Genetic dissection of male infertility-related molecular pathways by transcriptional profiling of testicular biopsies from patients with non-obstructive azoospermia. A. Tajima¹, Y. Sakamoto¹, H. Okada², A. Tanaka³, K. Shichiri⁴, K. Tanaka², I. Inoue¹. 1) Div Genetic Diagnosis, IMS, Univ Tokyo, Tokyo, Japan; 2) Dep Obstet & Gynecol, Niigata Univ School of Medicine, Niigata, Japan; 3) Saint Mother Obstet & Gynecol Clinic, Fukuoka, Japan; 4) Tachikawa Hospital, Niigata, Japan.
   Male-factor infertility accounts for about half the cases in which assisted reproductive techniques are recommended. Many factors such as spermatogenic failure could cause male infertility, however, the etiologies and pathogenesis of this disease remain poorly understood. To identify new genes and/or pathways underlying male infertility with spermatogenic dysfunction, we performed a microarray-based gene-expression profiling in infertile testes. Infertile testicular biopsies were obtained under informed consent from non-obstructive azoospermia (NOA; n=47) and obstructive azoospermia (OA; n=11) patients. We found 2,611 transcripts as differentially expressed between NOA and OA testes, using Agilent Human 1A(v2) Oligo microarrays. Gene ontology-based profiling of the 2,611 transcripts revealed a significant association with biological processes involved in male gamete generation. To find novel NOA subclasses, the 2,611 transcripts were further examined with non-negative matrix factorization (NMF) method, a recently introduced clustering approach in class discovery. The NMF analysis provided three robust NOA subclasses, among which there were statistically significant differences in NOA-related clinical characteristics such as testicular pathological score. Subsequent statistical analysis showed that 149 transcripts (P<0.05) were differentially expressed among three NOA subclasses. The among-subclass differences in testicular expression for 53 transcripts with highly statistical significance (P<0.01) were confirmed by quantitative real-time RT-PCR method. These findings indicate that our strategy is successful in disclosing a group of transcripts related to spermatogenic defects and NOA classification. The 149 transcripts would therefore be diagnostic markers for NOA phenotypes, as well as potential candidates for susceptibility to NOA.