Targeted next-generation sequencing panel screening of 668 Chinese patients with non-obstructive azoospermia

Abstract

Purpose

We aimed (1) to determine the molecular diagnosis rate and the recurrent causative genes of patients with non-obstructive azoospermia (NOA) using targeted next-generation sequencing (NGS) panel screening and (2) to discuss whether these genes help in the prognosis for microsurgical testicular sperm extraction (micro-TESE).

Methods

We used NGS panels to screen 668 Chinese men with NOA. Micro-TESE outcomes for six patients with pathogenic mutations were followed up. Functional assays were performed for two NR5A1 variants identified: p.I224V and p.R281C.

Results

Targeted NGS panel sequencing could explain 4/189 (2.1% by panel 1) or 10/479 (2.1% by panel 2) of the patients with NOA after exclusion of karyotype abnormalities and Y chromosome microdeletions. Almost all mutations detected were newly described except for NR5A1 p.R281C and TEX11 p.M156V. Two missense NR5A1 mutations—p.R281C and p.I244V—were proved to be deleterious by in vitro functional assays. Mutations in TEX11, TEX14, and NR5A1 genes are recurrent causes of NOA, but each gene explains only a very small percentage (less than 4/668; 0.6%). Only the patient with NR5A1 mutations produced viable spermatozoa through micro-TESE, but other patients with TEX11 and TEX14 had poor micro-TESE prognoses.

Conclusions

A targeted NGS panel is a feasible diagnostic method for patients with NOA. Because each gene implicated explains only a small proportion of such cases, more genes should be included to further increase the diagnostic rate. Considering previous reports, we suggest that only a few genes that are directly linked to meiosis can indicate poor micro-TESE prognosis, such as TEX11, TEX14, and SYCE1.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-021-02154-9.

Introduction

Up to 1 in 10 couples have infertility problems, and male factors account for 20–25% of these cases [1, 2]. Azoospermia is defined as the absence of spermatozoa in the ejaculate after centrifugation of semen specimens. Non-obstructive azoospermia (NOA) is an extreme phenotype of quantitative spermatogenic impairment, affecting approximately 10–15% of infertile men [3]. The etiology of NOA includes both acquired and congenital factors. Genetic analyses such as karyotyping and evaluating azoospermia factor (AZF) microdeletions have become routine clinical practice in andrology [4, 5]. About 13.7% of patients with azoospermia have an abnormal karyotype [6], and about 10.04% of patients with NOA have AZF microdeletions [7]. However, many patients with NOA are still diagnosed as having idiopathic infertility. Without a clear diagnosis, it is hard to counsel such patients about the causes of their infertility, their prognosis for microsurgical testicular sperm extraction (micro-TESE), and the health risk to the man and his offspring.

Some causative genes for NOA have been identified in recent years [8]. Unlike previous genome-wide association studies, next-generation sequencing (NGS)-based studies mainly focus on rare variants with high penetrance, also known as monogenic variants [8]. Genetic defects of more than 400 genes can affect male mouse fertility [9], and at least 14 human genes have been confirmed to have pathogenic effects, including TEX11, TEX14, SYCE1, SYCP3, MEIOB, AR, NR5A1, KLHL10, FANCM, SOHLH1, STAG3, TEX15, TDRD9, and ZMYND15 [10–23]. Because most of them are inherited as recessive forms, the interpretation rate is expected to be very low for each gene. As NGS becomes more affordable, it has the potential to become a routine diagnostic method for NOA [24, 25].

Although this approach is promising, there are still a few issues of concern. First, genes in an NGS panel must be selected carefully to avoid misdiagnosis. Only about 18% of all described male genes associated with infertility are at least moderately linked to the phenotype [26], as evaluated by an existing gene-disease scoring system [27]. Second, knowing the diagnostic rate and hot-spot genes or variants can help optimize NGS panels and provide priorities for the development of targeted therapies. Third, a detailed mutational and phenotypic spectrum is needed for diagnosis and for predicting the prognosis of micro-TESE. Here we carefully selected monogenic genes for targeted NGS panels (Table 1), supported by mouse modeling and clinical cases, and used them to screen 668 Chinese patients with NOA to clarify the following issues: (1) the mutational landscape for these genes among such patients with NOA and (2) whether those gene mutations correspond to or can help predict the outcomes of micro-TESE.

Table 1.

Non-obstructive azoospermia causative genes in targeted next-generation sequencing panel

Gene	Panel 1	Panel 2	Gene function	Inheritance	OMIM
TEX11	√	√	Chromosome synapsis and formation of crossovers	XLR	300311
AR	√	√	Androgen receptor	XLR	313700
NR5A1	√	√	Transcription factors involved in sex determination	AD	184757
KLHL10	√	√	Mediate protein ubiquitination during spermiogenesis	AD	608778
SYCP3	√	√	Component of the synaptonemal complex	AD	604759
FANCM		√	DNA repair	AR	609644
MEIOB		√	Meiosis specific with OB domain	AR	617670
TEX14		√	Intercellular bridges	AR	605792
SOHLH1		√	Important regulators of spermatogenesis	AD	610224
STAG3		√	Cohesion of sister chromatids, DNA repair	AR	608489
SYCE1		√	Component of the synaptonemal complex	AR	611486
TEX15		√	Chromosome synapsis and DNA repair	AR	605795
TDRD9		√	piRNA-mediated retrotransposon silencing	AR	618110
ZMYND15		√	Transcriptional repressor in spermiogenesis	AR	614312

XLR X-linked recessive, AR autosomal recessive, AD autosomal dominance

Methods

Study subjects

This study included 668 Chinese patients with NOA who presented to the Department of Urology and Andrology of Renji Hospital, Shanghai Jiao Tong University, School of Medicine from November 2017 to June 2020. All patients had undergone a comprehensive andrological examination including semen analyses; serum hormone analyses for the levels of follicle-stimulating hormone (FSH) luteinizing hormone (LH), testosterone (T), prolactin (PRL), and estradiol (E2); testicular volume measurement via B-mode ultrasonography; karyotyping analysis; and Y chromosome microdeletion screening (sY84, sY86, sY127, sY134, sY254, sY255) [5]. Patient A4524 had undergone bilateral orchiopexy for undescended testes. Patients with obstructive azoospermia characterized by physical obstruction of the post-testicular genital tract were excluded. Other exclusion criteria include histories of orchitis or parotitis, chemotherapy for neoplastic disease, chromosomal anomalies, or Y chromosome microdeletions (AZFa, AZFb, AZFbc, and AZFc). All study participants signed informed consents.

Next-generation sequencing

Genomic DNA was extracted from blood samples using the DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA, USA). The samples have been screened successively in two NGS panel versions (panel 1 and panel 2, see Table 1). The first 189 samples were screened using panel 1 (5 genes), and the rest 479 samples were sequenced by panel 2 (14 genes). Target capture was performed for coding exons and flanking introns (± 10 bp) using IDT xGen Lockdown Probes (Integrated DNA Technologies, Coralville, IA, USA). The extent of each library was assessed using a Qubit 2.0 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). The quality and size of libraries were measured using a 2100 Bioanalyzer High Sensitivity DNA Assay (Agilent Technologies, Santa Clara, CA, USA) according to the reagent kit guide. The libraries were applied to 2 × 75 bp paired-end sequencing on the Illumina NextSeq 500 platform (Illumina Inc., San Diego, CA, USA).

Data analysis

FASTQ format raw data were filtered and aligned to the human reference genome (hg19/GRCh37) using BWA v. 0.7.13 [28]. Variants including single nucleotide variants (SNVs) and short insertions and deletions (InDels) were genotyped from recalibrated BAM files by VarDict [29]. Benign or likely benign variants identified by InterVar [30] were filtered, and the remaining variants were classified as pathogenic (P), likely pathogenic (LP), or a variant of unknown significance (VUS) according to the guidelines from the American College of Medical Genetics and Genomics (ACMG) [31]. Copy number variants (CNVs) were first called using the DNAcopy R package [32] and checked based on sequencing depth using the Integrative Genomics Viewer [33]. Nonpolymorphic CNVs were classified as P, LP, or VUS by applying the ACMG guidelines [34]. Patients were considered to achieve a definitive diagnosis when any P/LP variant was found in genes under autosomal dominant or X-linked inheritance, or P/LP homozygous variant were found in autosomal recessive genes, or at least one P/LP variant was found among compound heterozygous mutations under autosomal recessive genes.

Variant validation

All variants reported in this study were checked manually by the Integrative Genomics Viewer [33] followed by Sanger sequencing validation to avoid false positives. Sanger sequencing was used to distinguish the cis/trans relationship of pairs of heterozygous mutations in autosomal recessive genes using saliva samples from the parents.

Hematoxylin and eosin (HE) staining

Testicular tissues acquired by diagnostic biopsy or during micro-TESE were fixed in 4% paraformaldehyde solution for 12 h, embedded in paraffin wax, and sectioned at 5 μm thickness. Sections were stained with HE solution, and the images were captured by optical microscopy under ×40 magnification (evos FL auto 2, Thermo Fisher Scientific, Waltham, MA, USA).

In vitro functional assays of NR5A1 mutations

Western blot analysis and gene expression assays were performed for two NR5A1 missense variants—p.I224V and p.R281C—identified in this study. Analysis of three additional variants—p.G35E, p.R191C, and p.D238N—described in a previous study [35] were also repeated to increase credibility to our experiment. Wild-type (WT) human NR5A1 cDNA was cloned into a pCDNA3.0 vector to allow expression of Flag-tagged protein. Based on this vector, five NR5A1 mutations were generated by site-directed mutagenesis kits (Vazyme Biotech Co., Nanjing, China). The entire coding sequences of all mutant plasmids were confirmed by direct sequencing. Plasmids (2 μg/well) were transfected into HEK293T cells using the lipofectamine 2000 transfection reagent (Thermo Fisher Scientific, Waltham, MA, USA). Cells were collected 48 h later, and whole cell extracts were prepared in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) buffer. Protein products were separated on SDS–PAGE gels and blotted onto nitrocellulose membrane. The blot was probed with anti-Flag and anti-β-actin antibodies.

The NR5A1-regulated genes CYP11A1, CYP17A1, and CYP19A1 [36] were used for quantitative reverse transcription polymerase chain reaction (RT–qPCR) analysis. Total RNA was isolated from the HEK293T cells containing WT or mutant NR5A1 cDNA using TRIzol reagent (Invitrogen). Total RNA (1 μg) for each sample was converted into cDNA by RT with oligo(dT) primers. Three replicates were used for each qPCR analysis using a QuantStudio™ 7 Flex Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA). Empty vector served as a control. The primer sequences used for plasmid construction and qPCR are listed in Supplementary Table S1.

Results

Gene variants

For all 668 patients with NOA, the first 189 samples were screened using panel 1 (5 genes) to a mean depth of 1137.73 ± 429.48 ×, and the remaining 479 samples were sequenced using panel 2 (14 genes) to a mean depth of 533.20 ± 54.36×. In all,14 patients were considered to achieve a definitive diagnosis. The positive molecular diagnosis rates were 4/189 (2.1%) for panel 1 and 10/479 (2.1%) for panel 2. The gene variants detected contained 14 SNVs, 1 InDel, and 1 CNV. For four sequenced autosomal dominant genes NR5A1, KLHL10, and SYCP3 (both included in panels 1 and 2) and SOHLH1 (only included in panel 2), only NR5A1 (4/668; 0.6%) and KLHL10 (1/668; 0.2%) were found to have causative mutations. For eight of the autosomal recessive genes—FANCM, MEIOB, TEX14, STAG3, SYCE1, TEX15, TDRD9, and ZMYND15 (included in panel 2 only)—compound heterozygous mutations, homozygous mutations, or homozygous deletions were found in TEX14 (3/479, 0.6%), ZMYND15 (1/479, 0.2%), and SYCE1 (1/479, 0.2%). For two X-linked recessive genes, AR and TEX11 (included in both panel 1 and 2), four SNVs were found in TEX11 (4/668, 0.6%). In addition, strong candidate mutations were found in four patients including four SNVs in AR and KLHL10 (Table 2).

Table 2.

Patients with definite diagnosis and strong candidate variants

Patient ID	Age	Gene	Transcript	Exon/intron	Variants (zygosity)	Pop Freq	Clin Sig
Patients with definite diagnosis
A2380	34	KLHL10	NM_001329596	Exon 3	c.1012G>A, p.A338T, (Het)	0	LP (PM1+PM2+PP2+PP3)
A4524	20	NR5A1	NM_004959	Exon 2	c.39C>A, p.C13X, (Het)	0	LP (PVS1+PM2)
A5197	31	NR5A1	NM_004959	Intron 3	c.244+1G>A, NA, (Het)	0	LP (PVS1+PM2)
A1064	35	NR5A1	NM_004959	Exon 4	c.730A>G, p.I244V, (Het)	0	LP (PS3+PM1+PM2+PP2)
A2719	32	NR5A1	NM_004959	Exon 4	c.841C>T, p.R281C, (Het) *	0	LP (PS3+PM1+PM2+PP2)
A2799	30	TEX11	NM_031276	Exon 26	c.2240C>A, p.S747X, (Hem)	0	P (PVS1+PM2+PP3)
A4999	30	TEX11	NM_031276	Exon 16	c.1337G>T, p.R446M, (Hem)	0	LP (PM1+PM2+PP2+PP3)
A961	31	TEX11	NM_031276	Exon 7	c.466A>G, p.M156V, (Hem) *	0.0006	LP (PM1+PM2+PP2+PP5)
A2153	34	TEX11	NM_031276	Exon 16	c.1246C>T, p.Q416X, (Hem)	0	LP (PVS1+PM2)
A4124	30	TEX14	NM_001201457	Exon 10	c.1113dupG, p.F372fs, (Het)	0	LP (PVS1+PM2)
		TEX14	NM_001201457	Exon 10	c.1102C>T, p.H368Y, (Het)	8.24E-06	LP (PM1+PM2+PM3+PP3)
A5352	22	TEX14	NM_001201457	Exon 2	c.76C>T, p.Q26X, (Hom)	8.25E-06	LP (PVS1+PM2)
A6326	26	TEX14	NM_001201457	Exon 14	c.1898C>A, p.S633X, (Hom)	0	P (PVS1+PM2+PP3)
A5343	28	ZMYND15	NM_001136046	Exon 3	c.827G>A, p.R276Q, (Het)	0.0001	VUS (PM2+PM3+PP3)
		ZMYND15	NM_001136046	Intron 11	c.1837+1G>C, NA, (Het)	8.24E-06	LP (PVS1+PM2)
A5181	33	SYCE1	NM_001143763	Whole gene	Homozygous deletion	/	P
Patients with strong candidate variants
A4640	32	AR	NM_000044	Exon 1	c.1325A>T, p.E442V, (Hem)	0	VUS (PM2+PP2)
A2864	33	AR	NM_000044	Exon 1	c.1157G>T, p.R386L, (Hem)	5.13E-05	VUS (PM1+PM2+PP2)
A3600	28	KLHL10	NM_001329596	Exon 2	c.53C>T, p.P18L, (Het)	8.12E-06	VUS (PM1+PM2+PP2)
A2664	36	KLHL10	NM_001329596	Exon 4	c.1121T>C, p.V374A, (Het)	0	VUS (PM1+PM2+PP2)

Pop Freq population allele frequency (ExAC database), Clin Sig clinical significance, LP likely pathogenic, P pathogenic, VUS variant of uncertain significance, Het heterozygous, Hom homozygous, Hem hemizygous. Reported mutations are marked as asterisk

Clinical characteristics

Eight of the patients had been examined for testicular histopathology: A2799, A961, A4124, A5352, and A5181 showed partial or complete maturation arrest (MA); A3600 and A2664 had Sertoli cell only (SCO) phenotype; and A5197 had normal spermatogenesis (Table 3, Fig. 1). More than half of the patients (10/18; 56%) had small testes (< 10 mL), and half of them (9/18; 50%) had higher than normal FSH values, indicating gonadal dysgenesis (Table 3). Testicular histopathology diagnoses for patients A2664 (SCO with a KLHL10 mutation), A961 (MA with a TEX11 mutation), and A5197 (normal spermatogenic function with an NR5A1 mutation) are shown in Fig. 1.

Table 3.

Clinical phenotype, hormone profile, testicular volume, and treatment prognosis

Patient ID	Gene	Histopathology	LH (mIU/mL)	FSH (mIU/mL)	PRL (mIU/L)	E2 (pg/mL)	T (ng/mL)	Testicular volume (L/R, mL)	Medicine treatment (3 months)	Micro-TESE
Patients with definite diagnosis
A2380	KLHL10	/	5.01	10.71	210.32	27	2.68	9.7/9.9↓	/	/
A4524 *	NR5A1	/	4.51	17.91↑	182.90	35	3.94	6.6/10.1↓	Few sperm	/
A5197	NR5A1	Normal	19.04↑	50.92↑	357.63	21	2.69	4.1/5.4↓	/	Succeed
A1064	NR5A1	/	7.91	9.35	/	10	5.84	13.4/14.1	Few sperm	/
A2719	NR5A1	/	21.98↑	36.85↑	/	35	4.19	0.7/0.7↓	/	/
A2799	TEX11	Complete MA	4.00	6.11	/	34	2.57↓	12.6/14.2	/	No sperm
A4999	TEX11	/	7.10	10.19	191.24	20	2.91	10.2/11.1	/	/
A961	TEX11	Complete MA	3.95	16.70↑	231.80	22	2.49↓	10.1/10.9	/	No sperm
A2153	TEX11	/	5.92	12.81↑	210.81	30	3.17	8.1/7.9↓	/	/
A4124	TEX14	Partial MA	3.71	11.82↑	156.38	40	4.75	10.2/10.6	/	Few sperm
A5352	TEX14	Complete MA	7.90	13.85↑	209.12	31	3.90	8.2/9.7↓	No sperm	/
A6326	TEX14	/	6.68	7.06	283.30	37	7.00	8.2/7.3↓	/	/
A5343	ZMYND15	/	5.96	5.76	182.13	51	9.81	9.3/9.8↓	/	/
A5181	SYCE1	Complete MA	1.96	2.65	329.36	10	1.81↓	15.4/15.4	/	/
Patients with strong candidate variants
A4640	AR	/	6.81	21.56↑	278.20	25	33.70↑	9.5/8.0↓	/	/
A2864	AR	/	7.05	27.02↑	185.11	15	0.96↓	4.1/4.8↓	/	/
A3600	KLHL10	SCO	4.02	6.24	372.50	13	1.83↓	12.1/11.9	/	No sperm
A2664	KLHL10	SCO	4.10	9.91	201.19	18	2.58↓	10.1/11.3	/	No sperm

TV testicular volume (left/right, normal ≥ 10mL); Reference ranges are marked in brackets as follows: LH (1.80–8.40 mIU/mL), FSH (1.30–11.80 mIU/mL), PRL (86.92–392.20 mIU/mL), E2 (0–56 pg/mL) and T (2.60–7.40 ng/mL); normal normal spermatogenic function, MA meiotic arrest, SCO Sertoli cell only. Abnormal values are all marked with upward (higher than normal) or downward (lower than normal) arrows. Slash means untested or untreated. A4524 has history of cryptorchidism (marked as asterisk)

Fig. 1.

HE-stained sections of testicular tissue. Three figures demonstrate testicular phenotypes of varying severity in patients with NOA, including a normal spermatogenic function (A5197 with NR5A1 mutation), b maturation arrest (A961 with TEX11 mutation), and c Sertoli cell only (A2664 with KLHL10 mutation). Bar represents 75 μm. SC, Sertoli cells; SPC I and II, primary and secondary spermatocytes; SPT, spermatids

Deleterious function of NR5A1 mutations

Five of the analyzed mutations—p.I224V, p.R281C, p.G35E, p.R191C, and p.D238N—did not decrease the expression of NR5A1 protein (Fig. 2a), but they all affected the expression of downstream genes CYP11A1, CYP17A1, and CYP19A1 (Fig. 2b). The p.I224V and p.R281C mutations caused the most severe effects, reducing downstream gene expression to empty vector (EV) levels. The degrees of deleterious effect of p.G35E, p.R191C, and p.D238N were consistent with a previous report [35], indicating robustness of this experiment.

Fig. 2.

In vitro functional assay of NR5A1 mutations. p.I224V and p.R281C were identified in this study. Analysis of three previously described variants p.G35E, p.R191C, and p.D238N were also repeated to increase credibility to our experiment. a Western blot analysis. b Assays of NR5A1 transcriptional activity by using downstream gene CYP11A1, CYP17A1, and CYP19A1. EV, empty vector; WT, wild type

Outcomes of micro-TESE

Most patients chose not to use medication or micro-TESE treatments after being informed of their genetic diagnosis, and they gave up hope of fatherhood or used donor semen. Three of them (A4524, A1064, and A5352) chose to use 3 months of medication: a combination of vitamin E (300 mg/day) and clomiphene citrate (50 mg/day). Of these, patients A4524 and A1064 with NR5A1 mutations successfully produced a few spermatozoa by micro-TESE, but patient A5352 with a TEX14 mutation remained with a diagnosis of NOA after treatment. Only two of six patients with L/LP or VUS mutations produced viable spermatozoa through micro-TESE. Of these, patient IA5197 with an NR5A1 mutation successfully produced sufficient spermatozoa for intracytoplasmic sperm injection (ICSI). Only very few spermatozoa (1–2 per high power field) that were insufficient for ICSI were obtained for patient A4124 with a TEX14 mutation. Two patients, A3600 and A2664 with SCO phenotypes (KLHL10 mutations), failed to obtain any spermatozoa through micro-TESE (Fig. 1, Table 3).

Discussion

Diagnostic rate

To date, there have been several studies using NGS or whole exome sequencing for NOA or severe oligozoospermia (SO) in cohort screening, yielding diagnostic rates of 6/314 (1.9%) [37] and 1/314 (0.3%) [38]. In this study, targeted panel sequencing could explain 4/189 (2.1% for panel 1) or 10/479 (2.1% for panel 2) of the patients with NOA after exclusion of those with karyotypic abnormalities and Y chromosome microdeletions. Some studies have not been discussed here because they did not classify variants properly according to the ACMG guidelines [31]. Fakhro et al. [39] reported two variants of NANOS2 and FKBP6 at 5% and 9%, respectively, in 75 infertile man. Araujo et al. [40] conducted genetic screening of 16 Brazilian patients with MA and SCO, and 5/10 (50%) of the reported mutations had a population frequency of more than 1%, which is benign evidence.

Our low diagnosis rates using the NGS panels (2.1% or 2.1%) might be because of the high heterogeneity of NOA. First, NOA can arise from nongenetic causes, and this will lower the genetic diagnostic rates. More stringent categorization into MA or SCO phenotypes by testicular histopathology could certainly increase the genetic diagnosis rate, but this would diminish the noninvasive advantage of genetic testing. Second, the low diagnostic rate by NGS panel is linked to limited number of genes selected for testing. According to the available screening studies [37–39] and this study, it appears that each gene explains only a very small percentage of cases with NOA. Therefore, increasing the diagnostic rate of NGS panels for NOA needs to exclude patients with nongenetic causes as much as possible, at the same time expanding the numbers of genes in NGS panel.

The frequently reported gene variants in patients with NOA are likely to be pathogenic. As NGS-based studies in NOA are still limited, almost all of the mutations we detected here have not been reported except for NR5A1 p.R281C [41] and TEX11 p.M156V (reported as p.M171V of NM_001003811.2) [14]. It should be noted that SYCE1 deletions are likely to have high prevalence in the Chinese population [42]. Therefore, this study provides important evidence for the clinical diagnosis of NOA in China and highlights the importance of sharing data across different clinical centers.

Clinical phenotypes

Inferring the phenotype—the prognosis for micro-TESE—by genotyping is currently challenging. Lines of evidence to be considered are as follows: (1) how genes are involved in spermatogenesis; (2) the phenotypes of specific gene knockout mice; and (3) the phenotypes of previously reported cases. For example, several tested genes are directly involved in meiosis, and TEX14, TEX11, and SYCE1 were found to have pathogenic mutations in this study. The Tex14 protein localizes to male germ-cell intercellular bridges [11]; TEX11 interacts with SYCP2, an integral component of lateral elements of the synaptonemal complex [43]; and SYCE1 encodes central element protein 1 of the synaptonemal complex [44]. In the absence of these key proteins, spermatogenesis would halt at meiotic division [11, 12, 14, 43]. Previously reported cases all showed MA or SCO phenotypes for TEX14 [39, 40, 45], TEX11 [14], and SYCE1 [12]. Here, the outcomes of testicular biopsy or micro-TESE were all poor for patients with TEX14, TEX11 (Fig. 1b), or SYCE1 mutations. Therefore, we suggest that TEX11, TEX14, and SYCE1, which are directly associated with meiosis, predict a poor micro-TESE prognosis.

The genes AR and NR5A1 affect spermatogenesis by participating in hormonal regulation and action [36, 46]. The AR gene encodes the androgen receptor, which mediates the role of androgens in spermatogenesis and sexual development [47]. NR5A1 encodes a transcription factor involved in the regulation of reproduction, steroidogenesis, and sexual differentiation [48]. As a result, patients with mutations in these two genes can exhibit a wide phenotypic spectrum, ranging from severe to mild sexual developmental abnormalities [36, 49] to isolated azoospermia [46, 50, 51]. AR mutations are not predictive of poor micro-TESE prognosis, also supported by case reports [52, 53]. It should be noted that the AR variants reported here are two types of VUS lacking functional studies.

The inheritance of NR5A1 mutations is dose dependent, but is not classical autosomal recessive or autosomal dominant [16]. In our study, heterozygous LOF mutations p.C13X and c.244+1G>A were identified in patients A4524 and A5197, who had a history of cryptorchidism and who both showed small testes (4.1/5.4 mL) Both patients produced spermatozoa following medication (a combination of vitamin E and clomiphene citrate) [54] or micro-TESE. That the patients could be treated medically indicates that NR5A1 variants do not affect spermatogenesis. A heterozygous missense mutation, p.R281C, was found in A2719 who had severe gonadal dysplasia with 0.7/0.7 mL testicular volumes. This patient has had no further treatment. Buonocore et al. [41] identified p.R281C in a patient with disorders of sex development and speculated that the 280/281 position of the NR5A1-encoded protein forms a key region interacting with corepressors. The in vitro functional analysis of p.R281C showed an obvious effect on downstream gene expression (Fig. 2b). Although this effect of p.I244V on downstream genes is equivalent to p.R281C (Fig. 2b), patient A1064 presented as isolated NOA with a normal hormone profile, and micro-TESE was successful. This suggests that the effects of NR5A1 mutations and the severity of clinical phenotypes are not simple correspondences. The three medically treated cases we report here also demonstrate that NR5A1 mutations are not predictive of a poor micro-TESE prognosis.

As of now, patient A5343 is the second reported case with NOA caused by a ZMYND15 mutation, while the first such case was from a consanguineous family and had an MA phenotype identified by bilateral testis biopsies [23]. ZMYND15 encodes a transcriptional repressor, and its inactivation results in early activation of haploid genes and depletion of late spermatids. Zmynd15-null male mice also presented with MA [55]. More patients with NOA associated with ZMYND15 mutations are needed to expand the phenotypic spectrum.

KLHL10 is expressed in the cytoplasm of elongating and elongated spermatids, and mutations of this gene can disrupt spermiogenesis and lead to reduction in the numbers of late spermatids [56]. Two patients with KLHL10 variant reported in this study (A3600 and A2664) had SCO phenotypes (Fig. 1b for patient A2664). However, the previously reported cases all had SO phenotypes [17], so whether these two VUS mutants (p.P18L and p.V374A) are causative of the SCO phenotype remains uncertain and requires further study. The phenotypic spectrum of the KLHL10 mutations also includes SO, which indicates that KLHL10 mutation is not associated with poor micro-TESE prognosis.

Limitations

This work had limitations as follows: (1) although the cohort size was relatively large (668), the number of samples used to establish relationships between genotype and phenotype was insufficient because of the low rate of molecular diagnosis of NOA. Moreover, some patients with NOA will choose donated semen instead of micro-TESE and ICSI; this further reduces the opportunity to observe the relationship between mutations and micro-TESE outcomes. (2) The testicular histopathology analysis was not sufficiently detailed to provide a Johnsen's score of spermatogenesis [57]. (3) This study did not perform functional analyses of all VUS mutations. We will continue to enlarge the NOA cohort and refine the histopathology and functional analysis in the future.

Conclusions

We used NGS panels to screen 668 patients with NOA. Targeted panel sequencing could explain 4/189 (2.1% with panel 1) or 10/478 (2.1% with panel 2) of patients with NOA after exclusion of karyotypic abnormalities and Y chromosome microdeletions. Almost all mutations detected in this study were newly described except for NR5A1 p.R281C and TEX11 p.M156V. Two missense NR5A1 mutations, p.R281C and p.I244V, were proved to be deleterious by in vitro functional assays. Mutations in TEX11, TEX14, and NR5A1 genes were causes, but each gene explains only a very small percentage (4/668 or 0.6%). Although more genes should be included to increase the diagnostic rate, we suggest that only a few genes such as TEX11, TEX14, and SYCE1 that are directly related to meiosis can indicate a poor micro-TESE prognosis.

Funding

This study was supported by grants from the National Natural Science Foundation of China (81971376), a grant from the science and technology project of Pudong New Area Health and Family Planning Commission, Shanghai, China (XG8300000-2017-364), a grant from the Health Commission of Pudong New Area, Shanghai, P. R. China (PW2020D-7), a grant from Shanghai Municipal Health Commission for advanced and suitable technology promotion projects (2019SY056), and Clinical Research Plan of SHDC (No. SHDC2020CR4035).