Evidence of distinct RELN and TGFB1 genetic associations in familial and non-familial otosclerosis in a British population

Evidence of distinct RELN and TGFB1 genetic associations in familial and non-familial... Otosclerosis is a common form of hearing loss which typically presents in young adults. The disease has a familial, mono- genic form and a non-familial form with a more complex aetiology. A previous genome wide association study identified evidence that variants within RELN are associated with the condition. Other genes in which an association has been reported include BMP2, COL1A1, FGF2, PPP2R5B and TGFB1. However, follow up studies have often failed to replicate initial posi- tive results. The aim of this study was to establish if an association exists between eight single nucleotide polymorphisms (SNPs) in these six previously implicated genes and otosclerosis in a British case–control cohort (n = 748). Evidence of an association between rs1800472 in TGFB1 and otosclerosis was found (p = 0.034), this association was strongest amongst non- familial cases (p = 0.011). No evidence of an association was detected with variants in COL1A1, FGF2, BMP2, and PPP2R5B. No association between variation in RELN and otosclerosis was observed in the whole cohort. However, a significant asso- ciation (p = 0.0057) was detected between one RELN SNP (rs39399) and otosclerosis in familial patients. Additionally, we identify expression of one RELN transcript in 51 of 81 human stapes tested, clarifying previous conflicting data as to whether RELN is expressed in the affected tissue. Our findings strengthen the association of TGFB1 (rs1800472) with otosclerosis and support a relationship between RELN and familial otosclerosis only, which may explain previous variable replications. Introduction been estimated to be 0.3–0.4% amongst Caucasians (Declau et  al. 2001). The age of onset varies from the first to the Otosclerosis is a common form of progressive, adult hearing fifth decade, but most commonly presents in the third (Caw - loss. The disease is characterised by dystrophy localized to thorne 1955) and this is maintained within families (Chole the bone of the otic capsule (Chole and McKenna 2001). A and McKenna 2001; Morrison 1967). conductive hearing loss develops, as otosclerotic foci invade The disease exists in familial and non-familial forms. the stapedio-vestibular joint at the oval window, or the junc- The familial form, accounting for between 25–50% of tion of the incus and stapes. Affected individuals may also cases, displays an autosomal dominant inheritance pattern develop a profound sensorineural loss (Ishai et al. 2016). with incomplete penetrance (Morrison 1967). Eight different The only available treatment is two separate surgeries to genetic loci have been reported based on genetic mapping replace the defective stapes bone with a micro-prosthetic in families. However, none of the causal genes within these device, which carries the risk of further hearing loss (Mann loci have yet been identified (Ealy and Smith 2010). This et al. 1996). The clinical prevalence of the condition has is likely to be due to the reduced power of linkage studies in the presence of incomplete penetrance. A more recent study using a whole exome sequencing approach identified * Sally J. Dawson multiple individuals with rare mutations in SERPINF1 in a sally.dawson@ucl.ac.uk cohort of familial patients (Ziff et al. 2016). Functional stud- UCL Ear Institute, University College London, ies suggest these mutations affect expression of an alterna- London WC1X 8EE, UK tively spliced transcript which is normally highly expressed Department of ENT Surgery, The Princess Margaret in stapes bone. Non-familial otosclerosis is considered as a Hospital, Windsor SL4 3SJ, UK complex disorder which exists as isolated cases with a mixed Royal National Throat Nose and Ear Hospital, environmental and genetic aetiology (Ealy and Smith 2010). London WC1X 8EE, UK Vol.:(0123456789) 1 3 358 Human Genetics (2018) 137:357–363 Whether mutations within the same genes contribute to a powered study are required, such as the cohort used here, to common pathology in both familial and non-familial types ascertain the reliability of the RELN association. is currently unclear. Here, we genotyped eight previously associated variants The strongest evidence linking any single gene to the dis- in the following six candidate genes: RELN, BMP2, FGF2, ease came from a genome wide association study (GWAS) COL1A1, TGFB1, and PPP2R5B in a large UK case–control in a cohort of patients from Belgium, Holland and France otosclerosis cohort (n = 748). In addition, we also investi- (Schrauwen et al. 2009a). Schrauwen et al. (2009a) per- gate the expression of RELN transcripts in human stapes formed a case–control discovery experiment utilizing and resolve the previous controversies in reelin expression 555,000 single nucleotide polymorphisms (SNPs). The in human stapes. group discovered two highly associated SNPs that were replicated in additional independent populations. The region with the strongest association signal was found on Materials and methods −10 chromosome 7q22.1 (rs39395, p = 6.23 × 10 ) within the RELN gene. A secondary region of association was found Patient recruitment at chromosome 11q13.1 at the PPP2R5B locus (rs616322, −5 p = 2.20 × 10 ). RELN codes for the protein reelin, which Individuals with a confirmed clinical diagnosis of otoscle - has a role in neuronal migration. It enables detachment of rosis were recruited from the Royal National Throat Nose neurons from radial glial cells (Dulabon et al. 2000). Fail- and Ear Hospital (London, UK), Princess Margaret Hos- ure of this migration is associated with neurological and pital (Windsor), Sunderland Royal Hospital (Sunderland), psychiatric disorders, including autism (Fatemi et al. 2005). Freeman Hospital (Newcastle upon Tyne) and Ninewells However, there is no evidence that the bone dysregulation Hospital (Dundee). The study was approved by the London that underlies otosclerosis has any relationship to reelin’s Bloomsbury NRES Ethics committee (11/LO/0489). The known biological function. Therefore, the association sug- diagnosis of otosclerosis was made on clinical and audio- gested between RELN and otosclerosis is a surprising one. metric examination, and then confirmed during surgery. Schrauwen et  al. (2009a, b) have detected expression of Patients and controls were recruited by informed consent. RELN in human and mouse stapes, supporting a role for Controls had otosclerosis ruled out on the basis of a pure RELN in stapes bone (Schrauwen et al. 2009a). A subsequent tone audiogram. Blood or saliva samples were obtained for study could not detect RELN in adult human stapes, although genomic DNA isolation (Oragene Saliva Kit, DNA Gen- the RELN protein was present (Csomor et al. 2012). otek) by standard methods. Patients completed a question- Subsequent replication of this genetic association naire recording family history of otosclerosis and relevant between RELN and otosclerosis has been inconsistent. The medical history. Female patients answered additional ques- same group who performed the GWAS found supportive tions recording history of pregnancy, oral contraceptives and evidence when analysing 591 patients from Germany, Italy, hormone replacement therapy. For stratified analysis based Switzerland and Romania (Schrauwen et al. 2010). Simi- on questionnaire data, patients were categorised as famil- larly, an independent replication using 149 Tunisian cases ial or non-familial based on reporting an additional first or and 152 controls found evidence of association with three second-degree family member with a clinical diagnosis of RELN SNPs (Khalfallah et al. 2010). However, other stud- otosclerosis as reported by the proband. ies have failed to replicate the association. A case–control association study excluded any association between the Genotyping of SNPs rs39335 SNP and the disease in a southern Italian popula- tion of 92 cases and controls (Iossa et al. 2013). Similarly, a SNPs were genotyped using Applied Biosystems TaqMan study in an Indian population failed to show an association Pre-Designed SNP Genotyping assays on an Applied Bio- between four separate RELN SNPs (rs3914131, rs3914132, systems 7500 Real Time PCR System according manufac- rs4641319, and rs10227303), when 170 otosclerosis cases turer’s instructions. Assays IDs were: rs3178250 (BMP2), were compared with 170 controls (Priyadarshi et al. 2010). C__27466541_10; rs1800012 (COL1A1), C___7477170_30; Furthermore, a study in a Hungarian population of 153 cases rs17407577 (FGF2), C__34186732_10; rs39335 (RELN), and 300 controls genotyped thirteen SNPs across six candi- C____758635_10; rs39399 (RELN), C__2415500_10; date genes (COL1A1, TGFB1, BMP2, BMP4, AGT, RELN). rs3914132 (RELN), C__7594810_10; rs1800472 (TGFB1), Only the association between TGFB1 and otosclerosis was C___8708464_20; rs616322 (PPP2R5B), C____633165_10. replicated (Sommen et al. 2014). A number of these stud- ies were underpowered for detection of a RELN associa- tion. Therefore, tests of such an association in an adequately 1 3 Human Genetics (2018) 137:357–363 359 Sequence-specific primers were designed to specifi- Association testing cally amplify RELN transcripts RELN-201 (Ensembl Tran- script ID: ENST00000343529.9) and RELN-203 (Ensembl Tests for genetic association with disease were performed using logistic regression analysis using the SNPstats plat- Transcript ID: ENST00000428762.5) based on the sites of primers used in previous studies to detect RELN expression form (Sole et al. 2006) under different genetic inheritance models (dominant, recessive and additive), and to investigate and with reference to the exon/intron boundaries. Primer sequences: RELN-201F, 5′-GCA ATA CAG CGT CAA CAA covariate interactions with gender and ethnicity. A p value of < 0.05 was considered significant. None of the SNPs CGG-3′; RELN-201R, 5′ -G TT T GC G AG T G A GG A CG A CCT-3′; RELN-203F, 5′ -A CC AT G T GG A GG T CG T CC showed a deviation from Hardy–Weinberg equilibrium in cases or controls. To investigate whether the relationship TAGTA-3′ and RELN-203R, 5′ -CAC T CG G T C TT G AGA AGG GCTTT -3′. For all cDNA samples, a GAPDH PCR between the SNPs and otosclerosis was gender-dependent we ran the overall association test using gender as a cat- assay was performed as a control for cDNA integrity and a non-template control for background contamination. MG-63 egorical co-variate. With seven of the eight SNPs, no statisti- cally significant interaction with gender was observed. For cDNA was used as a positive control. PCR reactions con- tained 400 ng cDNA, 0.5 µM each primer, 0.3 mM dNTPs, rs17407577 in FGF2 a weak association with gender was observed (p = 0.034). In our cohort, 94% of the sample was 2  mM MgCl2, 5  µl of 5× GoTaq PCR buffer and 1.25U GoTaq polymerase (Promega) in a total volume of 20 µl. of white European origin, no interaction was detected when ethnicity was included as a categorical variable. The tests of Thermal cycling was performed under the following reac- tion conditions: denaturation at 93 °C for 15 s, annealing at association were performed separately in white Europeans alone and with participants of all ethnicity and gave similar optimum temperature for 30 s, extension at 72 °C for 30 s, for up to 40 cycles, with an initial denaturation step of 93 °C results. for 3 min on an Eppendorf Mastercycler Gradient machine. The resulting DNA bands were visualised by agarose gel Power calculations electrophoresis. Power calculations were performed using the Genetic Power Calculator (http://zzz.bwh.harva rd.edu/gpc/) (Purcell et al. Results and discussion 2003). All calculations were performed assuming a disease prevalence of 0.35%, setting the relative risk and disease Genome wide and candidate gene association studies rep- allele frequency based on previous studies under an additive model (Sommen et al. 2014). resent powerful genetic tools. However, such studies can generate false positive associations. Replication in independ- RELN expression in human stapes ent populations is, therefore, vital for discriminating true associations from false positives. An important limitation Stapes suprastructures were collected from 76 individuals to the replication of genetic associations within otosclerosis has been the lack of well characterised patient cohorts of undergoing laser-assisted stapedotomy surgery at the Royal National Throat Nose and Ear Hospital (London, UK). A sufficient size for replication studies. We have genotyped diagnosis of otosclerosis was confirmed during surgery in all cases. Five additional stapes suprastructures and one whole stapes were obtained from individuals who did not have oto- Table 1 Power calculation for each SNP within the British case–con- trol cohort sclerosis and who were undergoing surgical procedures dur- ing which the stapes were removed. These included surgery SNP Gene DAF OR Power (%) due to head trauma, glomus tumour and total petrosectomy. rs39399 RELN 0.49 1.45 99.9 Human stapes were preserved in AllProtect™ Solution (Qia- rs3914132 RELN 0.78 1.54 99.7 gen) and stored at − 80 °C. Stapes were homogenized in rs39335 RELN 0.15 1.35 81.8 QIAzol Lysis reagent and RNA purified using the RNeasy rs3178250 BMP2 0.76 2.03 99.9 lipid tissue mini kit (Qiagen) according to the manufactur- rs1800012 COL1A1 0.82 1.55 45.0 er’s protocol, including an on column DNase digest. RNA rs17407577 FGF2 0.03 2.00 76.1 was also extracted from homogenised human MG-63 cells, rs1800472 TGFB1 0.97 2.33 48.9 a human osteosarcoma cell line using the RNeasy mini kit rs616322 PPP2R5B 0.08 1.46 70.0 (Qiagen) in accordance with the manufacturer’s protocol. RNA was reverse transcribed into cDNA by Omniscript Power calculations were performed based on previously reported data Reverse Transcriptase Kit (Promega) for RT-qPCR. (Chen et al. 2002; Sommen et al. 2014) DAF disease allele frequency, OR odds ratio 1 3 360 Human Genetics (2018) 137:357–363 eight previously implicated SNPs in a novel British cohort of might be protective against otosclerosis is unclear (Thys 374 patients and 374 controls. Association with disease was et al. 2007). Our results, combined with previous similar tested by logistic regression analysis under various genetic association studies, suggest that the mutant form of the pro- models. Power calculations (Table  1) estimated that the tein is protective against otosclerosis. study had good power to detect most SNP effects (between The lack of association between any of the three RELN 70 and 99.9% power for six SNPs) with only TGFB1 and SNPs and otosclerosis is surprising given the high power COL1A1 having limited power to detect an effect, at 48.9 of the study to detect an effect for these SNPs. Otosclerosis and 45.0%, respectively. has both familial and non-familial forms and approximately Results for tests of genetic association with otosclerosis 44% of our patient cohort reported other family members in all cases and controls are shown in Table  2. Only the with a diagnosis of otosclerosis. Since it is possible that rs1800472 variant within the TGFB1 gene (p = 0.021) was there are distinct genetic susceptibilities involved in the two found to be significantly associated with otosclerosis. This forms of the disease we performed an analysis when strati- SNP was selected for a case–control association study due fying the patient cohort by familial or non-familial disease, to several genetic indications of involvement with the dis- see Tables 3 and 4 respectively. In familial cases a strong ease. Thys et al. (2009), performed a case–control associa- association was detected between rs39399 and otosclerosis tion study in a Belgian-Dutch cohort which gave significant (p = 0.0057, Table 3) that was not present in non-familial results (p = 0.0044) (Thys et  al. 2009). Within the same cases (Table  4). As with the association seen in TGFB1, study, the association was replicated in a French population, it is the major allele that is associated with disease, rep- with even greater significance (p = 0.00019). Our study rep- licating the direction of association detected in the origi- resents the third subsequent case–control association study nal GWAS. Our results suggest that an association exists which has confirmed the association (Khalfallah et al. 2010; between rs39399 and the familial form of otosclerosis, which Sommen et al. 2014). The evidence for the association is increased further by the absence of any conflicting stud- Table 3 Association testing in familial otosclerosis cases (n = 160) ies which have failed to demonstrate an association. Fur- and controls (n = 374) thermore, much of this supportive evidence has come from SNP Gene p OR (95% CI) studies which are underpowered to detect the SNP’s effect. All studies have found the disease to be associated with the rs39399 RELN 0.0057 1.73 (1.18–2.54) rs3914132 RELN 0.17 1.33 (0.89–1.99) major allele suggesting the variant is protective. The SNP is relatively rare with a minor allele frequency of 0.024 and rs39335 RELN 0.30 1.25 (0.82–1.90) rs3178250 BMP2 0.54 0.89 (0.61–1.30) 0.045 in cases and controls respectively. The rs1800472 SNP is a missense variant that causes a rs1800012 COL1A1 0.30 1.25 (0.82–1.90) rs17407577 FGF2 0.84 0.96 (0.65–1.41) threonine to isoleucine substitution at amino acid position 263 in TGFB1, this change has been shown to effect activ - rs1800472 TGFB1 0.28 1.48 (0.71–3.08) rs616322 PPP2R5B 0.25 0.78 (0.52–1.18) ity of the protein (Thys et al. 2009). Analysis of the T263I substitution with a luciferase reporter assay showed that the Significant values are indicated in bold. p values are shown for a test protective variant of TGFB1 is more active than the major of association under the dominant model. All odds ratios quoted are allele. The reason why a more active version of the protein for the minor allele Table 2 Association testing of the eight SNPs in the full case–control Table 4 Association testing in non-familial otosclerosis cases cohort (n = 748) (n = 206) and controls (n = 374) SNP Gene p OR (95% CI) SNP Gene p OR (95% CI) rs39399 RELN 0.32 1.17 (0.86–1.60) rs39399 RELN 0.40 0.85 (0.58–1.24) rs3914132 RELN 0.33 1.16 (0.86–1.58) rs3914132 RELN 0.55 1.00 (0.70–1.43) rs39335 RELN 0.38 1.16 (0.84–1.59) rs39335 RELN 0.52 1.04 (0.71–1.51) rs3178250 BMP2 0.65 0.93 (0.70–1.25) rs3178250 BMP2 0.92 0.98 (0.69–1.39) rs1800012 COL1A1 0.21 1.13 (0.84–1.53) rs1800012 COL1A1 0.24 1.24 (0.86–1.78) rs17407577 FGF2 0.45 0.85 (0.55–1.30) rs17407577 FGF2 0.90 0.99 (0.59–1.67) rs1800472 TGFB1 0.021 0.51 (0.28–0.91) rs1800472 TGFB1 0.015 0.40 (0.18–0.88) rs616322 PPP2R5B 0.38 0.86 (0.62–1.20) rs616322 PPP2R5B 0.71 0.89 (0.60–1.33) Significant values are indicated in bold. p values are shown for a test Significant values are indicated in bold. p values are shown for a test of association under the dominant model. All odds ratios quoted are of association under the dominant model. All odds ratios quoted are for the minor allele for the minor allele 1 3 Human Genetics (2018) 137:357–363 361 requires further investigation in familial cohorts. No associa- Previous association studies with RELN and otosclero- tion between rs1800472 in TGFB1 and disease was detected sis have not discriminated between cases with familial and in familial cases (Table 3), however, in non-familial cases non-familial disease. If our finding of RELN association an association was again detected this time with a lower p with familial disease only is substantiated it is possible that value (p = 0.015, Table 4) and greater odds ratio than was the proportion of patients with familial disease in cohorts detected in the combined otosclerosis cases, suggesting that used in previous studies may have influenced the ability to this SNP may be playing a role in the non-familial form of detect this association and account for variable replication the disease only. between studies. For example, variant rs3914132 has now It has been well documented that some women with oto- been included in five follow up case–control association sclerosis associate the onset or worsening of hearing loss studies following its identification in the original GWAS. with pregnancy and it has been postulated that this may be Two of those studies have replicated the initial association due to the influence of oestrogen (Torsiglieri et al. 1990). (Khalfallah et al. 2010; Schrauwen et al. 2009a). However, Approximately 40% of the females in our cohort who had two others have failed to display the replication (Priyadarshi experienced a pregnancy self-reported a worsening of et al. 2010; Sommen et al. 2014). The results also suggest a hearing with pregnancy in questionnaire data. To explore hypothesis that the otosclerosis phenotype maybe reached, whether these females represent a distinct genetic aetiology through an oestrogen dependent or independent pathway. from the main cohort we tested association with disease in Those cases which are oestrogen dependent, progress dispro- females stratified by this self-reported association with preg- portionately during pregnancy. Our data suggest that RELN nancy (Table 5). may be exclusively involved in the oestrogen independent No association with otosclerosis was detected with any mechanism. If so, the results of previous case–control asso- of the eight SNPs in those women who reported worsening ciation studies may have been determined by the make-up of hearing in pregnancy or those who had not been pregnant. each cohort which may have been influenced by differences A significant relationship was seen with the RELN SNPs, in recruitment pipelines. rs3914132 (p = 0.0095) and rs39335 (p = 0.038), in those Another factor in considering whether RELN is involved women who did not report a deterioration in their hearing in the pathogenesis of otosclerosis has been the lack of when compared with controls. These data should be treated an obvious role for the reelin protein in stapes bone and with caution as the stratified sample sizes are small and, the uncertainty as to whether it is expressed in human therefore, more prone to error. If substantiated by replication stapes. Two previous studies reported conflicting results in other cohorts our results suggest that RELN may play a in detecting RELN transcripts in stapes bone (Cso- role in the pathogenesis of otosclerosis, but only in specific mor et  al. 2012; Schrauwen et  al. 2009b). The current subtypes. Indeed our data when taken together suggest that release of human genome assembly (Ensembl: GRCh38. otosclerosis has a complex pathophysiology, with non-famil- p10, accessed November 2017) shows that the predicted ial and familial cases having different underlying genetic major transcripts of RELN diverge in the use of different mechanisms, each involving separate genetic pathways. microexons 64 (no exon, 4 bp exon or 6 bp exon) at the 3′ end of the transcript. Examination of the primers used in previous studies suggested that the assays used may Table 5 Association testing in women stratified by reported hearing be detecting different transcripts (RELN-201 in Schrau- loss associated with pregnancy wen et al. 2009a) and (RELN-203 in Csomor et al. 2012), SNP Gene No pregnan- Pregnant Pregnant cies change no change p value p value p value (n = 55) (n = 65) (n = 99) Table 6 RELN expression in human stapes Sample (n) No. (%) of samples in which tran- rs39399 RELN 0.28 0.95 0.094 script is detected rs3914132 RELN 0.89 0.22 0.0073 GAPDH RELN-201 RELN-203 rs39335 RELN 0.75 0.76 0.038 rs3178250 BMP2 0.86 0.63 0.70 MG-63 cells (1) 1 (100%) 1 (100%) 1 (100%) rs1800012 COL1A1 0.75 0.60 0.47 Female stapes suprastructures 52 (100%) 32 (62%) 0 (0%) rs17407577 FGF2 0.19 0.60 0.39 (52) a a a rs1800472 TGFB1 0.36 0.67 0.17 Male stapes suprastructures 19 (100%) 11 (58%) 0 (0%) rs616322 PPP2R5B 0.93 0.21 0.20 (19) Control suprastructures (5) 5 (100%) 4 (80%) 0 (0%) Significant values are indicated in bold. p values are shown for a test Total stapes suprastructures 76 (100%) 47 (62%) 0 (0%) of association under the dominant model unless otherwise indicated (76) Recessive model 1 3 362 Human Genetics (2018) 137:357–363 Fig. 1 A 2% agarose gel showing RELN-201 RT-PCR amplicons detected in human stapes cDNA. RELN-201 was detected in MG-63 cells and five out of ten human stapes suprastructures shown (a repre- sentative sample of the cohort), illustrated by a 209 bp fragment (arrowed) which might account for the different results. We therefore Acknowledgements We are indebted to the patients and their families for their willingness to participate in our research study. We are also designed assays to detect both RELN-201 and RELN-203 grateful for the UK CRN staff, physicians and hospital staff whose in human stapes. The results (Table 6; Fig. 1) show that the efforts in collecting these samples were essential to this research. RELN-201 transcript is detected in the majority of stapes suprastructures removed from both otosclerosis patients Funding This work was supported by Action on Hearing Loss (refs G71, G83 and 512:UEI:SD). and controls. However, the RELN-203 transcript was not detected in any human stapes, but was detected in the posi- Compliance with ethical standards tive control, cDNA from the MG-63 cell line. Our results clarify the expression of RELN in human stapes and are Conflict of interest On behalf of all authors, the corresponding author similar to those reported by Schrauwen et al. (2009a, b) states that there is no conflict of interest. suggesting that reelin is expressed in the majority but not all of human stapes tested (Schrauwen et al. 2009a). Open Access This article is distributed under the terms of the Crea- No statistically significant association was seen between tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- four SNPs in COL1A1, FGF2, PPP2R5B, or BMP2 and tion, and reproduction in any medium, provided you give appropriate otosclerosis across all comparisons. This provides further credit to the original author(s) and the source, provide a link to the evidence against the involvement of these genes in oto- Creative Commons license, and indicate if changes were made. sclerosis disease pathogenesis. For some of the SNPs in question, such as rs1800012 in COL1A1, this is the third consecutive case–control association study (Khalfallah et al. 2010; Sommen et al. 2014) in which an association References has failed to replicate the initial positive association (Chen et al. 2002). However, due to the low odds ratio reported Cawthorne T (1955) Otosclerosis. J Laryngol Otol 69:437–456 Chen W, Campbell CA, Green GE et al (2002) Linkage of otosclerosis for this association this study has poor power to detect to a third locus (OTSC3) on human chromosome 6p21.3-22.3. J such a weak effect. For other variants the reliability of Med Genet 39:473–477 the negative association is increased by the power of this Chole RA, McKenna M (2001) Pathophysiology of otosclerosis. Otol study. For example, no effect was detected for rs3178250 Neurotol 22:249–257 Csomor P, Sziklai I, Karosi T (2012) Controversies in RELN/reelin in BMP2 despite the study having a 99.9% power to detect expression in otosclerosis. Eur Arch Otorhinolaryngol 269:431– it (Table 1). 440. https ://doi.org/10.1007/s0040 5-011-1653-4 In conclusion, we detected an association between a func- Declau F, Van Spaendonck M, Timmermans JP, Michaels L, Liang J, tional variant in TGFB1 and clinically confirmed otoscle- Qiu JP, Van de Heyning P (2001) Prevalence of otosclerosis in an unselected series of temporal bones. Otol Neurotol 22:596–602 rosis in a British population. This replicates two previous Dulabon L, Olson EC, Taglienti MG, Eisenhuth S, McGrath B, Walsh studies which have found the same association in the same CA, Kreidberg JA, Anton ES (2000) Reelin binds alpha3beta1 direction. Further, we detect an association between different integrin and inhibits neuronal migration. Neuron 27:33–44 SNPs within RELN and familial otosclerosis and in women Ealy M, Smith RJ (2010) The genetics of otosclerosis. Hear Res 266:70–74. https ://doi.org/10.1016/j.heare s.2009.07.002 whose hearing loss does not progress during pregnancy. A Fatemi SH, Snow AV, Stary JM, Araghi-Niknam M, Reutiman TJ, previously described association between four other genes Lee S, Brooks AI, Pearce DA (2005) Reelin signaling is impaired (COL1A1, BMP2, PPP2R5B, and FGF2) was not replicated, in autism. Biol Psychiatry 57:777–787. https://doi.or g/10.1016/j. providing further negative evidence for any involvement of biops ych.2004.12.018 these genes in the pathogenesis of the condition. 1 3 Human Genetics (2018) 137:357–363 363 Iossa S, Corvino V, Giannini P et al (2013) The rs39335 polymorphism Schrauwen I, Ealy M, Fransen E et al. (2010) Genetic variants in the of the RELN gene is not associated with otosclerosis in a southern RELN gene are associated with otosclerosis in multiple European Italian population. Acta Otorhinolaryngol Ital 33:320–323 populations. Hum Genet 127:155–162. https ://doi.org/10.1007/ Ishai R, Halpin CF, Shin JJ, McKenna MJ, Quesnel AM (2016) Long-s0043 9-009-0754-2 term incidence and degree of sensorineural hearing loss in oto- Sole X, Guino E, Valls J, Iniesta R, Moreno V (2006) SNPStats: a sclerosis. Otol Neurotol 37:1489–1496. https ://doi.org/10.1097/ web tool for the analysis of association studies. Bioinformatics MAO.00000 00000 00123 4 22:1928–1929. https ://doi.org/10.1093/bioin forma tics/btl26 8 Khalfallah A, Schrauwen I, Mnaja M et al (2010) Genetic variants in Sommen M, Van Camp G, Liktor B, Csomor P, Fransen E, Sziklai RELN are associated with otosclerosis in a non-European popula- I, Schrauwen I, Karosi T (2014) Genetic association analysis in tion from Tunisia. Ann Hum Genet 74:399–405. https://doi.or g/1 a clinically and histologically confirmed otosclerosis popula- 0.1111/j.1469-1809.2010.00595 .x tion confirms association with the TGFB1 gene but suggests an Mann WJ, Amedee RG, Fuerst G, Tabb HG (1996) Hearing loss as association of the RELN gene with a clinically indistinguishable a complication of stapes surgery. Otolaryngol Head Neck Surg otosclerosis-like phenotype. Otol Neurotol 35:1058–1064. https 115:324–328. https ://doi.org/10.1016/S0194 -5998(96)70046 -3://doi.org/10.1097/MAO.00000 00000 00033 4 Morrison AW (1967) Genetic factors in otosclerosis. Ann R Coll Surg Thys M, Schrauwen I, Vanderstraeten K et al (2007) The coding poly- Engl 41:202–237 morphism T263I in TGF-beta1 is associated with otosclerosis in Priyadarshi S, Panda KC, Panda AK, Ramchander PV (2010) Lack two independent populations. Hum Mol Genet 16:2021–2030. of association between SNP rs3914132 of the RELN gene and https ://doi.org/10.1093/hmg/ddm15 0 otosclerosis in India. Genet Mol Res 9:1914–1920. https ://doi. Thys M, Schrauwen I, Vanderstraeten K et  al (2009) Detection org/10.4238/vol9-3gmr8 90 of rare nonsynonymous variants in TGFB1 in otosclerosis Purcell S, Cherny SS, Sham PC (2003) Genetic Power Calculator: patients. Ann Hum Genet 73:171–175. https ://doi.or g/10.111 design of linkage and association genetic mapping studies of 1/j.1469-1809.2009.00505 .x complex traits. Bioinformatics 19:149–150 Torsiglieri AJ Jr, Tom LW, Keane WM, Atkins JP Jr (1990) Otolar- Schrauwen I, Ealy M, Huentelman MJ et al. (2009a) A genome-wide yngologic manifestations of pregnancy. Otolaryngol Head Neck analysis identifies genetic variants in the RELN gene associated Surg 102:293–297. https://doi.or g/10.1177/0194599890 10200 317 with otosclerosis. Am J Hum Genet 84:328–338. https ://doi. Ziff JL, Crompton M, Powell HR, Lavy JA, Aldren CP, Steel KP, Saeed org/10.1016/j.ajhg.2009.01.023 SR, Dawson SJ (2016) Mutations and altered expression of SER- Schrauwen I, Thys M, Vanderstraeten K et al (2009b) No evidence for PINF1 in patients with familial otosclerosis. Hum Mol Genet association between the renin-angiotensin-aldosterone system and 25:2393–2403. https ://doi.org/10.1093/hmg/ddw10 6 otosclerosis in a large Belgian-Dutch population. Otol Neurotol 30:1079–1083. https://doi.or g/10.1097/MAO.0b013e3181 ab305 8 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Human Genetics Springer Journals

Evidence of distinct RELN and TGFB1 genetic associations in familial and non-familial otosclerosis in a British population

Free
7 pages
Loading next page...
 
/lp/springer_journal/evidence-of-distinct-reln-and-tgfb1-genetic-associations-in-familial-JCz2bp4R3G
Publisher
Springer Berlin Heidelberg
Copyright
Copyright © 2018 by The Author(s)
Subject
Biomedicine; Human Genetics; Molecular Medicine; Gene Function; Metabolic Diseases
ISSN
0340-6717
eISSN
1432-1203
D.O.I.
10.1007/s00439-018-1889-9
Publisher site
See Article on Publisher Site

Abstract

Otosclerosis is a common form of hearing loss which typically presents in young adults. The disease has a familial, mono- genic form and a non-familial form with a more complex aetiology. A previous genome wide association study identified evidence that variants within RELN are associated with the condition. Other genes in which an association has been reported include BMP2, COL1A1, FGF2, PPP2R5B and TGFB1. However, follow up studies have often failed to replicate initial posi- tive results. The aim of this study was to establish if an association exists between eight single nucleotide polymorphisms (SNPs) in these six previously implicated genes and otosclerosis in a British case–control cohort (n = 748). Evidence of an association between rs1800472 in TGFB1 and otosclerosis was found (p = 0.034), this association was strongest amongst non- familial cases (p = 0.011). No evidence of an association was detected with variants in COL1A1, FGF2, BMP2, and PPP2R5B. No association between variation in RELN and otosclerosis was observed in the whole cohort. However, a significant asso- ciation (p = 0.0057) was detected between one RELN SNP (rs39399) and otosclerosis in familial patients. Additionally, we identify expression of one RELN transcript in 51 of 81 human stapes tested, clarifying previous conflicting data as to whether RELN is expressed in the affected tissue. Our findings strengthen the association of TGFB1 (rs1800472) with otosclerosis and support a relationship between RELN and familial otosclerosis only, which may explain previous variable replications. Introduction been estimated to be 0.3–0.4% amongst Caucasians (Declau et  al. 2001). The age of onset varies from the first to the Otosclerosis is a common form of progressive, adult hearing fifth decade, but most commonly presents in the third (Caw - loss. The disease is characterised by dystrophy localized to thorne 1955) and this is maintained within families (Chole the bone of the otic capsule (Chole and McKenna 2001). A and McKenna 2001; Morrison 1967). conductive hearing loss develops, as otosclerotic foci invade The disease exists in familial and non-familial forms. the stapedio-vestibular joint at the oval window, or the junc- The familial form, accounting for between 25–50% of tion of the incus and stapes. Affected individuals may also cases, displays an autosomal dominant inheritance pattern develop a profound sensorineural loss (Ishai et al. 2016). with incomplete penetrance (Morrison 1967). Eight different The only available treatment is two separate surgeries to genetic loci have been reported based on genetic mapping replace the defective stapes bone with a micro-prosthetic in families. However, none of the causal genes within these device, which carries the risk of further hearing loss (Mann loci have yet been identified (Ealy and Smith 2010). This et al. 1996). The clinical prevalence of the condition has is likely to be due to the reduced power of linkage studies in the presence of incomplete penetrance. A more recent study using a whole exome sequencing approach identified * Sally J. Dawson multiple individuals with rare mutations in SERPINF1 in a sally.dawson@ucl.ac.uk cohort of familial patients (Ziff et al. 2016). Functional stud- UCL Ear Institute, University College London, ies suggest these mutations affect expression of an alterna- London WC1X 8EE, UK tively spliced transcript which is normally highly expressed Department of ENT Surgery, The Princess Margaret in stapes bone. Non-familial otosclerosis is considered as a Hospital, Windsor SL4 3SJ, UK complex disorder which exists as isolated cases with a mixed Royal National Throat Nose and Ear Hospital, environmental and genetic aetiology (Ealy and Smith 2010). London WC1X 8EE, UK Vol.:(0123456789) 1 3 358 Human Genetics (2018) 137:357–363 Whether mutations within the same genes contribute to a powered study are required, such as the cohort used here, to common pathology in both familial and non-familial types ascertain the reliability of the RELN association. is currently unclear. Here, we genotyped eight previously associated variants The strongest evidence linking any single gene to the dis- in the following six candidate genes: RELN, BMP2, FGF2, ease came from a genome wide association study (GWAS) COL1A1, TGFB1, and PPP2R5B in a large UK case–control in a cohort of patients from Belgium, Holland and France otosclerosis cohort (n = 748). In addition, we also investi- (Schrauwen et al. 2009a). Schrauwen et al. (2009a) per- gate the expression of RELN transcripts in human stapes formed a case–control discovery experiment utilizing and resolve the previous controversies in reelin expression 555,000 single nucleotide polymorphisms (SNPs). The in human stapes. group discovered two highly associated SNPs that were replicated in additional independent populations. The region with the strongest association signal was found on Materials and methods −10 chromosome 7q22.1 (rs39395, p = 6.23 × 10 ) within the RELN gene. A secondary region of association was found Patient recruitment at chromosome 11q13.1 at the PPP2R5B locus (rs616322, −5 p = 2.20 × 10 ). RELN codes for the protein reelin, which Individuals with a confirmed clinical diagnosis of otoscle - has a role in neuronal migration. It enables detachment of rosis were recruited from the Royal National Throat Nose neurons from radial glial cells (Dulabon et al. 2000). Fail- and Ear Hospital (London, UK), Princess Margaret Hos- ure of this migration is associated with neurological and pital (Windsor), Sunderland Royal Hospital (Sunderland), psychiatric disorders, including autism (Fatemi et al. 2005). Freeman Hospital (Newcastle upon Tyne) and Ninewells However, there is no evidence that the bone dysregulation Hospital (Dundee). The study was approved by the London that underlies otosclerosis has any relationship to reelin’s Bloomsbury NRES Ethics committee (11/LO/0489). The known biological function. Therefore, the association sug- diagnosis of otosclerosis was made on clinical and audio- gested between RELN and otosclerosis is a surprising one. metric examination, and then confirmed during surgery. Schrauwen et  al. (2009a, b) have detected expression of Patients and controls were recruited by informed consent. RELN in human and mouse stapes, supporting a role for Controls had otosclerosis ruled out on the basis of a pure RELN in stapes bone (Schrauwen et al. 2009a). A subsequent tone audiogram. Blood or saliva samples were obtained for study could not detect RELN in adult human stapes, although genomic DNA isolation (Oragene Saliva Kit, DNA Gen- the RELN protein was present (Csomor et al. 2012). otek) by standard methods. Patients completed a question- Subsequent replication of this genetic association naire recording family history of otosclerosis and relevant between RELN and otosclerosis has been inconsistent. The medical history. Female patients answered additional ques- same group who performed the GWAS found supportive tions recording history of pregnancy, oral contraceptives and evidence when analysing 591 patients from Germany, Italy, hormone replacement therapy. For stratified analysis based Switzerland and Romania (Schrauwen et al. 2010). Simi- on questionnaire data, patients were categorised as famil- larly, an independent replication using 149 Tunisian cases ial or non-familial based on reporting an additional first or and 152 controls found evidence of association with three second-degree family member with a clinical diagnosis of RELN SNPs (Khalfallah et al. 2010). However, other stud- otosclerosis as reported by the proband. ies have failed to replicate the association. A case–control association study excluded any association between the Genotyping of SNPs rs39335 SNP and the disease in a southern Italian popula- tion of 92 cases and controls (Iossa et al. 2013). Similarly, a SNPs were genotyped using Applied Biosystems TaqMan study in an Indian population failed to show an association Pre-Designed SNP Genotyping assays on an Applied Bio- between four separate RELN SNPs (rs3914131, rs3914132, systems 7500 Real Time PCR System according manufac- rs4641319, and rs10227303), when 170 otosclerosis cases turer’s instructions. Assays IDs were: rs3178250 (BMP2), were compared with 170 controls (Priyadarshi et al. 2010). C__27466541_10; rs1800012 (COL1A1), C___7477170_30; Furthermore, a study in a Hungarian population of 153 cases rs17407577 (FGF2), C__34186732_10; rs39335 (RELN), and 300 controls genotyped thirteen SNPs across six candi- C____758635_10; rs39399 (RELN), C__2415500_10; date genes (COL1A1, TGFB1, BMP2, BMP4, AGT, RELN). rs3914132 (RELN), C__7594810_10; rs1800472 (TGFB1), Only the association between TGFB1 and otosclerosis was C___8708464_20; rs616322 (PPP2R5B), C____633165_10. replicated (Sommen et al. 2014). A number of these stud- ies were underpowered for detection of a RELN associa- tion. Therefore, tests of such an association in an adequately 1 3 Human Genetics (2018) 137:357–363 359 Sequence-specific primers were designed to specifi- Association testing cally amplify RELN transcripts RELN-201 (Ensembl Tran- script ID: ENST00000343529.9) and RELN-203 (Ensembl Tests for genetic association with disease were performed using logistic regression analysis using the SNPstats plat- Transcript ID: ENST00000428762.5) based on the sites of primers used in previous studies to detect RELN expression form (Sole et al. 2006) under different genetic inheritance models (dominant, recessive and additive), and to investigate and with reference to the exon/intron boundaries. Primer sequences: RELN-201F, 5′-GCA ATA CAG CGT CAA CAA covariate interactions with gender and ethnicity. A p value of < 0.05 was considered significant. None of the SNPs CGG-3′; RELN-201R, 5′ -G TT T GC G AG T G A GG A CG A CCT-3′; RELN-203F, 5′ -A CC AT G T GG A GG T CG T CC showed a deviation from Hardy–Weinberg equilibrium in cases or controls. To investigate whether the relationship TAGTA-3′ and RELN-203R, 5′ -CAC T CG G T C TT G AGA AGG GCTTT -3′. For all cDNA samples, a GAPDH PCR between the SNPs and otosclerosis was gender-dependent we ran the overall association test using gender as a cat- assay was performed as a control for cDNA integrity and a non-template control for background contamination. MG-63 egorical co-variate. With seven of the eight SNPs, no statisti- cally significant interaction with gender was observed. For cDNA was used as a positive control. PCR reactions con- tained 400 ng cDNA, 0.5 µM each primer, 0.3 mM dNTPs, rs17407577 in FGF2 a weak association with gender was observed (p = 0.034). In our cohort, 94% of the sample was 2  mM MgCl2, 5  µl of 5× GoTaq PCR buffer and 1.25U GoTaq polymerase (Promega) in a total volume of 20 µl. of white European origin, no interaction was detected when ethnicity was included as a categorical variable. The tests of Thermal cycling was performed under the following reac- tion conditions: denaturation at 93 °C for 15 s, annealing at association were performed separately in white Europeans alone and with participants of all ethnicity and gave similar optimum temperature for 30 s, extension at 72 °C for 30 s, for up to 40 cycles, with an initial denaturation step of 93 °C results. for 3 min on an Eppendorf Mastercycler Gradient machine. The resulting DNA bands were visualised by agarose gel Power calculations electrophoresis. Power calculations were performed using the Genetic Power Calculator (http://zzz.bwh.harva rd.edu/gpc/) (Purcell et al. Results and discussion 2003). All calculations were performed assuming a disease prevalence of 0.35%, setting the relative risk and disease Genome wide and candidate gene association studies rep- allele frequency based on previous studies under an additive model (Sommen et al. 2014). resent powerful genetic tools. However, such studies can generate false positive associations. Replication in independ- RELN expression in human stapes ent populations is, therefore, vital for discriminating true associations from false positives. An important limitation Stapes suprastructures were collected from 76 individuals to the replication of genetic associations within otosclerosis has been the lack of well characterised patient cohorts of undergoing laser-assisted stapedotomy surgery at the Royal National Throat Nose and Ear Hospital (London, UK). A sufficient size for replication studies. We have genotyped diagnosis of otosclerosis was confirmed during surgery in all cases. Five additional stapes suprastructures and one whole stapes were obtained from individuals who did not have oto- Table 1 Power calculation for each SNP within the British case–con- trol cohort sclerosis and who were undergoing surgical procedures dur- ing which the stapes were removed. These included surgery SNP Gene DAF OR Power (%) due to head trauma, glomus tumour and total petrosectomy. rs39399 RELN 0.49 1.45 99.9 Human stapes were preserved in AllProtect™ Solution (Qia- rs3914132 RELN 0.78 1.54 99.7 gen) and stored at − 80 °C. Stapes were homogenized in rs39335 RELN 0.15 1.35 81.8 QIAzol Lysis reagent and RNA purified using the RNeasy rs3178250 BMP2 0.76 2.03 99.9 lipid tissue mini kit (Qiagen) according to the manufactur- rs1800012 COL1A1 0.82 1.55 45.0 er’s protocol, including an on column DNase digest. RNA rs17407577 FGF2 0.03 2.00 76.1 was also extracted from homogenised human MG-63 cells, rs1800472 TGFB1 0.97 2.33 48.9 a human osteosarcoma cell line using the RNeasy mini kit rs616322 PPP2R5B 0.08 1.46 70.0 (Qiagen) in accordance with the manufacturer’s protocol. RNA was reverse transcribed into cDNA by Omniscript Power calculations were performed based on previously reported data Reverse Transcriptase Kit (Promega) for RT-qPCR. (Chen et al. 2002; Sommen et al. 2014) DAF disease allele frequency, OR odds ratio 1 3 360 Human Genetics (2018) 137:357–363 eight previously implicated SNPs in a novel British cohort of might be protective against otosclerosis is unclear (Thys 374 patients and 374 controls. Association with disease was et al. 2007). Our results, combined with previous similar tested by logistic regression analysis under various genetic association studies, suggest that the mutant form of the pro- models. Power calculations (Table  1) estimated that the tein is protective against otosclerosis. study had good power to detect most SNP effects (between The lack of association between any of the three RELN 70 and 99.9% power for six SNPs) with only TGFB1 and SNPs and otosclerosis is surprising given the high power COL1A1 having limited power to detect an effect, at 48.9 of the study to detect an effect for these SNPs. Otosclerosis and 45.0%, respectively. has both familial and non-familial forms and approximately Results for tests of genetic association with otosclerosis 44% of our patient cohort reported other family members in all cases and controls are shown in Table  2. Only the with a diagnosis of otosclerosis. Since it is possible that rs1800472 variant within the TGFB1 gene (p = 0.021) was there are distinct genetic susceptibilities involved in the two found to be significantly associated with otosclerosis. This forms of the disease we performed an analysis when strati- SNP was selected for a case–control association study due fying the patient cohort by familial or non-familial disease, to several genetic indications of involvement with the dis- see Tables 3 and 4 respectively. In familial cases a strong ease. Thys et al. (2009), performed a case–control associa- association was detected between rs39399 and otosclerosis tion study in a Belgian-Dutch cohort which gave significant (p = 0.0057, Table 3) that was not present in non-familial results (p = 0.0044) (Thys et  al. 2009). Within the same cases (Table  4). As with the association seen in TGFB1, study, the association was replicated in a French population, it is the major allele that is associated with disease, rep- with even greater significance (p = 0.00019). Our study rep- licating the direction of association detected in the origi- resents the third subsequent case–control association study nal GWAS. Our results suggest that an association exists which has confirmed the association (Khalfallah et al. 2010; between rs39399 and the familial form of otosclerosis, which Sommen et al. 2014). The evidence for the association is increased further by the absence of any conflicting stud- Table 3 Association testing in familial otosclerosis cases (n = 160) ies which have failed to demonstrate an association. Fur- and controls (n = 374) thermore, much of this supportive evidence has come from SNP Gene p OR (95% CI) studies which are underpowered to detect the SNP’s effect. All studies have found the disease to be associated with the rs39399 RELN 0.0057 1.73 (1.18–2.54) rs3914132 RELN 0.17 1.33 (0.89–1.99) major allele suggesting the variant is protective. The SNP is relatively rare with a minor allele frequency of 0.024 and rs39335 RELN 0.30 1.25 (0.82–1.90) rs3178250 BMP2 0.54 0.89 (0.61–1.30) 0.045 in cases and controls respectively. The rs1800472 SNP is a missense variant that causes a rs1800012 COL1A1 0.30 1.25 (0.82–1.90) rs17407577 FGF2 0.84 0.96 (0.65–1.41) threonine to isoleucine substitution at amino acid position 263 in TGFB1, this change has been shown to effect activ - rs1800472 TGFB1 0.28 1.48 (0.71–3.08) rs616322 PPP2R5B 0.25 0.78 (0.52–1.18) ity of the protein (Thys et al. 2009). Analysis of the T263I substitution with a luciferase reporter assay showed that the Significant values are indicated in bold. p values are shown for a test protective variant of TGFB1 is more active than the major of association under the dominant model. All odds ratios quoted are allele. The reason why a more active version of the protein for the minor allele Table 2 Association testing of the eight SNPs in the full case–control Table 4 Association testing in non-familial otosclerosis cases cohort (n = 748) (n = 206) and controls (n = 374) SNP Gene p OR (95% CI) SNP Gene p OR (95% CI) rs39399 RELN 0.32 1.17 (0.86–1.60) rs39399 RELN 0.40 0.85 (0.58–1.24) rs3914132 RELN 0.33 1.16 (0.86–1.58) rs3914132 RELN 0.55 1.00 (0.70–1.43) rs39335 RELN 0.38 1.16 (0.84–1.59) rs39335 RELN 0.52 1.04 (0.71–1.51) rs3178250 BMP2 0.65 0.93 (0.70–1.25) rs3178250 BMP2 0.92 0.98 (0.69–1.39) rs1800012 COL1A1 0.21 1.13 (0.84–1.53) rs1800012 COL1A1 0.24 1.24 (0.86–1.78) rs17407577 FGF2 0.45 0.85 (0.55–1.30) rs17407577 FGF2 0.90 0.99 (0.59–1.67) rs1800472 TGFB1 0.021 0.51 (0.28–0.91) rs1800472 TGFB1 0.015 0.40 (0.18–0.88) rs616322 PPP2R5B 0.38 0.86 (0.62–1.20) rs616322 PPP2R5B 0.71 0.89 (0.60–1.33) Significant values are indicated in bold. p values are shown for a test Significant values are indicated in bold. p values are shown for a test of association under the dominant model. All odds ratios quoted are of association under the dominant model. All odds ratios quoted are for the minor allele for the minor allele 1 3 Human Genetics (2018) 137:357–363 361 requires further investigation in familial cohorts. No associa- Previous association studies with RELN and otosclero- tion between rs1800472 in TGFB1 and disease was detected sis have not discriminated between cases with familial and in familial cases (Table 3), however, in non-familial cases non-familial disease. If our finding of RELN association an association was again detected this time with a lower p with familial disease only is substantiated it is possible that value (p = 0.015, Table 4) and greater odds ratio than was the proportion of patients with familial disease in cohorts detected in the combined otosclerosis cases, suggesting that used in previous studies may have influenced the ability to this SNP may be playing a role in the non-familial form of detect this association and account for variable replication the disease only. between studies. For example, variant rs3914132 has now It has been well documented that some women with oto- been included in five follow up case–control association sclerosis associate the onset or worsening of hearing loss studies following its identification in the original GWAS. with pregnancy and it has been postulated that this may be Two of those studies have replicated the initial association due to the influence of oestrogen (Torsiglieri et al. 1990). (Khalfallah et al. 2010; Schrauwen et al. 2009a). However, Approximately 40% of the females in our cohort who had two others have failed to display the replication (Priyadarshi experienced a pregnancy self-reported a worsening of et al. 2010; Sommen et al. 2014). The results also suggest a hearing with pregnancy in questionnaire data. To explore hypothesis that the otosclerosis phenotype maybe reached, whether these females represent a distinct genetic aetiology through an oestrogen dependent or independent pathway. from the main cohort we tested association with disease in Those cases which are oestrogen dependent, progress dispro- females stratified by this self-reported association with preg- portionately during pregnancy. Our data suggest that RELN nancy (Table 5). may be exclusively involved in the oestrogen independent No association with otosclerosis was detected with any mechanism. If so, the results of previous case–control asso- of the eight SNPs in those women who reported worsening ciation studies may have been determined by the make-up of hearing in pregnancy or those who had not been pregnant. each cohort which may have been influenced by differences A significant relationship was seen with the RELN SNPs, in recruitment pipelines. rs3914132 (p = 0.0095) and rs39335 (p = 0.038), in those Another factor in considering whether RELN is involved women who did not report a deterioration in their hearing in the pathogenesis of otosclerosis has been the lack of when compared with controls. These data should be treated an obvious role for the reelin protein in stapes bone and with caution as the stratified sample sizes are small and, the uncertainty as to whether it is expressed in human therefore, more prone to error. If substantiated by replication stapes. Two previous studies reported conflicting results in other cohorts our results suggest that RELN may play a in detecting RELN transcripts in stapes bone (Cso- role in the pathogenesis of otosclerosis, but only in specific mor et  al. 2012; Schrauwen et  al. 2009b). The current subtypes. Indeed our data when taken together suggest that release of human genome assembly (Ensembl: GRCh38. otosclerosis has a complex pathophysiology, with non-famil- p10, accessed November 2017) shows that the predicted ial and familial cases having different underlying genetic major transcripts of RELN diverge in the use of different mechanisms, each involving separate genetic pathways. microexons 64 (no exon, 4 bp exon or 6 bp exon) at the 3′ end of the transcript. Examination of the primers used in previous studies suggested that the assays used may Table 5 Association testing in women stratified by reported hearing be detecting different transcripts (RELN-201 in Schrau- loss associated with pregnancy wen et al. 2009a) and (RELN-203 in Csomor et al. 2012), SNP Gene No pregnan- Pregnant Pregnant cies change no change p value p value p value (n = 55) (n = 65) (n = 99) Table 6 RELN expression in human stapes Sample (n) No. (%) of samples in which tran- rs39399 RELN 0.28 0.95 0.094 script is detected rs3914132 RELN 0.89 0.22 0.0073 GAPDH RELN-201 RELN-203 rs39335 RELN 0.75 0.76 0.038 rs3178250 BMP2 0.86 0.63 0.70 MG-63 cells (1) 1 (100%) 1 (100%) 1 (100%) rs1800012 COL1A1 0.75 0.60 0.47 Female stapes suprastructures 52 (100%) 32 (62%) 0 (0%) rs17407577 FGF2 0.19 0.60 0.39 (52) a a a rs1800472 TGFB1 0.36 0.67 0.17 Male stapes suprastructures 19 (100%) 11 (58%) 0 (0%) rs616322 PPP2R5B 0.93 0.21 0.20 (19) Control suprastructures (5) 5 (100%) 4 (80%) 0 (0%) Significant values are indicated in bold. p values are shown for a test Total stapes suprastructures 76 (100%) 47 (62%) 0 (0%) of association under the dominant model unless otherwise indicated (76) Recessive model 1 3 362 Human Genetics (2018) 137:357–363 Fig. 1 A 2% agarose gel showing RELN-201 RT-PCR amplicons detected in human stapes cDNA. RELN-201 was detected in MG-63 cells and five out of ten human stapes suprastructures shown (a repre- sentative sample of the cohort), illustrated by a 209 bp fragment (arrowed) which might account for the different results. We therefore Acknowledgements We are indebted to the patients and their families for their willingness to participate in our research study. We are also designed assays to detect both RELN-201 and RELN-203 grateful for the UK CRN staff, physicians and hospital staff whose in human stapes. The results (Table 6; Fig. 1) show that the efforts in collecting these samples were essential to this research. RELN-201 transcript is detected in the majority of stapes suprastructures removed from both otosclerosis patients Funding This work was supported by Action on Hearing Loss (refs G71, G83 and 512:UEI:SD). and controls. However, the RELN-203 transcript was not detected in any human stapes, but was detected in the posi- Compliance with ethical standards tive control, cDNA from the MG-63 cell line. Our results clarify the expression of RELN in human stapes and are Conflict of interest On behalf of all authors, the corresponding author similar to those reported by Schrauwen et al. (2009a, b) states that there is no conflict of interest. suggesting that reelin is expressed in the majority but not all of human stapes tested (Schrauwen et al. 2009a). Open Access This article is distributed under the terms of the Crea- No statistically significant association was seen between tive Commons Attribution 4.0 International License (http://creat iveco mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- four SNPs in COL1A1, FGF2, PPP2R5B, or BMP2 and tion, and reproduction in any medium, provided you give appropriate otosclerosis across all comparisons. This provides further credit to the original author(s) and the source, provide a link to the evidence against the involvement of these genes in oto- Creative Commons license, and indicate if changes were made. sclerosis disease pathogenesis. For some of the SNPs in question, such as rs1800012 in COL1A1, this is the third consecutive case–control association study (Khalfallah et al. 2010; Sommen et al. 2014) in which an association References has failed to replicate the initial positive association (Chen et al. 2002). However, due to the low odds ratio reported Cawthorne T (1955) Otosclerosis. J Laryngol Otol 69:437–456 Chen W, Campbell CA, Green GE et al (2002) Linkage of otosclerosis for this association this study has poor power to detect to a third locus (OTSC3) on human chromosome 6p21.3-22.3. J such a weak effect. For other variants the reliability of Med Genet 39:473–477 the negative association is increased by the power of this Chole RA, McKenna M (2001) Pathophysiology of otosclerosis. Otol study. For example, no effect was detected for rs3178250 Neurotol 22:249–257 Csomor P, Sziklai I, Karosi T (2012) Controversies in RELN/reelin in BMP2 despite the study having a 99.9% power to detect expression in otosclerosis. Eur Arch Otorhinolaryngol 269:431– it (Table 1). 440. https ://doi.org/10.1007/s0040 5-011-1653-4 In conclusion, we detected an association between a func- Declau F, Van Spaendonck M, Timmermans JP, Michaels L, Liang J, tional variant in TGFB1 and clinically confirmed otoscle- Qiu JP, Van de Heyning P (2001) Prevalence of otosclerosis in an unselected series of temporal bones. Otol Neurotol 22:596–602 rosis in a British population. This replicates two previous Dulabon L, Olson EC, Taglienti MG, Eisenhuth S, McGrath B, Walsh studies which have found the same association in the same CA, Kreidberg JA, Anton ES (2000) Reelin binds alpha3beta1 direction. Further, we detect an association between different integrin and inhibits neuronal migration. Neuron 27:33–44 SNPs within RELN and familial otosclerosis and in women Ealy M, Smith RJ (2010) The genetics of otosclerosis. Hear Res 266:70–74. https ://doi.org/10.1016/j.heare s.2009.07.002 whose hearing loss does not progress during pregnancy. A Fatemi SH, Snow AV, Stary JM, Araghi-Niknam M, Reutiman TJ, previously described association between four other genes Lee S, Brooks AI, Pearce DA (2005) Reelin signaling is impaired (COL1A1, BMP2, PPP2R5B, and FGF2) was not replicated, in autism. Biol Psychiatry 57:777–787. https://doi.or g/10.1016/j. providing further negative evidence for any involvement of biops ych.2004.12.018 these genes in the pathogenesis of the condition. 1 3 Human Genetics (2018) 137:357–363 363 Iossa S, Corvino V, Giannini P et al (2013) The rs39335 polymorphism Schrauwen I, Ealy M, Fransen E et al. (2010) Genetic variants in the of the RELN gene is not associated with otosclerosis in a southern RELN gene are associated with otosclerosis in multiple European Italian population. Acta Otorhinolaryngol Ital 33:320–323 populations. Hum Genet 127:155–162. https ://doi.org/10.1007/ Ishai R, Halpin CF, Shin JJ, McKenna MJ, Quesnel AM (2016) Long-s0043 9-009-0754-2 term incidence and degree of sensorineural hearing loss in oto- Sole X, Guino E, Valls J, Iniesta R, Moreno V (2006) SNPStats: a sclerosis. Otol Neurotol 37:1489–1496. https ://doi.org/10.1097/ web tool for the analysis of association studies. Bioinformatics MAO.00000 00000 00123 4 22:1928–1929. https ://doi.org/10.1093/bioin forma tics/btl26 8 Khalfallah A, Schrauwen I, Mnaja M et al (2010) Genetic variants in Sommen M, Van Camp G, Liktor B, Csomor P, Fransen E, Sziklai RELN are associated with otosclerosis in a non-European popula- I, Schrauwen I, Karosi T (2014) Genetic association analysis in tion from Tunisia. Ann Hum Genet 74:399–405. https://doi.or g/1 a clinically and histologically confirmed otosclerosis popula- 0.1111/j.1469-1809.2010.00595 .x tion confirms association with the TGFB1 gene but suggests an Mann WJ, Amedee RG, Fuerst G, Tabb HG (1996) Hearing loss as association of the RELN gene with a clinically indistinguishable a complication of stapes surgery. Otolaryngol Head Neck Surg otosclerosis-like phenotype. Otol Neurotol 35:1058–1064. https 115:324–328. https ://doi.org/10.1016/S0194 -5998(96)70046 -3://doi.org/10.1097/MAO.00000 00000 00033 4 Morrison AW (1967) Genetic factors in otosclerosis. Ann R Coll Surg Thys M, Schrauwen I, Vanderstraeten K et al (2007) The coding poly- Engl 41:202–237 morphism T263I in TGF-beta1 is associated with otosclerosis in Priyadarshi S, Panda KC, Panda AK, Ramchander PV (2010) Lack two independent populations. Hum Mol Genet 16:2021–2030. of association between SNP rs3914132 of the RELN gene and https ://doi.org/10.1093/hmg/ddm15 0 otosclerosis in India. Genet Mol Res 9:1914–1920. https ://doi. Thys M, Schrauwen I, Vanderstraeten K et  al (2009) Detection org/10.4238/vol9-3gmr8 90 of rare nonsynonymous variants in TGFB1 in otosclerosis Purcell S, Cherny SS, Sham PC (2003) Genetic Power Calculator: patients. Ann Hum Genet 73:171–175. https ://doi.or g/10.111 design of linkage and association genetic mapping studies of 1/j.1469-1809.2009.00505 .x complex traits. Bioinformatics 19:149–150 Torsiglieri AJ Jr, Tom LW, Keane WM, Atkins JP Jr (1990) Otolar- Schrauwen I, Ealy M, Huentelman MJ et al. (2009a) A genome-wide yngologic manifestations of pregnancy. Otolaryngol Head Neck analysis identifies genetic variants in the RELN gene associated Surg 102:293–297. https://doi.or g/10.1177/0194599890 10200 317 with otosclerosis. Am J Hum Genet 84:328–338. https ://doi. Ziff JL, Crompton M, Powell HR, Lavy JA, Aldren CP, Steel KP, Saeed org/10.1016/j.ajhg.2009.01.023 SR, Dawson SJ (2016) Mutations and altered expression of SER- Schrauwen I, Thys M, Vanderstraeten K et al (2009b) No evidence for PINF1 in patients with familial otosclerosis. Hum Mol Genet association between the renin-angiotensin-aldosterone system and 25:2393–2403. https ://doi.org/10.1093/hmg/ddw10 6 otosclerosis in a large Belgian-Dutch population. Otol Neurotol 30:1079–1083. https://doi.or g/10.1097/MAO.0b013e3181 ab305 8 1 3

Journal

Human GeneticsSpringer Journals

Published: May 4, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off