SEARCH

SEARCH BY CITATION

Keywords:

  • Age-related hearing loss;
  • auditory brainstem response;
  • BXD recombinant inbred strains;
  • complex trait analysis;
  • heritability;
  • qPCR;
  • QTL

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

The DBA/2J inbred strain of mice has been used extensively in hearing research as it suffers from early-onset, progressive hearing loss. Initially, it mostly affects high frequencies, but already at 2–3 months hearing loss becomes broad. In search for hearing loss genes other than Cadherin 23 (otocadherin) and fascin-2, which make a large contribution to the high-frequency deficits, we used a large set of the genetic reference population of BXD recombinant inbred strains. For frequencies 4, 8, 16 and 32 kHz, auditory brainstem response hearing thresholds were longitudinally determined from 2–3 up to 12 weeks of age. Apart from a significant, broad quantitative trait locus (QTL) for high-frequency hearing loss on chromosome 11 containing the fascin-2 gene, we found a novel, small QTL for low-frequency hearing loss on chromosome 18, from hereon called ahl9. Real-time quantitative polymerase chain reaction of organs of Corti, isolated from a subset of strains, showed that a limited number of genes at the QTL were expressed in the organ of Corti. Of those genes, several showed significant expression differences based on the parental line contributing to the allele. Our results may aid in the future identification of genes involved in low-frequency, early-onset hearing loss.

The genetic basis of age-related hearing loss (AHL) has been well studied in the mouse (Friedman et al. 2007; Ohlemiller 2006). Longitudinal measurements of hearing level are relatively easy, and the contribution of genetic and environmental factors can be assessed more readily than in humans. Many mouse strains develop AHL (Zheng et al. 1999), including the C57BL/6J mouse, which is often used as the background strain in transgenic studies. In this strain, the first AHL locus (ahl) was described and mapped to a site on chromosome 10 (Johnson et al. 1997). Additional studies in inbred mouse strains identified more quantitative trait loci (QTLs) for AHL, including ahl2 (Johnson & Zheng 2002), ahl3 (Morita et al. 2007; Nemoto et al. 2004), ahl4 (Zheng et al. 2009), ahl5 and 6 (Drayton & Noben-Trauth 2006) and, more recently, ahl8, which is located on the distal end of chromosome 11 (Johnson et al. 2008). In addition, Mashimo et al. (2006) identified Phl1 and Phl2, two QTLs for progressive hearing loss. To date, the responsible genes were identified only for ahl and ahl8. For ahl, the responsible gene is Cadherin 23, also called otocadherin, which is required for the normal organization of hair cell stereocilia (Di Palma et al. 2001). For ahl8, a mutation in the fascin-2 gene, which also encodes a component of stereocilia, contributes to the early-onset hearing loss in DBA/2J mice (Shin et al. 2010).

The BXD strains are a panel of recombinant inbred strains derived from the parental strains C57BL/6J and DBA/2J (from hereon called B6 and D2, respectively). D2 is a widely used mouse strain in which hearing loss starts much earlier than in B6. In addition, low frequencies are also affected in D2, in contrast to the high-frequency hearing loss of B6 (Willott & Erway 1998). The BXD strains have been used to identify QTLs for the early-onset hearing loss of D2 (Johnson et al. 2008; Willott & Erway 1998). At frequencies above 8 kHz, up to 75% of total variance can be explained by mutations in cadherin 23 (ahl), fascin-2 (ahl8) and their interaction in (B6.CAST-Cdh23Ahl+ × D2) × D2 backcross mice. However, at 8 kHz, approximately two thirds of the variance remains unexplained, while up to half of the variance in auditory brainstem response (ABR) to clicks (2–8 kHz range) could not be attributed to ahl or ahl8 (Johnson et al. 2008). In addition, preliminary evidence has suggested the presence of an additional third locus in D2 (Erway et al. 1993). In this study, we therefore performed a search for the existence of additional QTLs underlying early-onset hearing loss in BXD strains, incorporating recordings at lower frequencies and making use of the advanced intercross BXD mouse strains that have recently become available (Peirce et al. 2004).

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

Animals

Parental (C57BL/6J and DBA/2J) and 33 BXD lines were received from The Jackson Laboratory (http://www.jax.org) or from Oak Ridge Laboratory (BXD43, BXD62, BXD65, BXD68, BXD69, BXD73, BXD75, BXD87 and BXD90), and were bred in the facility of the Neuro-Bsik consortium of the VU University Amsterdam. Hearing level thresholds were longitudinally assessed at either 2 or 3 weeks to minimize the burden of anesthesia at a very young age and subsequently at 4, 6 and 12 weeks of age under ketamine/xylazine anesthesia (60/10 mg/kg i.p.) in both male and female mice. The average number of mice per strain was 7 (range 2–13). To facilitate recovery from anesthesia, mice were injected with atipamezole (25 µg s.c.) at the end of the experiments. Recordings were performed during the light phase. On average, seven mice were used per strain, generally derived from two different litters. Mice were housed in same-sex groups after weaning, from 3 to 4 weeks onward. The experiments were approved by the Erasmus University Animal Welfare Committee (DEC).

Auditory brainstem response

The ABR was used to obtain hearing level thresholds as described previously (Spoor et al. 2012). Briefly, hearing level thresholds were measured in response to 1 millisecond tone pips at 4, 8, 16 and 32 kHz at intensities ranging from −10 to 110 dB sound pressure level (SPL; re 20 μPa) at 5-dB resolution. At each sound intensity tested, 500 brainstem evoked responses with artifacts below 30 μV were averaged; the minimum threshold was defined as the lowest SPL at which a reproducible peak could be identified. If no response was elicited at 110 dB, the threshold was classified as 115 dB.

Heritability calculation

As a quantitative trait for early-onset hearing loss, we used the difference between the ABR threshold at 12 weeks and the lowest threshold measured at any age. For each frequency, this maximum hearing loss (MHL) was calculated per mouse and subsequently averaged per strain. Narrow-sense heritability (h2) of the MHL was calculated as the ratio of the within- and between-strain variance: (F − 1)/(F − 1 + 2 × k), in which F is the between-groups divided by within-groups mean squares ratio, which can be retrieved from a normal one-way analysis of variance (anova) (Hegmann & Possidente 1981), k is a function of the replicate number of mice per strain and is defined by: (N − ((Σ n2)/N))/(S − 1), in which N is the total number of mice, n is the number of mice for each strain and S is the number of different strains (Lynch & Walsh 1998), as implemented previously (Heimel et al. 2008; Loos et al. 2009).

QTL linkage mapping

Linkage mapping of MHL values of BXD plus parental strains to genotypes was performed by WebQTL (http://www.genenetwork.org/) scripts, which uses a set of 3795 single nucleotide polymorphism (SNP) and microsatellite markers (Chesler et al. 2004; Wang et al. 2003). For each part of the genome, a likelihood ratio statistic (LRS) was calculated by Haley–Knott interval mapping (Haley & Knott 1992). Suggestive and significant LRS values were determined by a permutation test (Churchill & Doerge 1994), consisting of 1000 permutations in which trait values were randomly reassigned across all strains, followed by a comparison of the permuted and original outcome data to assess significance. A significant LRS value was defined as a 5% probability of falsely rejecting the null hypothesis that there is no linkage anywhere in the genome. A suggestive threshold represents the LRS value that yields, on average, one false-positive QTL per genome scan (P = 0.63). Candidate genes were analyzed within the 1-LOD (=4.61 LRS) drop-off region with respect to the maximum LRS, which theoretically corresponds to a 97% confidence interval (Visscher & Goddard 2004). Pearson correlation coefficients and respective probabilities were calculated across strain means to approximate genetic correlations. Linear regression was used to quantify independent contributions of the gene loci to the observed MHL. For the analysis of SNPs and (non)-synonymous mutations in genes under the QTL peak, the SNP browsers from the Phenome database of Jackson Laboratories (http://phenome.jax.org/db/q?rtn=snp/ret1) and GeneNetwork (http://www.genenetwork.org/webqtl/main.py?FormID=snpBrowser) were used, the latter of which harbors data from multiple sequencing projects, including the Wellcome Trust Sanger Institute and the UCLA genome browser.

Expression analysis of candidate genes

Real-time quantitative polymerase chain reaction (qPCR) for genes on the chromosome 18 and 10 loci (Tables 1 and 2; Table S3, Data S1) was carried out on cochlea tissue of selected strains of 12 weeks of age (BXD11, n = 7; BXD13, n = 8; BXD32, n = 6; BXD73, n = 5; B6, n = 12; D2, n = 10). RNA extraction (tissue pooled from two mice) and complementary DNA (cDNA) synthesis (300-ng RNA equivalent in a 25-µl reaction) were performed as described previously (Loos et al. 2009; Spijker et al. 2004). The PCR measurements (10 µl; ABI PRISM 7900; Applied Biosystems, Foster City, CA, USA) were performed with transcript-specific primers (300 nM; Table S4) on cDNA corresponding to approximately 1.2 ng of total DNAse-I-treated RNA. Cycle of threshold (Ct) values were used to calculate the relative level of gene expression (log 2 scale) normalized to the geometric mean of the replicated reference controls GAPDH and β-actin, as described before (Spijker et al. 2004). Control genes, such as Cdh23 (Siemens et al. 2004), Otof (Engel et al. 2006) and the cochlea-specific subunit of the acetylcholine receptor Chnra9, were measured twice. Only primer sets for which a single product could be detected were used, as determined by dissociation analysis of the end product. Presence or absence of gene expression in cochlea was determined using a pooled cDNA sample of cochlea vs. total brain cDNA. When primers detected a true product in the total sample in a replicate PCR, but not in the cochlea sample, or when the Ct value of the cochlea sample was within 2 cycles of the water control, the gene was called ‘not expressed’.

Table 1. Genes at ahl9, the significant QTL on chromosome 18
Gene symbolNameStart (Mb)Length (kb)# Coding SNPsExpressed
  1. Displayed are genes within the 1-LOD drop-off region (except top and bottom gene) with their gene symbol, description, start site, length and number of synonymous (S) and non-synonymous (NS) coding SNPs that are similar in normal hearing strains B6 and CBA/J, but differ from D2. Those SNPs that differ between B6 and D2 are listed between parentheses (for a full list of SNPs, see Table S2). ‘Expressed’ indicates whether expression could be measured above background in cochlea by qPCR.

 S/NS 
Ccdc11Coiled-coil domain containing 1174.4475.130/0N
Myo5bMyosin 5b74.60328.860/0 (19/6)Y
1700120E14RikRIKEN cDNA 1700120E14 gene74.6635.450/0N
Acaa2Acetyl-Coenzyme A acyltransferase 274.9427.000/0Y
LipgLipase, endothelial75.1021.940/0N
9030625G05RikRIKEN cDNA 9060325G05 gene75.138.780/0Y
Rpl17Ribosomal protein L1775.162.910/0Y
BC031181cDNA sequence BC03118175.174.030/0Y
DymDymeclin75.18268.200/0 (2/0)Y
2010010A06RikRIKEN cDNA 2010010A06 gene75.458.660/0N
Smad7MAD Homolog 7 (Drosophila)75.5328.570/0Y
Gm672 (Ctif)Gene model 672/CBP80/20-dependent translation initiation factor75.59266.480/0 (1/0)Y
2900040J22RikRIKEN cDNA 2900040J22 gene75.631.310/0N
1700034B16RikRIKEN cDNA 1700034B16 gene75.941.270/0N
Table 2. Gene expression analysis of genes at the significant QTL on chromosome 18 and controls
Gene symbolExpression levelanovaAverage ranks B6 or D2 alleleSpearman rank MHL at 4 kHzPearson product MHL at 4 kHzMHL at4 kHz
Strain P-value (F5,42)B6 or D2* P-value (F1,24)B6 or D2 P-value (F1,46)3x B63x D2Corr.P-valueCorr.P-valueLRS score
  1. Displayed are genes (see Table 1) used in the qPCR analysis across six strains with divergent hearing loss, their gene symbol, the average expression level, the anova results (α = 0.001) across strains, and the presence of the B6 or the D2 allele (* without parental lines; α = 0.01), the mean of ranks when strain averages were used for the analysis of having the B6 or D2 allele, the Spearman rank and Pearson product–moment correlation (and P-value) with MHL at 4 kHz and the LRS score of this trait. XH, Relative expression level (log 2) across strains (Expression) ≥ 10; H, 10 > Expression ≥ 8; M, 8 > Expression ≥ 5; L, 5 > Expression. Significant (bold; anova, P < 0.001; correlation, P < 0.05) and trends for (italics; correlation, 0.05 < P < 0.1) differences are highlighted. Note that no significant results were obtained in any analysis for Cdh23, a gene on chromosome 10 outside the suggestive QTL for MHL.

Chr18           
Myo5bM2.52E 03 (4.4)1.93E − 02 (6.3)1.09E − 04 (17.9)5.02.0−0.770.084−0.680.14119.5
Acaa2M-H1.14E − 04 (6.7)6.98E − 03 (8.7)1.74E − 06 (30.0)5.02.0−0.730.103−0.810.05120.9
9030625G05RikL4.23E − 05 (7.7)1.89E − 01 (1.8)4.37E − 03 (9.05)4.56.00.260.5620.620.18620.7
Rpl17XH2.19E − 08 (14.8)3.38E − 01 (1.0)1.65E − 02 (6.2)6.54.0−0.710.110−0.400.42720.7
BC031181XH1.31E − 05 (8.5)4.11E − 04 (16.8)3.44E − 07 (35.4)5.02.0−0.940.035−0.950.00420.7
DymH3.81E − 02 (2.6)1.06E − 01 (2.8)4.02E − 03 (9.2)5.02.0−0.830.063−0.640.16820.7
Smad7M-H5.86E − 15 (41.3)3.05E − 04 (17.8)2.72E − 09 (54.6)5.02.0−0.770.084−0.820.04720.7
Gm672 (Ctif)H1.84E − 04 (6.3)1.05E − 02 (7.7)5.55E − 06 (26.4)5.02.0−0.830.063−0.800.05620.7
Control           
Cdh23M5.96E − 01 (0.7)3.62E − 01 (0.9)3.21E − 01 (1.0)4.32.7−0.540.222−0.360.489NA
OtofH5.32E − 10 (19.5)2.17E − 03 (11.8)1.38E − 04 (17.3)4.32.7−0.540.222−0.670.146NA
Chrna9L2.80E − 04 (6.7)4.38E 02 (4.9)4.54E 01 (4.3)4.32.7−0.310.703−0.590.220NA

Statistical analysis

Quantitative measurements were analyzed by Pearson correlation to the trait and by anova for the factor allele (i.e. B6 or D2). Non-parametric ranks are indicated for strain averages carrying the B6 or D2 allele. All data are presented as average ± standard error of the mean.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

The hearing of a total number of 35 (including parental) strains was evaluated in longitudinal recordings from a postnatal age of either 2 or 3 weeks up to the last measurement at 12 weeks of age (Fig. S1 and Table S1). Maximum hearing loss varied greatly among strains and frequencies (Fig. 1). Pearson correlations between MHLs at 4, 8 and 16 kHz were high, ranging from r = 0.77 to 0.89. A negative correlation (r ≈ −0.24) between MHL at 32 kHz and each of the other tested frequencies was found. At 32 kHz, most BXD strains already showed high thresholds at a young age (Fig. S1). Even though these mice would typically become deaf at this frequency before or at an age of 12 weeks, the MHL was nevertheless low. This ceiling effect was responsible for the negative correlation with the MHL at other frequencies.

image

Figure 1. Strain means of MHL at four different frequencies. MHL was defined as the difference between the ABR threshold at 12 weeks and the lowest threshold measured at any time point. Insets contain Pearson's correlation coefficients between MHL data of the tested sound frequencies.

Download figure to PowerPoint

Heritability and correlations with previously published phenotypes

Narrow-sense heritability was high: 0.30 at 4 kHz, 0.45 at 8 kHz, 0.56 at 16 kHz and 0.21 at 32 kHz. Our data correlated well with other auditory studies on BXD mice within the database of WebQTL, not only with spiral ganglion cell density measurements in different basoapical sections of the cochlea (r from −0.49 to −0.78, P ≤ 0.03 for several basoapical sections at 4, 8 and 16 kHz; Willott & Erway 1998) but also with post-mortem magnetic resonance imaging measurements on total volume of the lateral lemniscus (r = −0.70, P = 0.02 at 16 kHz; r ∼ −0.85, P < 0.001 at 4 and 8 kHz; Badea et al. 2009). Higher hearing loss values were correlated with lower cell count or lateral lemniscus volume.

Quantitative trait mapping

A whole genome scan, which was performed separately for each of the four frequencies, yielded two significant peaks of interest (Fig. 2), in each case contributed by the D2 allele. First of all, we found a novel QTL at chromosome 18, which will be called ahl9. It was significant at 4 kHz (LRS = 20.9), but not at higher frequencies, and contains only a small number of genes (Fig. 3 and Table 1). Detailed analysis of the region indicated a small, non-significant peak upstream of ahl9 (Fig. 3). Because there was considerable recombination surrounding the 1-LOD interval, with 60% of the BXD lines analyzed having the D allele, we assumed that this extended region represents a separate QTL.

image

Figure 2. Complete genome linkage analysis for early-onset hearing loss. All four frequencies are shown in different colors (red = 4 kHz; blue = 8 kHz; green = 16 kHz and black = 32 kHz), with approximate suggestive and significance levels indicated by horizontal, dashed lines. Two QTLs at chromosome 11 and 18 reached significance (LRS > 18). These QTLs, highlighted by circles, are determined by the presence of the D2 allele.

Download figure to PowerPoint

image

Figure 3. Significant locus at chromosome 18, ahl9. (a) Magnified view of the significant QTL at chromosome 18 (peak at LRS = 20.9; additive allele effect of D2 = 7.5) for early-onset, low-frequency hearing loss (data from 4 kHz). Horizontal, dashed lines indicate suggestive (*; LRS = 10.7) and significance (**; LRS = 18.8) levels. (b) Zoomed-in view of (a) showing an overview of the genes present on ahl9, the significant QTL for early-onset hearing loss at 4 kHz.

Download figure to PowerPoint

Second, a locus was mapped to the distal end of chromosome 11, confirming the recently published fascin-2 gene (Shin et al. 2010). This locus was only significant at 16 kHz (LRS = 20.3 at 16 kHz) and produced peaks of smaller magnitude for 8 and 4 kHz (LRS = 11.7 and 7.6, respectively). Furthermore, we found a QTL at chromosome 10. It barely missed significance at 4 kHz (LRS = 17.1) and, similar to the locus on chromosome 18, harbors only a small number of genes (Fig. S2 and Table S3). Exclusion of parental strains did not alter significance. Pair-scan analysis showed no significant epistatic interactions among the significant and suggestive loci at chromosome 18 and 10, respectively, or between the newly found loci and the known locus of the fascin-2 gene. Composite interval mapping, which allows controlling for variation caused by the fascin-2 locus, showed a small drop in LRS value to 17.3, just below significance, of the locus on chromosome 18 at 4 kHz. Vice versa, the LRS value of the fascin-2 locus at 16 kHz also drops below significance, to 16.9, after controlling for the new locus at chromosome 18.

At 4 kHz, the new QTL showed a higher correlation with MHL (r = 0.67) than the known mutation in the fascin-2 gene (r = 0.44). Calculation of the partial correlation indicated that the QTL at chromosome 18 could explain 39.1% of the variance in MHL at 4 kHz that remained after accounting for the effects of fascin-2.

Additional analysis using absolute hearing levels instead of MHL is given in Data S1.

Candidate gene analysis

First, the QTL was checked for (non)-synonymous mutations. A list of genes and number of SNPs is shown in Table 1. We looked for the presence or absence of these SNPs in inbred mouse strains without early-onset hearing loss (Zheng et al. 1999), in particular the CBA/J, DBA/1J and DBA/2HaSmnJ strains. Several coding mutations that were present in D2 and absent in B6 were also present in strains with good hearing function and were therefore discarded. Mutations absent in B6 and the above mentioned normal hearing strains, and hence possibly related to hearing loss, were present in the 3′ UTR of Ctif and in the 5′ UTR of Smad7 (Table S2), all of which could possibly interfere with mRNA stability. No additions or changes in stop codons were found in any of the genes.

Second, as mutations could possibly interfere with regulation of gene expression, we analyzed SNPs in intronic regions. Intronic SNPs were present in all genes listed, except for 1700120E14Rik, Lipg, Rpl17, 2010010A06Rik and 2900040J22Rik. The analysis of SNP differences between B6 and D2 that were also present in the good hearing strain CBA/J greatly reduced the list of hearing-loss-related intronic SNPs, although this analysis does not take into account that predisposing alleles at other loci may be required for the hearing loss. A full list of coding (non-synonymous and synonymous) as well as non-coding SNPs is available in Table S2, which also includes the suggestive locus at chromosome 10.

Cochlear expression of candidate genes

We quantitatively measured the cochlear expression of the genes found at the QTL of interest in a selected number of strains, including parental strains (B6 and D2) and strains with divergent MHL values (BXD 11, 13, 32 and 73). Only a subset of these genes is expressed in the cochlea (Figs. 4 and S3; Tables 1 and S2) with considerable variation by strain (anova; Table 2). First, we determined whether the presence of either the B6 or the D2 allele would influence the relative expression of a gene. A significant contribution (P < 0.01) of the parental allele in BXD strains was observed for Acaa2, BC031181 and Smad7, and a trend (P = 0.011) for Ctif (Table 2). We next analyzed the cochlear gene expression for Spearman rank correlation with MHL at 4 kHz (Figs. 5 and S3). Only the expression of BC031181 showed significant correlation to the hearing loss trait (Table 2). However, several other genes approached significance (P < 0.1), including Dym, Ctif, Myo5b and Smad7. In the case of Dym, Ctif and Myo5b, MHL at 4 kHz was also significantly correlated with whole-brain gene expression. However, only in the case of Dym the correlation was positive, whereas for Ctif and Myo5b it was negative (Table S5). Gene expression of Otof, Chrna9 and Cdh23 was used as a control for hair cell quantity. Otof and Chrna9 showed variable expression between strains (Fig. 4), but the expression of these two genes and of Cdh23 was not significantly correlated with MHL (Table 2).

image

Figure 4. Gene expression of genes present within the 1-LOD interval of ahl9 and controls. Relative gene expression levels for genes at the significant ahl9 locus at chromosome 18 are shown. The bars represent different BXD strains in the sequence (from left to right): BXD13, BXD11, B6, BXD73, BXD32 and D2. Shading indicates allele inheritance from either B6 (dark gray) or D2 (white gray). Significant differences (anova) between strains (hatched line) or for the presence of the B6 or D2 allele (black lines; values without parental strains) are indicated (see Table 2). Cadherin 23 (Cdh23; chromosome 10), which is expressed by all BXD strains and is located outside of the suggestive QTL, was used as a PCR control. Otof and Chrna9 (chromosome 5) were used as controls for hair cell quantity. *P < 0.01; **P < 0.001; ***P < 0.0001.

Download figure to PowerPoint

image

Figure 5. Low-frequency hearing loss correlations with gene expression. Spearman rank correlations (r) between MHL at 4 kHz and genes at the ahl9 locus that are significant (P < 0.05; BC031181) or show a trend [P < 0.1; Myo5b, Dym, Smad7 and Ctif (Gm762); see Table 2]. The parental origin of the allele is indicated for each sample.

Download figure to PowerPoint

SNPs in intronic sequences could contribute to altered promotor function, and hence altered levels of gene expression. Therefore, intronic SNPs not present in normal hearing strains at the chromosome 18 locus were further selected for having SNPs in the first exon. These were present in Myo5b (16x), Acaa2 (2x), BC031181 (3x), Dym (1x) and Ctif (6x).

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

In this study on early-onset hearing loss in BXD recombinant inbred strains, we found a new significant QTL on chromosome 18, designated ahl9. Using expression studies we showed that the majority of genes on this QTL are expressed within the cochlea, and that some of these genes were differentially expressed depending on the parental strain the allele was derived from, and/or were correlated with MHL at 4 kHz.

Validity of the approach

We confirmed the significant, high-frequency QTL on chromosome 11, called ahl8 (Johnson et al. 2008), for which a mutation in fascin-2 is responsible (Shin et al. 2010). Moreover, our data showed high correlations with previous data on spiral ganglion cell density in BXD strains (Willott & Erway 1998). The inclusion of 4-kHz tones in our study largely explains why previous studies did not find the QTL on chromosome 18. The fascin-2 locus explains only a modest fraction of variance in the MHL at 4 kHz, and after accounting for these effects, ahl9 on chromosome 18 explained about 40% of the remaining variance. Additional differences are that our study comprises a larger number of BXD strains, including several advanced intercrosses, and incorporates recordings at earlier ages than were used in detecting ahl8 (Johnson et al. 2008).

Expression data

Fourteen out of 25 genes tested were expressed in the organ of Corti. Many of the analyzed genes showed strain variability; however, only a few genes showed expression patterns related to the presence of the parental allele. Expression patterns in other cochlear structures relevant to hearing were not investigated. Notably, genes with a profile dependent on the allele of origin were all at the peak of the locus, but were interspersed with genes that showed expression independent of the parental allele. Most likely, this was not due to experimental error, as the parent-independently expressed genes showed high expression in our samples. From this, we may conclude that genes at this locus have a differentiated set of enhancers that regulate gene expression levels independently from each other. All parent-dependently expressed genes showed a high correlation with MHL at 4 kHz. The correlation of Otof expression with MHL was considerably lower than that of genes at the loci, and not significant. Moreover, only expression of Smad7 and Chrna9 correlated positively with that of Otof. Thus, despite the limited set of strains in this analysis and possible strain differences in hair cell numbers, as suggested by differences in Otof expression, our gene expression analysis indicated several candidate genes on ahl9.

Candidate genes and pathways involved in hearing loss

The novel QTLs on chromosome 18 and 10 contained only few genes. Several non-synonymous mutations are present in different genes on the loci. Only two of them were specific for D2, whereas most were present in several normal hearing strains. This procedure is a first crude analysis, which reduced the number of candidate SNPs at ahl9 substantially. However, this analysis does not take into account that predisposing alleles at other loci may be required for manifestation of the hearing loss effect. In addition, several of the exon SNPs were found to be synonymous or had mutations in the 5′ or 3′ UTR of the mRNA. All of these could be related to changing the stability of the mRNA and hence could lead to differences in measured expression levels. None of the genes on the new loci has previously been associated with hearing loss or had strong homology to known proteins involved in hearing loss, albeit that several members of the myosin family are involved in hearing loss (reviewed in Friedman et al. 1999; Petit & Richardson 2009). Myosin 5b is involved in receptor trafficking, especially in polarized epithelial cells (Mattila et al. 2012; Roland et al. 2011; Schuh 2011), suggesting that it may also play a role in the trafficking of proteins and vesicles in (polarized) cochlear hair cells. It is also expressed in spiral ganglion neurons (Lu et al. 2011). Mutations in myosin 5b cause microvillus inclusion disease, which is characterized by lack of microvilli on the surface of enterocytes and occurrence of intracellular vacuolar structures containing microvilli (Muller et al. 2008). However, decreased hearing has not been reported in these patients.

Several genes on the newly found significant and suggestive loci are involved in fatty acid and lipid metabolism and sterol regulation, including Acaa2 and Lipg on chromosome 18, and Zdhhc17 and Osbpl8 on chromosome 10. Caloric restriction can reduce the advancement of AHL (Seidman 2000). One of the proposed mechanisms lies in the accumulation of reactive oxygen species due to fatty acid metabolism (Wolf 2006), which may accumulate over time and cause auditory sensory cell damage, eventually resulting in apoptosis of auditory neurons and hair cells. In humans, NADPH-oxidase activity, as measured by the amount of plasma superoxide anion radicals, is correlated mainly with low-frequency hearing thresholds (Hwang et al. 2012). Reducing oxidative stress by using lipoic acid as a dietary supplement has been shown to reduce hearing loss in D2 mice in an age-related manner (Ahn et al. 2008). Similarly, oral treatment of pravastatin, an inhibitor of the rate-limiting enzyme of cholesterol synthesis, which reduced serum cholesterol levels, protected against noise-induced cochlear injury (Park et al. 2012). Dietary restriction can protect against hearing loss via antioxidant proteins named sirtuins, more specifically Sirt3 in the mouse cochlea (Someya et al. 2010).

Our genetical genomics approach in the BXD strains suggests the involvement of a selected set of candidate genes in low-frequency, early-onset hearing loss. The list of most likely candidates comprises only six genes that showed (correlated) expression differences, and in case of ahl9, a limited set of possible causative SNPs: Myo5b, Acaa2, BC031181, Dym, Smad7 and Ctif, although it should be noted that the low expression of several genes in the organs of Corti might have hampered this analysis. Future studies that identify the causative genes at these new QTLs will help to understand the mechanisms underlying low-frequency, early-onset hearing loss.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

We would like to thank Maarten Loos for his help with calculating heritability and primer design; Rolinka van der Loo and Iris van Zutphen for coordinating animal deliveries. Madi Salehi and Annie Offenberg performed part of the ABR measurements. This work was supported by a Neuro-Bsik grant (BSIK 03053; SenterNovem, The Netherlands) and the Heinsius Houbolt fund. The authors declare no conflict of interest.

Appendix

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information

The Neuro-Bsik Mouse Phenomics Consortium is composed of the laboratories of A.B. Brussaard, J.G.G. Borst, Y. Elgersma, N. Galjart, G.T.J. van der Horst, C.N. Levelt, C.M. Pennartz, A.B. Smit, B.M. Spruijt, M. Verhage and C.I. de Zeeuw and companies Noldus Information Technologies B.V. (http://www.noldus.com/) and Synaptologics B.V. (http://www.synaptologics.com/).

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Appendix
  9. Supporting Information
FilenameFormatSizeDescription
gbb845-sup-0001-AppendixS1.docxWord 2007 document10K Data S1: Additional analysis using absolute hearing levels instead of MHL.
gbb845-sup-0002-FigureS1.docWord document834K Figure S1: Average ABR thresholds as a function of time and sound frequency for all tested strains. For some strains, no data are available at either 2 or 3 weeks of age, because data collection was more sparse at these early ages. Additional information about the ABR measurements in the different strains is provided in Table S1.
gbb845-sup-0003-FigureS2.docWord document37K Figure S2: Highly suggestive locus at chromosome 10. (a) Suggestive QTL at chromosome 10, highest peak at 4 kHz (LRS = 17.1; additive allele effect = 6.8). Frequencies 8 and 16 kHz show peaks with decreasing magnitude (LRS = 16.3 and 14.0, respectively). Approximate suggestive (*; LRS = 10.7) and significant (**; LRS = 18.8) values are indicated by horizontal, dashed lines. Presence of the D2 allele increases trait values. This locus has a higher correlation with MHL values at 4 kHz (r = 0.62) than the known mutation in the fascin-2 gene (r = 0.44). Calculation of the partial correlation showed that the locus at chromosome 10 could explain 28.4% in the variation of the MHL at 4 kHz after accounting for the effects of fascin-2. (b) Zoomed-in view of the genes present at the locus on chromosome 10.
gbb845-sup-0004-FigureS3.docWord document567K Figure S3: Gene expression analysis for genes of the suggestive chromosome 10 locus. (a) Relative gene expression levels are shown. The bars represent different BXD strains in the sequence (from left to right): BXD13, BXD11, B6, BXD73, BXD32 and D2. Shading indicates allele inheritance from either B6 (dark gray) or D2 (light gray). Significant differences (anova) between strains (hatched line) or for the presence of the B6 or D2 allele (black lines; values without parental strains) are indicated (see Table S3). *P < 0.01; **P < 0.001; ***P < 0.0001. (b) Spearman rank correlations (r) of MHL at 4 kHz with gene expression of Zdhhc17 and Osbpl8 (P < 0.1; see Table S3). The parental origin of the allele is indicated for each sample.
gbb845-sup-0005-tS1.xlsExcel spreadsheet44K Table S1: Extended overview of ABR threshold data. The number of mice used in our study, including distribution of sexes, is given for each strain. Second, the average best hearing threshold and hearing threshold at the endpoint (12 weeks) are provided per frequency. The difference between these two gives the MHL used in our analysis.
gbb845-sup-0006-tS2.xlsExcel spreadsheet104K Table S2: Analysis of SNPs within the significant and suggestive QTLs on chromosome 18 and 10, respectively. Genes on ahl9 (chromosome 18) and chromosome 10 were analyzed for SNPs in exons (left, green) and introns (right, blue). Indicated is the location (Mb) of the SNP, the sequence of the SNP (D2) and the reference population (B6), the type of SNP (SC, synonymous coding; NSC, non-synonymous coding; 5′ UTR, exon: 5′ UTR; 3′ UTR, exon: 3′ UTR; NA, sequence not annotated in Genome browser; Nonsplice Site, intron not at a splice site) and whether the SNP was also present in the normal hearing strain, CBA/J, DBA/1J and DBA/2HaSmnJ. Note that for chromosome 10 for all genes, except Nav3, the sequence of B6 and good hearing strains was mostly similar and therefore only the total number of SNPs is indicated.
gbb845-sup-0007-tS3.docWord document55K Table S3: Gene expression analysis of genes at the suggestive QTL on chromosome 10. Displayed are genes (see Table 1) used in the qPCR analysis across six strains with divergent hearing loss, their gene symbol, the average expression level, the anova results (α = 0.001) across strains and the presence of the B6 or the D2 allele (*without parental lines; α = 0.01), the mean of ranks when strain averages were used for the analysis of having the B6 or D2 allele, the Spearman rank and Pearson product–moment correlation (and P-value) with MHL at 4 kHz and the LRS score of this trait. XH, Relative expression level (log 2) across strains (Expression) ≥ 10; H, 10 > Expression ≥ 8; M, 8 > Expression ≥ 5; L, 5 > Expression. Significant (bold; anova, P < 0.001; correlation, P < 0.05) and trends for (italics; correlation, 0.05 < P < 0.1) differences are highlighted. Note that no significant results were obtained in any analysis for Cdh23, a gene on chromosome 10 outside the suggestive QTL for MHL.
gbb845-sup-0008-tS4.docWord document62K Table S4: Primer sequences used for expression analysis of genes at ahl9 (chromosome 18) and chromosome 10. Genes outside the 1-LOD interval are shown in gray.
gbb845-sup-0009-tS5.xlsExcel spreadsheet151K Table S5: Genome-wide Pearson correlation of MHL at 4 kHz with total brain gene expression. Using GeneNetwork.org, the dataset of MHL at 4 kHz was uploaded and correlated using a Pearson product–moment correlation with expression of two total brain databases: UCHSC BXD Whole Brain M430 2.0 (Nov06) RMA and INIA Brain mRNA M430 (Jun06) RMA. Genes present at ahl9 are given in bold and dark blue, those nearby in light blue. Note that only the correlation of Dym is significant and in the corresponding direction as in the cochlear samples (see Fig. 5).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.