Skip to main content
NCBI home page
Human Reproduction (Oxford, England) logoLink to Human Reproduction (Oxford, England)
. 2015 Jul 24;30(10):2439–2451. doi: 10.1093/humrep/dev180

Pathway analysis supports association of nonsyndromic cryptorchidism with genetic loci linked to cytoskeleton-dependent functions

Julia Spencer Barthold 1,2,*, Yanping Wang 1, Thomas F Kolon 3, Claude Kollin 4, Agneta Nordenskjöld 4, Alicia Olivant Fisher 1, T Ernesto Figueroa 2, Ahmad H BaniHani 2, Jennifer A Hagerty 2, Ricardo Gonzaléz 2,10, Paul H Noh 2,11, Rosetta M Chiavacci 5, Kisha R Harden 5, Debra J Abrams 5, Cecilia E Kim 5, Jin Li 5, Hakon Hakonarson 5,6,7, Marcella Devoto 6,7,8,9
PMCID: PMC4573451  PMID: 26209787

Abstract

STUDY QUESTION

What are the genetic loci that increase susceptibility to nonsyndromic cryptorchidism, or undescended testis?

SUMMARY ANSWER

A genome-wide association study (GWAS) suggests that susceptibility to cryptorchidism is heterogeneous, with a subset of suggestive signals linked to cytoskeleton-dependent functions and syndromic forms of the disease.

WHAT IS KNOWN ALREADY

Population studies suggest moderate genetic risk of cryptorchidism and possible maternal and environmental contributions to risk. Previous candidate gene analyses have failed to identify a major associated locus, although variants in insulin-like 3 (INSL3), relaxin/insulin-like family peptide receptor 2 (RXFP2) and other hormonal pathway genes may increase risk in a small percentage of patients.

STUDY DESIGN, SIZE, DURATION

This is a case–control GWAS of 844 boys with nonsyndromic cryptorchidism and 2718 control subjects without syndromes or genital anomalies, all of European ancestry.

PARTICIPANTS/MATERIALS, SETTING, METHODS

All boys with cryptorchidism were diagnosed and treated by a pediatric specialist. In the discovery phase, DNA was extracted from tissue or blood samples and genotyping performed using the Illumina HumanHap550 and Human610-Quad (Group 1) or OmniExpress (Group 2) platform. We imputed genotypes genome-wide, and combined single marker association results in meta-analyses for all cases and for secondary subphenotype analyses based on testis position, laterality and age, and defined genome-wide significance as P = 7 × 10−9 to correct for multiple testing. Selected markers were genotyped in an independent replication group of European cases (n = 298) and controls (n = 324). We used several bioinformatics tools to analyze top (P < 10−5) and suggestive (P < 10−3) signals for significant enrichment of signaling pathways, cellular functions and custom gene lists after multiple testing correction.

MAIN RESULTS AND THE ROLE OF CHANCE

In the full analysis, we identified 20 top loci, none reaching genome-wide significance, but one passing this threshold in a subphenotype analysis of proximal testis position (rs55867206, near SH3PXD2B, odds ratio = 2.2 (95% confidence interval 1.7, 2.9), P = 2 × 10−9). An additional 127 top loci emerged in at least one secondary analysis, particularly of more severe phenotypes. Cytoskeleton-dependent molecular and cellular functions were prevalent in pathway analysis of suggestive signals, and may implicate loci encoding cytoskeletal proteins that participate in androgen receptor signaling. Genes linked to human syndromic cryptorchidism, including hypogonadotropic hypogonadism, and to hormone-responsive and/or differentially expressed genes in normal and cryptorchid rat gubernaculum, were also significantly overrepresented. No tested marker showed significant replication in an independent population. The results suggest heterogeneous, multilocus and potentially multifactorial susceptibility to nonsyndromic cryptorchidism.

LIMITATIONS, REASONS FOR CAUTION

The present study failed to identify genome-wide significant markers associated with cryptorchidism that could be replicated in an independent population, so further studies are required to define true positive signals among suggestive loci.

WIDER IMPLICATIONS OF THE FINDINGS

As the only GWAS to date of nonsyndromic cryptorchidism, these data will provide a basis for future efforts to understand genetic susceptibility to this common reproductive anomaly and the potential for additive risk from environmental exposures.

STUDY FUNDING/COMPETING INTERESTS

This work was supported by R01HD060769 (the Eunice Kennedy Shriver National Institute for Child Health and Human Development (NICHD)), P20RR20173 (the National Center for Research Resources (NCRR), currently P20GM103464 from the National Institute of General Medical Sciences (NIGMS)), an Institute Development Fund to the Center for Applied Genomics at The Children's Hospital of Philadelphia, and Nemours Biomedical Research. The authors have no competing interests to declare.

Keywords: cryptorchidism, genetics, cytoskeleton, testis, pathways

Introduction

Cryptorchidism, or undescended testis (UDT), affects up to 4% of boys during childhood and is associated with other reproductive defects including epididymal anomalies, early germ cell depletion and testicular germ cell tumor (TGCT) (Kollin and Ritzen, 2014). Although the etiology remains unknown, the available data suggest a complex interplay of moderate genetic risk and maternal and/or environmental factors (Schnack et al., 2008; Jensen et al., 2010). Testicular hormones (insulin-like 3 (INSL3) and androgen) and their receptors, relaxin/insulin-like family peptide receptor 2 (RXFP2) and androgen receptor (AR), are required for growth and remodeling of the fetal gubernaculum-cremaster muscle complex, which in turn is essential for successful testicular descent (Kaftanovskaya et al., 2011, 2012). However, variants in INSL3 and RXFP2 are associated with only 2–3% of nonsyndromic cryptorchidism cases, and exhibit a range of phenotypes including unilateral and spontaneously resolving presentations, suggesting significant genetic heterogeneity (Ferlin et al., 2008; Foresta et al., 2008). Likewise, mutational analyses of other potential candidates including homeobox A10 (HOXA10), homeobox A11 (HOXA11), estrogen receptor 1 (ESR1), steroidogenic factor 1 (SF1) or AR yielded inconsistent, population-specific results (Kolon et al., 1999; Bertini et al., 2004; Yoshida et al., 2005; Galan et al., 2007; Wang et al., 2007, 2008).

Similarly, unbiased analyses have provided limited insight into cryptorchidism etiology. In a single prior genome-wide association study (GWAS) of the testicular dysgenesis syndrome (TDS; comprising related reproductive phenotypes including hypospadias, subfertility, cryptorchidism and/or TGCT), signals in the HOXD cluster linked to both TDS and the cryptorchidism subphenotype were not replicated, while a weaker signal in transforming growth factor beta receptor 3 (TGFBR3) showed only suggestive association in discovery (P = 2.4 × 10−4) and replication (P = 0.05) analyses (Dalgaard et al., 2012), which we also observed in an initial analysis of our data (Barthold et al., 2015). In contrast, separate European-based genome-wide analyses of hypospadias, a closely related reproductive anomaly, showed convincing association of hypospadias with multiple developmental loci (Geller et al., 2014; van der Zanden et al., 2011). These data also provided evidence for genetic heterogeneity dependent upon the severity of the hypospadias phenotype. A genome-wide analysis of copy number variation (CNV) using array comparative genomic hybridization found rare cases of vesicle-associated membrane protein 7 (VAMP7) duplication and orthodenticle homolog 1 (OTX1) deletion in individuals with cryptorchidism as the primary reproductive trait (Jorgez et al., 2014; Tannour-Louet et al., 2014).

Familial evidence of moderate genetic risk exceeding that for many complex diseases (Hemminki et al., 2008) suggests that genome-wide analysis of common genomic variants might help to define the pattern of genetic susceptibility to this disease. We hypothesized that a larger, more homogeneous case population would provide increased power to detect genome-wide significant loci associated with nonsyndromic cryptorchidism, and that these loci would contain gene candidates that influence developmental patterning of the gubernaculum by the fetal testis. In contrast to genome-wide analyses of hypospadias, the present data do not support the existence of multiple strong associations, but rather suggest weaker effects of multiple loci, with overrepresentation of genes linked to cytoskeleton-related pathways, syndromic cryptorchidism and hypogonadotropic hypogonadism (HH). In addition, we observed a stronger association of unique loci in subphenotype analyses, suggesting complex, multilocus genetic susceptibility.

Materials and Methods

Subjects and genotyping

Cases were subjects with cryptorchidism who underwent surgical correction at Nemours/Alfred I. DuPont Hospital for Children (Nemours) or The Children's Hospital of Philadelphia (CHOP). Exclusion criteria included multiple congenital anomalies and/or diagnosis of a syndrome; other genital anomalies (hypospadias, chordee or other penile anomalies); abdominal wall defects or syndromic urogenital malformation sequences. Control genotypes were obtained from males >6 years of age with no known history of testicular disease, penile anomaly, diagnosis of a syndrome or any additional medical disorder associated with cryptorchidism.

Data including age of diagnosis, race, ethnicity, laterality and the position of affected testes were collected and stored in a Research Electronic Data Capture (REDCap) database. Blood samples or excess tissue from UDT cases were collected and stored at −80°C or in RNAlater (Qiagen). Subphenotypes were categorized based on prior reports (Barthold et al., 2012, 2015). UDT were defined as distal if both testes were situated beyond the external inguinal ring and proximal if at least one testis was positioned within the inguinal canal or abdomen. To segregate individuals with the congenital and acquired forms of the disease, we defined early and late diagnoses as surgery performed at ≤2 and >2 years of age, respectively. Informed consent was obtained for all participants and studies were approved by the relevant Institutional Review Board.

We extracted DNA (5 PRIME, Gaithersburg, USA), and performed whole genome amplification (REPLI-g Mini Kit, Qiagen) in cases of low DNA yield. Samples of adequate purity, defined as OD 260/280 of 1.8–2.0 based on Nanodrop1000 spectrophotometer analysis, were entered into the standard genotyping workflow at the Center of Applied Genomics at CHOP. In the discovery phase, we genotyped two sets of cases to match available control genotype data. In Group 1, 559 cases and 1772 controls were genotyped using the Illumina HumanHap550 v1, HumanHap550 v3 or Human610-Quad v1, platforms that have over 535K single nucleotide polymorphisms (SNPs) in common. In Group 2, 353 cases and 1149 controls were genotyped using the Illumina Human OmniExpress 12v1 or 12v1.1, platforms that share over 719K SNPs.

Discovery phase data analysis

We used PLINK (v1.07; http://pngu.mgh.harvard.edu/purcell/plink/) (Purcell et al., 2007; Anderson et al., 2010) to analyze data from Groups 1 and 2 separately as previously described (Barthold et al., 2015). Exclusion criteria for individuals included: (i) discordance between reported sex and sample X and Y chromosome SNP data; (ii) missing genotype rate >3%; (iii) higher or lower than expected heterozygosity rate (greater than ±3 SD from the mean) and (iv) duplicates or relatives (based on estimate of proportion of alleles shared identical by descent >0.1875). Exclusion criteria for SNPs included: (i) missing genotype rate >5%; (ii) Hardy–Weinberg equilibrium (HWE) deviation in controls (P < 0.00001); (iii) significantly different missing genotype rates between cases and controls (P < 0.00001); and (iv) low minor allele frequency (MAF < 0.01). We controlled for European population substructure using multidimensional scaling (MDS) analysis in PLINK, employing data from the Stanford Human Genome Diversity Project (HGDP, http://www.hagsc.org/hgdp/files.html (Rosenberg, 2006)) and removed all samples that deviated from the means of the first or second MDS components by >3 SDs. Separately for Groups 1 and 2, we converted genotype data to the IMPUTE2 (version 2.3.0, https://mathgen.stats.ox.ac.uk/impute/impute_v2.html) file format using GTOOL (version 0.7.5, http://www.well.ox.ac.uk/~cfreeman/software/gwas/gtool.html) and performed imputation using the 1000 Genome reference population (September 2013) (Howie et al., 2012). We performed association analysis of imputed data using SNPTEST (version 2.5β, https://mathgen.stats.ox.ac.uk/genetics_software/snptest/snptest.html) with logistic regression using MDS components 1 and 2 as covariates (Howie et al., 2009, 2011). Imputed SNPs with MAF <0.01, control HWE P < 0.00001 and imputation quality score <0.8 were removed from the analysis. We combined the results from Groups 1 and 2 using META (version 1.5, http://www.stats.ox.ac.uk/~jsliu/meta.html) for the full analysis (all cases and controls) and for subphenotype analyses based on testicular position, laterality and age at surgery. Markers with P-values >0.05 for either of the separate Group 1 or Group 2 analyses were excluded from further consideration. Based on the total number of full and subphenotype analyses (n = 7) we used a P-value of 7 × 10−9 as a cutoff for genome-wide significance, and we defined top and suggestive signals as those with single marker P-values of <10−5 and <10−3, respectively. For each top marker, we identified the corresponding topological domain, based on experiments using human embryonic stem cells (Dixon et al., 2012), and report the closest protein coding gene(s) as well as all genes within the topological domain based on annotations in the UCSC Genome Browser (http://genome.ucsc.edu/). In addition, we used the GWAS3D tool (http://jjwanglab.org/gwas3d/; Li et al., 2013) to filter all suggestive SNPs identified in full and/or subphenotype analyses for potential regulatory activity. These SNPs, or those in linkage disequilibrium (r2> 0.8) are mapped to genes or genomic regions. We did not restrict the analysis to a specific cell type, as none provides a relevant reproductive target tissue, but otherwise used the default settings of the tool.

Pathway analysis

Following an initial analysis that failed to identify major cryptorchidism-associated loci, we adopted a pathway analysis approach based on reasoning that markers showing suggestive, even nominal, association with a given phenotype may collectively provide insight into the functional basis of disease (Wang et al., 2010). We used several bioinformatics tools for analysis of association signals, including Ingenuity® Pathway Analysis (IPA; QIAGEN, Redwood City, CA; www.qiagen.com/ingenuity), Improved Gene Set Enrichment Analysis for Genome-Wide Association Studies (i-GSEA4GWAS; http://gsea4gwas.psych.ac.cn/; Zhang et al., 2010) and the Database for Annotation, Visualization and Integrated Discovery (DAVID, v6.7; http://david.abcc.ncifcrf.gov/home.jsp; Huang da et al., 2009a,b). We generated several lists for use in IPA pathway analyses: (i) the closest gene within 100 kb of top markers, (ii) suggestive SNPs in the full genome-wide analysis, (iii) suggestive SNPs in the full and/or subphenotype genome-wide analyses and (iv) GWAS3D-mapped genes containing regulatory signals. Either genes or SNPs can be used as input in IPA; when SNPs are used, the program generates a nonredundant list of genes containing the SNP(s) of interest, and scores the resultant gene list against hand-curated biological function categories and canonical pathways. For the DAVID bioinformatics tool, which identifies overrepresentation of Gene Ontology annotations and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways, we used the IPA-generated gene list defined by suggestive intragenic SNPs from full and/or subphenotype analyses. For i-GSEA4GWAS, we used all suggestive full and/or subphenotype SNPs and corresponding P-values (P< 10−3) as input with default settings, except we set mapping parameters at 100 kb upstream and downstream of genes and specified a minimum of 10 genes per gene set.

In addition, we created biologically relevant custom gene lists and scored these against top and suggestive signals in IPA. For cryptorchidism-associated genes, we manually reviewed Online Mendelian Inheritance in Man (OMIM; http://www.omim.org/) clinical synopses, Mouse Genome Informatics (MGI; http://www.informatics.jax.org/) phenotypes and PubMed references (http://www.ncbi.nlm.nih.gov/pubmed/) and excluded genes associated with HH in OMIM. This list mapped to 182 genes in IPA, and we used core analysis of this dataset to define reference cryptorchidism-associated annotations. For other custom lists we used fetal rat gubernaculum gene expression data, including androgen- and INSL3-responsive and differentially expressed genes (in the cryptorchid LE/orl versus wildtype rat strain) identified in prior studies (Barthold et al., 2008, 2013, 2014; Johnson et al., 2010). We also compiled lists of HH-associated genes (OMIM), actin-binding and calponin homology (CH) domain proteins, which include a class of AR coactivators (Ting and Chang, 2008; UniProtKB, http://www.uniprot.org/) and genes encoding known AR-interacting proteins or coregulators (http://androgendb.mcgill.ca/ARinteract.pdf and (Jasavala et al., 2007)).

For all pathway analyses, we report only higher confidence results as defined by P-values of <0.05 after correction for multiple testing using the methods available for each tool: Benjamini–Hochberg (B-H) for IPA and DAVID, and false discovery rate (FDR) for i-GSEA4GWAS.

Replication

We used DNA samples from 298 boys with cryptorchidism and 324 unaffected controls collected in a prior study of cryptorchidism in Sweden (Kollin et al., 2006, 2007, 2012) for replication. We analyzed selected discovery phase candidate loci with associated P-values <10−4 in the full and/or subphenotype analysis based on location within or near (<100 kb) a functionally relevant gene, and the availability of a validated TaqMan® SNP genotyping assay for the given target. Genotyping was performed in a 384-well format using the ABI 7900 Real Time PCR instrument as previously described (Barthold et al., 2015). Genotyped or imputed SNP data from the discovery phase, and replication and combined data were analyzed using an online association analysis tool (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl).

Results

Quality control

As we reported previously (Barthold et al., 2015), 56 cases, 158 controls and 28 749 of 535 752 markers were removed in Group 1, and 12 cases, 45 controls and 89 489 of 719 629 markers were removed in Group 2, leaving 844 cases and 2718 controls in the final analysis. Using Quanto v.1.2.4 (http://biostats.usc.edu/Quanto.html), we estimated that at type I error rate of 7 × 10−9 (corresponding to the standard genome-wide significance threshold of 5 × 10−8 after correction for seven genome-wide analyses), we had >80% power to detect odds ratios (OR) of 1.5 or more for common risk alleles (frequency range 0.3–0.6). We identified XXY individuals in seven cases (0.8%), similar to the 1.3% incidence identified in another series (Ferlin et al., 2008), and these samples were excluded from analysis. Quantile–quantile plots of the distribution of P-values from SNPTEST analysis in Groups 1 and 2 (Supplementary Fig. S1) showed an expected distribution, and genomic inflation factor (λ) values of 1.014 and 1.017, respectively.

Single marker analysis

In the meta-analysis of imputed case and control data we found one or more top signals (P< 10−5) within 20 loci, 13 within 100 kb of the closest gene (Supplementary Table SI). Strongest signals in the full analysis included rs62443778 (OR 1.8, 95% confidence interval (CI) 1.4, 2.3; P = 4.5 × 10−7), an intronic SNP in phosphodiesterase 10A (PDE10A), a gene expressed in fetal gubernaculum that is highly responsive to INSL3 stimulation (Johnson et al., 2010), rs117605123 near zinc finger and AT hook domain containing (ZFAT) (OR 3.3, 95% CI 2.1, 5.2; P = 4.8 × 10−7) and rs56329627 near SH3 and PX domains 2B (SH3PXD2B) (OR 1.6, 95% CI 1.3, 2.0; P = 8.7 × 10−7). Another marker near SH3PXD2B, rs55867206, showed stronger association in the proximal subphenotype analysis, reaching genome-wide significance (OR = 2.2, 95% CI 1.7, 2.9; P = 1.9 × 10−9) even after correction for the total number of full and subphenotype genome-wide analyses we performed (7; P < 7 × 10−9). SH3PXD2B encodes tyrosine kinase substrate with four SH3 domains (TKS4), a protein required for normal embryonic development that regulates formation of actin-rich membrane protrusions, extracellular matrix degradation and cellular motility (Buschman et al., 2009; Lanyi et al., 2011; Bogel et al., 2012). In addition, SH3PXD2B mutations cause Frank-ter-Haar syndrome, which can include cryptorchidism (Maas et al., 2004; Haznedaroglu et al., 2014).

Top signals in the subphenotype analyses included 13 at the same loci identified in the full analysis (boxed cells in Supplementary Table SII), and additional markers at 127 unique loci; a total of 102 mapped to within 100 kb of at least one gene. Of the latter, the strongest (P < 10−6) signals include those in death-associated protein kinase 2 (DAPK2), MAM domain containing glycosylphosphatidylinositol anchor 2 (MDGA2), guanine deaminase (GDA), arylsulfatase B (ARSB), NUAK family, SNF1-like kinase, 1 (NUAK1), phosphatidic acid phosphatase type 2 domain containing 1A (PPAPDC1A), copine IV (CPNE4), G protein-coupled receptor 83 (GPR83), epithelial membrane protein 1 (EMP1), v-ets avian erythroblastosis virus E26 oncogene homolog (ERG), thiamin pyrophosphokinase 1 (TPK1), dermatopontin (DPT), protein tyrosine phosphatase, receptor type, D (PTPRD), and dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1A (DYRK1A). Additional top signals were most commonly associated with bilateral, proximal and early diagnosis subphenotypes and the majority were unique to a single analysis (Supplementary Table SII). Of note, an associated top signal in unc13 homolog A (UNC13A) occurs within a topological domain that also contains INSL3.

Pathway analysis

To provide initial insight into pathways and biological functions linked to cryptorchidism genes, and as a point of reference for our GWAS data, we used IPA to functionally annotate genes involved in the pathogenesis of clinical syndromes and animal models associated with cryptorchidism (Supplementary Table SIIIa–d). Enriched canonical pathways in this analysis include many involved in development, including Wnt/β-catenin, transforming growth factor beta (TGF-β), bone morphogenetic protein (BMP) and fibroblast growth factor (FGF) signaling, and retinoic acid receptor (RAR) activation. We identified Axon Guidance and Protein Kinase A Signaling as significantly overrepresented canonical pathways common to the IPA analyses of both GWAS signals and established syndromic/animal model cryptorchidism-associated genes (Table I and Supplementary Table SIV). Several developmental and/or related molecular and cellular functions including cellular assembly and organization, cellular function and maintenance, cell morphology, cellular movement, cellular development and cell death and survival were most consistently and significantly enriched in the GWAS signal datasets that we analyzed in IPA, categories that were also strongly overrepresented in the analysis of established cryptorchidism genes (Table I and Supplementary Table SV). Among the top (P < 10−5) signals identified in full and/or subphenotype analyses, 30–38% of the closest genes carry annotations in one or more of these categories. We observed similar results in our analyses of suggestive full and/or subphenotype signals (P < 10−3) using other pathway analysis tools: i-GSEA4GWAS identified enrichment of functional annotations related to actin cytoskeleton (P < 0.001) and cytoplasmic trafficking (P≤ 0.02) (Supplementary Table SVI) and DAVID identified cell adhesion, neuronal development, cell motion, axon guidance and GTPase signaling categories (all P < 0.01; Supplementary Table SVII). These convergent results provide corroborative evidence that suggestive cryptorchidism-associated loci are enriched for cellular processes that are dependent on cytoskeletal organization and function.

Table I.

Ingenuity® Pathway Analysis (IPA) pathway analysis of top and suggestive signals in a genome-wide association study (GWAS) of nonsyndromic cryptorchidism.

Reference cryptorchidism-associated genesa
(n = 182)
 Full analysis
 Full and/or subphenotype analysis
P < 10−3
 P < 10−5
 P < 10−3
IPA-mapped intragenic markers
(n = 492)
 Closest gene <100 kb
(n = 107)
 IPA-mapped intragenic markers
(n = 1960)
 GWAS3D regulatory markers mapped to genes
(n = 1603)
No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb
Canonical signaling pathways
 Axonal guidance 15 1.3 × 10−3 – – – – 61 1.3 × 10−3 61 5.9 × 10−6
 Protein kinase A 14 5.2 × 10−4 – – – – 60 1.7 × 10−4 58 7.5 × 10−4
Molecular and cellular functions
 Cellular assembly and organization 52 1.4 × 10−8 94 1.4 × 10−4 32 2.4 × 10−3 321 9.4 × 10−9 311 2.6 × 10−10
 Cellular function and maintenance 60 1.4 × 10−8 91 5.4 × 10−4 38 2.4 × 10−3 304 9.4 × 10−9 391 2.6 × 10−10
 Cellular movement 54 4.8 × 10−7 50 2.5 × 10−3 10 0.049 306 9.1 × 10−10 305 1.3 × 10−9
 Cell morphology 74 9.6 × 10−11 11 2.0 × 10−4 36 3.2 × 10−3 397 6.4 × 10−8 387 1.5 × 10−9
 Cellular development 108 2.0 × 10−15 102 1.4 × 10−4 39 0.017 458 1.1 × 10−8 448 1.9 × 10−8
 Cellular growth and proliferation 112 6.1 × 10−17 138 1.4 × 10−4 44 0.049 477 9.3 × 10−5 460 2.8 × 10−5
 Cell death and survival 93 2.8 × 10−14 55 0.019 41 0.019 453 7.9 × 10−4 437 2.3 × 10−3
 DNA replication, recombination and repair 21 2.7 × 10−5 8 0.017 1 0.049 13 1.7 × 10−4 13 1.0 × 10−4
 Cell cycle 50 2.4 × 10−10 8 0.014 6 0.049 50 0.040 12 0.028
Most commonly represented functional annotations
 Organization of cytoskeleton 45 1.8 × 10−8 65 8.0 × 10−4 24 3.0 × 10−3 225 9.4 × 10−9 219 2.6 × 10−10
 Microtubule dynamics 41 1.4 × 10−8 58 5.4 × 10−4 22 2.4 × 10−3 196 1.4 × 10−8 191 5.3 × 10−10
 Organization of cytoplasm 47 3.2 × 10−8 66 5.3 × 10−3 24 3.0 × 10−3 236 1.9 × 10−7 230 5.2 × 10−9
 Formation of cellular protrusions 31 1.3 × 10−6 45 2.3 × 10−3 18 3.2 × 10−3 152 1.7 × 10−7 148 1.5 × 10−8
Open in a new tab

aGenes associated with syndromes containing cryptorchidism in the clinical synopsis (OMIM: Online Mendelian Inheritance in Man) excluding those associated with hypogonadotropic hypogonadism (HH).

bBenjamini–Hochberg corrected P-values, functional annotation results are uncorrected P-values.

Several suggestive loci, including actinin 2 (ACTN2), supervillin (SVIL) and vav 3 guanine nucleotide exchange factor (VAV3), encode cytoskeleton-associated proteins that participate in AR signaling (Huang et al., 2004; Ting et al., 2004; Lyons and Burnstein, 2006), a pathway that is critical for gubernacular development and testicular descent (Kaftanovskaya et al., 2012). Actinin 2 (ACTN2) and actinin 4 (ACTN4) interact with AR and/or other nuclear receptors via conserved N-terminal LXXLL motifs (nuclear receptor (NR) boxes) located within actin-binding calponin homology (CH) domains (Huang et al., 2004; Khurana et al., 2012). We tested for overrepresentation of actin-associated and/or CH domain proteins in our results by scoring suggestive signals against UniProtKB-generated custom gene lists, and observed significant enrichment of both categories (Table II). Of 16 CH-domain proteins containing suggestive signals, 14 contain LXXLL motifs that show homology to ACTN proteins, occur within CH domains and/or are classifiable as NR boxes (Bramlett and Burris, 2002) (Fig. 1). Alignment of the LXXLL domain of these proteins shows homology of flanking sequences with other proteins that are known AR coregulators (Gottlieb et al., 2012) or AR-interacting proteins (Jasavala et al., 2007) as well as with nuclear envelope spectrin repeat protein 2 (nesprin-2), encoded by spectrin repeat containing, nuclear envelope 2 (SYNE2), an actinin-like CH domain protein associated with cryptorchidism susceptibility in the LE/orl rat (unpublished observations). Overrepresentation of these loci suggests the possibility that the encoded cytoskeletal proteins may increase cryptorchidism susceptibility via their participation in AR signaling.

Table II.

IPA enrichment analysis of selected gene/protein categories in GWAS signals associated with nonsyndromic cryptorchidism.

No. in list mapped by IPA Data source(s) for custom list Reference cryptorchidism-associated genesa
(n = 182)
 Full analysis
 Full and/or subphenotype analysis
P< 10−3
 P < 10−5
 P < 10−3
IPA-mapped intragenic markers (n = 492)
 Closest gene <100 kb (n = 107)
 IPA-mapped intragenic markers (n = 1960)
 GWAS3D regulatory markers mapped to genes (n = 1603)
No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb No. of genes P-valueb
Custom lists
 Cryptorchidism-associated genesa 182 OMIM, MGI, PubMed excluded 12 1.7 × 10−3 – – 34 5.7 × 10−6 34 1.2 × 10−6
 Calponin homology (CH) domain proteins 74 UniProtKB – – 7 2.3 × 10−3 – – 16 2.3 × 10−4 17 3.0 × 10−5
 Actin-binding (not CH domain) proteins 300 UniProtKB – – – – – – 49 2.5 × 10−6 49 4.7 × 10−7
 Hypogonadotropic hypogonadism-associated genes 89 OMIM excluded – – – – 13 0.029 14 7.1 × 10−3
 Androgen receptor-interacting proteins or coactivators 640 Gottleib et al. (2012)
Jasavala et al. (2007) 		16 4.2 × 10−4 – – – – 52 – 52 –
Transcript expression in fetal rat gubernaculum
 Developmentally down-regulated 645 Barthold et al. (2014) 15 1.0 × 10−3 25 3.8 × 10−3 8 0.049 94 7.0 × 10−8 85 5.2 × 10−7
 Developmentally up-regulated 1008 Barthold et al. (2014) – – – – – – 89 – 95 0.012
 DHT-responsive 1657 Barthold et al. (2013) 35 5.7 × 10−6 54 1.7 × 10−3 19 5.1 × 10−3 186 2.5 × 10−6 180 4.7 × 10−7
 INSL3-responsive 2182 Johnson et al. (2010) 36 4.2 × 10−4 – – 24 3.7 × 10−3 230 8.4 × 10−6 215 1.5 × 10−5
 Differentially expressed in wt v LE/orl (cryptorchid) rat 3621 Barthold et al. (2014) 62 8.2 × 10−7 93 0.014 29 0.039 353 2.2 × 10−5 329 4.0 × 10−5
Open in a new tab

OMIM (http://www.omim.org/); MGI, Mouse Genome Informatics (http://www.informatics.jax.org/); UniProtKB, UniProt Knowledgebase (http://www.uniprot.org/); HH, hypogonadotropic hypogonadism; DHT, dihydrotestosterone; INSL3, insulin-like 3; −, not significant.

aGenes associated with human syndromes (excluding HH) or mouse models.

bBenjamini–Hochberg corrected P-value.

Figure 1.

Figure 1

Open in a new tab

Alignment of calponin homology (CH) domain proteins with LXXLL motifs of potential relevance to androgen receptor (AR) signaling and/or cryptorchidism susceptibility. Shown are proteins with one (1X) or two (2X) CH domains and LXXLL motifs (L = leucine, X = any amino acid (AA), yellow highlighting). Proteins with suggestive intragenic signals are highlighted in gray; AR coregulators and/or AR-interacting proteins in bold and blue font, respectively (Jasavala et al, 2007; Gottlieb et al., 2012); and a cryptorchid orl rat susceptibility locus shown in red font (Barthold et al, unpublished observations). *AA numbers indicate the positions of the LXXLL motif and the most N-terminal CH domain (CH1), with numbers in bold indicating that the LXXLL motif resides within this CH domain. #Nuclear receptor (NR) box classification from Bramlett and Burris (2002).

Scoring of other custom lists in IPA against suggestive signals showed significant enrichment of cryptorchidism- and HH-associated genes, and also of transcripts that are hormone responsive, cryptorchidism-associated, and developmentally regulated in the developing rat fetal gubernaculum (Tables II and III). Although suggestive markers were mapped to over 50 genes encoding AR coregulators or AR-interacting proteins, these were not significantly overrepresented in the IPA analysis (Table III). However, in pathway analysis limited to genes mapped by GWAS3D as containing potential regulatory variants, we found that the significance of our results was generally enhanced (Tables I and II), suggesting that these loci should be prioritized in future mapping studies of this complex trait.

Table III.

Genes in custom lists containing at least suggestive (P < 10−3) markers in a GWAS of nonsyndromic cryptorchidism.

Custom list Gene* No. Corrected P-value
Actin-binding proteins ABL1, ACTR2, ACTR3B, ADD2, ADD3, ALDOA, AVIL, CLEC9A, CTTN, CYFIP1, DAAM1, DAG1, DMTN, EPB41L3, FBXO25, FGD4, FHOD3, FMN1, FMN2, FXYD5, HIP1, INF2, INO80, KLHL3, KLHL20, MTSS1, MYH15, MYO3B, MYO7A, MYO10, MYO18A, NCALD, NEBL, NEXN, PACRG, PALLD, PARK2, PHACTR3, POF1B, SHROOM2, SHROOM3, SNTB2, SNTG1, SPIRE1, SVIL, SYNPO2, TNNI1, VCL, WASF1 49 2.5 × 10−6
Cryptorchidism genes—syndromes and mouse models ADAMTS16, ANKRD11, ARNT2, BMP5, CCBE1, DHCR7, DHODH, DIS3L2, DNAJC5, ERCC8, GATA4, GLI3, GPC3, GPC6, HPSE2, IGF1R, LBR, LRP2, MID1, NALCN, PDE4D, PIEZO2, PTDSS1, RARB, ROR2, SCG5, SKI, SRCAP, STS, TBC1D20, TGFB2, TP63, TUSC3, YWHAE 34 5.8 × 10−6
Calponin homology domain proteins ACTN1, ACTN2, ARHGEF7, CLMN, DMD, IQGAP2, LIMCH1, MAPRE1, MICAL2, NAV2, NAV3, PARVB, PARVG, VAV2, VAV3, UTRN 16 2.3 × 10−4
Hypogonadotropic hypogonadism genes ANK1, CEP290, CGNL1, GLI2, HS6ST1, INSR, PGM2, PPARGC1B, POMC, PYY, SEMA3A, TTC8, WDPCP 13 0.029
Androgen receptor-interacting proteins or coactivators ACTN2, ACTR2, ALDOA, BAG3, CASP1, CASP8, CCDC50, CCND3, CCNE1, CDK4, CDK6, CDK8, CTPS2, CTTN, DSP, DTNBP1, EEF2, EFCAB8, ERG, FUBP1, GATA3, GNG7, GOLGA5, HIP1, IDE, KDM4C, MAK, MAPK1, MYBBP1A, MYO18A, NCOR1, PDCD6IP, PIAS4, POU2F3, PZP, RBBP5, RFC3, RFC5, RNH1, SGTA, SMARCC1, SORBS3, SRCAP, SRGAP2, STAU1, SVIL, TACC3, TBC1D4, TES, TP53, VAV3, ZMIZ1 52 0.47
Open in a new tab

Bold: P ≤ 10−6, blue: P ≤ 10−5, gray: P ≤ 10−4.

ABL proto-oncogene 1 non-receptor tyrosine kinase (ABL1), actinin 1 (ACTN1), actinin 2 (ACTN2), ARP2 actin-related protein 2 homolog (ACTR2), ARP3 actin-related protein 3 homolog B (ACTR3B), ADAM metallopeptidase with thrombospondin type 1 motif, 16 (ADAMTS16), adducin 2 (ADD2), adducin 3 (ADD3), ankyrin 1 (ANK1), aldolase A (ALDOA), ankryrin repeat domain-containing 11 (ANKRD11), rho guanine nucleotide exchange factor 7 (ARHGEF7), aryl-hydrocarbon receptor nuclear translocator 2 (ARNT2), advillin (AVIL), BCL2-associated athanogene 3 (BAG3), bone morphogenetic protein 5 (BMP5), caspase 1 (CASP1), caspase 8 (CASP8), collagen and calcium binding EGF domains 1 (CCBE1), coiled-coil domain containing 50 (CCDC50), cyclin D3 (CCND3), cyclin E1 (CCNE1), cyclin-dependent kinase 4 (CDK4), cyclin-dependent kinase 6 (CDK6), cyclin-dependent kinase 8 (CDK8), centrosomal protein 290 kDa (CEP290), cingulin-like 1 (CGNL1), C-type lectin domain family 9, member A (CLEC9A), calmin (CLMN), CTP synthase 2 (CTPS2), cortactin (CTTN), cytoplasmic FMR1 interacting protein 1 (CYFIP1), dishevelled associated activator of morphogenesis 1 (DAAM1), dystroglycan 1 (DAG1), 7-dehydrocholesterol reductase (DHCR7), dihydroorotate dehydrogenase (quinone) (DHODH), DIS3 like 3′-5′ exoribonuclease 2 (DIS3L2), dystrophin (DMD), dematin actin binding protein (DMTN), DnaJ (Hsp40) homolog, subfamily C, member 5 (DNAJC5), desmoplakin (DSP), dystrobrevin binding protein 1 (DTNBP1), eukaryotic translation elongation factor 2 (EEF2), EF-hand calcium binding domain 8 (EFCAB8), erythrocyte membrane protein band 4.1-like 3 (EPB41L3), excision repair cross-complementation group 8 (ERCC8), v-ets avian erythroblastosis virus E26 oncogene homolog (ERG), F-box protein 25 (FBXO25), FYVE, RhoGEF and PH domain containing 4 (FGD4), formin homology 2 domain containing 3 (FHOD3), formin 1 (FMN1), formin 2 (FMN2), far upstream element (FUSE) binding protein 1 (FUBP1), FXYD domain containing ion transport regulator 5 (FXYD5), GATA binding protein 3 (GATA3), GATA binding protein 4 (GATA4), GLI-Kruppel family member GLI2 (GLI2), GLI-Kruppel family member GLI3 (GLI3), guanine nucleotide binding protein (G protein), gamma 7 (GNG7), golgin A5 (GOLGA5), glypican 3 (GPC3), glypican 6 (GPC6), huntingtin interacting protein 1 (HIP1), heparanase 2 (HPSE2), heparan sulfate 6-O-sulfotransferase 1 (HS6ST1), insulin-degrading enzyme (IDE), insulin-like growth factor 1 receptor (IGF1R), inverted formin, FH2 and WH2 domain containing (INF2), INO80 complex subunit (INO80), insulin receptor (INSR), IQ motif containing GTPase activating protein 2 (IQGAP2), lysine (K)-specific demethylase 4C (KDM4C), kelch-like family member 20 (KLHL20), kelch-like family member 3 (KLHL3), lamin B receptor (LBR), LIM and calponin homology domains 1 (LIMCH1), low density lipoprotein receptor-related protein 2 (LRP2), male germ cell-associated kinase (MAK), mitogen-activated protein kinase 1 (MAPK1), microtubule-associated protein, RP/EB family, member 1 (MAPRE1), microtubule associated monooxygenase, calponin and LIM domain containing 2 (MICAL2), midline 1 (MID1), metastasis suppressor 1 (MTSS1), MYB binding protein (P160) 1a (MYBBP1A), myosin, heavy chain 15 (MYH15), myosin X (MYO10), myosin 18A (MYO18A), myosin 3B (MYO3B), myosin 7A (MYO7A), sodium leak channel, non selective (NALCN), neuron navigator 2 (NAV2), neuron navigator 3 (NAV3), neurocalcin delta (NCALD), nuclear receptor corepressor 1 (NCOR1), nebulette (NEBL), nexilin (F actin binding protein) (NEXN), PARK2 co-regulated (PACRG), palladin, cytoskeletal associated protein (PALLD), parkin RBR E3 ubiquitin protein ligase (PARK2), parvin B (PARVB), parvin G (PARVG), programmed cell death 6 interacting protein (PDCD6IP), phosphodiesterase 4D (PDE4D), phosphoglucomutase 2 (PGM2), phosphatase and actin regulator 3 (PHACTR3), protein inhibitor of activated STAT, 4 (PIAS4), piezo-type mechanosensitive ion channel component 2 (PIEZO2), premature ovarian failure, 1B (POF1B), pro-opiomelanocortin-alpha (POMC), POU class 2 homeobox 3 (POU2F3), peroxisome proliferator-activated receptor gamma, coactivator 1 beta (PPARGC1B), phosphatidylserine synthase 1 (PTDSS1), peptide YY (PYY), pregnancy-zone protein (PZP), retinoic acid receptor, beta (RARB), retinoblastoma binding protein 5 (RBBP5), replication factor C (activator 1) 3, 38 kDa (RFC3), replication factor C (activator 1) 5, 36.5 kDa (RFC5), ribonuclease/angiogenin inhibitor 1 (RNH1), receptor tyrosine kinase-like orphan receptor 2 (ROR2), secretogranin V (SCG5), sema domain, immunoglobulin domain (Ig), short basic domain, secreted, (semaphorin) 3A (SEMA3A), small glutamine-rich tetratricopeptide repeat (TPR)-containing, alpha (SGTA), shroom family member 2 (SHROOM2), shroom family member 3 (SHROOM3), SKI proto-oncogene (SKI), SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily c, member 1 (SMARCC1), syntrophin, beta 2 (dystrophin-associated protein A1, 59 kDa, basic component 2) (SNTB2), syntrophin, gamma 1 (SNTG1), sorbin and SH3 domain containing 3 (SORBS3), spire-type actin nucleation factor 1 (SPIRE1), Snf2-related CREBBP activator protein (SRCAP), SLIT-ROBO Rho GTPase activating protein 2 (SRGAP2), staufen double-stranded RNA binding protein 1 (STAU1), steroid sulfatase (microsomal), isozyme S (STS), supervillin (SVIL), synaptopodin 2 (SYNPO2), transforming, acidic coiled-coil containing protein 3 (TACC3), TBC1 domain family, member 20 (TBC1D20), TBC1 domain family, member 4 (TBC1D4), testin LIM domain protein (TES), transforming growth factor, beta 2 (TGFB2), troponin I type 1 (skeletal, slow) (TNNI1), tumor protein p53 (TP53), tumor protein p63 (TP63), tetratricopeptide repeat domain 8 (TTC8), tumor suppressor candidate 3 (TUSC3), utrophin (UTRN), vav 2 guanine nucleotide exchange factor (VAV2), vav 3 guanine nucleotide exchange factor (VAV3), vinculin (VCL), WAS protein family, member 1 (WASF1), WD repeat containing planar cell polarity effector (WDPCP), tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, epsilon (YWHAE), zinc finger, MIZ-type containing 1 (ZMIZ1).

Replication

We chose top (PDE10A, neurofibromin 2 (NF2), FERM domain-containing 1 (FRMD1), paired box 3 (PAX3), FMN1, SH3PXD2B, ABL proto-oncogene 1, non-receptor tyrosine kinase (ABL1), DYRK1A and ERG) and suggestive (adenylate cyclase 10 (ACDY10), deleted in azoospermia-like (DAZL), microphthalmia-associated transcription factor (MITF), neuropilin 1 (NRP1), sidekick 1 (SDK1), SMAD family member 6 (SMAD6) and torsin 1 interacting protein 2 (TOR1AIP2)) signals for replication based on pathway analysis results, protein–protein interaction data and/or functional annotation (Supplementary Table SVIII). Power analysis using Quanto v.1.2.4 indicated that minimum detectable ORs in the replication cohort given the risk allele frequencies in the discovery cohort (varying from 0.07 to 0.91) for type I error of 0.05 and power of at least 80%, ranged from 1.4 to 1.9. Genotyping of case and control samples collected from Caucasian subjects in Sweden failed to show significant association of any intragenic markers with cryptorchidism. However, for rs73378978 in formin 1 (FMN1), we noted similar risk allele frequency in discovery and replication cases (0.14 for both) and controls (0.10 and 0.11, respectively), suggesting that we lacked sufficient power for replication. FMN1 encodes an actin-binding protein that participates in cellular developmental processes and can interact with filamin B (Dettenhofer et al., 2008; Hu et al., 2014), another AR-interacting actin-binding protein (Fig. 1).

Discussion

Using standard criteria to define genetic association, including correction for genome-wide testing and replication in an independent population, we failed to identify a major susceptibility locus for cryptorchidism. Our findings could reflect a highly heterogeneous, polygenic susceptibility to the disease, environmental causes, and/or a more prominent role for rare variants or CNVs in cryptorchidism etiology. Genetic heterogeneity or lack of power to detect small effect risk alleles could account for failure to replicate association of selected loci in a small, independent and more homogeneous population.

Current data in OMIM indicate that cryptorchidism is a documented feature of over 400 discrete clinical syndromes, and known syndromic genes were overrepresented in our list of suggestive signals, suggesting that at least some of these may contribute to the genetic heterogeneity of the nonsyndromic phenotype. Strikingly, a similarly powered GWAS of hypospadias, which occurs in 10% of boys with cryptorchidism, confirmed a previously validated association with a major locus, diacylglycerol kinase, kappa (DGKK; van der Zanden et al., 2011), and with multiple additional genome-wide significant signals replicated in an independent population (Geller et al., 2014). Differences in the developmental mechanisms of these genital anomalies may in part relate to the fact that closure of the urethral plate occurs in a narrow gestational interval of the first trimester in response to a surge in fetal serum testosterone. In contrast, the process of testicular descent is prolonged, may extend beyond birth, and is regulated by INSL3 and gonadotrophins in addition to androgens. The latter process may implicate a wider range of pathways and targets that contribute to susceptibility, including those involved in gonadotrophin signaling in the brain. The multiple suggestive signals we identified in genes linked to HH were not included in a previous genetic analysis of nonsyndromic cryptorchidism (Laitinen et al., 2011) and are relevant candidate loci for further study, particularly based on the hypothesis that some cryptorchid boys have ‘forme fruste’ HH (Hadziselimovic et al., 2004).

We observed novel and/or stronger signals in some of our subphenotype analyses, particularly for more severe cases (proximal, bilateral and/or early diagnosis), despite reduction in power due to smaller sample size. These observations are consistent with prior detailed analysis of our GWAS data at the TGFBR3 locus, which showed suggestive association of independent signals based on phenotypic severity (Barthold et al., 2015). To add further complexity, fetal exposure to anti-androgenic or estrogenic chemicals inhibits testicular descent and/or gubernacular development in animal models, raising the possibility that endocrine-disrupting chemical(s) (EDC) contribute to cryptorchidism, but supporting human data are difficult to collect and remain limited (Virtanen and Adamsson, 2012). However, independent studies showing reduced anogenital distance in boys with cryptorchidism (Jain and Singal, 2013; Thankamony et al., 2014) suggest the possibility that global inhibition of androgen secretion and/or action may play a role in pathogenesis. The multifactorial nature of the disease may prevent discovery of causal loci solely via a GWAS approach, without a dramatic increase in sample size that would provide adequate power for gene-environment interaction and subphenotype analyses.

For many complex diseases for which large effect loci are rare, pathway analysis can be useful for hypothesis generation by providing novel insight into gene- and protein-interaction networks that collectively determine disease susceptibility. Comparative pathway analysis of our GWAS data and of known syndromic/animal model cryptorchidism-associated genes identified common functional themes that suggest a role for cytoskeleton-associated genes in cryptorchidism pathogenesis. The most consistently overrepresented annotations in our analyses, such as cellular adhesion, morphology and motility; axon guidance, cytoplasmic trafficking and other cytoskeleton-dependent processes, may reflect the cellular proliferation, migration and myogenesis that are required for fetal gubernacular development (Kaftanovskaya et al., 2011, 2012). Additionally, we noted further enrichment of these pathways when we limited our analysis to regulatory SNP-associated genes as defined by GWAS3D, suggesting that these pathway data can be leveraged for post-GWAS validation of susceptibility loci. In view of evidence suggesting that topological domains may define functionally significant higher order chromatin interaction regions (Dixon et al., 2012), domains defined by our top signals (one of which includes INSL3) may also contain susceptibility genes. We are aware that many suggestive signals are likely to be false positive in the present analysis, and that these initial studies do not resolve the question of whether true signals are more commonly in or near associated genes or in more distal regulatory regions; further studies, including those correlating alleles with tissue-specific gene expression data, will be needed to validate risk genes. Interestingly, we found essentially no overlap between our results and differentially expressed spermatozoal genes in men with and without prior cryptorchidism (Nguyen et al., 2009), suggesting that sperm may not be a preferred target tissue for genotype-specific gene expression studies designed to identify cryptorchidism causal variants in regulatory regions.

Our pathway results are also congruent with prior analyses of hormone-responsive (Johnson et al., 2010; Barthold et al., 2013) and differentially expressed wildtype Long Evans versus inbred cryptorchid LE/orl (Barthold et al., 2008, 2014) transcripts in the fetal rat gubernaculum. Cytoskeleton-related transcripts normally down-regulated during embryonic (E) days 17–19 showed persistently high expression in the LE/orl gubernaculum while the normal increase in myogenic transcripts did not occur. These gene expression differences were associated with abnormal muscle patterning in the fetal gubernaculum by E21. Additionally, our pathway analysis of transcript data in these prior studies suggested the possibility of altered AR signaling in the LE/orl gubernaculum. In additional to the relevant functional annotations that we observed for many top signals (e.g. Supplementary Table SVIII), our analysis of genes containing suggestive signals showed enrichment of genes that are differentially expressed in the fetal gubernaculum as a result of stimulation by androgens or INSL3, and/or genetic variation in the LE/orl rat. These genes were also overrepresented in our list of syndromic cryptorchidism genes in man and mouse (Supplementary Table SIII), suggesting that loci involved in developmental and hormone-responsive signaling in the fetal gubernaculum may underlie cryptorchidism susceptibility.

Gubernacular mesenchyme, but not differentiated cremaster muscle, is the target of AR signaling during fetal development, but the cellular mechanisms of androgen-stimulated gubernacular development are not well understood (Kaftanovskaya et al., 2012). Cellular functions of potential relevance that are regulated by AR signaling in other tissues include actin cytoskeleton reorganization, focal adhesion formation, translocation of cytoplasmic AR to the nucleus and stimulation of cell motility (Castoria et al., 2011; Liao et al., 2013; Leach et al., 2014; Stournaras et al., 2014). Actin-binding proteins function as AR coregulators, with particular relevance to muscle development (Ting and Chang, 2008). Hic-5 is an example of an AR coregulator that regulates cytoskeletal organization in androgen-sensitive prostate mesenchymal cells via mechanisms that do not involve actin binding (Leach et al., 2014). Additionally, cytoskeletal gene expression is altered in the androgen-insensitive tfm mouse (O'Shaughnessy et al., 2007) and annotation of AR-binding proteins shows enrichment of functions related to the cytoskeleton (Jasavala et al., 2007). Enrichment of cytoskeletal functions in our pathway analysis therefore suggests the possibility that variants in genes encoding downstream, gubernacular mesenchyme-specific participants in AR signaling could account for risk of cryptorchidism in the absence of other androgen deficiency phenotypes. Some coregulators can be shared between androgen and other steroid receptors, notably estrogen receptors (Huang et al., 2004; Lanzino et al., 2005; Khurana et al., 2012), allowing for possible competition for their use in target tissues. If genetic variants contribute to altered function or expression of these molecules, resultant distortion of the androgen-estrogen balance could potentially increase susceptibility to the estrogenic and/or anti-androgenic effects of EDCs.

In summary, we did not find any genome-wide significant signals in a cryptorchidism GWAS of moderate size, but pathway analysis of suggestive signals suggests a potential role for genes involved in cytoskeletal functions, HH and syndromic forms of the disease, and overlap with a polygenic animal model of the disease. These findings are summarized in Fig. 2, which provides a general overview of syndromic and nonsyndromic cryptorchidism, known pathways downstream of hormonal signaling in the fetal gubernaculum (Emmen et al., 2000; Kubota et al., 2002; Kaftanovskaya et al., 2011, 2012), and possible areas of overlap with the findings of the present study. These results contribute to existing data that include gubernacular gene expression and genetic analysis of a polygenic rat model of cryptorchidism, which collectively, via a systems biology approach, may inform future studies that more precisely define the genetic basis of nonsyndromic cryptorchidism.

Figure 2.

Figure 2

Open in a new tab

Prospective targets for future causal gene discovery in nonsyndromic cryptorchidism. (A) Potential overlap between syndromic (orange) and nonsyndromic (red) cryptorchidism, based on suggestive signals in hypogonadotropic hypogonadism (HH) and malformation syndrome genes in the present genome-wide association study (GWAS). (B) Known pathways downstream of hormonal signaling in the fetal gubernaculum (blue). INSL3 (insulin-like 3) and androgens stimulate cellular proliferation (Emmen et al., 2000; Kubota et al., 2002) and myogenesis (Kaftanovskaya et al., 2011, 2012) via RXFP2 (relaxin/insulin-like family peptide receptor 2), and AR. AR interacts with tissue-specific coactivators, some of which are ABPs (actin-binding proteins). Pathway analysis of suggestive GWAS signals (green) shows potential overlap with gubernacular hormone pathways. Other proteins, such as those in WNT, transforming growth factor B, HOX and NOTCH signaling pathways that may also contribute to gubernacular development are not shown. EDCs, endocrine-disrupting chemicals; cAMP, cyclic adenosine monophosphate; PKA, protein kinase A.

Supplementary data

Supplementary data are available at http://humrep.oxfordjournals.org/.

Authors' roles

J.S.B., H.H., M.D.: study design and data analysis; Y.W.: sample preparation, genotyping and data analysis; J.S.B., T.F.K., C.K., A.N., A.O.F., T.E.F., A.H.B., J.A.H., R.G., P.H.N., R.M.C., K.R.H., D.J.A.: recruitment, sample collection and phenotyping; C.E.K.: genotyping; J.L.: data analysis.

Funding

This work was supported by R01HD060769 (the Eunice Kennedy Shriver National Institute for Child Health and Human Development (NICHD)), P20RR20173 (the National Center for Research Resources (NCRR), currently P20GM103464 from the National Institute of General Medical Sciences (NIGMS)), an Institute Development Fund to the Center for Applied Genomics at The Children's Hospital of Philadelphia, and Nemours Biomedical Research.

Conflict of interest

None declared.

Supplementary Material

Supplementary Data
supp_30_10_2439__index.html (1.8KB, html)

Acknowledgements

The authors would like to sincerely thank all our participants and their families for their gracious participation in this study.

References

  1. Anderson CA, Pettersson FH, Clarke GM, Cardon LR, Morris AP, Zondervan KT. Data quality control in genetic case-control association studies. Nat Protoc 2010;5:1564–1573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barthold JS, McCahan SM, Singh AV, Knudsen TB, Si X, Campion L, Akins RE. Altered expression of muscle- and cytoskeleton-related genes in a rat strain with inherited cryptorchidism. J Androl 2008;29:352–366. [DOI] [PubMed] [Google Scholar]
  3. Barthold JS, Hossain J, Olivant-Fisher A, Reilly A, Figueroa TE, Banihani A, Hagerty J, Gonzalez R, Noh PH, Manson JM. Altered infant feeding patterns in boys with acquired nonsyndromic cryptorchidism. Birth Defects Res A Clin Mol Teratol 2012;94:900–907. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barthold JS, Wang Y, Robbins A, Pike J, McDowell E, Johnson KJ, McCahan SM. Transcriptome analysis of the dihydrotestosterone-exposed fetal rat gubernaculum identifies common androgen and insulin-like 3 targets. Biol Reprod 2013;89:143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barthold JS, Robbins A, Wang Y, Pugarelli J, Mateson A, Anand-Ivell R, Ivell R, McCahan SM, Akins RE Jr. Cryptorchidism in the orl rat is associated with muscle patterning defects in the fetal gubernaculum and altered hormonal signaling. Biol Reprod 2014;91:41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Barthold JS, Wang Y, Kolon TF, Kollin C, Nordenskjold A, Fisher AO, Figueroa TE, BaniHani AH, Hagerty JA, Gonzalez R et al. Phenotype specific association of the TGFBR3 locus with nonsyndromic cryptorchidism. J Urol 2015;193:1637–1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bertini V, Bertelloni S, Valetto A, Lala R, Foresta C, Simi P. Homeobox HOXA10 gene analysis in cryptorchidism. J Pediatr Endocrinol Metab 2004;17:41–45. [DOI] [PubMed] [Google Scholar]
  8. Bogel G, Gujdar A, Geiszt M, Lanyi A, Fekete A, Sipeki S, Downward J, Buday L. Frank-ter Haar syndrome protein Tks4 regulates epidermal growth factor-dependent cell migration. J Biol Chem 2012;287:31321–31329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bramlett KS, Burris TP. Effects of selective estrogen receptor modulators (SERMs) on coactivator nuclear receptor (NR) box binding to estrogen receptors. Mol Genet Metab 2002;76:225–233. [DOI] [PubMed] [Google Scholar]
  10. Buschman MD, Bromann PA, Cejudo-Martin P, Wen F, Pass I, Courtneidge SA. The novel adaptor protein Tks4 (SH3PXD2B) is required for functional podosome formation. Mol Biol Cell 2009;20:1302–1311. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Castoria G, D'Amato L, Ciociola A, Giovannelli P, Giraldi T, Sepe L, Paolella G, Barone MV, Migliaccio A, Auricchio F. Androgen-induced cell migration: role of androgen receptor/filamin A association. PLoS One 2011;6:e17218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dalgaard MD, Weinhold N, Edsgard D, Silver JD, Pers TH, Nielsen JE, Jorgensen N, Juul A, Gerds TA, Giwercman A et al. A genome-wide association study of men with symptoms of testicular dysgenesis syndrome and its network biology interpretation. J Med Genet 2012;49:58–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dettenhofer M, Zhou F, Leder P. Formin 1-isoform IV deficient cells exhibit defects in cell spreading and focal adhesion formation. PLoS One 2008;3:e2497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dixon JR, Selvaraj S, Yue F, Kim A, Li Y, Shen Y, Hu M, Liu JS, Ren B. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 2012;485:376–380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Emmen JM, McLuskey A, Adham IM, Engel W, Grootegoed JA, Brinkmann AO. Hormonal control of gubernaculum development during testis descent: gubernaculum outgrowth in vitro requires both insulin-like factor and androgen. Endocrinology 2000;141:4720–4727. [DOI] [PubMed] [Google Scholar]
  16. Ferlin A, Zuccarello D, Zuccarello B, Chirico MR, Zanon GF, Foresta C. Genetic alterations associated with cryptorchidism. JAMA 2008;300:2271–2276. [DOI] [PubMed] [Google Scholar]
  17. Foresta C, Zuccarello D, Garolla A, Ferlin A. Role of hormones, genes, and environment in human cryptorchidism. Endocr Rev 2008;29:560–580. [DOI] [PubMed] [Google Scholar]
  18. Galan JJ, Guarducci E, Nuti F, Gonzalez A, Ruiz M, Ruiz A, Krausz C. Molecular analysis of estrogen receptor alpha gene AGATA haplotype and SNP12 in European populations: potential protective effect for cryptorchidism and lack of association with male infertility. Hum Reprod 2007;22:444–449. [DOI] [PubMed] [Google Scholar]
  19. Geller F, Feenstra B, Carstensen L, Pers TH, van Rooij IA, Korberg IB, Choudhry S, Karjalainen JM, Schnack TH, Hollegaard MV et al. Genome-wide association analyses identify variants in developmental genes associated with hypospadias. Nat Genet 2014;46:957–963. [DOI] [PubMed] [Google Scholar]
  20. Gottlieb B, Beitel LK, Nadarajah A, Paliouras M, Trifiro M. The androgen receptor gene mutations database: 2012 update. Hum Mutat 2012;33:887–894. [DOI] [PubMed] [Google Scholar]
  21. Hadziselimovic F, Emmons LR, Buser MW. A diminished postnatal surge of Ad spermatogonia in cryptorchid infants is additional evidence for hypogonadotropic hypogonadism. Swiss Med Wkly 2004;134:381–384. [DOI] [PubMed] [Google Scholar]
  22. Haznedaroglu E, Tanboga I, Sozkes S. Clinical and radiological evaluation of ter haar syndrome: case report of a patient with extreme longevity. J Rare Disord 2014;2:15–17. [Google Scholar]
  23. Hemminki K, Li X, Sundquist K, Sundquist J. Familial risks for common diseases: etiologic clues and guidance to gene identification. Mutat Res 2008;658:247–258. [DOI] [PubMed] [Google Scholar]
  24. Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet 2009;5:e1000529. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Howie B, Marchini J, Stephens M. Genotype imputation with thousands of genomes. G3 (Bethesda) 2011;1:457–470. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Howie B, Fuchsberger C, Stephens M, Marchini J, Abecasis GR. Fast and accurate genotype imputation in genome-wide association studies through pre-phasing. Nat Genet 2012;44:955–959. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Hu J, Lu J, Lian G, Ferland RJ, Dettenhofer M, Sheen VL. Formin 1 and filamin B physically interact to coordinate chondrocyte proliferation and differentiation in the growth plate. Hum Mol Genet 2014;23:4663–4673. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Huang SM, Huang CJ, Wang WM, Kang JC, Hsu WC. The enhancement of nuclear receptor transcriptional activation by a mouse actin-binding protein, alpha actinin 2. J Mol Endocrinol 2004;32:481–496. [DOI] [PubMed] [Google Scholar]
  29. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res 2009a;37:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009b;4:44–57. [DOI] [PubMed] [Google Scholar]
  31. Jain VG, Singal AK. Shorter anogenital distance correlates with undescended testis: a detailed genital anthropometric analysis in human newborns. Hum Reprod 2013;28:2343–2349. [DOI] [PubMed] [Google Scholar]
  32. Jasavala R, Martinez H, Thumar J, Andaya A, Gingras AC, Eng JK, Aebersold R, Han DK, Wright ME. Identification of putative androgen receptor interaction protein modules: cytoskeleton and endosomes modulate androgen receptor signaling in prostate cancer cells. Mol Cell Proteomics 2007;6:252–271. [DOI] [PubMed] [Google Scholar]
  33. Jensen MS, Toft G, Thulstrup AM, Henriksen TB, Olsen J, Christensen K, Bonde JP. Cryptorchidism concordance in monozygotic and dizygotic twin brothers, full brothers, and half-brothers. Fertil Steril 2010;93:124–129. [DOI] [PubMed] [Google Scholar]
  34. Johnson KJ, Robbins AK, Wang Y, McCahan SM, Chacko JK, Barthold JS. Insulin-like 3 exposure of the fetal rat gubernaculum modulates expression of genes involved in neural pathways. Biol Reprod 2010;83:774–782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Jorgez CJ, Rosenfeld JA, Wilken NR, Vangapandu HV, Sahin A, Pham D, Carvalho CM, Bandholz A, Miller A, Weaver DD et al. Genitourinary defects associated with genomic deletions in 2p15 encompassing OTX1. PLoS One 2014;9:e107028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kaftanovskaya EM, Feng S, Huang Z, Tan Y, Barbara AM, Kaur S, Truong A, Gorlov IP, Agoulnik AI. Suppression of insulin-like3 receptor reveals the role of beta-catenin and Notch signaling in gubernaculum development. Mol Endocrinol 2011;25:170–183. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Kaftanovskaya EM, Huang Z, Barbara AM, De Gendt K, Verhoeven G, Gorlov IP, Agoulnik AI. Cryptorchidism in mice with an androgen receptor ablation in gubernaculum testis. Mol Endocrinol 2012;26:598–607. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Khurana S, Chakraborty S, Zhao X, Liu Y, Guan D, Lam M, Huang W, Yang S, Kao HY. Identification of a novel LXXLL motif in alpha-actinin 4-spliced isoform that is critical for its interaction with estrogen receptor alpha and co-activators. J Biol Chem 2012;287:35418–35429. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kollin C, Ritzen EM. Cryptorchidism: a clinical perspective. Pediatr Endocrinol Rev 2014;11(Suppl 2):240–250. [PubMed] [Google Scholar]
  40. Kollin C, Hesser U, Ritzen EM, Karpe B. Testicular growth from birth to two years of age, and the effect of orchidopexy at age nine months: a randomized, controlled study. Acta Paediatr 2006;95:318–324. [DOI] [PubMed] [Google Scholar]
  41. Kollin C, Karpe B, Hesser U, Granholm T, Ritzen EM. Surgical treatment of unilaterally undescended testes: testicular growth after randomization to orchiopexy at age 9 months or 3 years. J Urol 2007;178:1589–1593; discussion 1593. [DOI] [PubMed] [Google Scholar]
  42. Kollin C, Stukenborg JB, Nurmio M, Sundqvist E, Gustafsson T, Soder O, Toppari J, Nordenskjold A, Ritzen EM. Boys with undescended testes: endocrine, volumetric and morphometric studies on testicular function before and after orchidopexy at nine months or three years of age. J Clin Endocrinol Metab 2012;97:4588–4595. [DOI] [PubMed] [Google Scholar]
  43. Kolon TF, Wiener JS, Lewitton M, Roth DR, Gonzales ET Jr., Lamb DJ. Analysis of homeobox gene HOXA10 mutations in cryptorchidism. J Urol 1999;161:275–280. [PubMed] [Google Scholar]
  44. Kubota Y, Temelcos C, Bathgate RA, Smith KJ, Scott D, Zhao C, Hutson JM. The role of insulin 3, testosterone, Mullerian inhibiting substance and relaxin in rat gubernacular growth. Mol Hum Reprod 2002;8:900–905. [DOI] [PubMed] [Google Scholar]
  45. Laitinen EM, Tommiska J, Virtanen HE, Oehlandt H, Koivu R, Vaaralahti K, Toppari J, Raivio T. Isolated cryptorchidism: no evidence for involvement of genes underlying isolated hypogonadotropic hypogonadism. Mol Cell Endocrinol 2011;341:35–38. [DOI] [PubMed] [Google Scholar]
  46. Lanyi A, Barath M, Peterfi Z, Bogel G, Orient A, Simon T, Petrovszki E, Kis-Toth K, Sirokmany G, Rajnavolgyi E et al. The homolog of the five SH3-domain protein (HOFI/SH3PXD2B) regulates lamellipodia formation and cell spreading. PLoS One 2011;6:e23653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Lanzino M, De Amicis F, McPhaul MJ, Marsico S, Panno ML, Ando S. Endogenous coactivator ARA70 interacts with estrogen receptor alpha (ERalpha) and modulates the functional ERalpha/androgen receptor interplay in MCF-7 cells. J Biol Chem 2005;280:20421–20430. [DOI] [PubMed] [Google Scholar]
  48. Leach DA, Need EF, Trotta AP, Grubisha MJ, DeFranco DB, Buchanan G. Hic-5 influences genomic and non-genomic actions of the androgen receptor in prostate myofibroblasts. Mol Cell Endocrinol 2014;384:185–199. [DOI] [PubMed] [Google Scholar]
  49. Li MJ, Wang LY, Xia Z, Sham PC, Wang J. GWAS3D: detecting human regulatory variants by integrative analysis of genome-wide associations, chromosome interactions and histone modifications. Nucleic Acids Res 2013;41:W150–W158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Liao W, Huang W, Guo Y, Xin M, Fu X. Testosterone promotes vascular endothelial cell migration via upregulation of ROCK-2/moesin cascade. Mol Biol Rep 2013;40:6729–6735. [DOI] [PubMed] [Google Scholar]
  51. Lyons LS, Burnstein KL. Vav3, a Rho GTPase guanine nucleotide exchange factor, increases during progression to androgen independence in prostate cancer cells and potentiates androgen receptor transcriptional activity. Mol Endocrinol 2006;20:1061–1072. [DOI] [PubMed] [Google Scholar]
  52. Maas SM, Kayserili H, Lam J, Apak MY, Hennekam RC. Further delineation of Frank-ter Haar syndrome. Am J Med Genet A 2004;131:127–133. [DOI] [PubMed] [Google Scholar]
  53. Nguyen MT, Delaney DP, Kolon TF. Gene expression alterations in cryptorchid males using spermatozoal microarray analysis. Fertil Steril 2009;92:182–187. [DOI] [PubMed] [Google Scholar]
  54. O'Shaughnessy PJ, Abel M, Charlton HM, Hu B, Johnston H, Baker PJ. Altered expression of genes involved in regulation of vitamin A metabolism, solute transportation, and cytoskeletal function in the androgen-insensitive tfm mouse testis. Endocrinology 2007;148:2914–2924. [DOI] [PubMed] [Google Scholar]
  55. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81:559–575. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Rosenberg NA. Standardized subsets of the HGDP-CEPH Human Genome Diversity Cell Line Panel, accounting for atypical and duplicated samples and pairs of close relatives. Ann Hum Genet 2006;70:841–847. [DOI] [PubMed] [Google Scholar]
  57. Schnack TH, Zdravkovic S, Myrup C, Westergaard T, Wohlfahrt J, Melbye M. Familial aggregation of cryptorchidism—a nationwide cohort study. Am J Epidemiol 2008;167:1453–1457. [DOI] [PubMed] [Google Scholar]
  58. Stournaras C, Gravanis A, Margioris AN, Lang F. The actin cytoskeleton in rapid steroid hormone actions. Cytoskeleton (Hoboken) 2014;71:285–293. [DOI] [PubMed] [Google Scholar]
  59. Tannour-Louet M, Han S, Louet JF, Zhang B, Romero K, Addai J, Sahin A, Cheung SW, Lamb DJ. Increased gene copy number of VAMP7 disrupts human male urogenital development through altered estrogen action. Nat Med 2014;20:715–724. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Thankamony A, Lek N, Carroll D, Williams M, Dunger DB, Acerini CL, Ong KK, Hughes IA. Anogenital distance and penile length in infants with hypospadias or cryptorchidism: comparison with normative data. Environ Health Perspect 2014;122:207–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Ting HJ, Chang C. Actin associated proteins function as androgen receptor coregulators: an implication of androgen receptor's roles in skeletal muscle. J Steroid Biochem Mol Biol 2008;111:157–163. [DOI] [PubMed] [Google Scholar]
  62. Ting HJ, Hu YC, Chang C. Actin monomer enhances supervillin-modulated androgen receptor transactivation. Biochem Biophys Res Commun 2004;319:393–396. [DOI] [PubMed] [Google Scholar]
  63. van der Zanden LF, van Rooij IA, Feitz WF, Knight J, Donders AR, Renkema KY, Bongers EM, Vermeulen SH, Kiemeney LA, Veltman JA et al. Common variants in DGKK are strongly associated with risk of hypospadias. Nat Genet 2011;43:48–50. [DOI] [PubMed] [Google Scholar]
  64. Virtanen HE, Adamsson A. Cryptorchidism and endocrine disrupting chemicals. Mol Cell Endocrinol 2012;355:208–220. [DOI] [PubMed] [Google Scholar]
  65. Wang Y, Barthold J, Kanetsky PA, Casalunovo T, Pearson E, Manson J. Allelic variants in HOX genes in cryptorchidism. Birth Defects Res A Clin Mol Teratol 2007;79:269–275. [DOI] [PubMed] [Google Scholar]
  66. Wang Y, Barthold J, Figueroa E, Gonzalez R, Noh PH, Wang M, Manson J. Analysis of five single nucleotide polymorphisms in the ESR1 gene in cryptorchidism. Birth Defects Res A Clin Mol Teratol 2008;82:482–485. [DOI] [PubMed] [Google Scholar]
  67. Wang K, Li M, Hakonarson H. Analysing biological pathways in genome-wide association studies. Nat Rev Genet 2010;11:843–854. [DOI] [PubMed] [Google Scholar]
  68. Yoshida R, Fukami M, Sasagawa I, Hasegawa T, Kamatani N, Ogata T. Association of cryptorchidism with a specific haplotype of the estrogen receptor alpha gene: implication for the susceptibility to estrogenic environmental endocrine disruptors. J Clin Endocrinol Metab 2005;90:4716–4721. [DOI] [PubMed] [Google Scholar]
  69. Zhang K, Cui S, Chang S, Zhang L, Wang J. i-GSEA4GWAS: a web server for identification of pathways/gene sets associated with traits by applying an improved gene set enrichment analysis to genome-wide association study. Nucleic Acids Res 2010;38:W90–W95. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
supp_30_10_2439__index.html (1.8KB, html)
supp_dev180_dev180supp_fig1.pdf (196.2KB, pdf)
supp_dev180_dev180supp_table1.xlsx (14.4KB, xlsx)
supp_dev180_dev180supp_table2.xlsx (27.5KB, xlsx)
supp_dev180_dev180supp_table3.xlsx (21.6KB, xlsx)
supp_dev180_dev180supp_table4.xlsx (12.5KB, xlsx)
supp_dev180_dev180supp_table5.xlsx (260.9KB, xlsx)
supp_dev180_dev180supp_table6.xlsx (55.7KB, xlsx)
supp_dev180_dev180supp_table7.xlsx (19.4KB, xlsx)
supp_dev180_dev180supp_table8.xlsx (14KB, xlsx)

Articles from Human Reproduction (Oxford, England) are provided here courtesy of Oxford University Press

RESOURCES