Abstract
Infertility is a global healthcare problem, and despite long years of assisted reproductive activities, a significant number of cases remain idiopathic. Our currently restricted understanding of basic mechanisms driving human gametogenesis severely limits the improvement of clinical care for infertile patients. Using exome sequencing, we identified a nonsense mutation leading to a premature stop in the TEX15 locus (c.2130T>G, p.Y710*) in a consanguineous Turkish family comprising eight siblings in which three brothers were identified as infertile. TEX15 displays testis-specific expression, maps to chromosome 8, contains four exons and encodes a 2789-amino acid protein with uncertain function. The mutation, which should lead to early translational termination at the first exon of TEX15, co-segregated with the infertility phenotype, and our data strongly suggest that it is the cause of spermatogenic defects in the family. All three affected brothers presented a phenotype reminiscent of the one observed in KO mice. Indeed, previously reported results demonstrated that disruption of the orthologous gene in mice caused a drastic reduction in testis size and meiotic arrest in the first wave of spermatogenesis in males while female KO mice were fertile. The data from our study of one Turkish family suggested that the identified mutation correlates with a decrease in sperm count over time. A diagnostic test identifying the mutation in man could provide an indication of spermatogenic failure and prompt patients to undertake sperm cryopreservation at an early age.
Introduction
As attested by a recent World Health Organization (WHO) report, infertility is a global healthcare problem. Indeed, about 48.5 million couples suffer from infertility worldwide (1). The cause of infertility can be of female, male or mixed origin, each cause accounting for about one-third of all cases.
The definition of a male factor is generally based on abnormal semen parameters. Semen quality is usually assessed by sperm count in the ejaculate, percentage of motile sperm and morphology (2). Male infertility can then be defined as the absence of sperm in ejaculate (azoospermia), inability to produce spermatozoa in sufficient numbers (oligozoospermia, with diverse severity), inadequate motility (asthenozoospermia), abnormal morphology (teratozoospermia) or a combination of these defects. Non-obstructive azoospermia (NOA) and severe oligozoospermia (SO) are the most frequent causes of male infertility in man. Unfortunately, a significant number of cases remain idiopathic. The current estimate is that about 30% of men seeking help at an infertility clinic are found to have oligozoospermia or azoospermia of unknown aetiology (3). So far, the only available treatment is in vitro fertilization with an intra-cytoplasmic sperm injection (ICSI), which represents an empirical approach generally offered as a standard treatment option.
Spermatogenesis is an extremely complex process that involves highly specialized mechanisms. Considering the concerted action of more than 2000 genes in spermatogenesis, a mutation in any of these genes may be responsible for a spermatic defect (4), and therefore, it is likely that many ‘idiopathic’ forms may have a genetic origin.
Recent technological improvements enable researchers to implement whole genome approaches such as single nucleotide polymorphism (SNP) microarrays, whole genome or exome analysis for genetic testing studies. So far, the application of SNP microarrays and homozygosity mapping has successfully led to the identification of a few genes involved in infertility (5–9).
Genome-wide association studies (GWAS) have been performed by different groups, and a dozen gene polymorphisms have been proposed to be associated with impaired spermatogenesis (10–20). However, associations are generally weak, and initial results are rarely confirmed by other studies. The need to study a very large population (>10 000) displaying phenotypic homogeneity remains a major paradigm of GWAS (4). Therefore, the analysis of large families of well-documented male infertility with some degree of consanguinity provides an alternative approach to identify genes involved in infertility.
Following up on studies of family cases of teratozoospermia and the identification of two genes involved in globozoospermia (5,6,21), characterized by round-headed spermatozoa lacking an acrosome, we are now focussing our research on NOA and SO (<1 × 106 spermatozoa/ml). We present here the analysis of a consanguineous Turkish family comprising six brothers; three infertile, two fertile, one of unknown fertility status and two fertile sisters. We revealed a nonsense mutation in TEX15, which co-segregated with the infertility phenotype, as the cause of spermatogenic defects in the family. Interestingly, the human phenotype is very similar to the one observed in the Tex15 KO mouse. In mice, disruption of the gene causes a drastic testis size reduction, as observed for the patients described here, and meiotic arrest in the first wave of spermatogenesis in males.
Results
Infertility phenotype
A consanguineous Turkish family comprising six brothers, three infertile, two fertile, one of unknown fertility status and two fertile sisters (Fig. 1A), was identified at the Bahceci Health Group, Istanbul, Turkey. Parents were first-degree cousins. Two of the infertile brothers were diagnosed as NOA (IV:1 and IV:3; Fig. 1A), while the younger third infertile brother (IV:5; Fig. 1A) was SO with, in 2010, little sperm in the ejaculate. All affected brothers presented, considering an average normal size of 18–20 ml (22), a drastic testicular size reduction of more than 50% (Fig. 1B). Histopathology of the index patient (IV:3) identified a maturation arrest at the primary spermatocyte stage (Fig. 1C).
Consanguineous Turkish family with eight siblings. (A) Pedigree structure showing the segregation of c.2130T>G, p.Tyr710* mutation. Filled symbols indicate affected members, and clear symbols indicate unaffected members. One female sibling died at the age of 2, the cause of death is unknown. Asterisk (*) points out the family member for whom no information about semen analysis was available. Whole exome sequencing was performed on two affected brothers indicated with arrows. (B) Testicular volume of affected members in the family. Normal testicular volume for man is given as 18–20 ml8. (C) Testicular histology of patient IV:3. Representative sections of human left and right testes biopsies displaying maturation arrest pattern. Arrows indicate primary spermatocytes; asterisks indicate interstitial spaces. Bars represent 50 μm in length.
Brothers IV:1 and IV:3, with NOA, had a testicular sperm extraction (TESE) operation, a limited number of spermatozoa were detected and ICSI was performed. One embryo was transferred for both cases; transfer on Day 2 was realized for IV:1, while Day 3 transfer was chosen for IV:3. Transferred embryos were of poor quality, and no pregnancy was obtained (Supplementary Material, Table S1).
Two ICSI attempts were performed for patient IV:5, the first one with testicular sperm since no motile spermatozoa were found in the ejaculate. A pregnancy was achieved, but resulted in early abortion. Ejaculated sperm were used for the second attempt in 2012, and a healthy child was born. In 2014, a new semen analysis revealed azoospermia; no spermatozoon was detected even after concentration.
Whole exome sequencing and data analysis
In order to identify the genetic cause of male infertility in the family, whole exome sequencing (WES) was applied for samples IV:1 and IV:3. For these samples, 6 and 6.7 GB DNA sequence were generated, respectively. In both the cases, >56% of the target exome was represented with >40-fold coverage (Supplementary Material, Table S2).
Given the known consanguinity in the family, we hypothesized that the disease should follow an autosomal recessive inheritance mode, and we thus tracked homozygous variants shared by both samples. We filtered variants according to coverage, minor allele frequency in populations (we filtered out when >1%), validation status of the reference SNP (not validated or validated by only one database) and novelty by comparing them with our in-house exome database.
Data analysis revealed 17 homozygous variations in 15 genes shared by the two samples (Supplementary Material, Table S3). Amazonia (http://amazonia.transcriptome.eu/), BioGPS (http://biogps.org/) and EMBL-EBI expression atlases (http://www.ebi.ac.uk/gxa/home) were scanned for all 15 genes that passed our filtering process. This revealed TEX15 (MIM: 605795) as the most plausible candidate gene due to its specific expression in germ cells and the existence of a mouse KO model with a male sterility phenotype similar to the one observed in our patients. In particular, KO male mice presented a drastic testis size reduction and a meiotic arrest (23). The human TEX15 maps to chromosome 8 and consists of four exons and spans a genomic region of 59.06 Kb on the reverse strand. The gene encodes a 2789-amino acid protein with only two predicted unknown domains named ‘TEX15’ (PF15326) according to the PFAM (24) and SMART (25) databases (Fig. 3C).
Mutation confirmation and screening
A homozygous single base substitution (NM_031271.3: c.2130T>G, p.Y710*) causes a premature stop at the first exon of TEX15, shared by two infertile brothers. The c.2130T>G variation, supported by 112 and 134 reads, respectively, for the samples sequenced, was not observed in any variation database. The variant and the segregation in the family were confirmed through Sanger sequencing (Fig. 2). The mutation co-segregated with the infertility phenotype; the two NOA brothers and the oligozoospermic sibling were homozygous for the mutation, while parents were carriers as well as two fertile brothers. Fertile sisters were wild type; however, the youngest brother with an unknown fertility status was also homozygous for the mutation (Fig. 1).
Plots of results from Sanger sequencing for mutant, carrier and wild-type samples. Allele in question is indicated by orange arrow. TEX15 is in the reverse strand, and mutation causes tyrosine codon (TAT) change into stop codon (TAG).
The mutation was absent in 107 (214 chromosomes) fertile males of Turkish origin. In addition, all exons and intronic flanking sequences of TEX15 were amplified and sequenced in order to validate variants and to look for other mutations in 85 unrelated individual NOA cases and 13 SO cases with similar phenotypes (primers listed in Supplementary Material, Table S4A). No mutations were found.
Expression of TEX15
In order to determine the expression profile of TEX15 in man, we performed reverse transcriptase-polymerase chain reaction (RT-PCR) and immunohistology. Twenty-one total RNA samples from different organs, including testis, and 9 testicular tissues derived from 3 Sertoli cells only, 4 maturation arrests at different stages, one normal and one prepubescent patient were analysed by RT-PCR (primers listed in Supplementary Material, Table S4B). The results confirmed the restricted expression of TEX15 to the testis (Fig. 3A) and very low levels of expression in Sertoli cells (not visible on the figure). TEX15 expression was predominant in the germ line of the testis (Fig. 3B). This latter result was confirmed by immunohistology on the nine testicular tissues, which showed a high level of expression in the germ line and only little expression in Sertoli cells and interstitial cells (Fig. 4). These results are similar to those obtained in the mouse species (23). To reinforce the hypothesis that the function of the human and the mouse Tex15 is similar, we compared both proteins and observed a high degree of amino acid identity over the entire sequence (52.46% identity; see Fig. 3C).
TEX15 expression studies at mRNA and protein levels. GADPH was used as internal control for RT-PCR, individual control without reverse transcriptase added to each sample and labelled as RT(−). (A) RT-PCR results of TEX15 and GADPH for different human tissues. L: ladder, Agl: adrenal gland, Bm: bone marrow, CrC: brain cerebellum, BrW: brain-whole, FBr: foetal brain, Flv: foetal liver, Kd: kidney, Lv: liver, Ln: lung, Pl: placenta, Pr: prostate, Sgl: salivary gland, Sm: skeletal muscle, Spl: spleen, Thy: thymus, Ut: uterus, Cl: colon, Si: small intestine, Sc: spinal cord, Ov: ovary and Ts: testis. (B) RT-PCR results of TEX15 and GADPH for human testicular samples with different histology. Wells 1–9: RT(+) and Wells 10–18: RT(−). Samples 1–3: Sertoli cell only, samples 4–7: maturation arrest at different stages, sample 8: normal fertile and sample 9: prepubescent. (C) Schematic representation of exons and protein sequence of TEX15. The exonic map of TEX15 is shown in the upper part of the figure based on the RefSeq transcript (NM_031271.3). Within the four exons, the coding sequence is highlighted in grey. The protein sequence (Uniprot: Q9BXT5, TEX15_HUMAN) is represented in the lower part with the two PFAM domains as black rectangles (TEX15, PF15326). The human protein shares 52.46% identify with the mouse protein. The TEX15 domains share, respectively, 66.24 and 65.09% with the mouse protein. The nonsense mutation (c.2130T>G, p.Y710*) is highlighted in both the transcript and the protein sequence, respectively.
Immunohistochemistry for TEX15. (A) Positive control: mouse testis; insert: isotype control. (B–D) Testicular tissue from three Sertoli cell only patients. No germ cells are present. (E–H) Testicular tissue from four patients with maturation arrest. (I) Fertile adult testicular tissue. (J) Prepubescent testicular tissue. Germ cell nuclei are intensively stained, while most, but not all, somatic cells show faint expression. Scale bars: 50 µm. Arrow: germ cell; arrowhead: Sertoli cell.
Case–control association study
Due to its known role during spermatogenesis, TEX15 was a good candidate gene for infertility that led Aston et al. to test for a possible association between infertility and six non-synonymous SNPs located in the TEX15 coding sequence (11). This case–control association study, performed in a cohort of men of European descent with NOA, SO and control individuals, revealed a marginal association between the major alleles of two SNPs of TEX15, rs323344 [NM_031271.3: c.4009T>G (p.L1337V)] and rs323345 [NM_031271.3: c.3932A>G (p.N1311S)], and infertility. A follow-up study conducted for these two SNPs in a European Balkan population failed to replicate this association (26). Chinese Han populations of NOA, SO and controls were also genotyped for all six non-synonymous SNPs of TEX15, and while no significant association signal was obtained for rs323344 [NM_031271.3: c.4009T>G (p.L1337V)] and rs323345 [NM_031271.3: c.3932A>G (p.N1311S)], a slight association for the minor alleles of two other SNPs of TEX15, [NM_031271.3: c.3103A>G (p.I1035V)] and rs323347 [NM_031271.3:c. 310T>C (p.C104R)], was detected in SO patients (27). The major limitation of these three studies, as for all association studies performed so far in the field of infertility, resides in the phenotypic heterogeneity and the limited number of patients and controls analysed (4). In order to contribute to the analysis of the role of these SNPs in infertility, even if our cohort is also limited in number, we performed a case–control association study in Turkish males with NOA (n = 85) and control males (n = 107). For all four SNPs of TEX15 tested, a slight increase in the frequency of the alleles previously associated was observed, but far from being significant, possibly due to the small size of the cohort (results summarized in Supplementary Material, Table S5).
Discussion
Studying a consanguineous Turkish family, we have identified, using WES, a nonsense mutation in TEX15 leading in early translational termination in the first quarter of the protein and causing an infertility phenotype. Our data strongly suggest that the identified homozygous mutation in TEX15 in three infertile brothers is the cause of a spermatogenic defect that is transmitted according to an autosomal recessive trait. Interestingly, the human phenotype is very similar to the one observed in the Tex15 KO mouse. In mice, disruption of the gene causes a drastic testis size reduction and meiotic arrest in the first wave of spermatogenesis in males. Further analysis revealed that in mice Tex15 is required for chromosomal synapsis. It regulates the loading of DNA proteins onto sites of double-strand DNA breaks, and thus its absence causes a failure in meiotic recombination (23). Interestingly, female KO mice were fertile. In man, we also observed a drastic testis size reduction and a meiotic maturation arrest. One major difference between the phenotype in man and in mice is that all male KO mice for Tex15 are infertile, while in the family studied here two brothers were NOA while the third one was initially diagnosed as SO and became azoospermic with time. The youngest brother with an unknown fertility status was also homozygous for the mutation (Fig. 1). Despite a detailed explanation of the medical issue to the youngest brother, we were unable to convince him to have a spermiogram, which could have prompted him to cryopreserve some of his spermatozoa. Indeed, considering that the younger infertile mutated brother was SO and he became azoospermia with time, it is likely that the phenotype becomes more manifest with age.
In both NOA patients, some foci of spermatogenesis could be found via TESE. Therefore, it remains to be established whether mutations in TEX15 correlate with a decrease in sperm count over time.
Although patient IV:3 displayed a maturation arrest, a very limited number of spermatozoa were extracted in the biopsy material, but no pregnancy could be achieved after ICSI. The mosaic appearance of the testis is well established, and a single small biopsy material may not always be representative of the entire testis picture (28). In a recent study, a mosaic pattern of spermatogenesis was found in 13.79% of men with maturation arrest (29). Mutation testing of TEX15 could also be offered and possibly used as an indicator of sperm retrieval with a more detailed search in biopsy material for patients with maturation arrest.
Considering that the mutation introduced a premature stop codon, either it can lead to the production of a truncated protein missing 74.5% of its sequence including two predicted domains or the protein may be entirely absent due to a nonsense-mediated mRNA decay. Since TEX15 protein is testis-restricted, we could not, for ethical reasons, get access to testis biopsies performed only for research purposes in order to verify the production of a truncated form of TEX15 or its absence.
The identification and characterization of the genetic bases of male infertility have large implications not only for understanding the cause of infertility but also in determining the prognosis, selection of treatment options and management of such couples. It allows the development of a molecular diagnostic test for patients with a similar phenotype.
Materials and Methods
Patient recruitment
This project has been approved by the Comité de Protection de la Personne (CPP) of Strasbourg University Hospital, France (CPP 09/40—W AC-2008-438 1W DC-2009-I 002), as well as by the ‘Istanbul University, Faculty of Medicine, Ethics committee for clinical research Faculty of Medicine’ (2012/1671-1265). A written consent had been obtained for all participants before samples were collected.
The family comprised six brothers; three infertile, two fertile, one of unknown fertility status and two fertile sisters. All affected brothers had a normal karyotype, and no Y-chromosome micro-deletions were found.
A total of 85 unrelated, Turkish individual NOA cases and 13 cases of SO were also collected. All patients were diagnosed through a complete andrological and urological diagnostic workflow for couple infertility, as they were seeking medical assistance for achieving pregnancy. NOA cases were selected based on two spermiograms, which were performed with at least a 15–20 days interval, and no spermatozoa were detected after concentration. Similarly, SO cases were selected with a threshold of <1 × 106 spermatozoa/ml after semen analysis. NOA cases were sorted following histological scoring of testis biopsy. Exclusion criteria were patients with mono or bilateral cryptorchidism, varicocele, previous testis trauma, mixed azoospermia (obstructive associated with non-obstructive), recurrent infections, iatrogenic infertility, hypogonadotrophic hypogonadism, karyotype anomalies or Y chromosome micro-deletions. A total of 107 fertile Turkish males who fathered at least one child naturally were included in the study as a control group.
DNA extraction
As a DNA source, saliva samples from family members, all controls and SO cases were used, whereas blood samples were collected from individual NOA cases. Genomic DNA was extracted from saliva using Oragene DNA self-collection kit (DNA genotek, Ottowa, Canada) or from peripheral blood using IPrep™ Pure Link® gDNA Blood kit (Invitrogen, CA, USA), according to the manufacturer's instructions.
Whole exome sequencing
Whole exome sequencing of patients IV:1 and IV:3 (Fig. 1A) was performed by the Institute of Genetics and Molecular and Cellular Biology (IGBMC) microarray and sequencing platform, member of the ‘France Génomique programme’. For this purpose, ∼1 µg DNA was sheared to 150–200 bp with Covaris Sonolab v4.3.3 (Covaris, Woburn, MA, USA). Fragments were subjected to library preparation, and for each sample to be sequenced, an individual indexed library was prepared. DNA libraries were then enriched (SureSelectXT2 Target Enrichment System) and sequenced with Illumina Hiseq 2500 following Illumina's instructions.
Data processing and analysis
Image analysis and base calling were performed using CASAVA v1.8.2 (Illumina).
Bad-quality parts of reads (Phred score < 10) were trimmed off using SolexaQA (v.2.0) (30), and the reads were then mapped onto the reference genome GRCh37/hg19 using BWA (v0.7.5a) (31). Duplicate reads were marked using Picard (v1.68). Realignment around indels and base quality recalibration were performed using GATK (v2.5-2) (32) following developer's recommendations. Samtools (v0.1.19) (33) was used to exclude multi-mapped reads from downstream analysis. Variant calling was done using GATK UnifiedGenotyper (v2.5-2) (32). Variant quality score recalibration was done using GATK (v2.5-2) (32) to assign a well-calibrated probability to each variant call in a call set.
Detected variants were ranked by VaRank (34), which incorporates the annotations retrieved by the Alamut Batch software (Interactive Biosoftware, France) as well as allele frequency from our internal exome database. The annotation took into account the functional annotation and external data such as the HGVS nomenclature (genomic, cDNA and proteic), dbSNP (http://www.ncbi.nlm.nih.gov/SNP/), 1000Genomes (http://www.1000genomes.org/) and the NHLBI Exome Variant Server (http://evs.gs.washington.edu/EVS/).
Mutation screening and case–control association study
The variant and the segregation in the family as well as the absence of variation in healthy matched Turkish controls were confirmed through Sanger sequencing. The identified mutation was submitted to ClinVar NCBI database (http://www.ncbi.nlm.nih.gov/clinvar/). All exons and exon/intron boundaries of TEX15 (NM_031271.3) were amplified and sequenced for validating variants and excluding other mutations in a group of Turkish infertile men with similar phenotypes (primers listed in Supplementary Material, Table S4A). DNA amplicons were purified, and double-strand sequencing of each DNA fragment was performed by GATC (Cologne, Germany).
For association studies between TEX15 SNPS and infertility, allelic distributions between control (n = 107) and infertile (n = 85) individuals were compared using Fisher's exact test (2BY2 Statistical Genetics Utility program, Rockfeller University, New York).
RT-PCR and immunohistology
Total RNA samples from different human tissues were purchased from Clontech (BD Biosciences Clontech, Palo Alto, CA); 20 RNA samples were included in Human Total RNA Master Panel II (Cat. 636643), and ovary RNA was obtained from Human Ovary Total RNA (Cat. 636555).
Human testicular tissue samples were obtained from adult patients undergoing a vasectomy reversal in the fertility centre of the UZ Brussels. During surgery, a testicular tissue sample measuring about 50 mm3 was taken to evaluate the spermatogenic status. A small piece of this testicular sample was used for research purposes after written informed consent. The sample was transported to the laboratory on ice, washed to remove any residual blood and cryopreserved according to uncontrolled slow freezing (USF) method using 1.5 M dimethylsulphoxide (DMSO) and 0.15 M sucrose as cryoprotectant (35). One piece of tissue was fixed in acetic acid/formaldehyde/alcohol (AFA) (VWR, Belgium) for at least 1 h and transported to the anatomo-pathology department of the UZ Brussel for subsequent fixation and embedding in paraffin.
Prepubertal tissue was obtained from a patient who underwent testicular tissue banking as part of a fertility preservation programme. A maximum of 10% of the biopsied tissue was donated to research. One fragment of the prepubertal tissue was fixed in AFA, while the other fragments were cryopreserved.
Human testicular tissue was thawed as described before (35). RNA was extracted using the RNeasy Kit (Qiagen, Venlo, The Netherlands) in combination with the Qiagen Shredder Kit following the manufacturer's recommendations. On column, DNase treatment (RNase-free DNase Set; Qiagen) was performed for all the samples. The total amount of RNA was measured with a NanoDrop spectrophotometer ND-1000 (NanoDrop Technologies, Inc., Wilmington, DE). The RNA was then stored at −80°C for later use.
cDNA synthesis was done according to the Invitrogen protocol starting from 1 µg RNA for pooled human tissue and 40 pg RNA for testicular tissue samples. DNase treatment was applied to each reaction in order to remove DNA contamination. Same samples were amplified in parallel with primers for the control housekeeping gene GADPH (MIM 138400), and individual controls without reverse transcriptase for each sample were included in polymerase chain reaction (PCR) reactions (primers for RT-PCR listed in Supplementary Material, Table S4B). Thirty cycles of RT-PCR were followed by gel electrophoresis and ethidium bromide staining.
For immunohistochemical evaluation, paraffin-embedded samples were sectioned at 5 μm. Sections were deparaffinized in xylene (3 × 10 min) and subsequently rehydrated in a decreasing ethanol series (2 × 100, 90 and 70% isopropanol). After washing with phosphate-buffered saline (PBS; 70011-036; Life-technologies, Gent, Belgium) for 5 min, endogenous peroxidase activity was blocked by incubating the tissue sections with hydrogen peroxide (0.3% H2O2 in methanol). After washing in PBS, a heat-mediated antigen retrieval was performed by incubating the slides in 0.01 M citric acid in a water bath (95°C) for 75 min in order to break the methylene bridges, expose the antigenic sites and allow the antibodies to bind. In order to avoid non-specific background staining, sections were incubated with CAS BlockTM blocking agent (008120; Life-technologies).
After the blocking steps, TEX15 primary antibodies (1/200; HPA036800; Sigma) were applied to the sections and incubated in a humidified chamber. No primary antibody was applied to the negative control.
After three wash steps, sections were incubated with a peroxidase-labelled secondary antibody (Dako Real Envision Detection System; K5007; Dako, Heverlee, Belgium) for 1 h at room temperature. After washing and visualisation with 3, 3′-diaminobenzidine (DAB; Dako Real EnvisionTM Envision System), slides were counterstained with haematoxylin.
Funding
The study was funded by Agence Nationale de la Recherché (ANR-11-BSV2-002 ‘TranspoFertil’) and l'Agence de BioMédecine (‘AMP, diagnostic prénatal et diagnostic génétique’). This work was supported by the French Centre National de la Recherche Scientifique (CNRS), Institut National de la Santé et de la Recherche Médicale (INSERM), the Ministère de l'Education Nationale et de l'Enseignement Supérieur et de la Recherche, the University of Strasbourg and the University Hospital of Strasbourg.
Acknowledgements
We would like to thank Stéphanie Legras for her help in the analysis of the sequencing data and Robert Drillien for his critical reading of the manuscript. We are grateful to the IGBMC platforms.
Conflict of Interest statement. None declared.
References
