31,32. The severe hearing loss associated with the substitution of Pro for Gln101 may be explained by additional factors. In wild boar SLC26A4, Gln101 plays a critical role at an anion binding site and a substitution of Leu results in a reported 50% reduction of anion transport activity33. In human SLC26A4, we assume that the substitution of Pro for Gln101 also destabilizes TM1 and reduces anion transport activity. Additionally, Tyr105 and Asn457 in the same anion binding pocket of the wild boar SLC26A4 structure interacts with the same Cl− as does Gln10133 (Supplementary Fig. S2). In the human SLC26A4, Ala98 and Gln101 are conserved at the corresponding positions (Fig. 5 and Supplementary Fig. S2). It is likely that three residues Gln101, Tyr105, and Asn457 of human SLC26A4 bind to the anion directly as they do for wild boar SLC26A4 and play essential roles for the anion exchange mechanism33. In fact, p.(Tyr105Cys)34,35, p.(Asn457Ile)36, p.(Asn457Asp)37, p.(Asn457Lys)26, and p.(Asn457Tyr)38 of human SLC26A4 are reported in human families segregating hearing loss. Based on structural predictions and the sites of the reported variants, anion binding to Gln101, Tyr105, and Asn457 seem critical for auditory function. Results of experiments designed to test these predictions could resolve the uncertainty of variant pathogenicity. Novel variants predicted to alter splicing The three canonical splice site variants were identified and predicted to cause frameshifts resulting from exon skipping or use of a newly created splice site (Table 1, Supplementary Table S5). Intronic variants close to the canonical splice sites of two different hearing loss genes were identified in MYO6 (c.261 + 7 A > G) and in COL9A1 (c.2504-5T > C) in three affected siblings in family IHT02 but were not predicted to affect splicing (Table 1). However, analysis of the sequence indicated that the MYO6 c.261 + 7 A > G variant creates an intronic “gt” sequence at + 7 position with respect to exon 4 which may act as a donor splice site and lead to exon skipping39 or intron retention. Any deleterious effect of the MYO6 or COL9A1 variant remains to be determined experimentally. In contrast, an intronic GIPC3 variant and a TECTA synonymous variant were predicted to be highly damaging to splicing by various in-silico tools including SpliceAI (https://spliceailookup.broadinstitute.org/, accessed July 2024). The GIPC3 c.788-9G > A variant was predicted by varSEAK (https://varseak.bio/) to produce an alternate acceptor “ag” within the ag-exclusion zone (AGEZ), a region between the branch point and 3′ splice site which is typically devoid of AGs to prevent activation of cryptic acceptor sites40. Usage of an alternate acceptor site within AGEZ results in the inclusion in the mature transcript of nine intronic nucleotides at the newly formed splice junction, subsequently introducing a frameshift 22 bp downstream of the inclusion site. The c.788-9G > A segregated with the phenotype of five affected individuals in family HLAM03 (Table 1). Segregating in family HLSNA1 was a synonymous TECTA c.4248 C > T; p.(Gly1416=) variant, predicted to create a novel 3´ donor splice site within exon 13 which, if used, will delete 54 nucleotides or 18 amino acids from TECTA. Future experiments using mRNA isolated from affected individual VI:3 (Fig. 1) could be used further to test the pathogenicity. Variants of LHFPL5 and PCDH15 are associated with moderate to severe hearing loss A homozygous frameshift variant (c.250delC) of LHFPL5 was detected in 4 of 6 affected individuals of family HLGM26 (Fig. 1; Table 1). One family participant had moderate hearing loss while the others had severe and profound hearing loss (Fig. 2). The same LHFPL5 variant was reported to segregate with severe to profound hearing loss in different individuals from six other families from Pakistan6,41–43. Other variants of LHFPL5 are associated with hearing loss in Algerian, Emirati, Indian, Iranian, Omani, Pakistani, Palestinian, Tunisian, and Turkish populations. However, none of the alleles (Human Gene Mutation Database, https://www.hgmd.cf.ac.uk/, accessed July 2025) have been reported to cause a mild to moderate hearing loss in affected individuals. Thus, our study is the first to report individuals who have a moderate degree of hearing loss (PTA 45 dB HL) due to a frameshift variant of LHFPL5. One described pathogenic in-frame deletion c.2361_2363del in PCDH15 was segregating in family HLAI-03 who participated in this study. The four siblings were reported to have a progressive hearing loss (Fig. 2). The youngest 14 years old individual had a moderate hearing loss while the 22 years old had a severe hearing loss. We also detected a novel variant of PCDH15 c.2858_2862dup; p.(Ala955PhefsTer16) in family HLMN2. The frameshift variant introduces a premature stop codon in exon 21 of a 38-exon gene, potentially leading to a reduction of PCDH15 mRNA due to nonsense-mediated decay (NMD). However, this variant in PCDH15 is located in the last exon 21 of a shorter splice isoform (NCBI RefSeq: NM_001354430.2) and therefore a potentially toxic shorter version of full-length PCDH15 may be translated44. The two affected individuals exhibited moderate to severe (PTA 59 dB HL and PTA 76 dB HL) hearing loss as 10- and 12-year-old children (Fig. 2). Electroretinograms to evaluate vision were not available for any of the affected individuals. Nyctalopia and photophobia, early signs of Usher syndrome-associated retinopathy were absent in all affected members of the family at the time of enrollment in the study. Many variants of PCDH15 are known to cause either Usher syndrome type 1F45 or profound nonsyndromic hearing loss DFNB2346–49. However, there are a few exceptions as some alleles are reported to cause moderate hearing loss50,51, but none of these reported variant resulted in a progressive hearing loss as observed in affected individuals of family HLMN2. Further expansion of ESPN alleles detected in moderate to severe hearing loss In a previous study5, we identified variants in many genes associated for the first time with moderate to severe hearing loss instead of the usually described profound hearing loss. However, one of these phenotypic expansions for ESPN was only supported by its presence in affected members of a single family. Here, we identify two additional alleles of ESPN which were detected in 11 affected individuals with moderate to severe hearing loss in two different families (Table 1, Supplementary Table S1). Both variants are predicted to be loss of function alleles as they introduce premature stop codons in ESPN. Thus, our work supports the hypothesis that some putative null alleles of ESPN can cause moderate to severe hearing loss. X-linked hearing loss in members of one family Three affected male participants of family HLHI10 were hemizygous for a novel frameshift variant c.747del; p.(Lys249AsnfsTer4) in POU3F4 (NCBI RefSeq: NM_000307.5). Pathogenic alleles of POU3F4 are the most common cause of X-linked hearing loss, and the hearing loss associated with variants in this gene ranges from severe to profound52. A total of 135 different variants in POU3F4 have been reported which can cause hearing loss to autism spectrum disorder (Human Gene Mutation Database, accessed February 2026). Most of these variants have been identified in Caucasian populations while 12 are from the Asian populations53–55. Only 4 families have been reported in Pakistan to have hearing loss due to variants of POU3F46,56,57. Discussion Profound human hearing loss is genetically heterogeneous and has been comprehensively studied in many countries6. In contrast, the spectrum of genetic variants that cause recessively inherited moderate or severe human hearing loss is understudied. In a cohort of 54 multiplex families ascertained in Pakistan, we identified both reported and new variants in genes responsible for recessively inherited moderate to severe hearing loss or progressive hearing loss. Our study has broadened the genetic spectrum of recessively inherited moderate to severe hearing loss and has expanded the phenotype of LHFPL5 and PCDH15 to include a moderate or progressive hearing loss. Of note, in family HLGM26, we found four of the six affected participants to be homozygous for a pathogenic LHFPL5 frameshift variant. Progression of hearing loss was not reported but may be present since the younger eight years old affected individual had a significantly lesser degree of hearing loss (45 dB HL) as compared to the 21 years old affected individual (99 dB HL). Interestingly, the same allele has been reported to cause only profound hearing loss in multiple affected individuals from Pakistan. After combining the data for the 54 families reported in this study with 70 previously published families for which we had exome data (Fig. 4), we analyzed a total of 124 families. Pathogenic variants of SLC26A4 emerged as the most prevalent cause of recessively inherited moderate to severe hearing loss identified in 30% of the families. In addition, among all gene variants, only two pathogenic alleles of SLC26A4 were detected in members of more than one family. These included SLC26A4 c.1337 A > G; p.(Gln446Arg) allele associated with hearing loss in three families and SLC26A4 c.2145G > T; p.(Lys715Asn) detected as an allele in two families (Table 1). As noted earlier5, neither SLC26A4 variant type nor a specific allele is correlated with the severity of the phenotype as the same variant can be associated with mild to profound hearing loss (Table 1). For example, the variant c.1337 A > G; p.(Gln446Arg) was found to cause moderate hearing loss (PTA 65 dB HL) in one individual of family HLHI07 while the same variant was associated with profound hearing loss (PTA 99 dB HL) in a member of family HLMR1. This suggests variable expressivity or a role for the environment or genetic background as modifiers of hearing loss severity5,58. MYO15A- and OTOF-associated hearing loss explains 7.4% and 5.6% of the participating families respectively. These incidence rates for SLC26A4, MYO15A and OTOF related hearing loss are similar to those reported for singleton affected individuals with moderate to severe hearing loss in Pakistan59. Interestingly, no pathogenic variants in TMC1 or CLDN14 were identified as a cause of hearing loss for any participant in the current cohort of 54 families, which are a significant contributor to profound hearing loss in Pakistan. However, we previously identified TMC1 and CLDN14 pathogenic variants in 7 and 3 of 70 families, respectively5,60, associated with moderate to severe hearing loss. Some variants identified in our study affected the same splice site or codons reported earlier to be pathogenic, but with different nucleotide changes. These include SLC26A4 c.1544 + 1G> T61, CDH23 c.4567 A > G; p.(Asn1523Asp)62 and MYO15A c.2200_2200dupG; p.(Asp734GlyfsTer246)63 variants. In one instance, a missense variant in OTOF c.3263T > C; p.(Leu1088Pro)64 was previously reported at the same position we detected a novel frameshift variant. We evaluated the variant effects of three substitutions p.(Pro458Leu), p.(Asn1523Ile), p.(Arg1577Pro) of CDH23 protein and two substitutions of SLC26A4 p.(Ala98Asp), p.Gln101Pro using the reported protein structures, protein modeling and stability predictions. CDH23 Pro458 is involved in the conserved motif immediately after Ca2+-binding motif, and the Arg1523 is in the Ca2+-binding motif. Substitutions of both variants are predicted to impact Ca2+-binding and protein function. Asn1523 is located in a β-sheet of EC15 and its substitution is predicted to partially damage β-sheet structure. Predictions for each substitution indicates that they are “likely pathogenic” (Supplementary Table S2). In the case of SLC26A4, both variants p.(Ala98Asp) and p.(Gln101Pro) are predicted to destabilize transmembrane local structure. Additionally, based on a study of wild boar SLC26A4 the p.(Gln101Pro) substitution is predicted to reduce anion transport33. Variants associated with human hearing loss are reported in this anion binding site26,34–37, indicating a critical function of this site for human hearing. In this current study, we identified variants of many previously reported hearing loss genes but did not implicate any novel hearing loss genes associated with moderate to severe hearing loss segregating in Pakistani families. This perhaps indicates that the majority of the genes causing moderate to severe hearing loss have already been identified in Pakistani cohorts in the communities from which we ascertained subjects6. On the other hand, we previously reported five families without a genetic diagnosis for the hearing loss after exome sequencing5 suggesting there are novel coding or noncoding variants that remain to be discovered, some of which could be in unreported genes. Although we expanded the allelic diversity associated with moderate to severe hearing loss, there are limitations to our study. Presently, we have no access to additional clinical data for affected individuals with SLC26A4 variants to distinguish between Pendred syndrome or nonsyndromic hearing loss DFNB4 with possible enlarged vestibular aqueduct. Second, in most cases, alternative causes of hearing loss cannot be excluded for those without exome analysis. This is especially relevant for family HLMN3 in which the DNA from the individual with a mild hearing loss underwent exome sequencing and was homozygous for a nonsense variant in MYO3A. Her sibling with the same homozygous variant had a more severe hearing loss. Third, pathogenicity of all variants including missense, synonymous and intronic variants were assessed only using in silico tools and segregation analyses. Experimental functional assays were not performed to determine if there are effects on gene function or splicing. Similarly, though our research indicates that identical variants can result in variable degrees of hearing loss even among affected relatives of the same family, we did not identify the factors explaining the variable expressivity. The presence of milder phenotypes, despite pathogenic variants typically linked to profound hearing loss, suggests potential genetic or environmental modifiers that remain to be elucidated which can be windows into novel therapies. Understanding these modifiers may provide valuable insight into the complex mechanisms regulating auditory function and their variable expressivity as well as suggest novel therapies. Future studies using genome sequencing, integrating functional assays, transcriptomic profiling, and epigenetic analyses may help identify modifying factors, ultimately improving personalized therapeutic approaches for individuals with hereditary hearing loss. Methodology Ethics approval This study conforms to the ethical guidelines laid down in the Declaration of Helsinki and received prior approval from the Institutional Review Board (IRB #00005281 to S.N.) at the School of Biological Sciences, University of the Punjab, Lahore, and the Combined National Institutes of Health IRB (protocol OH93DC0016 to T.B.F.). Family ascertainment and sample collection We ascertained 54 Pakistani consanguineous families with a history of hearing loss inherited as a recessive disorder. Among these, 50 families are from various regions of the Punjab province while 4 families, HLHT01/02, HLNK1, HLNK2 and IHT02, were ascertained from the Khyber Pakhtunkhwa province. Only affected members HLMS1819, HLMS2022 and HLSHI01 participated from 3 of the 54 families. Most probands were identified from special education schools or schools for deaf children. Audiometry was performed under ambient noise conditions, and pure tone thresholds were measured at frequencies of 0.25, 0.5, 1, 2, 4, and 8 kHz. Air conduction pure tone averages (PTA0.5–4 kHz) were calculated for all available affected members of the families by averaging pure tone thresholds at four frequencies of 0.5, 1, 2, and 4 kHz. Hearing loss was classified according to the better hearing ear as mild (20–40 dB HL), moderate (41–70 dB HL), severe (71–94 dB HL), and profound (above 95 dB HL)65. Hearing loss was considered asymmetric if the difference of air conduction PTA0.5–4 kHz between the right and the left ear was greater than 15 dB HL. Audiometry could not be performed for individuals in 9 families (HLHT01/02, HLHI05, HLHI10, HLHI12, HLMR10, HLSHI01, HLMS1819, HLMS2022 and HLNK2). However, school records, parental and personal assessments indicated that hearing loss was not profound in affected individuals in these families (Table 1, Supplementary Table 1) as they responded to sounds when spoken to by a person standing behind them. Pediatric conditioned play audiometry66 for affected younger children was performed for participants between 4 and 5 years old in families HLHKA01 and HLGM27, which confirmed that hearing loss was not profound. The age of onset and the progression in severity of hearing loss were determined by questioning the participants as well as their parents since previous audiograms were unavailable. Individuals with moderate or severe degree hearing loss were recruited, while those with profound hearing loss were only included if the disorder was reported to have worsened from a milder hearing loss or if other younger affected members in the family had better hearing thresholds as compared to the older affected individuals. Romberg and tandem gait tests were performed to evaluate vestibular function. Medical histories were obtained. Individuals manifesting obvious syndromes or reporting environmental causes for their phenotype were excluded. Written informed consent forms were obtained from all the participants in the study or from the legal guardians of individuals under the age of 18 years. Blood samples were collected from the participants in ethylenediaminetetraacetic acid containing vials. Genomic DNA was isolated using sucrose lysis and a salting-out method67. Screening and exome sequencing DNA samples from affected members of participating families were screened by Sanger sequencing for pathogenic variants of GJB2 (NCBI RefSeq: NM_004004.6) for the sole protein-coding exon 2 and the non-coding founder variant of HGF (NCBI RefSeq: NM_000601.6) located in intron 45 (Supplementary Table S6). The families confirmed to be negative for the pathogenic variants of GJB2 and HGF were then subjected to further screening. Exome sequencing was performed on DNA samples of one or more affected individuals from each family while for family IHT02, a sample of an unaffected individual was also included. The work was completed at multiple locations including the Laboratory of Molecular Genetics, National Institutes on Deafness and Communication Disorders (NIDCD), National Institutes of Health (NIH), USA (21 families), the Baylor-Hopkins Center for Mendelian Genomics (BHCMG), USA (5 families), 3billion Inc., South Korea (21 families) and Macrogen, Inc., South Korea (3 families). BHCMG and NIH utilized an Agilent SureSelect HumanAllExonV4_51Mb Kit_S03723314 (​Agilent Technologies, Inc., USA) and the Illumina HiSeq 2000 platform (Illumina Inc., USA). 3billion Inc. used the xGen Exome Research Panel v2 (Integrated DNA Technologies, USA), while Agilent SureSelect Exome Capture, and NovaSeq platform (Illumina NovaSeq 6000, USA) were used by Macrogen for exome sequencing. All sequencing was bidirectional with 100X coverage. Researchers in Pakistan received the Variant Call files (VCF) data generated from the various sites for further analyses. Some VCF files were annotated using the web-based wANNOVAR (https://wannovar.wglab.org/) against the reference human genome assembly GRCh38/hg38, (VCF of 12 families from NIH). The VCF data of the remaining 38 families were uploaded to the Franklin software (https://franklin.genoox.com/clinical-db/home) for assembly GRCh37/hg19 and filtered based on multiple criteria. Variants that were homozygous, hemizygous, or compound heterozygous with population allele frequencies below 0.01 in the gnomAD v2.1.1 or gnomAD v4.1.0 (Genome Aggregation Database, https://gnomad.broadinstitute.org/)68, 1000 genomes (https://www.internationalgenome.org/)69, and GME databases (Greater Middle East Variome)70 were retained for further analyses. All exonic and splice-site variants up to ± 10 base pairs from an exon-intron junction were considered. The deleteriousness of a variant was estimated through predictions by various software including SpliceAI (https://spliceailookup.broadinstitute.org/, accessed 2024)71, REVEL (https://sites.google.com/site/revelgenomics/ accessed 2024)72, SIFT (http://sift-dna.org, accessed 2024)73 and PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/)74 and those with high deleterious scores were further examined. We also checked the CADD prediction scores (https://cadd.gs.washington.edu/, accessed 2024)75 and those with CADD scores higher than 15 were further prioritized. Allele frequencies of the short-listed variants were examined in our in-house exome dataset comprising 300 unrelated Pakistani individuals with normal hearing thresholds. The final shortlisted variants were classified according to the American College of Medical Genetics (ACMG) criteria76. Conservations of the amino acids affected by missense variants were analyzed by Multiz Alignment of 100 vertebrates on the UCSC genome browser (https://genome.ucsc.edu/, accessed 2024)77 as well as by Clustal Omega 1.2 (https://www.ebi.ac.uk/jdispatcher/msa/clustalo)78 of the respective protein orthologues from diverse vertebrates. Exome data generated at 3billion were also analyzed by the company in parallel using EVIDENCE79, their automated prioritization system. For single nucleotide variants (SNV), 3billion algorithm selected variants with frequency < 5% gnomAD v2.1.1 (https://gnomad.broadinstitute.org/). Variants with a frequency higher than 5% were only retained if they were reported likely pathogenic or pathogenic by ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/). Prior to variant classification, the 3billion algorithm calculated phenotypic similarity scores by converting the proband’s clinical phenotype into standardized HPO terms and comparing these features to known disease-associated phenotypes. This symptom-based prioritization enabled the identification of variants with a higher likelihood of pathogenicity. Following this initial filtering, ACMG and the American Molecular Pathology guideline were employed to classify variants. For large genomic deletions or insertions, copy number variant (CNV) analyses were performed using CoNIFER 0.2.280 and 3bCNV (3billion Inc. internally developed software) which uses depth of coverage information. This involves the number of sequencing reads mapped to specific genomic regions, which is used as an indicator for identifying potential copy number variations. Analysis identifies regions exhibiting significantly reduced or elevated read depths and marks these as containing potential deletions or duplications, respectively. Segregation analyses Variants identified after exome sequencing were further evaluated for segregation in each family by Sanger sequencing of genomic DNA samples from the participants. Respective fragments for analyses of the variants were obtained using PCR with specifically designed primers (Supplementary Table S6), which were commercially synthesized (Macrogen, Inc., South Korea). Cycle sequencing was performed using BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific, USA). Samples were evaluated on a 3500 Genetic Analyzer (Applied Biosystems, USA) and sequences were analyzed using DNASTAR software (SeqMan 5.00© 1989–2001 DNASTAR Inc., USA). Splicing prediction Variants predicted by SpliceAI to affect splicing were further examined using Transcript-inferred Pathogenicity (TraP) v3 scores (http://trap-score.org/)81. Both tools use the genomic sequence of the pre-mRNA transcript as input. The possible effects of the variants on gene splicing were also obtained from varSEAK (https://varseak.bio/)82. Percentage calculations of genes contributing to hearing loss in the cohort The contributions of reported genes to hearing loss in families of Pakistan affected with moderate to severe hearing loss were calculated for a cohort of 54 families ascertained for this study, and separately by combining these data with the results of families we reported previously5,23,60. Duplicates were removed and only those with a definitive diagnosis or exome sequencing were included for a total of 124 families. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3