To analyze gene and protein expression patterns of Samd4A/SAMD4A in various mice tissues, RT-PCR and Western blot analysis were conducted (at least triplicates for each tissue). The effects of gene knockout were investigated after generating Samd4a knockout mice as described below and by analyzing testes samples (n = 5). All animal care and procedures were approved by the Institutional Animal Care and Use Committees (IACUCs) of Sunchon National University (protocol number: SCNU IACUC-2022–10) and The Ohio State University (protocol number: 2007A0183). Mice were raised under ad libitum feeding conditions in mice housing facilities before euthanization via carbon dioxide inhalation followed by cervical dislocation. For isolation of murine total RNAs, the testis, muscle, liver, brain, lung, kidney, adipose tissue, heart, spleen, and small intestine were collected from three-month-old FVB mice using Trizol reagent (Invitrogen, Carlsbad, CA, United States) as described previously (Ahn et al., 2018). Total RNAs from the adult human testis, muscle, liver, brain, lung, kidney, thymus, and heart were purchased from Agilent Technologies (Santa Clara, CA, United States) and adult human RNA from adipose tissue was purchased from Clontech Laboratories (Mountain View, CA, United States). Three-month-old C57BL/6 mice were euthanized for protein extraction from the same tissues as described in our reports (Ahn et al., 2018).

CRISPR guide RNA (gRNA) (5′ - CTG​ACG​ATA​AGC​TCA​ATG​GC 3′) was designed to target mouse Samd4a exon three of A-through C-forms and exon two of D- and E-forms. Restriction by the gRNA and Cas9 protein was validated in vitro by incubation of the reaction mixture including gRNA, Cas9 protein, substrates, 10x BSA, and buffer at 37°C for 1 h 30 min, addition of RNase and incubation at 37°C for 20 min, and addition of stop solution and incubation at 37°C for 20 min, followed by separation on a 2% agarose gel. The gRNA was microinjected into 1-cell stage mouse embryos. Twenty-nine pups were born and genotyped by PCR using a primer set (forward: 5ʹ -ACC​GAT​CCC​ATC​CAC​AGT​TGA - 3ʹ and reverse: 5ʹ - ACC​AAA​CCC​GAA​TGC​GTT​CCA - 3ʹ). This genotyping and subsequent sequencing revealed that four F₀ founders had eight bp deletion which gave rise to frameshift mutation. Those four F₀ founders were mated to produce heterozygous and homozygous Samd4 knockout mice. Sequencing analysis of the litters was conducted for the target site to confirm the heterozygous and homozygous mutant mice.

For analyses of bulk RNA-sequencing (RNA-seq), raw data were retrieved from public databases and processing was performed as described previously in our report (Ahn et al., 2021). In detail, the raw FASTQ files were checked with FastQC v0.11.7 and then adapter-trimmed and filtered by Trimmomatic v0.38 with default parameters (Bolger et al., 2014). Cleaned reads were quantified against decoy-aware mouse and human transcriptome using Salmon v1.3.0 through the mapping-based mode (Patro et al., 2017). The gene expression values in transcripts per million (TPM) were collected from the Salmon output files (quant.sf) and imported into the R software v4.1.3 for boxplot visualization.

To measure the quantity of RNA, a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE) was used. The RNA samples were stored at -80 °C until use. Approximately 1 µg of RNA was reverse-transcribed to cDNA in a 20 µl total reaction using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen). The thermal cycle of the reverse transcription was 65°C for 5 min, 37°C for 52 min, and 70°C for 15 min. Exactly 1 µl of cDNA samples was used as templates for PCR in a 25 µl total reaction with AmpliTaq Gold DNA polymerase (Applied Biosystems, Carlsbad, CA, United States). The conditions for this reaction were 95°C for 1 min 30 s 33 cycles of 94°C for 30 s, 55°C for 1 min, 72°C for 1 min, with an additional extension step at 72°C for 10 min. PCR products were separated by using 1% agarose gel electrophoresis. Forward and reverse primers for amplifying murine Samd4a and human SAMD4A transcripts are listed in Supplementary Table S1. DNAs extracted from the gel were sequenced by The Ohio State University Sequencing Core Facility using an Applied Biosystems 3730 DNA analyzer (Foster City, CA, United States).

The construction of expression vectors containing murine Samd4a isoforms was conducted as follows. Murine Samd4a C- and E-forms were amplified using sets of forward (m-h-SAMD4A/F3-full and mSamd4a/F2, respectively) and reverse (m-h-SAMD4A/R1) primers (Supplementary Table S1). Both isoforms were then cloned into a TA cloning vector, pCR2.1 (Invitrogen). After digestion of the cloning vectors with BamHI and NotI, the two fragments of C- and E-forms were ligated into an expression vector, pcDNA3.1 (Invitrogen) via the same enzyme sites. The expression vectors were transfected into HEK293 cells to overexpress the C- and E-forms using lipofectamine 2000 (Invitrogen) and collect positive protein controls 2 days after transfection. Western blotting was conducted as described in our previous report (Ahn et al., 2018). In detail, protein extracts from mouse tissues and positive controls (C- and E-forms) were separated on 12% SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Amersham Biosciences Hybond-P; Amersham Biosciences, Piscataway, NJ, United States). The membranes were blocked for 30 min in 4% nonfat dry milk in 1× Tris-buffered saline containing 0.05% Tween-20 (TBST) and then incubated overnight at 4°C with a rabbit polyclonal SAMD4A primary antibody raised against the C-terminal end (HPA065309, Atlas antibodies, Stockholm, Sweden). Next day, the membranes were washed with 1×TBST and incubated with the corresponding HRP-conjugated secondary antibody (rabbit IgG; HAF 008, R&D Systems, Minneapolis, MN, United States) for 1 h at room temperature. After washing with 1×TBST, Amersham ECL plus Western Blotting Detection Reagents (GE Healthcare Biosciences, Pittsburgh, PA, United States) was applied on the membrane, and the blots were developed onto Hyperfilm (GE Healthcare Biosciences).

For analyses of developmental stages of testicular cells, original BAM files generated by 10x Genomics platform and submitted to the NCBI SRA were retrieved from Amazon Web Service (AWS) (Accession number GSE109033 for the mouse; GSE109037 for the human). These single-cell RNA-seq (scRNA-seq) data from testicular cells were converted to FASTQ files using the bamtofastq utility (v1.3.1) in order to proceed with the 10x Genomics’ Cell Ranger pipeline. The cellranger count was used to align sequencing reads to the reference transcriptome (mm10 and GRCh38) for each sample. Using the cellranger aggr function, the outputs of the cellranger count for individual samples were combined with batch correction. These processed data were imported to the R package Seurat v4.1.0 for data normalization and unsupervised cell clustering, and UMAP analyses were performed based on statistically significant principal components (Satija et al., 2015; Hao et al., 2021). After identification of cell clusters based on the top 10 differentially expressed genes and stage-specific marker genes listed previously (Hermann et al., 2018), raw count matrices were imported to Monocle3 v1.0.0 (Trapnell et al., 2014; Cao et al., 2019) for dynamic trajectory analyses which aligned cells in pseudotime. The numbers of cells analyzed in this study are listed in Supplementary Table S2.

After harvesting and photographing testes and measuring weights, the whole testes of wild-types, heterozygous Samd4a knockouts, and homozygous Samd4a knockouts were processed for histological analysis. In particular, testes were fixed and embedded in paraffin followed by sectioning into 5 µm slices. After deparaffinization, the testicular tissue sections were subjected to hematoxylin and eosin (H and E) staining to examine morphological changes of testicular cells in the seminiferous tubules. Periodic Acid-Schiff Reaction (PAS) stain for histochemical assay of the testes was processed for determining normal and abnormal stages of the seminiferous epithelia cycle. The terminal deoxyribonucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay was applied to detect apoptotic cells in testes. Photographing was performed using a camera (SM-N920L, Samsung, South Korea). Microscopic images of stained slides were captured by an imaging system, DP73 from Olympus or AxioCam MRc5 from Zeiss.

To test the normality of the data, the Shapiro-Wilk normality test was conducted with log-transformed or non-transformed expression values. For non-normal data, the non-parametric Kruskal–Wallis test was performed followed by the Dunn’s post-hoc test. For normal data, one-way ANOVA was conducted followed by the Tukey’s post-hoc test. Student’s t-test in R was used to compare two means. The minimum level of significance was set as p < 0.05. All statistical tests were performed using R packages, stats v4.1.3 and FSA v0.9.3.

Analyses of RNA-seq data revealed that the total expression level of Samd4a in the murine testis (∼47.5 TPM) was markedly high with an approximately 8.4-fold higher expression compared to an average TPM (∼5.7) of the rest of tissues (Figure 1A). Among analyzed transcript isoforms, one isoform (NCBI RefSeq accession: NM_001163433.1), E-form, was identified as testis-specific with an approximately 350-fold higher expression in the murine testis compared to an average TPM of the rest of tissues (Figure 1B), unlike other isoforms with variable expression levels in several tissues (Supplementary Figure S1). Using semiquantitative RT-PCR, transcripts were amplified to verify alternative splicing with four primer sets, mF1-mR1, mF2-mR1, mF3-mR2, and mF4-mhR1 (Figure 1C). As a result, with the first primer set (mF1-mR1), expression of Samd4a C-form was detected in all analyzed tissues including the testis, and the A- and B-forms were not observed in the testis (Figure 1D). On the other hand, using the second primer set (mF2-mR1), Samd4a E-form (1,065 bp) was detected only in the testis which contains the sterile alpha motif (SAM) domain, and the D-form (1,215 bp) was deficient in all analyzed tissues (Figure 1D). The F-form amplified using the third primer set (mF3-mR2) and the total transcript amplified using the fourth set of primers (mF4-mhR1) appeared to be upregulated in the testis (Figure 1D).

To further investigate expression patterns of Samd4a isoforms in different mouse tissues, Western blot analysis was performed including over-expressed Samd4a C- and E-forms. The Samd4a E-form protein (approx. 67 kDa, 610 amino acid residues) was predominantly expressed in the murine testis, and the C-form (approx. 69 kDa, 623 amino acid residues) was barely detectable in the testis and lung (Figure 1E). Other murine isoforms [A-form (approx. 84 kDa, 761 amino acid residues), B-form (approx. 78 kDa, 711 amino acid residues) and F-form (approx. 33 kDa, 302 amino acid residues)] were not detected. The murine E-form contains an intact SAM domain (63 amino acid residues, NCBI accession: cd09557) which binds to the Smaug recognition element in mRNAs, but the C-form lacks eight amino acids in the beginning of the SAM domain due to skipping of the exon corresponding to the exon three of the E-form (Figure 1C). In the human, the SAMD4A C-form protein (approx. 68 kDa, 617 amino acid residues) contains the 100% homologous SAM domain (SupplementaryFigure S2). Our data clearly showed that E-form mainly expressing in testis is predominant among different isoforms of Samd4A in mice.

Our analyses of GTEx dataset resulted in a median-based identification of the testis as the first-ranked tissue of SAMD4A expression among 47 human tissues with sample sizes ranged from 35 to 560 (Supplementary Figure S2A). Within reproductive human tissues, including the epididymis (caput, corpus, and cauda), and testis, expression levels of TPM of both the total and C-form were 9.6-fold and 21.3-fold higher in the testis compared to averages of the rest of reproductive tissues, respectively (Supplementary Figure S2B,C). The C-form in the human containing the full length of SAM domain (Supplementary Figure S2D) can be an orthologous alternative splicing isoform of the mouse Samd4a E-form. The data from RT-PCR with four different primer sets for the human (hereafter, parenthesized with abbreviations of forward and reverse primers) showed ubiquitous expression of A- and B-forms (mhF3-hR1), predominant C-form expression in the testis (hF4-hR1), as well as, testicular expression of D-form (hF4-hR1), E-form (hF2-hR1), and the total transcript (hRealF1-mhR1) (Supplementary Figure S2E). These alternative splicing patterns obtained from the NCBI Gene database were rechecked in the BLAT web page (https://genome.ucsc.edu). In summary, the SAMD4A C-form transcript exhibited a higher expression in the testis of the human compared to other isoforms, which was similar in the mouse, suggesting the conserved expression of the orthologous isoforms (Samd4a E-form and SAMD4A C-form) containing the SAM domain.

To identify murine cell types in the testis which express Samd4a, expression values at multiple developmental stages were analyzed. Based on a microarray dataset (GSE4193), the Samd4a expression was substantially upregulated in round spermatids (Figure 2A). Also, according to GSE926, in the developing mouse testis, Samd4a expression was markedly increased in ages between 30 and 35 days postpartum (dpp) (Figure 2B) which correspond to a period before the formation of spermatozoa (mature sperm) begins around 35 dpp (Itman et al., 2006). In addition, a UMAP projection of integrated single-cell transcriptome revealed 10 clusters of murine testicular cells including germ cells and somatic cells (Figure 2C). Given the clusters, the expression of Samd4a was projected to be sequestered in early, mid-, and late round spermatids (Figure 2D). Furthermore, pseudotime profiles were examined for marker genes known to be expressed in spermatocytes (Sycp3) and spermatids (Prm1 and Tnp1) (Hermann et al., 2018) (Figure 2E). According to cell density and regression lines, the expression pattern of Samd4a was similar to that of the spermatid markers which continued to increase along the pseudotime, but Samd4a expression slightly decreased from the late pseudotime around 40 after passing the plateau (Figure 2E). Last, compared to markers of early (Speer4e), mid- (Acot10), and late (1700027A15Rik) round spermatids (Hermann et al., 2018), the Samd4a expression tended to spread throughout the transition from early to late round spermatids (Figure 2F). Taken together, the expression of murine Samd4a was round spermatid-specific among the testicular cells.

Comparison of SAMD4A expression values of various human cell types based on processed single-cell RNA-seq data from the Human Protein Atlas revealed that, among 76 analyzed cell types including testicular cells, the SAMD4A expression was markedly high in early and late spermatids (Figure 3A). To confirm the spermatid-specific expression of SAMD4A, single-cell RNA-seq data of human testicular cells were analyzed and, as a result, these cells were clustered into 10 groups in a UMAP plot (Figure 3B). As with the murine Samd4a, the expression of human SAMD4A was specific to early through late round spermatids (Figure 3C). Further pseudotime analyses revealed analogous expression patterns between the spermatid markers (PRM1 and TNP1) and SAMD4A, compared to the spermatocyte marker (SYCP3) (Figure 3D). Also, unlike human spermatid stage-specific markers (C17orf98 for early, PRSS58 for mid, and FSCN3 for late round spermatids) (Hermann et al., 2018), the SAMD4A expression covered early through late round spermatids (Figure 3E). Overall, the round spermatid-specific expression of human SAMD4A was similar to that of murine orthologous Samd4a, suggesting their conserved expression patterns.

By targeting exon two of testis-specific E-form of Samd4a using the CRISPR/Cas9 system, 8-nucleotide deletion was achieved, resulting in a frameshift mutation that created a premature stop codon before the SAM domain (Figure 4A). The sizes of testes in homozygous Samd4a knockouts (Ho) were substantially reduced compared to the wild-types (WT) and heterozygotes (He) (Figure 4B). The average weight of testes from Ho (25.16 mg ± 2.20) was approximately 3.3 times lighter than those of the WT (83.34 mg ± 3.52) and He (82.40 mg ± 6.01) (Figure 4C). Only the Ho knockouts exhibited shrinkage of seminiferous tubules (41,348μm² ± 2817 for WT, 40,243μm² ± 2019 for He, and 18,342μm² ± 412 for Ho, Supplementary Figure S3) and abnormal morphologies of germ cells in the lumen within the seminiferous tubules (Figure 4D). Representative H&E-stained images further revealed that the lumen of seminiferous tubules of Ho knockout mice was filled with multi-nucleated giant germ cells (MGCs) and degenerated germ cells (DGs) including vacuolated or enlarged spermatids, pyknotic spermatids, and apoptotic spermatids with small nuclear fragments (Figure 4D). Also, unlike the WT and He knockout mice, spermatozoa were absent in the lumen of seminiferous tubules from the Ho knockout mice (Figure 4D). It suggested that Samd4a deficiency could lead to morphological abnormalities in the mouse testis.

Based on our findings on the absence of spermatozoa in homozygous Samd4a knockout mice, we further investigated which stages of spermatogenesis were affected by Samd4a deficiencies using PAS-stained testis sections. From the Stage I through V, it appeared that there were no differences in the types of germ cells from spermatogonia to round spermatids between the wild-type and homozygous knockout mice (Figure 5A). In Stages VI-VIII, MGCs appeared first in the homozygous knockout mice, implying the role of Samd4a at this stage. From Stage IX, elongated spermatids developed in the wild-type mice; whereas, round spermatids failed to develop into elongated spermatids in homozygous knockout mice. This absence of elongated spermatids in the homozygous knockout mice was maintained until Stage XII. During the stages without elongated spermatids in the testes of homozygous knockout mice, MGCs and DGs were observed unlike the wild-type mice. In addition, the TUNEL assay revealed that, in the homozygous Samd4a knockout mice, the number of apoptotic cells was significantly greater than that in the wild-type mice (p < 0.001) (Figure 5B). Taken together, it indicated that Samd4a deficiency in mice resulted in impairment of spermatogenesis, especially spermiogenesis due to abnormalities in testicular germ cells.

Azoospermia characterized by no sperm in the ejaculate, is generally separated into two conditions, obstructive azoospermia (OA) and non-obstructive azoospermia (NOA). The distinction between OA and NOA is that normal spermatogenesis occurs in OA, but not in NOA. As the SAMD4A is mainly expressed in late-spermatogenic stages, it was investigated whether SAMD4A expression can reflect different types of cells at late stages of spermatogenesis between OA and NOA patients. Expression levels of PLPPR3, as a marker gene for spermatogonia stem cells (Sohni et al., 2019), were not different between OA and NOA patients. However, expression levels of spermatid marker genes (TNP1 and PRM1) (Kleene, 2013; Hermann et al., 2018) and a sperm marker gene (ACRV1) (Foster et al., 1994) were significantly lower in the NOA than OA (Figure 6A). Also, SAMD4A expression was dramatically reduced in the testes of NOA patients compared to the OA (p < 0.001) (Figure 6A).

The condition of NOA is subcategorized into pre-meiotic arrest (NOA-pre), meiotic arrest (NOA-mei), and post-meiotic arrest (NOA-post). NOA-pre and NOA-mei patients have spermatogonia and/or spermatocytes but not spermatids, whereas, NOA-post patients have spermatogonia, spermatocytes, and spermatids in testes. The NOA-pre and NOA-mei conditions were compared using a dataset (GSE108886) and NOA-mei and NOA-post using another dataset (GSE45885). Although comparisons of NOA-pre and NOA-mei revealed that there was no expression difference in late stages of spermatogenesis (TNP1, PRM1 and ACRV1), both groups have a significantly lower expression of those genes compared to the OA (Figure 6B). The SAMD4A expression was also significantly lower in testes of both NOA-pre and NOA-mei, compared to the OA (Figure 6B). Next, comparisons of NOA-mei and NOA-post revealed that NOA-mei showed lower expression levels of the late marker genes than those of NOA-post which were statistically comparable to the control levels (Figure 6C). Similarly, the expression level of SAMD4A was significantly higher in the NOA-post than the NOA-mei, but statistically comparable to the control (Figure 6C).