The present study was approved by the ethics committee of Peking University Shenzhen Hospital (Shenzhen, China; approval no. 20090018). The current study was approved on July 18, 2009, initiated on August 1, 2009 and terminated on December 1, 2014.

The patient inclusion criteria was the same as described previously (13), with certain modifications. A total of 1,880 azoospermic patients were recruited for the present study from the Center of Reproductive Medicine, Tongji Medical College, Huazhong University of Science and Technology (Wuhan, China) between January 2007 and October 2011. Among the subjects, 776 patients fulfilled the following criteria for IA diagnosis: i) No sperm detected in the pellets of semen samples on three different occasions; ii) no obstruction, inflammation or injury of the reproductive system or pelvic cavity; and iii) no karyotypic abnormality or Y chromosome microdeletion. A total of 709 fertile men from the Center of Physical Examination, Peking University Shenzhen Hospital were recruited as controls. Each had fathered at least one child without assisted reproductive techniques, including in vitro fertilization, intracytoplasmic sperm injection or intracytoplasmic morphologically selected sperm injection. Following panel re-sequencing and quality control steps, 776 patients aged 24–46 years (mean, 30.6 years) and 709 fertile men aged 29–51 years (mean, 35.6 years) were available for further analysis. Informed written consent was obtained from each subject.

Panel re-sequencing was performed as described previously (13). Genomic DNA samples (5 µg) isolated from peripheral blood samples using the E.Z.N.A. Blood DNA kits (Omega Bio-Tek, Inc., Norcross, GA, USA) were sent to the Beijing Genomics Institute (Shenzhen, China) for exome capture and sequencing. The capture procedure was performed in solution with a NimbleGen custom array (Roche Applied Science, Madison, WI, USA) that enriched the exonic sequences of 654 infertility- or subfertility- associated genes. The majority these genes have been reviewed by Matzuk and Lamb (14). Additionally, the present study selected other genes that have been previously demonstrated to cause male reproductive defects in mouse models in studies published between November 2008 and December 2010 (15). Panel re-sequencing was performed using the Illumina platform (Illumina, Inc., San Diego, CA, USA) with 90 bp pair-end reads.

FASTQ sequence files were aligned against the human reference genome (NCBI build 37.1, hg19) with Short Oligonucleotide Analysis Package aligner software (version 2.21; www.soap.genomics.org.cn). Duplicated paired-end reads were removed from the merged data sets. Single nucleotide variants that were different from the hg19 reference genome were filtered out if they met any of the following criteria: i) A Phred-like quality score of ≤20; ii) overall depth of ≤8×; estimated copy number of ≥2; iii) or genomic distance between two adjacent variants of <5 bp. In addition, the quality score of the major and minor allele at heterozygous loci were ≥20. The variants were then annotated using an in-house functional prediction tool, and were compared with the dbSNP Build 135 (www.ncbi.nlm.nih.gov/projects/SNP/) and 1000 Genomes (www.1000genomes.org; as of August 2010). Validation of novel missense variants by Sanger sequencing. To validate the novel missense variants identified by deep sequencing, polymerase chain reaction (PCR) amplifications were performed and the PCR products were sequenced in both directions with a 3730 DNA Analyzer (Applied Biosystems; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The primers for PCR and Sanger sequencing validation of the AR gene are presented in Table I.

As described in Panel re-sequencing, genomic DNA samples (5 µg) were isolated from peripheral blood samples using the E.Z.N.A. Blood DNA kits (Omega Bio-Tek, Inc.). PCR was performed in a volume of 50 µl containing 1 µM of each forward and reverse primer, 50 ng DNA, and 25 µl EmeraldAmp PCR Master mix (Takara Bio Inc., Otsu, Japan). Products were amplified in a thermocycler (MyCycler; Bio-Rad, Hercules CA, USA) with the following conditions: 30 cycles of 10 sec at 98°C, 30 sec at 60°C, and 40 sec at 72°C. Amplicons were extracted from gels and sequenced in both directions with a 3730 DNA Analyzer (Applied Biosystems; Thermo Fisher Scientific, Inc.). The primers for PCR and Sanger sequencing validation of the AR gene are presented in Table I.

Multiple protein alignments were performed with MegAlign 7.1.0 (DNASTAR, Inc., Madison, WI, USA). The amino acid sequences of the androgen receptor in humans, chimpanzees, rhesus monkeys, cows, rats, mice and chickens was determined. The identification numbers of androgen receptor protein were as follows: Human, NP_000035.2; chimpanzee, XP_009437511.1; rhesus, NP_001028083.1; cow, NP_001231056.1; rat, NP_036634.1; mouse, NP_038504.1; and chicken, NP_001035179.1).

Sorting Intolerant From Tolerant (SIFT; sift.jcvi.org/) and PolyPhen 2.0 (genetics.bwh.harvard.edu/pph2/) analysis were used for the evaluation of coding single nucleotide polymorphisms (16,17). SIFT is based on the premise that protein evolution is correlated with protein function. SIFT scores ≤0.05 are predicted by the algorithm to be damaging, whereas scores >0.05 are considered tolerant. Predictions are based on a combination of phylogenetic, structural and sequence annotation information characterizing a substitution and its position in the protein. PolyPhen scores >0.85 are predicted by the algorithm to be probably damaging, scores >0.15 are considered possibly damaging, whereas scores <0.15 are considered benign.

A human AR expression plasmid was provided as a gift from Dr Chawnshang Chang (George H. Whipple Lab for Cancer Research, Departments of Pathology and Urology, University of Rochester Medical Center, Rochester, NY, USA). Site-directed mutagenesis was performed to generate AR expression plasmids bearing the C290R, S495N or R630W variant, as described previously (13). DNA sequencing was performed to confirm the introduced variants. The PCR primers used for site-directed mutagenesis and plasmid construction are presented in Table II. Products were amplified in a thermocycler (Bio-Rad) using an Expand High Fidelity PCR system (Roche, Basel, Switzerland). The 50 µl PCR reaction was conducted with 50 ng templates, 1 µM primer pairs, 200 µM dNTPs and 2 units of DNA polymerase. The extension reaction was initiated by pre-heating the reaction mixture to 94°C for 3 min; 16 cycles of 94°C for 1 min, 52°C for 1 min and 68°C for 8 min; followed by incubation at 68°C for 15 min. DNA sequencing was conducted with a 3730 DNA Analyzer (Applied Biosystems) to confirm the introduced variants.

Luciferase analysis was performed as described previously, with certain modifications (18). HeLa and TM4 cells (American Type Culture Collection, Manassas, VA, USA) were cultured in Dulbecco's modified Eagle's medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (Gibco; Thermo Fisher Scientific, Inc.), 100 U/ml penicillin (Gibco; Thermo Fisher Scientific, Inc.) and 100 µg/ml streptomycin (Gibco; Thermo Fisher Scientific, Inc.) at 37°C, 95% humidity and 5% CO₂ atmosphere. Cells were seeded in 24-well tissue culture plates for 24 h prior to transfection. Equivalent concentrations (100 ng) of wild-type (WT) or mutant AR expression plasmids were cotransfected with mouse mammary tumor virus (pMMTV) long terminal repeat plasmids (a gift from Dr Chawnshang Chang) into HeLa and TM4 cells using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. Cells were treated with or without 100 nM testosterone (Sigma-Aldrich) after 6 h of transfection and harvested 24 h after treatment. Firefly and Renilla luciferase expression was assessed using a Dual-Luciferase Reporter Assay System (cat no. E1910; Promega Corporation, Madison, WI, USA). Renilla luciferase activity was normalized to that of firefly luciferase. The Modulus/9200-003 luminometer (Turner Biosystems, Sunnydale, CA, USA) was used in this study. Following normalization for transfection efficiency, induction factors were calculated as the ratio of the mean luciferase value for testosterone-stimulated versus non-testosterone-stimulated (ethanol vehicle-treated) samples.

All ultrasonographic measurements were conducted by the same ultrasonographer. Patients with novel AR variants were examined with scrotal ultrasonography to test the size of testes. All examinations were performed using ultrasound scanner (EUB-7500, Hitachi Medical Corporation, Tokyo, Japan) equipped with a linear array transducer (6.5 MHz; Model EUP-L65; Hitachi Medical Corporation).

All experiments were repeated at least three times. Data of the luciferase assay are expressed as the mean ± standard deviation. SPSS version 17.0 statistical software (SPSS, Inc., Chicago, IL, USA) was used for statistical analysis. Analysis of variance (ANOVA) and Dunnett's t-test were used to compare the difference in the means between the variants and WT. P<0.05 was considered to indicate a statistically significant difference. Data were log-transformed before ANOVA to satisfy the equal variances assumption.

To examine whether AR genetic defects were associated with IA, the present study screened for AR exonic variants in 776 patients with IA and 709 fertile men using massive parallel sequencing technology. As demonstrated in Table III, seven missense variants and two synonymous variants were detected in AR. Two of these missense variants (F827L and M887V) have been previously reported in the ExAC database. Another variant at amino acid 290 (C290W) has been previously described in the ExAC database, however, this variant was different from the variant identified in the present study (C290R). The three patient-specific missense variants (C290R, S495N and R630W) have not been previously reported in the dbSNP135 database, 1000 Genome Project dataset or ExAC database. These three novel patient-specific missense variants were further confirmed by Sanger sequencing (Fig. 1). Alignment of the amino acid sequence of AR to its orthologs in different species demonstrated that the R630W variant affected a highly conserved amino acid (Fig. 2). Based on this conservation behavior, the variant R630W was predicted to be possibly damaging to the protein, according to SIFT and PolyPhen-2.0 analysis (Table IV). The locations of these three missense variants within the AR protein are demonstrated in Fig. 3.

To evaluate whether the identified patient-specific missense variants of AR inhibit its regulatory function, luciferase reporter constructs containing the androgen responsive elements (MMTV promoter) were transfected into HeLa and TM4 cell lines. AR WT, C290R and S495N, but not R630W variants, significantly increased MMTV promoter activity compared with empty vector in both cell types (P<0.05; Fig. 4). These results indicate that C290R and S495N did not affect the transcriptional regulation activity of AR. However, the transcriptional regulation activity of AR was inhibited by the R630W variant.

Scrotal color Doppler ultrasonography of the three patients with novel AR variants revealed small testes in the scrotal sac, with a homogeneous echotexture and wide hypoechogenicity. No solid or cystic lesions were observed. Table V summarizes the clinical and hormone data of these three patients, none of which had a family history of male infertility, CAIS or PAIS.