Transcriptomic analysis of the non obstructive azoospermia noa to address gene expression regulation in human testis

Systems Biology in Reproductive Medicine

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/iaan20

Transcriptomic analysis of the Non-Obstructive
Azoospermia (NOA) to address gene expression
regulation in human testis

Govindkumar Balagannavar, Kavyashree Basavaraju, Akhilesh Kumar
Bajpai, Sravanthi Davuluri, Shruthi Kannan, Vasan S. Srini, Darshan S.
Chandrashekar, Neelima Chitturi & Kshitish K. Acharya

To cite this article: Govindkumar Balagannavar, Kavyashree Basavaraju, Akhilesh Kumar
Bajpai, Sravanthi Davuluri, Shruthi Kannan, Vasan S. Srini, Darshan S. Chandrashekar,
Neelima Chitturi & Kshitish K. Acharya (2023) Transcriptomic analysis of the Non-Obstructive
Azoospermia (NOA) to address gene expression regulation in human testis, Systems Biology in
Reproductive Medicine, 69:3, 196-214, DOI: 10.1080/19396368.2023.2176268
To link to this article: />
View supplementary material

Published online: 08 Mar 2023.

Submit your article to this journal

Article views: 229

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

/>
SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE
2023, VOL. 69, NO. 3, 196–214
/>
RESEARCH ARTICLE

Transcriptomic analysis of the Non-Obstructive Azoospermia (NOA) to
address gene expression regulation in human testis

Govindkumar Balagannavara,b, Kavyashree Basavarajua,c, Akhilesh Kumar Bajpaic, Sravanthi Davuluric,
Shruthi Kannana, Vasan S. Srinid, Darshan S. Chandrashekara, Neelima Chitturic, and Kshitish K. Acharyaa,c

aInstitute of Bioinformatics and Applied Biotechnology (IBAB), Bengaluru, Karnataka, India; bResearch Scholar, Manipal Academy of
Higher Education (MAHE), Manipal, Karnataka, India; cBdataA: Biological data Analyzers’ Association (virtual organization http://
startbioinfo.com/BdataA/), India; dManipal Fertility, Manipal Hospital, Bengaluru, Karnataka, India

ABSTRACT ARTICLE HISTORY
Received 9 August 2022
There is a need to understand the molecular basis of testes under Non-Obstructive Revised 27 January 2023
Azoospermia (NOA), a state of failed spermatogenesis. There has been a lack of attention to Accepted 30 January 2023
the transcriptome at the level of alternatively spliced mRNAs (iso-mRNAs) and the mechan-
ism of gene expression regulation. Hence, we aimed to establish a reliable iso-mRNA profile KEYWORDS
of NOA-testes, and explore molecular mechanisms – especially those related to gene expres- Spermatogenesis; non-
sion regulation. We sequenced mRNAs from testicular samples of donors with complete obstructive Azoospermia;
spermatogenesis (control samples) and a failure of spermatogenesis (NOA samples). We transcription regulation;
identified differentially expressed genes and their iso-mRNAs via standard NGS data analy- transcription factors;
ses. We then listed these iso-mRNAs hierarchically based on the extent of consistency of dif- regulatory network; gene
ferential quantities across samples and groups, and validated the lists via RT-qPCRs (for 80 expression regulation; RNA-
iso-mRNAs). In addition, we performed extensive bioinformatic analysis of the splicing fea- seq; molecular interactions;
tures, domains, interactions, and functions of differentially expressed genes and iso-mRNAs. transcriptomics; male

Many top-ranking down-regulated genes and iso-mRNAs, i.e., those down-regulated more infertility
consistently across the NOA samples, are associated with mitosis, replication, meiosis, cilium,
RNA regulation, and post-translational modifications such as ubiquitination and phosphoryl-
ation. Most down-regulated iso-mRNAs correspond to full-length proteins that include all
expected domains. The predominance of alternative promoters and termination sites in
these iso-mRNAs indicate their gene expression regulation via promoters and UTRs. We
compiled a new, comprehensive list of human transcription factors (TFs) and used it to iden-
tify TF-’TF gene’ interactions with potential significance in down-regulating genes under the
NOA condition. The results indicate that RAD51 suppression by HSF4 prevents SP1-activa-
tion, and SP1, in turn, could regulate multiple TF genes. This potential regulatory axis and
other TF interactions identified in this study could explain the down-regulation of multiple
genes in NOA-testes. Such molecular interactions may also have key regulatory roles during
normal human spermatogenesis.

Abbreviations: NOA: Non-Obstructive Azoospermia; OA: Obstructive Azoospermia; VA:
Varicocele; iso-mRNAs: alternatively spliced transcript isoforms; NGS: Next-Generation
Sequencing; scRNAseq: single-cell RNA sequencing; TFgene: transcription-factor-coding
gene; Tu250: top up-regulated 250; Td250: top down-regulated 250; Tu3000genes: Top 3000
up-regulated; Td3000genes: Top 3000 down-regulated; GO: Gene Ontology; Bpv: Benjamini
P-value; AP: Alternative promoter; AT: Alternative terminator; AP/AT: Alternative promoter or
terminator; A5’-SS: Alternative 5’ Splice Sites; A3’-SS: Alternative 3’ Splice Sites; RI: Intron
Retention; SE: Skipping Exon; sub_RI: Skipping Exon; SE/sub_RI: Skipping Exon/Sub-Intron
Retention; SPC: Single protein coding; TFs: Transcription Factors; TFgenes: Trancription-fac-
tor-coding-genes; cTFs: co-Transcription Factors; RT-qPCR: Reverse transcription-quantitative
polymerase chain reaction

Introduction approximately 3 to 9 men in every 1000 may have the
Non-Obstructive Azoospermia (NOA) disorder. NOA
The proportion of men with infertility may have is difficult to deal with clinically (Esteves 2015; Chiba
increased in recent years (Agarwal et al. 2015; Kumar et al. 2016). Understanding the genetic and molecular

and Singh 2015). A literature review indicated that

CONTACT Kshitish Acharya Institute of Bioinformatics and Applied Biotechnology (IBAB), Bengaluru, Karnataka 560 100, India
Supplemental data for this article can be accessed online at />
ß 2023 Informa UK Limited, trading as Taylor & Francis Group

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 197

basis of the NOA condition can help to explore for mass-scale differential expression profiling of
potential novel solutions to address the issues associ- NOA-testes (e.g., Okada et al. 2008; Cappallo-
ated with this disease (Omolaoye et al. 2022). Obermann et al. 2010; Malcher et al. 2013; Baksi et al.
Identifying alternatively spliced transcript isoforms 2018). However, microarray-based gene expression
(iso-mRNAs) with such associations is particularly profiling studies, particularly the earlier ones, have
essential (Modrek and Lee 2002; Stastna and Van Eyk several limitations – including coverage of a limited
2012) for beginning a system-level understanding of number of genes among the probes used in the array
the disease condition within the testis tissue. (Draghici et al. 2006; Jaksik et al. 2015).

The non-obstructive azoospermia condition repre- In recent years, the Next-Generation Sequencing
sents a failed state of spermatogenesis, with the reduc- (NGS) methods have been extensively used for tran-
tion or absence of multiple types of male germ cells scriptomics (Lea et al. 2011; Hrdlickova et al. 2017) as
in the testes, particularly the late spermatid and sper- they have the potential to screen a maximum number
matozoa that are either absent or present in highly of genes as well as identify key iso-mRNAs (Ozsolak
reduced numbers (Ramasamy et al. 2009; Esteves and Milos 2011). It has been well-established that
2015; Chiba et al. 2016). The significance of transcrip- alternative splicing forms an important component in
tomics in understanding the molecular basis of sperm- spermatogenesis and testicular disorders (Song et al.
atogenesis and male infertility conditions is also 2020). However, exploring the transcriptome at the
well-known (Omolaoye et al. 2022). Thus, a compre- level of specific mRNA isoforms requires high-depth
hensive list of genes and transcripts expressed differ- NGS of a reasonable number of samples (https://gen-
entially in the testes of NOA patients may also ome.ucsc.edu/ENCODE/protocols/dataStandards/
include several molecules playing a key role in normal ENCODE_RNAseq_Standards_V1.0.pdf;

human spermatogenesis. We think that the genes con- Giannopoulou et al. 2015), which has not been done
sistently down-regulated in NOA could particularly till date using either bulk (non-single-cell) tissue RNA
help identify novel molecules potentially associated or scRNA from the testis under the NOA condition.
with spermatogenesis in men. Research efforts towards
developing new and improved contraceptives for men In recent years, single-cell RNA sequencing
(Robertson et al. 2020; Dominiak et al. 2021) could (scRNAseq) has gained popularity. Normal human
also be assisted by such a list of genes and their corre- testis samples have been subjected to scRNAseqs
sponding iso-mRNAs. We propose that a hierarchical (Zhu et al. 2016; Jan et al. 2017; Guo et al. 2018;
listing of testicular iso-mRNAs, based on their Hermann et al. 2018; Wang et al. 2018; Sohni et al.
strength of expression-based association with the 2019; Shami et al. 2020). Such scRNA-seq datasets have
NOA condition, can enable not only prioritizing also been meta-analyzed (Soraggi et al. 2021). NOA
research on specific key molecules in multiple applied testes have also been subjected to scRNAseq (Wang
contexts such as novel contraceptive development but et al. 2018; Zhao et al. 2020). However, it should be
also basic research related to the molecular mecha- noted that while scRNAseq makes several new prom-
nisms associated with various stages of normal ises, the associated methods also have disadvantages
spermatogenesis. (Lowe et al. 2017; Chambers et al. 2019; Kim et al.
2019; L€ahnemann et al. 2020). The technical variability
The mechanism of regulation of gene expression is of gene expression profiles derived using this approach
particularly an important functional aspect (Hoopes is higher than those from bulk tissue RNA-seq (Hicks
2008) that has been less studied in the context of et al. 2018; Kim et al. 2019). In addition, we note that
NOA. Logically, exploring the testicular transcriptome the cell identification methods introduce an additional
under the NOA condition would also help identify level of variation during the interpretation of
key transcription factors important in testicular gene scRNAseq results. We realize that reliable profiling of
expression regulation under normal conditions. Many mRNAs using bulk testis samples will help to corrobor-
studies (e.g., Mohsin et al. 2022) have used transcrip- ate the scRNA data, and such transcriptomic profiling
tomic profiling followed by molecular network ana- can help analyze molecular interactions in the context
lysis to predict the interaction of transcription factors of many tissue-level molecular functions.
with other molecules.
Hence, we decided to use bulk-tissue RNA-
Several studies have been conducted on gene sequencing to obtain potential insights into the func-

expression in normal testis of mice and humans (e.g., tions of iso-mRNAs expressed differentially in the
Sassone-Corsi 1997; Reddi et al. 2002; Acharya et al. testis of NOA patients, particularly regarding gene
2006). Many scientists used the microarray techniques expression regulation. We postulate that hierarchical

198 G. BALAGANNAVAR ET AL.

Figure 1. A schematic representation of the workflow, along with summary findings. The quality of the RNA extracted from
testis samples and the quality of reads obtained from RNA-seq were found to be good (mean quality score of 33.95, ±2.75). The
sequencing depth was also good across samples (a mean of 56 million reads). Abbreviation used: SRA – Sequence Read Archive
(repository); NOA: nonobstructive azoospermia; DE: differential expression or differentially expressed; TF: transcription factors.

listing of the iso-mRNAs based on the extent of their Results
consistency of differential expression and analyzing
molecular interactions for the top-ranking iso-mRNAs As summarized in Figure 1, sequencing of testicular
can help predict key functions and transcription RNA samples and the resulting analysis in combin-
factors. ation with existing RNA-seq data helped us generate a
comprehensive list of genes and transcripts arranged
Specifically, we seek answers to the following ques- hierarchically based on their expression-based associ-
tions: (a) Which genes and iso-mRNAs exhibit expres- ation with the Non-Obstructive Azoospermia (NOA).
sion-based association with the NOA? Moreover,
which ones are more strikingly and consistently up- Clustering analysis of samples
or down-regulated under the NOA condition across
the samples? (b) What testicular functions are likely The expression levels of all genes in the testes of
to be affected in NOA, particularly considering the donors with (test samples) and without (control sam-
specific spliced variants of mRNAs (iso-mRNAs)? (c) ples) Non-Obstructive Azoospermia (NOA) were
Which of the differentially transcribed transcription- compared. The results showed that NOA samples
factor-coding genes (TFgenes) and the corresponding grouped separately from the control samples (with
iso-mRNAs, could be significant in targeting other normal spermatogenesis) – irrespective of possible
genes, particularly the other TFgenes, which are differ- sub-types within each set (Figure 2). This observation
entially transcribed in NOA-testes? supported the decision to group normal testis samples

with those from other conditions where spermatogen-
In addition, we also performed extensive data esis occurs normally, viz., Varicocele (VA), and
analysis to address the following questions about Obstructive Azoospermia (OA) as ‘control samples’.
the testicular TFs and iso-mRNAs in NOA: (d)
Among the iso-mRNAs of each corresponding gene The results of the comparison of cell-type-specific
expressed in the NOA-testes, are the ‘principal tran- gene expression patterns across samples also con-
script isoforms’ produced differentially, or the firmed the grouping of control samples. For example,
minor ones? (e) Are any of the key domain-coding there was a significant reduction in the mean number
regions removed among the iso-mRNAs that are of germ-cell-specific genes expressed across NOA-tes-
over- or under-represented in NOA-testes? (f) In tis samples compared to normal testis samples
which cell types are the key TFgenes expressed (Mann-Whitney U test; p < 0.0004) (also see
under the normal condition?

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 199

Figure 2. Gene expression-based comparison of tissue samples used in the study. The relative quantities of all mRNAs across
samples were used to quantify the similarities or differences between the samples. The samples with similar overall trends in the
expression pattern of genes were clustered together, which reflected a physiological (and hence, pathological condition) similarity
of samples. (A) PCA results. Two normal samples taken from public repositories clustered peculiarly and were not considered for
hierarchical clustering or other analysis. (B) Hierarchical clustering of samples after removing the two outlying normal ones. In
both types of clustering, the NOA samples were clustered together but separately from the control samples (including OA and vari-
cocele samples).

Supplemental Table 1). This trend was also seen in cytometry results and the cell-type specific gene
Sertoli cells, in some cases, even though the difference expressions need to be explored via a separate study.
in the percentage was less (see Supplemental Table Nevertheless, in the absence of traditional character-
1A). A detailed comparison of individual NOA and ization of NOA subtypes in the current study, the
control samples for the expression of such earlier cell-type marker data will give a hint of the possible
reported cell-type-specific genes confirmed a signifi- germ-cell distribution across samples.
cant reduction in their number in many NOA samples

(see Supplemental Table 1B). We noticed an average Differential expression of genes and transcripts
reduction of about 10, 29, 43, and 54% of genes spe-
cific for Sertoli cells, spermatogonia, spermatocytes, There were substantially more down-regulated tran-
and spermatids, respectively, were not detected in scripts compared to up-regulated ones. This trend in
NOA samples. However, there were significant varia- differential expression is probably because of the
tions among the current NOA samples in the relative absence of, or a very low number of, germ cells in
occurrence of these cell-type-specific genes. A much NOA-testis, particularly the post-meiotic cell types.
higher standard deviation was observed among the
average cell-type-specific genes/markers across data- The differential transcription was also found to be
sets among the NOA samples (6.47) compared to the more prominent, and generally more reliable, among
normal ones (2.34). The mean percentage variations the down-regulated transcripts than among the up-
among NOA samples across the four comparisons regulated ones. For example, the highest False
were 6.6, 26.3, 21.4, and 18, for Sertoli cells, spermato- Discovery Rate (FDR) among the ‘top up-regulated
gonia, spermatocytes, and spermatids, respectively (see 250’ (Tu250) genes was 0.0068, while it was 7.69E-09
Supplemental Table 1B). Across samples, there was no among the ‘top down-regulated 250’ (Td250) genes
consistent trend of the abundance of any cell-type (Supplemental Table 2).
specific gene set that could help the sample-clustering
into further sub-types. Hence, the current results of Validation of differential expression of transcripts
analysis of cell type marker genes across samples can-
not help to identify the classical sub-types of NOA, No contradictions were observed among eighty tran-
such as the ‘Sertoli cell only’ or ‘post-meiotic arrest’. scripts selected from the top 200 up- or down-regu-
Correlations between typical histological or flow lated ones (Supplemental Figure 1). This observation

200 G. BALAGANNAVAR ET AL.

indicated the reliability of the RNA-seq data and the represented by the corresponding down-regulated
differential expression profiles generated. transcripts (from Biomart) (see Supplemental Table
3). Thus, the overall trend in the types of functions
Data mining and analysis of the top 500 associated with up- and down-regulated genes was
differentially transcribed genes reproduced at the transcript level. It should also be

noted that an analysis done with the APPRIS database
The results provide a relative rank for the strength of also showed that most of the down-regulated tran-
scripts were the main transcript isoforms (see
association, based on differential expression in NOA Supplemental Table 4).

condition, for many genes already established to be Data mining and analysis of the top 6000
differentially-regulated genes
associated with NOA. The Tu250 and Td250 genes
The Top 3000 down-regulated (Td3000genes), as well
were analyzed in detail. In addition, the results also as the top 3000 up-regulated (Tu3000genes) genes,
were also analyzed in detail (Supplemental Table 2),
specify the iso-mRNAs of these genes. mainly to check the consistency of the trends
observed among the Tu250 and Td250 genes. Many
A literature review and GO analysis indicated that of the significant molecular functions, biological proc-
esses, cellular components, and pathways enriched
the up-regulated genes are not associated with sperm- among the top 250 up- or down-regulated (Tu250
and Td250) genes were also enriched among the cor-
atogenesis. This trend was expected as most of the responding larger set (3000 genes each; see
Supplemental Tables 5, 6, and 7). This observation
genes up-regulated in the testes of NOA patients indicated the reliability and utility of the hierarchical
listing of the differentially transcribed genes and a
would be due to a relatively higher number of somatic good representation of the differentially expressed
genes by the top 250. Even though a few GO terms
cells compared to germ cells, which are reduced com- and pathways found among the top 3,000 up- or
down-regulated genes were not among the top 250 of
pared to normal testes. the corresponding set, some discrepancies across the
sets were expected due to the varying number of
PAFAH1B1 (Ensemble transcript ID: genes.

ENST00000576586), LDHC (ENST00000280704), SPAG9 The Td3000 genes are likely to represent the

molecular level of spermatogenic failure more cor-
(ENST00000357122), PRM1 (ENST00000312511), TNP2 rectly than the Tu3000 genes. This interpretation is
based on two observations: (a) There was a higher
(ENST00000312693), TNP1 (ENST00000236979) and consistency in the down-regulation status, across sam-
ples, among Td3000 genes compared to the Tu30000
PRM2 (ENST00000241808) are such important genes genes. The highest p-value was 0.025 (FDR: 0.064),
and the lowest fold change and percentage consistency
down-regulated in NOA that seem to be well studied in were 0.26 and 0%, respectively, for Tu3000. However,
among the Tu3000, there were only 207 genes with a
the context of NOA as well as spermatogenesis. These 100% consistency, which had 3.29E-03 (FDR: 0.011)
P-value, and !0.49 fold change, respectively. On the
genes have more than 200 research articles corresponding contrary, the highest P-value was 3.38E-04 (FDR:
1.66E-03), while the lowest fold change and percent-
to each in general and more than ten articles each in the age consistency, were 2.02 and 80% for Td3000,
respectively. And, there were 2810 genes among
context of spermatogenesis or NOA. In addition, there Td3000, with a 100% consistency, which had 1.81E-05
(FDR: 1.27E-04) P-value, and 2.02 fold change,
were also many well-studied genes corresponding to the respectively. (b) The Td3000 also represented

Td250 transcripts, which do not seem to -have been

studied earlier in the context of spermatogenesis

or NOA.

GO based association with spermatogenesis was

detected in 43 of the Td250 genes (Figure 3A). As per

GO analysis, many of the other Td250 genes were known

to be directly involved in sperm functions such as fertil-

ization (see Figure 3B) or post-meiotic events such as

acrosome formation and sperm capacitation. Around 17

relevant biological processes (BPs) (GOTERM_BP_all)

and 12 pathways were enriched with Td250 genes.

Among these genes, cell-division-related multiple bio-

logical processes, molecular functions, and cellular com-

ponents were enriched with very high significance (FDR

< 0.00010). Similarly, many spermatogenesis- or sperm-

related GO terms were also represented well by the

Td250genes. Among general functions, protein binding

was common in both Td250 and Tu250 gene-sets,

whereas kinase activity and ATP binding were enriched

molecular functions among Td250 genes only.

About 93% of the GO terms found over-repre-

sented for the Td250 genes, via DAVID, were also

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 201

Figure 3. Results of integrated functional analysis of the top 250 down-regulated (Td250) genes. The first two histogram bars
within each node indicate the number of papers related to NOA or spermatogenesis (Annokey), in that order, for each gene. The
last bar in each node indicates the number of papers in general (GenCLiP). Genes with a known GO association with the spermato-
genesis process are marked with blue font. (A) Of Td250 genes, 43 were associated with spermatogenesis function with a fold
enrichment score and FDR of 9.4 and 6.43E-26, respectively. (B) Of Td250 genes, 20 are involved in 12 pathways: PW1-PPAR signal-
ing pathway, PW2-Glucagon signaling pathway, PW3-Biosynthesis of amino acids, PW4-Glycolysis, PW5-Synthesis of PE, PW6-
Separation of Sister Chromatids, PW7-Gluconeogenesis, PW8-Mitotic Prometaphase, PW9-Deactivation of the beta-catenin transacti-
vating complex, PW10-Antigen processing: Ubiquitination & Proteasome degradation, PW11-Resolution of Sister Chromatid
Cohesion, PW12-RHO GTPases Activate Formins. The first two histogram bars within each node indicate the number of papers
related to NOA or spermatogenesis (Annokey), in that order, for each gene. In contrast, the last bar in each node indicates the
number of papers in general (GenCLiP). Genes with a known association with the spermatogenesis process are marked with blue
font.

spermatogenic functions, unlike Tu3000 genes, as spermatogenesis was an interesting observation. In
described below. most cases, however, the Td3000genes involved in
such specific processes are not yet reported to be asso-
The BP terms enriched among the Td3000genes ciated with spermatogenesis or fertilization-related
include obvious biological processes related to testis processes. This observation is probably because of the
functions. About 232 genes are associated with sperm non-discovery of the association of many genes with
development or related processes. Two such broad spermatogenesis and NOA.
prominent processes were ’spermatogenesis’ (count:
163 genes, Benjamini P-value or Bpv: 2.66E-42) and To obtain a better insight into the involvement of
’spermatid development’ (42 genes, BPv: 1.08E-13). genes in key processes during spermatogenesis, we
Another set of prominent BP terms included sperm analyzed the overlapping biological processes and
motility and related functions (88 genes, BPv: 1.13E- pathways known for the top NOA down-regulated

01). Other specific processes that seem to be nega- genes from the function-based core cluster indicated
tively affected in NOA are cilium assembly, cilium in (Figure 3B). In addition, we also considered the
morphogenesis, and related processes, cell division well-established protein-protein interactions among
and related processes such as meiosis and DNA the proteins coded by the selected genes.
repair, and several aspects of RNA synthesis, process
and post-translational modifications (Figure 4, also Several down-regulated genes among the top
see Supplemental Table 8). The suppression of genes rankers were found to be involved in multiple func-
representing a significant overlap of post-translation tions and pathways. For example, ABHD2 (Figure 3)
modifications with the mitotic phase of is linked to acrosome reaction and sperm capacitation.
IQCF1 is also linked to these functions but has an

202 G. BALAGANNAVAR ET AL.

Figure 4. Distribution of specific functions among the Td3000genes after removing some of the processes well known to be
associated with spermatogenesis (Supplemental Table 8). Numbers mentioned for each function inside the pie chart represent
the percentage of genes not known to be associated with (a) sperm motility-related processes (innermost circle), (b) fertilization-
related processes (middle circle), or (c) spermatogenesis-related processes (spermatogenesis, spermatid development, spermatid
nucleus elongation, and cell differentiation) (outermost circle).

additional association with sperm motility. TRIM36 is Interestingly, viral activity is indicated to be a
associated with the cytoskeleton, ubiquitination, and major function among the up-regulated genes, while
acrosome reaction. It interacts with the multi-protein- the antiviral mechanism is represented among the
interacting UBC associated with the regulation of down-regulated genes, with a few contradictions. The
mRNA stability and deactivation of the beta-catenin GO terms such as ’viral transcription’ (p-value <3.2E-
trans-activating complex. KLHL10 and PAFAH1B1 22, 63 genes) and ’viral nucleocapsid’ (p-value
were found to be involved in protein-ubiquitination as <0.0067, 11 genes), and the ’viral mRNA translation’
well as other aspects related to spermatid development (p-value <3.71E-31, 66 genes) pathway are enriched
and fertilization. among the top-3,000 up-regulated genes (Tu3000). On
the contrary, the pathway ’ISG15 antiviral mechanism’
Similarly, DNAAF1 and CPAF206 may be involved (Reactome p-value <3.16E-06, 24 genes) is over-repre-

in flagella formation during the latter part of sperm- sented among the top 3,000 down-regulated genes
atogenesis. In addition, some of the proteins coded by (Td3000). The ’MHC class II antigen presentation’
key down-regulated genes seem to interact well with pathway (Reactome p-value <4.65E-04, 28 genes) is
other down-regulated proteins. For example, also represented well among the down-regulated
RANGAP1, for which the role in spermatogenesis is genes. However, glutathione metabolism, a pathway
not known well except for an indication of a role via likely to be associated with innate immunity and anti-
SUMO1-ylation (Marchiani et al. 2014), interacts with viral processes (Diotallevi et al. 2017), was over-repre-
three proteins (KIF2B, KIF2C, and PAFAH1B1) sented among some up-regulated genes. Glutathione
known to be involved in the microtubule-based move- representation was statistically significant during both
ment, and all these four proteins are known to be KEGG and Reactome analysis of the top 250 as well
involved in pathways related to mitotic chromatid as 3000 up-regulated genes (7–18 genes with p-values
separation. These observations hint that RANGAP1, from <0.004 to <0.000005). Related molecular func-
along with KIF2B, KIF2C, and PAFAH1B1, may be tions were also similarly over-represented. Similarly,
involved in the microtubule-based movement of chro- the ’L13a-mediated translational silencing of
matids during the early (mitotic or meiotic) stages of Ceruloplasmin expression’ (Genes: 68, P-value: 1.59E-
spermatogenesis. Like UBC, PSMF1 is also associated 24) pathway with a potential role in immunity is also
with protein modifications, interacts with many other well represented among up-regulated genes. On the
proteins associated with the same function, and is other hand, NS1-mediated effects on host pathways,
associated with mRNA stability regulation.

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 203

rev-mediated nuclear export of HIV RNA, nuclear Supplemental Table 10). This list was prepared from
import of Rev protein, and vpr-mediated nuclear five articles identified via a thorough literature search
import of PICs were among other significant (P-value (Carro et al. 2010; Ravasi et al. 2010; Yusuf et al.
0.02 to 4.6E-05) pathways or GO terms enriched 2012; Lambert et al. 2018; Hu et al. 2019). Next, we
among down-regulated genes. used this list to identify genes coding for TFs and
cTFs that are significantly (logFC > 2) differentially
Analysis of alternatively spliced mRNA isoforms expressed in the testes of NOA patients. We noted
(iso-mRNAs) 433 down-regulated genes and 59 up-regulated genes

coding for TFs/cTFs (Supplemental Table 2). We then
About 61 and 42% of Td250 and Td3000 transcripts, used the TRRUST database to shortlist such TFs and
respectively, were the principal transcript isoforms. cTFs with known binding sites on the promoters of
Similarly, about 51 and 37% of the up-regulated tran- other differentially expressed TF/cTF-coding genes.
scripts were the principal isoforms among the Tu250 Cytoscape-based visualization of these interactions
and Tu3000, respectively. Only 8 of the Tu250 and 9 of revealed 35 TFs and 20 cTFs with known interactions,
the Td250genes, respectively, had a completely oppos- including multiple interesting TF-TFgene hubs of
ing expression among their iso-mRNAs. The transcript prominent interactions (Figure 6). CytoHubba (Chin
ENST00000618113 of the down-regulated SPAG9 gene et al. 2014) was used to calculate the node score.
was one such example. The cases of partial contradic-
tions (i.e., up-/down-regulation vs. non-differentiated Based on the averages scores of Closeness,
transcripts from a gene) were more, with 121 and 43 Betweenness, and Degree (Supplemental Table 11),
cases among the Tu250 and Td250 genes, respectively. SP1 formed the most conspicuous central node with
The regions with functional domains did not seem to an average node score of 372 in the network, with 15
be altered across the transcript isoforms of most genes. potential direct target genes that code for TFs (10) or
However, the transcription initiation and termination cTFs (5). Of these 15 potential targets, all of which
sites were altered in many cases. This observation indi- were down-regulated, seven were known transcription
cates a potential change in the NOA-associated gene activators. SP1 also seems to be part of a regulatory
expression regulation mechanisms, mainly at the tran- axis along with HSF4 and RAD51. HSF4 is a known
scription initiation and post-transcriptional phases inhibitor of RAD51, which activates SP1 in turn. We
involving promoters and the 5’ and 3’ UTRs. hypothesize that the up-regulation of HSF4 and the
resulting down-regulation of RAD51 under NOA-tes-
The types of alternative splicing events and transcripts tis is one of the causes of SP1 down-regulation, which
were different across up- and down-regulated transcripts in turn results in the down-regulation of at least seven
(Supplemental Figure 2 and Figure 5). ‘Alternative pro- other transcription factors. The current results help us
moter or terminator’ (AP/AT) and ‘Alternative 50’ Splice to postulate that multiple spermatogenic genes (see
Sites’ (A50-SS) combinations were more common in Table 2) might be regulated via this regulatory axis,
down-regulated transcripts (204, which is 16.32%) than in where up-regulation of HSF4 under the NOA condi-
up-regulated ones (106, 7.89%) or undifferentiated (27, tion results in suppression of RAD51 transcription,
6.89%) transcripts. The transcripts with a shared pair of that, in turn, results in a lack of activation of SP1,

event types were also found to have an uneven distribu- thus resulting in the down-regulation of this key TF.
tion across up- and down-regulated genes (Supplemental CREBBP could also be an important regulator, via
Table 9). For example, AP/AT and Skipping Exon/Sub- RAD51, of SP1. Down-regulation of CREBBP and
Intron Retention (SE/sub_RI) were found to be exclusive ATF5, which normally activate CREB1, may also have
to 50 (4%) down-regulated transcripts. Overall, all alterna- suppressed CREB1 under the NOA condition, and
tive splicing events were much more common among the eventually causing an up-regulation of SNAI2.
differentially expressed genes than the undifferentiated Another suppressor of this TF, EZH2, is also down-
ones. Single protein-coding genes were also proportion- regulated, even though the activating TF for the
ately very high among differentially transcribed genes. SNAI2 gene, MTA1, is up-regulated under NOA. The
cell-type expression analysis of the TFgenes showed
Regulatory network analysis that SP1 and CREBBP are spermatocyte and sperm-
atid enriched in normal adult testis.
We first prepared a comprehensive list of 3056 human
Transcription Factors (TFs) or co-Transcription In contrast, RAD51 is enriched in spermatogonia,
Factors (cTFs) by reviewing the literature (see Table 1, spermatocytes, and spermatids (see Table 2). SNAI2 is
enriched in spermatocytes and spermatids in normal

204 G. BALAGANNAVAR ET AL.

Figure 5. The number of transcripts among the top 3000 each of NOA up (green color), down (red color), and undifferenti-
ated (grey color) genes with a single alternative splicing event per transcript. The spliced events for the Td3000 and Tu3000
were obtained using the SpliceDetector tool. The overall size of each pie is proportional to the total number of iso-mRNAs (alterna-
tively spliced mRNAs). Iso-mRNAs represented each type of splicing event were more common in differentially expressed mRNAs
than the undifferentiated ones. There were only minor differences in the representation among up- vs. down-regulated iso-mRNAs
in the case of most of the splicing event-type. AP: Alternative promoter; AT: Alternative terminator; AP/AT: Alternative promoter or
terminator; A3’-SS: Alternative 3’ Splice Sites; A5’-SS: Alternative 5’ Splice Sites; RI: Intron Retention; SE: Skipping Exon; sub_RI:
Skipping Exon; SE/sub_RI: Skipping Exon/Sub-Intron Retention; SPC: Single protein-coding.

Table 1. Compilation of human transcription factors (TFs) and co-transcription factors (cTFs) that are known or predicted.

PubMed article ID Year; Journal Data type Total TFs / cTFs reported Active human gene IDs New union list of gene
Ids corresponding to

potential TFs/cTFs

20032975 2010; Nature Curated & predicted 928 826 3056

30204897 2019; NAR Curated & predicted 2690 2672

20211142 2010; Cell Experimental (qPCR) & 1988 1969

predicted

29425488 2018; Cell Curated 2765 1636
563
22458515 2012; Genome Biology Curated 803

A thorough literature search helped to identify the relevant research reports. The union list generated forms the largest number of (known and predicted)
human TFs to date.

men. TBP seems to be another TFgene with complex differentially expressed in the NOA-testis also indicate
innate regulation in normal spermatocytes but down- their significance during normal spermatogenesis.
regulated in NOA. Multiple activators were seen in Therefore, the possibility of the key TFs identified in
the network for CDK1, enriched among spermatogo- the current study having a causative role in inhibiting
nia, spermatocytes, and spermatids, and BIRC5, the development of multiple pre- and post-meiotic
enriched in spermatogonia, spermatocytes, and sper- cells in the NOA testis needs to be explored.
matids. MUC1 was unusual because it had multiple
suppressors, while STAT1 seemed to activate it. Interestingly, none of the NOA-down-regulated SP1
transcripts was found to be testis-specific or predomin-
Interestingly, none of the Td3000 genes, including ant (Mammalian Alternatively Spliced RNA Isoforms

TFgenes, seem to be targeted by CDK1, BIRC5, for Normal Tissues: MASRINT, .
MUC1, and TDP. The observations on the interac- in/masrint/humans). On the contrary, the NOA down-
tions among the TF-coding genes and their products regulated splice variants for RAD51 and CREBBP were

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 205

Figure 6. The regulatory network of transcripts differentially regulated in NOA condition, indicating the possible role of the HSF4,
RAD51, and CREBBP roles in activating the SP1 gene, which regulates many other genes. Interactions between up- and down-regu-
lated genes coding for transcription factors (TFs) and other down-regulated TF-coding genes were identified using the TRRUST database,
along with the mode of interaction (activation/suppression) and the network of interactions was visualized using cytoscope.

Table 2. Analysis results summary for the key NOA down-regulated TFs. The interactions between TFs and TF-coding-genes
(TFgenes) were identified via TRRUST.

Transcription factor No. of TFgenes No. of

(TF) coding gene targeted directly by TFgenes targeted No. of Td3000 genes Domains present in

(TFgene) Cell type expression the TF indirectly by the TF targeted by the TF the TF

SP1 SpermatocyteÃ, 15 12 47 ZF-C2H2
spermatidÃ

HSF4 NA 1 0 2 HSF_DNA-bind
RAD51 SpermatogoniaÃ,
spermatocyteÃÃ, 1 0 1 HHH_5
CREBBP
spermatidÃ
spermatocyteÃ,
spermatidà 4 4 5 KIX

CREB1 NA 3 3 8 BZIP, pKID
SNAI2 spermatocyteÃ,
spermatidà 0 0 1 ZF-C2H2

The expression pattern of genes across cell-types was deduced after analyzing already published scRNA datasets (ÃHermann et al. 2018; ÃÃ Hermann

et al. 2018 and Wang et al. 2018). The information about specific iso-mRNAs, their differential expression status across conditions, and the correspond-

ing SEBA score can be found in the Supplemental Table 2.

found to have a higher expression level in the normal Another interesting observation was about chromo-
testis than the control testis samples. In contrast, the somal representation among the genes. When the
NOA-up-regulated transcript for HSF4 had a lower down-regulated genes were considered as a percentage
expression in the normal testis. A total of nine genes in of total genes in each chromosome, while the Td3000
the network were found to have testis-specific isoforms were well distributed across chromosomes, the down-
( regulated TFs were represented well by some chromo-
Interestingly, the testis-exclusive transcript of CREM somes, particularly chromosomes 19, Y, 13, and 20,
and the testis predominantly expressed isoforms from when compared to other chromosomes.
BIRC5, CSNK2A1, MTA1, and PTTG1 seem to be tar-
geted by SP1. Overall, regulation of SP1 might play an Eighteen TF domains were found to be enriched
important role in successfully completing the spermato- across the 433 NOA down-regulated TFs (Table 3 and
genesis process. Supplemental Table 12). Among the 18 enriched
domains, five of the domains are enriched in more

206 G. BALAGANNAVAR ET AL.

Table 3. Domains enriched among the NOA-down regulated transcription factors (TFs).

PFAM ID Domain name P-Value Fold Enrichment FDR

PF00096 Zinc finger, C2H2 type 9.4E-16 4.00 2.8E-13
PF00439 Bromodomain 5.9E-10 13.76 8.8E-08
PF01352 KRAB box 8.5E-09 3.54 8.5E-07
PF13912 C2H2-type zinc finger 5.5E-08 3.54 4.1E-06
PF00505 HMG (high mobility group) box 9.9E-08 10.09 6.0E-06
PF00250 Fork head domain 9.6E-06 8.42 4.8E-04
PF02373 JmjC domain, hydroxylase 1.4E-04 11.47 6.1E-03
PF00385 Chromo (CHRromatin Organisation MOdifier) domain 2.1E-04 10.58 8.0E-03
PF01388 ARID/BRIGHT DNA binding domain 2.5E-04 15.29 8.3E-03
PF00105 Zinc finger, C4 type (two domains) 4.6E-04 6.98 1.3E-02
PF00447 HSF-type DNA-binding 5.3E-04 22.93 1.3E-02
PF00313 ’Cold-shock’ DNA-binding domain 5.3E-04 22.93 1.3E-02
PF00104 Ligand-binding domain of nuclear hormone receptor 5.8E-04 6.69 1.3E-02
PF02257 RFX DNA-binding domain 7.8E-04 20.38 1.7E-02
PF09011 HMG-box domain 1.1E-03 18.35 2.2E-02
PF00628 PHD-finger 1.3E-03 5.73 2.5E-02
PF02820 mbt repeat 1.5E-03 16.68 2.6E-02
PF00249 Myb-like DNA-binding domain 1.9E-03 9.17 3.2E-02

The down-regulated (Td3000) TFs were subjected to domain analysis using the DAVID database, and the main observations were
tabulated.

than 10 TFgene. In addition, there was a remarkable current work in its preprint form (Govindkumar et al.
over-representation of C2H2 zinc finger domains, bro- BioRxiv 2019a). The current results produced using
modomain, and KRAB box domains among many the bulk testis samples may fill a major gap in the
down-regulated TFs. These results indicate the pres- transcriptomic profiling of NOA-testes. In addition to
ence of common themes of molecular interactions general transcriptomic profiling, the current study
that may be involved in regulating the transcription of also provides a first-of-its-kind hierarchical list of iso-
genes during spermatogenesis. mRNAs based on their strength of association in

terms of differential expression in the NOA condition.
Discussion The scoring method used in this study for hierarchical
listing is similar to the approach our group followed
All applied research towards diagnostics, treatment, earlier for the meta-analysis of microarray data, where
and prognostics of male infertility, as well as male consistency in expression patterns across the studies
contraceptive development, would be facilitated by a was considered (Acharya et al. 2010).
clear understanding of the human spermatogenesis
process itself. Transcriptomic studies on testes from The current study successfully identified several
NOA patient donors are significant in this context. consistently differently expressed genes and tran-
The current study forms the first bulk testicular tran- scripts, and the analysis indicated the relevance of
scriptomic exploration using a reasonable number of most of them in the context of normal human sperm-
NOA samples, particularly via a good-depth NGS. atogenesis. The overall trend in functions represented
The expression profile reported here has a remarkably by up- and down-regulated genes was similar to the
larger number of genes than the earlier transcrip- previous observations. However, the number of
tomic, particularly the microarray, studies. The study reported genes that support the trends in functions
has also established a first-time, comprehensive splice promoted or suppressed under the NOA condition is
isoform profiling of mRNAs for the NOA condition. much higher than in any other study before.

Wu et al. (2020) reported RNA-seq of 3 NOA-tes- A higher number of down-regulated genes, vs. the
tes samples. They focused more on the epigenomic up-regulated ones, was expected as most spermato-
abnormality than the transcriptome itself. The authors genic cells would be absent or reduced in number, or
did not report any significant findings or information non-functional or sub-optimally functional, and sev-
about the transcriptome. Wang et al. (2018) used sin- eral key genes expressed during spermatogenesis
gle-patient data, while Zhao et al. (2020) focused spe- would also be dormant or under-expressed in the tes-
cifically on Sertoli cells during their scRNA analysis. tis of NOA patients. On the other hand, genes
Thus, there has been a need for a more detailed tran- expressed in somatic cells and pre-meiotic germ cells
scriptomic analysis of NOA samples. Unfortunately, within the testis would be proportionately enriched,
none of the recent testicular transcriptomic reports and the corresponding functions would seem pro-
seem to have noticed the preliminary report of the moted in NOA. Genes up-regulated in NOA had an
over-representation of translation, general metabolism,

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 207

steroidogenesis and androgen biosynthesis, lysosome ubiquitination, many other post-translational modifi-
activity, extracellular matrix, and cell-adhesion-related cations are already known to play an important role
functions. The mechanism of nonsense-mediated in sperm functions (e.g., Bhagwat et al. 2014).
decay, independent of or enhanced by the role of Integrated data analysis of top-ranking down-regu-
many such RNA processing events, indicated to be lated genes indicated a possibility of novel and specific
important in NOA among up- or down-regulated functional associations of several genes and tran-
genes, needs to be explored more. Exon Junction scripts. This observation again emphasizes the sug-
Complex was an interesting pathway striking among gested significance of prioritizing research on the top-
the up-regulated genes. listed genes and transcripts reported here.

The current study also revealed many novel func- Suppression of a huge number of genes in the testis
tional associations with up- as well as down-regulated of NOA patients could be because of the down-regula-
genes. Some of the examples include GO terms and tion of TFs that normally activate other genes or the
pathways related to inner and outer dynein arm up-regulation of suppressing TFs. The hub-genes
assembly and motile cilium assembly, acrosome observed in the newly generated network of TFs and
assembly, peptidyl-serine phosphorylation, sister chro- cTFs could play an important role in the down-regu-
matid cohesion, and synaptonemal complex assembly, lation of genes in the testis of patients, which in turn
Glycolysis/Gluconeogenesis, Glucagon signaling path- may be responsible for the lack of significant types of
way, Biosynthesis of amino acids, and protein ubiqui- male germ cells and spermatogenic failure. The hub
tination involved in/proteasome-mediated ubiquitin- genes likely have an important role in gene expression
dependent protein catabolic process as well as the regulation during normal spermatogenesis. Some of
positive regulation of ubiquitin protein ligase activity the hub genes identified in the current study as poten-
– for down-regulated genes. An example of functions tial key regulators of genes of spermatogenic signifi-
represented by up-regulated genes is the ’mRNA sur- cance were well studied in general, as described
veillance pathway’. Less than 25% of GO terms and below. However, the role of many of these genes in
pathways found to be associated with NOA by the gene expression regulation in normal and NOA testis
current analysis have been reported to be overrepre- is being indicated for the first time. However, such

sented among genes differentially expressed in NOA possibilities, including the proposed role of the HSF4-
by earlier studies (Okada et al. 2008; Cappallo- RAD51-SP1 axis, need to be tested further.
Obermann et al. 2010; Zhuang et al. 2015; Li et al.
2018) – including a meta-analysis report (Razavi et al. SP1 has been reported to be one of the key tran-
2017). Moreover, as per the current observations, the scriptional regulators for normal spermatogenesis
number of genes associated with each GO term is at (Thomas et al. 2007; Zhu et al. 2016). It plays a role in
least double that of the earlier reports. Such findings chromatin remodeling and DNA damage break repair
show the benefits of NGS-based transcriptomics, espe- (Beishline and Kelly 2012). Chromatin remodeling is a
cially when the top differentially expressed genes and key process during the protamine transition process in
transcripts are carefully chosen. the post-meiosis stage of spermatogenesis (Govin et al.
2004; Rathke et al. 2014). RAD51 is essential for sper-
Among the functions richly represented by the top matogonia maintenance and meiotic progression (Qin
3000 NOA-down-regulated genes (Td3000genes), et al. 2022), mainly for the prophase stage (Dai et al.
mitosis, DNA replication, and meiosis, and the proc- 2017). Heat shock family proteins are known to play a
esses related to these three events, have an obvious role in spermatogenesis (Chalmel et al. 2012; Hemati
connection with the spermatogenesis process. et al. 2020), but HSF4, which is explored in limited
However, many other processes over-represented by cases (Syafruddin et al. 2021), has been known to be an
Td3000genes are also important for spermatogenesis. activator as well as a repressor (Tanabe et al. 1999) and
For example, ubiquitination is a key biological process required for cell differentiation (Fujimoto et al. 2004).
involved in spermatogenesis (e.g., Gou et al. 2017; However, its implication in spermatogenesis and NOA
Nakagawa et al. 2017). Ubiquitination and mRNA sta- condition is highlighted only by the current study.
bility being linked via two proteins (PSMF1, UBC)
and down-regulation of several such multi-functional The lower expression of CREB1 in the NOA condi-
genes was also an interesting observation in the cur- tion observed in the current study supports earlier
rent study. This aspect needs to be explored further. suggestions of the decreased levels of CREB1 affecting
Other post-translation modifications are also found to fertility (Don and Stelzer 2002; Xu et al. 2011). The
be relevant in NOA in the current study. Apart from EZH2 gene is important in chromatin remodeling
with increased expression during the onset of

208 G. BALAGANNAVAR ET AL.

spermatid formation (Lambrot et al. 2012). SNAIL genes. Apart from noticing multiple TFs, particularly
family genes are also known to have important func- the 433 down-regulated and 59 up-regulated TFs, with
tions in the testis. For example, higher expression of potential significance in the human testis, the current
SNAI1 and the absence of the SNAI2 gene in the tes- analysis also helped us to postulate a few regulatory
tis leads to infertility (Micati et al. 2018). axes that may be responsible for the down-regulation
of a significant number of genes in the NOA-testis,
The C2H2 zinc finger, found predominantly among and are probably important during normal
NOA down-regulated TF-genes, is commonly found spermatogenesis.
in transcription factors with a strong and specific
binding ability to a long DNA recognition target Material and methods
sequence and is suggested to be involved in the organ-
ization of the chromatin architecture but studied min- An overview of the study plan, along with results, is
imally (Fedotova et al. 2017). They are also well- mentioned in Figure 1.
known as part of the human regulatory lexicon
(Weissig et al. 2003; Najafabadi et al. 2015). Clinical diagnosis

The current study has a restricted goal of compar- All NOA donors represented primary cases of the dis-
ing testicular transcriptomics in NOA vs. normal con- order. Even though hormonal profiling was done
ditions, broadly. However, not considering the sub- (FSH levels ranged from 5.8 to 31.7 mIU/mL), the
typing and genotypic of the NOA samples and not final diagnosis of the NOA condition was made using
performing parallel histological analysis is a limitation cytological examination of a small amount of the
of the current study, even though the current detailed biopsied samples, where absence of sperm indicated
analysis of sample-wise expression of cell-type-specific NOA. Detailed sub-typing of NOA was not attempted
genes may be useful for fellow researchers. Id should as the focus was to compare all NOA cases with con-
also noted that collecting a few seminiferous tubules trol cases with evidence of spermatogenesis occurring
from one location in the testis, the practical approach within testis. Diagnosis of OA was made based on (a)
to perform studies, could represent the majority situa- full epididymal size, (b) normal FSH levels associated
tions or the overall trend in the completion of the with azoospermia, and (c) cytological confirmation of
spermatogenesis process in the tissue. But the testis of the testis sample that the patients did not have failed

NOA samples is likely to heterogeneous; there could spermatogenesis. Cryptozoospermia cases were
be some regions within the large testis tissue which excluded. Semen analysis of varicocele samples
have a different type of progress in spermatogenesis. showed a sperm count of <10 million proper sperm
morphology in <4% sperm, sperm motility of <3%,
Overall, the expression-based association of specific and cytological examinations of the testicular samples
alternatively spliced transcript isoforms (iso-mRNAs) confirmed hypospermatogenesis. Two samples each
and genes with the NOA condition, and their func- from donors with OA and VA, from which testicular
tional analyses, indicate a potential association of spermatozoa were detected, were considered equal to
many of them with spermatogenesis under normal other normal testicular samples, in terms of spermato-
conditions. For example, we notice an association of genesis completion, and grouped as control samples.
many NOA-down-regulated genes with mitosis, sperm
motility, and other functions found exclusively or pre- Sample collection
dominantly in the testis. Selecting the top-ranking
genes and iso-mRNAs from the ’hierarchical list’ Testicular biopsies were collected from 12 donors hav-
being reported could be an efficient way to explore ing the Non-Obstructive Azoospermia (NOA) condi-
any novel functional associations with the spermato- tion and two each with Obstructive Azoospermia
genesis process and involvement in developing the (OA) and Varicocele (VA) conditions, following eth-
NOA condition. The current report of RNA analysis ical procedures approved by the Institutional Ethics
results using bulk (non-single-cell) NOA-testis tissue Committee (IEC) at the Institute of Bioinformatics
could form a foundation for efforts towards under- and Applied Biotechnology (IBAB) and the collaborat-
standing splicing events and system biology of the ing clinical center. The age of donors ranged from 28
NOA testes. In addition, the observations in this study to 38 years, with a mean of 32.2 and a standard devi-
may facilitate further understanding of the molecular ation of 4.15. While a smaller portion of each biopsy
basis of the normal spermatogenesis process in the
human testes, particularly in terms of the regulatory
role of proteins coded by the NOA-down-regulated

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 209

was used for microscopic examination, the rest of Apart from the statistical significance of the differ-

each sample was stored in an RNA-later solution ence (p-value) in the mean TPM values and the fold
(Ambion, cat # AM7024), according to the manufac- change of these two means, the consistency of differ-
turer’s guidelines, for RNA extraction. ential expression in the samples across the test and
control groups to derive a ’Strength of Expression-
RNA extraction, sequencing and analysis Based Association’ (SEBA) (rtbioinfo.
org/methods/StA) was also used to list the differen-
RNA was extracted using the RiboPure kit (Ambion tially expressed transcripts hierarchically.
cat # AM1924), and quality was checked with formal-
dehyde agarose gel electrophoresis and a micro-vol- RT-qPCR validation of top differentially regulated
ume spectrophotometer. transcripts

Testicular RNA from eight NOA, two OA, and two For further validation using RT-qPCR, we selected 80
Varicocele donors were used for NGS. In addition, transcripts among the top-scoring transcripts. Forty
two commercial RNA (Clonetech cat. no.: 636533, transcripts, where unique regions were available for
Asian: lot no: 1105041; Caucasian: lot. no.: primer designing, were selected among up-regulated
LOT1105214A) samples from normal testes were also transcripts, and similarly, 40 transcripts were also
used. Paired-end library preparation was carried out selected among the down-regulated transcript list.
using Illumina TruSeq RNA Library Prep Kit v2, and Primers were designed for each of these 80 transcripts.
sequencing was done using the Illumina HiSeq 2000. An effort was made to pick at least one of the primers
To increase the number of control samples, we for each transcript from an exon-exon junction. ExEx
screened the Sequence Read Archive (SRA) database Primer (Govindkumar et al. 2022, http://resource2.
for RNA-seq data sets corresponding to the normal ibab.ac.in/exprimer/) tool was used to design primers.
human testis, and those with reasonable depth
(>40 M) were downloaded (see Supplemental Table RNA extracted from seven NOA samples, one sam-
13). The FastQC (ra- ple each of OA and Varicocele condition, and one
ham.ac.uk/projects/fastqc/) tool was used to assess the commercial, normal RNA sample were used. cDNA
quality of reads, and seqtk ( was prepared using the ‘ImProm-IITM Reverse
seqtk) was used for trimming low-quality reads. Transcription System’ (Promega cat No: A3800)
Alignments, identification of transcripts and the chi- according to the kit protocol. RT-qPCR experiments
meric/transplice molecules, and their quantification were performed using SYBR green (Kappa cat no:

were performed by Kallisto software (Bray et al. KK4601) reaction mixture, using the manufacturer’s
2016). Human transcript (GRCh38.90) sequences from protocol, and the StepOnePlus instrument of Applied
Ensembl ( were Biosystems.
used as a reference. To challenge the homogeneity
among the groups, particularly among the control The 2–DDCT method (Schmittgen and Livak 2008)
samples, and heterogeneity across groups, all samples was used to identify the fold change in the expression
were clustered via Principal Component Analysis level of transcripts in NOA compared to the control
(PCA) and hierarchical clustering. Though the quality conditions, and the student’s t-test was used to assess
of raw data from 11 samples was good, two of these the significance of the difference.
samples stood out (Figure 2A). Hence, nine public
datasets were included in the final analysis, along with Functional analysis
the in-house generated RNA-seq data from 8 NOA
and six control samples. All samples were further Enrichment of functions among the top 250 each of
clustered hierarchically based on the expression values up- (Tu250) and down- (Td250) regulated genes was
of transcripts across samples, using an R package performed by identifying Gene Ontology (GO)
(ggplots, Wilkinson 2011). The clusters were visual- (Ashburner et al. 2000; The gene ontology consortium
ized using RColorBrewer (). 2019) Terms (GOTERM_BP_DIRECT, GOTERM_
After confirming the sample segregations via cluster- MF_DIRECT, GOTERM_CC_DIRECT) and pathways
ing RNA-seq data from the selected 23 samples, dif- over-represented among each set using DAVID 6.8
ferential expression analysis was done using the database ( To
Bioconductor package limma (Ritchie et al. 2015). check the generality of the observations made with
the top 250 gene sets (Tu250 or Td250), we repeated

210 G. BALAGANNAVAR ET AL.

the process with the top 3000 each of the up- and proteins in the secondary interactions corresponding
down-regulated genes, and the results were compared. to each core cluster were considered.

The significantly enriched BPs for the top 3000 Comparative analysis with single-cell RNA
down-regulated genes (Td3000genes) were grouped sequencing (scRNAseq) reports

based on similar functions/terms, excluding the
spermatogenesis, sperm motility, and fertilization pro- For the TFgenes identified as important for
cess, and a pie chart was generated. Then, the genes NOA/spermatogenesis in the current study, the earlier
of each functional group were compared with the rest scRNAseq datasets (Hermann et al. 2018; Wang et al.
of the groups, and the percentage of genes overlap- 2018; Soraggi et al. 2021) were analyzed to explore the
ping with other functional groups was calculated. cell-type-specific expression. In addition, we also
checked the number of cell-type-specific genes, from
To explore the relative extent of studies on some of each of the previous datasets, with a reasonable
the differentially expressed genes, we performed litera- amount of expression in each sample used in the cur-
ture searches using Annokey (Park et al. 2014). In rent study. A cut-off of 3 TPM was used for detection
addition, to find out the general extent of studies car- during this analysis.
ried out on each gene, GenCLIP (Wang et al. 2014),
HPO ( and OMIM (https:// Analysis of alternatively spliced mRNA isoforms
www.ncbi.nlm.nih.gov/omim) were used. (iso-mRNAs)

Integrated analysis of functions and interactions For Td3000 & Tu3000 genes, we obtained details of
isoform type from the APPRIS database (http://appris.
We shortlisted enriched biological processes (BPs) bioinfo.cnio.es) and the list of alternative splice events
(GO_BP_all) (Benjamini p-value <0.05), excluding using the SpliceDetector tool (Baharlou et al. 2018).
broad processes such as reproduction, cell differenti- We explored possible cases of iso-mRNAs having
ation, multi-organism process, etc., across the Td250 contradictory expression patterns across the normal
genes from the DAVID 6.8 database (https://david. and NOA conditions. We listed undifferentiated tran-
ncifcrf.gov/gene2gene.jsp). We obtained the protein- scripts (>4TPM and >0.05) corresponding to the top
protein interactions (PPIs) for these BP-related genes 250 NOA up- and down-regulated genes and marked
from the two sources of evidence (experimental and them as partially contradicting cases. Similarly, we
database) with a medium confidence score cut-off (0.4 listed the cases with complete contradiction corre-
and above) using the STRING database (https://string- sponding to the same 500 genes, where one mRNA
db.org/), which was selected based on the results of a isoform was up- or down-regulated, and there was
recent comparative study (Bajpai et al. 2020). The pri- another transcript from the same with significance for
mary functional network was then created using the the opposite type of expression. We obtained the

enriched BPs, PPIs, and pathways of the Td250 genes domain details for such genes from the APPRIS data-
using Cytoscape software ( The base (). The transcript level
six core functional protein clusters were manually GO terms were obtained from Biomart (Ensembl,
marked based on similar processes to visualize them and the
better. data was mapped with the gene-level enriched GO
terms (BPs) from the DAVID 6.8 database (https://
A secondary functional network was created to david.ncifcrf.gov/gene2gene.jsp).
analyze the proteins interacting with the core func-
tional protein clusters. The gene list was obtained for Gene regulatory network analysis
each cluster of the primary functional network and
retrieved the proteins interacting with these primary A literature search was performed to search for
proteins among the top 3000 NOA down-regulated articles reporting human transcription factors (TFs)
genes. Given the limitations of the web interface of and co-transcription factors (cTFs) using recom-
the STRING database (Bajpai et al. 2020), the com- mended search tools (Bajpai et al. 2011). The expres-
plete set of protein-protein interactions (9606.protein.- sion status of these known/predicted transcription
links.detailed.v11.0) was downloaded. However, only factors was compared with that of NOA differentially
the relevant interactions with a high confidence score regulated genes, which had a cut-off logFC ! 2 and
cut-off (0.9 and above) were selected via python
scripts. The top 10 enriched BPs shared among the

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 211

adj p-value 0.05. The TRRUST (Han et al. 2018), Methodology, Writing–original draft, Writing – review &
the database of TF-promoter interactions, was used to editing, Supervision, Project administration: AKK.
collect a list of TF/cTF interactions with their target
genes along with the mode of action, viz., activation Data availability statement
and suppression. Such analysis was done only among
the NOA down-regulated TFs/cTFs, as there were Below are the accession ID details for the sequencing data
very few up-regulated TFs/cTFs. However, the up- disclosed in this paper and deposited in the Sequence Read
regulated TFs or their genes with significant interac- Archive (SRA) AND GEO. SRA project ID: PRJNA556935;

tions with the down-regulated TFs or their genes were GEO ID: GSE216907
also considered. Cytoscape was used to visualize the
network, and cytoHubba was used to identify poten- References
tial key role players. Among the TFs/cTFs down-regu-
lated in NOA condition, domain enrichment analysis Acharya KK, Chandrashekar DS, Chitturi N, Shah H,
was performed using DAVID 6.8 (fcrf. Malhotra V, Sreelakshmi KS, Deepti H, Bajpai A,
gov/gene2gene.jsp). Davuluri S, Bora P, et al. 2010. A novel tissue-specific
meta-analysis approach for gene expression predictions,
Ethical approval initiated with a mammalian gene expression testis data-
base. BMC Genomics. 11(1):467.
This study was approved by the Institutional Ethics
Committee (IEC), IBAB. The study followed the eth- Acharya KK, Govind CK, Shore AN, Stoler MH, Reddi PP.
ical standards laid down in the 1964 Declaration of 2006. cis-requirement for the maintenance of round
Helsinki and its later amendments. Informed consent spermatid-specific transcription. Dev Biol. 295(2):781–
was obtained from all participants. Samples were not 790.
collected from any minor; this study related to sperm-
atogenesis used the help of adult married male donors Agarwal A, Mulgund A, Hamada A, Chyatte MR. 2015. A
only. unique view on male infertility around the
globe. Reproductive biology and endocrinology. Reprod
Acknowledgments Biol Endocrinol. 13:37.

Research projects at IBAB are supported by the Department Ashburner M, Ball CA, Blake JA, Botstein D, Butler H,
of IT, BT, and S&T, Government of Karnataka, India. Staff Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT,
members at the Bio-IT center at IBAB carried out the et al. 2000. Gene ontology: tool for the unification of
RNA-sequencing work. Vashishtkumar Balagannavar, Sailaja biology. The Gene Ontology Consortium. Nat Genet.
Oguru, and S Renushree, from BdataA, contributed to the 25(1):25–29.
work by writing scripts to analyze GO terms, the initial part
of biocuration, and the initial standardization of splice form Baharlou Houreh M, Ghorbani Kalkhajeh P, Niazi A,
analysis, respectively. Ebrahimi F, Ebrahimie E. 2018. SpliceDetector: a software
for detection of alternative splicing events in human and

Disclosure statement model organisms directly from transcript IDs. Sci Rep.
8(1):5063.
AKK was also the founder director of a private limited
company, Shodhaka Life Sciences Pvt. Ltd., during the Bajpai A, Davuluri S, Haridas H, Kasliwal G, Deepti H,
research period, and AKB and SD have also been employees Sreelakshmi KS, Chandrashekar D, Bora P, Farouk M,
of this company. VSS is affiliated with commercial health- Chitturi N, et al. 2011. In search of the right literature
care organizations as well, as listed above. The intellectual search engine(s). Available from Nature Precedings. <>
property of the diagnostic application of shortlisted poten-
tial molecules and primer pairs, including the new control Bajpai AK, Davuluri S, Tiwary K, Narayanan S, Oguru S,
pairs, is being protected via a patent application by IBAB Basavaraju K, Dayalan D, Thirumurugan K, Acharya KK.
and Ankur Andrology & Men’s Health. 2020. Systematic comparison of the protein-protein inter-
action databases from a user’s perspective. J Biomed Inf.
Authors’ contributions 103:103380.

Investigation, Validation, Formal analysis, Visualization, Baksi A, Vasan SS, Dighe RR. 2018. DNA Flow cytometric
Writing – review & editing: GB, KB; Investigation: AKB, analysis of the human testicular tissues to investigate the
SD, SK, DCS, NM; Resources: VSS; Conceptualization, status of spermatogenesis in Azoospermic patients. Sci
Rep. 8(1):11117.

Barcelo M, Mata A, Bassas L, Larriba S. 2018. Exosomal
microRNAs in seminal plasma are markers of the origin
of azoospermia and can predict the presence of sperm in
testicular tissue. Hum Reprod. 33(6):1087–1098.

Beishline K, Kelly C. Olofsson BA, Koduri S, Emrich J,
Greenberg RA, Azizkhan-Clifford J. 2012. Sp1 facilitates
DNA double-strand break repair through a nontranscrip-
tional mechanism. Mol Cell Biol. 32(18):3790–3799.

Bhagwat S, Dalvi V, Chandrasekhar D, Matthew T, Acharya

K, Gajbhiye R, Kulkarni V, Sonawane S, Ghosalkar M,
Parte P. 2014. Acetylated a-tubulin is reduced in

212 G. BALAGANNAVAR ET AL.

individuals with poor sperm motility. Fertil Steril. 101(1): Friedler S, Raziel A, Schachter M, Strassburger D, Bern O,
95–104.e3. Ron-El R. 2002. Outcome of first and repeated testicular
Bray NL, Pimentel H, Melsted P, Pachter L. 2016. Near- sperm extraction and ICSI in patients with non-obstruct-
optimal probabilistic RNA-seq quantification. Nat ive azoospermia. Hum Reprod. 17(9):2356–2361.
Biotechnol. 34(5):525–527.
Cappallo-Obermann H, von Kopylow K, Schulze W, Spiess Fujimoto M, Izu H, Seki K, Fukuda K, Nishida T, Yamada
AN. 2010. A biopsy sample reduction approach to iden- S, Kato K, Yonemura S, Inouye S, Nakai A. 2004. HSF4
tify significant alterations of the testicular transcriptome is required for normal cell growth and differentiation
in the presence of Y-chromosomal microdeletions that during mouse lens development. Embo J. 23(21):4297–
are independent of germ cell composition. Hum Genet. 4306.
128(4):421–431.
Carro MS, Lim WK, Alvarez MJ, Bollo RJ, Zhao X, Snyder Giannopoulou EG, Elemento O, Ivashkiv LB. 2015. Use of
EY, Sulman EP, Anne SL, Doetsch F, Colman H, et al. RNA sequencing to evaluate rheumatic disease patients.
2010. The transcriptional network for mesenchymal Arthritis Res Ther. 17(1):167.
transformation of brain tumours. Nature. 463(7279):318–
325. Gou LT, Kang JY, Dai P, Wang X, Li F, Zhao S, Zhang M,
Chalmel F, Lardenois A, Evrard B, Mathieu R, Feig C, Hua MM, Lu Y, Zhu Y, et al. 2017. Ubiquitination-defi-
Demougin P, Gattiker A, Schulze W, Jegou B, Kirchhoff cient mutations in human Piwi Cause male infertility by
C, et al. 2012. Global human tissue profiling and protein impairing histone-to-protamine exchange during sper-
network analysis reveals distinct levels of transcriptional miogenesis. Cell. 169(6):1090–1104.e13.
germline-specificity and identifies target genes for male
infertility. Hum Reprod. 27(11):3233–3248. Govin J, Caron C, Lestrat C, Rousseaux S, Khochbin S.
Chambers DC, Carew AM, Lukowski SW, Powell JE. 2019. 2004. The role of histones in chromatin remodelling dur-
Transcriptomics and single-cell RNA-sequencing. ing mammalian spermiogenesis. Eur J Biochem. 271(17):
Respirology. 24(1):29–36. 3459–3469.

Chiba K, Enatsu N, Fujisawa M. 2016. Management of non-
obstructive azoospermia. Reprod Med Biol. 15(3):165– Govindkumar B, Kavyashree B, Akhilesh Kumar B,
173. Sravanthi D, Shruthi K, Vasan SS, Madhusudhan M,
Chin CH, Chen SH, Wu HH, Ho CW, Ko MT, Lin CY. Darshan SC, Neelima N, Vashishtkumar B, et al. 2019.
2014. cytoHubba: identifying hub objects and sub-net- Listing candidate diagnostic markers and transcriptomic
works from complex interactome. BMC Syst Biol. 8(S4): exploration of the molecular basis of a type of male infer-
S11. tility (non-obstructive Azoospermia), via next generation
Dai J, Voloshin O, Potapova S, Camerini-Otero RD. 2017. sequencing methods. bioRxiv. 778670.
Meiotic Knockdown and complementation reveals essen-
tial role of RAD51 in mouse spermatogenesis. Cell Rep. Govindkumar B, Kavyashree B, Patel K, Sasidharan K, Siva
18(6):1383–1394. Arumugam T, Thomas L, Praveena BKG, Raksha HN,
Diotallevi M, Checconi P, Palamara AT, Celestino I, Coppo Menon R, Acharya KK. 2022. Ex-Ex Primer: an experi-
L, Holmgren A, Abbas K, Peyrot F, Mengozzi M, Ghezzi mentally validated tool for designing oligonucleotides
P. 2017. Glutathione fine-tunes the innate immune spanning spliced nucleic acid regions from multiple spe-
response toward antiviral pathways in a macrophage cell cies. J Biotechnol. 343:1–6.
line independently of its antioxidant properties. Front.
Immunol. 8:1239. Guo J, Grow EJ, Mlcochova H, Maher GJ, Lindskog C, Nie
Dominiak Z, Huras H, KreR cisz P, Krzeszowski W, X, Guo Y, Takei Y, Yun J, Cai L, et al. 2018. The adult
Szymanski P, Czarnecka K. 2021. Promising results in human testis transcriptional cell atlas. Cell Res. 28(12):
development of male contraception. Bioorg Med Chem 1141–1157.
Lett. 41:128005.
Don J, Stelzer G. 2002. The expanding family of Han H, Cho JW, Lee S, Yun A, Kim H, Bae D, Yang S,
CREB/CREM transcription factors that are involved with Kim CY, Lee M, Kim E, et al. 2018. TRRUST v2: an
spermatogenesis. Mol Cell Endocrinol. 187(1–2):115–124. expanded reference database of human and mouse tran-
Draghici S, Khatri P, Eklund AC, Szallasi Z. 2006. scriptional regulatory interactions. Nucleic Acids Res.
Reliability and reproducibility issues in DNA microarray 46(D1):D380–D386.
measurements. Trends Genet. 22(2):101–109.
Esteves SC. 2015. Clinical management of infertile men He WB, Tu CF, Liu Q, Meng LL, Yuan SM, Luo AX, He
with nonobstructive azoospermia. Asian J Androl. 17(3): FS, Shen J, Li W, Du J, et al. 2018. DMC1 mutation that
459–470. causes human non-obstructive azoospermia and prema-

Fedotova AA, Bonchuk AN, Mogila VA, Georgiev PG. ture ovarian insufficiency identified by whole-exome
2017. C2H2 Zinc Finger proteins: the largest but poorly sequencing. J Med Genet. 55(3):198–204.
explored family of higher eukaryotic transcription factors.
Acta Naturae. 9(2):47–58. Hemati A, Modarressi MH, Kolivand S, Azarnia M. 2020.
Fishel S, Green S, Hunter A, Lisi F, Rinaldi L, Lisi R, Heat shock factor 5 is essential for spermatogenesis in
McDermott H. 1997. Human fertilization with round and mice: detected by a new monoclonal antibody. Iran J
elongated spermatids. Hum Reprod. 12(2):336–340. Basic Med Sci. 23(3):293–297.

Hermann BP, Cheng K, Singh A, Roa-De La Cruz L, Mutoji
KN, Chen I, Gildersleeve C, Lehle H, Mayo JD,
Westernstro€er M, et al. 2018. The Mammalian spermato-
genesis single-cell transcriptome, from spermatogonial
stem cells to spermatids. Cell Rep. 25(6):1650–1667.e8.

Hicks SC, Townes FW, Teng M, Irizarry RA. 2018. Missing
data and technical variability in single-cell RNA-sequenc-
ing experiments. Biostatistics, 19(4):562–578.

Hoopes L. 2008. Introduction to the gene expression and
regulation topic room. Nature Educ. 1(1):160. http://

SYSTEMS BIOLOGY IN REPRODUCTIVE MEDICINE 213

www.nature.com/scitable/topic/gene-expression-and-regu- 2014. SUMO1 in human sperm: new targets, role in
lation-15. motility and morphology and relationship with DNA
Hrdlickova R, Toloue M, Tian B. 2017. RNA-Seq methods damage. Reproduction. 148(5):453–467.
for transcriptome analysis. Wiley interdisciplinary Micati DJ, Hime GR, McLaughlin EA, Abud HE, Loveland
reviews. RNA. 8(1):1364. KL. 2018. Differential expression profiles of conserved
Hu H, Miao YR, Jia LH, Yu QY, Zhang Q, Guo AY. 2019. Snail transcription factors in the mouse testis. Androl.
AnimalTFDB 3.0: a comprehensive resource for annota- 6(2):362–373.

tion and prediction of animal transcription factors. Modrek B, Lee C. 2002. A genomic view of alternative splic-
Nucleic Acids Res. 47(D1):D33–D38. ing. Nat Genet. 30(1):13–19.
Jaksik R, Iwanaszko M, Rzeszowska-Wolny J, Kimmel M. Mohsin M, Singh P, Khan S, Verma AK, Jha R, Alsahli
2015. Microarray experiments and factors which affect MA, Rahmani AH, Almatroodi SA, Alrumaihi F,
their reliability. Biol Direct. 10:46. Kaprwan N, et al. 2022. Integrated transcriptomic and
Jan SZ, Vormer TL, Jongejan A, Ro€ling MD, Silber SJ, de regulatory network analyses uncovers the role of let-7b-
Rooij DG, Hamer G, Repping S, van Pelt AMM. 2017. 5p, SPIB, and HLA-DPB1 in sepsis. Sci Rep. 12(1):11963.
Unraveling transcriptome dynamics in human spermato- Najafabadi HS, Mnaimneh S, Schmitges FW, Garton M,
genesis. Development. 144(20):3659–3673. Lam KN, Yang A, Albu M, Weirauch MT, Radovani E,
Jarvi K, Lo K, Fischer A, Grantmyre J, Zini A, Chow V, Kim PM, et al. 2015. C2H2 zinc finger proteins greatly
Mak V. 2010. CUA Guideline: the workup of azoosper- expand the human regulatory lexicon. Nat Biotechnol.
mic males. Can Urol Assoc J. 4(3):163–167. 33(5):555–562.
Jiang W, Sun H, Zhang J, Zhou Q, Wu Q, Li T, Zhang C, Nakagawa T, Zhang T, Kushi R, Nakano S, Endo T,
Li W, Zhang M, Xia X. 2015. Polymorphisms in Nakagawa M, Yanagihara N, Zarkower D, Nakayama K.
Protamine 1 and Protamine 2 predict the risk of male 2017. Regulation of mitosis-meiosis transition by the ubi-
infertility: a meta-analysis. Sci Rep. 5:15300. quitin ligase b-TrCP in male germ cells. Development.
Kim B, Lee E, Kim, JK. 2019). Analysis of technical and 144(22):4137–4147.
biological variability in single-cell RNA sequencing. Okada H, Tajima A, Shichiri K, Tanaka A, Tanaka K, Inoue
Methods Mol Biol. 1935:25–43.. I. 2008. Genome-wide expression of azoospermia testes
Krausz C, Riera-Escamilla A. 2018. Genetics of male infertil- demonstrates a specific profile and implicates ART3 in
ity. Nat Rev Urol. 15(6):369–384. genetic susceptibility. PLoS Genet. 4(2):e26.
Kumar N, Singh AK. 2015. Trends of male factor infertility, Omolaoye TS, Omolaoye VA, Kandasamy RK, Hachim MY,
an important cause of infertility: a review of literature. J Du Plessis SS. 2022. Omics and male infertility: highlight-
Hum Reprod Sci. 8(4):191–196. ing the application of transcriptomic data. Life, 12(2):280.
L€ahnemann D, K€oster J, Szczurek E, McCarthy DJ, Hicks Ozsolak F, Milos PM. 2011. RNA sequencing: advances,
SC, Robinson MD, Vallejos CA, Campbell KR, challenges and opportunities. Nat Rev Genet. 12(2):87–
Beerenwinkel N, Mahfouz A, et al. Scho€nhuth A, 2020. 98.
Eleven grand challenges in single-cell data science. Park DJ, Nguyen-Dumont T, Kang S, Verspoor K, Pope BJ.
Genome Biol. 21(1):31. 2014. Annokey: an annotation tool based on key term
Lambert SA, Jolma A, Campitelli LF, Das PK, Yin Y, Albu search of the NCBI Entrez Gene database. Source Code

M, Chen X, Taipale J, Hughes TR, Weirauch MT. 2018. Biol Med. 9(1):15.
The human transcription factors. Cell. 172(4):650–665. Qin J, Huang T, Wang J, Xu L, Dang Q, Xu X, Liu H, Liu
Lambrot R, Jones S, Saint-Phar S, Kimmins S. 2012. Z, Shao C, Zhang X. 2022. RAD51 is essential for sperm-
Specialized distribution of the histone methyltransferase atogenesis and male fertility in mice. Cell Death Discov.
Ezh2 in the nuclear apical region of round spermatids 8(1):118.
and its interaction with the histone variant H1t2. J Ramasamy R, Bakırcıoglu ME, Cengiz C, Karaca E, Scovell
Androl. 33(5):1058–1066. J, Jhangiani SN, Akdemir ZC, Bainbridge M, Yu Y, Huff
Lea K, Qu L, Whitley P, Schageman J, Gu J, Bramlett K. C, et al. 2015. Whole-exome sequencing identifies novel
2011. Transcriptome analysis using next-generation homozygous mutation in NPAS2 in family with nonob-
sequencing technology. J Biomol Tech. 22(Suppl):S44– structive azoospermia. Fertil Steril. 104(2):286–291.
S45. Ramasamy R, Lin K, Gosden LV, Rosenwaks Z, Palermo
Li Z, Chen S, Yang Y, Zhuang X, Tzeng CM. 2018. Novel GD, Schlegel PN. 2009. High serum FSH levels in men
biomarker ZCCHC13 revealed by integrating DNA with nonobstructive azoospermia does not affect success
methylation and mRNA expression data in non-obstruct- of microdissection testicular sperm extraction. Fertil
ive azoospermia. Cell Death Discov. 4:36. Steril. 92(2):590–593.
Lowe R, Shirley N, Bleackley M, Dolan S, Shafee T. 2017. Rathke C, Baarends WM, Awe S, Renkawitz-Pohl R. 2014.
Transcriptomics technologies. PLoS Comput Biol. 13(5): Chromatin dynamics during spermiogenesis. Biochim
e1005457. Biophys Acta. 1839(3):155–168.
Malcher A, Rozwadowska N, Stokowy T, Kolanowski T, Ravasi T, Suzuki H, Cannistraci CV, Katayama S, Bajic VB,
Jedrzejczak P, Zietkowiak W, Kurpisz M. 2013. Potential Tan K, Akalin A, Schmeier S, Kanamori-Katayama M,
biomarkers of nonobstructive azoospermia identified in Bertin N, et al. Hayashizaki Y, 2010. An atlas of com-
microarray gene expression analysis. Fertil Steril. 100(6): binatorial transcriptional regulation in mouse and man.
1686–1694.e947. Cell. 140(5):744–752.
Marchiani S, Tamburrino L, Ricci B, Nosi D, Cambi M, Razavi SM, Sabbaghian M, Jalili M, Divsalar A,
Piomboni P, Belmonte G, Forti G, Muratori M, Baldi E. Wolkenhauer O, Salehzadeh-Yazdi A. 2017.

214 G. BALAGANNAVAR ET AL.

Comprehensive functional enrichment analysis of male Tang D, Huang Z, He X, Wu H, Peng D, Zhang L, Zhang
infertility. Sci Rep. 7(1):15778. X. 2018. Altered miRNA profile in testis of post-cryp-

Reddi PP, Shore AN, Acharya KK, Herr JC. 2002. torchidopexy patients with non-obstructive azoospermia.
Transcriptional regulation of spermiogenesis: insights Reprod Biol Endocrinol. 16(1):78.
from the study of the gene encoding the acrosomal
protein SP-10. J Reprod Immunol. 53(1–2):25–36. The Gene Ontology Consortium. 2019. The Gene Ontology
(01)00104-8. Resource: 20 years and still GOing strong. Nucleic Acids
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Res. 47(D1):D330–D338.
Smyth GK. 2015. Limma powers differential expression
analyses for RNA-sequencing and microarray studies. Thomas K, Wu J, Sung DY, Thompson W, Powell M,
Nucleic Acids Res. 43(7):e47. McCarrey J, Gibbs R, Walker W. 2007. SP1 transcription
Robertson MJ, Kent K, Tharp N, Nozawa K, Dean L, factors in male germ cell development and differenti-
Mathew M, Grimm SL, Yu Z, Legare C, Fujihara Y, et al. ation. Mol Cell Endocrinol. 270(1–2):1–7.
2020. Large-scale discovery of male reproductive tract-
specific genes through analysis of RNA-seq datasets. Wang JH, Zhao LF, Lin P, Su XR, Chen SJ, Huang LQ,
BMC Biol. 18(1):103. Wang HF, Zhang H, Hu ZF, Yao KT, et al. 2014.
Sabetian S, Shamsir MS. 2016. Systematic analysis of protein GenCLiP 2.0: a web server for functional clustering of
interaction network associated with azoospermia. Int J genes and construction of molecular networks based on
Mol Sci. 17(11):1857. free terms. Bioinformatics. 30(17):2534–2536.
Sandhu KS, Acharya KK. 2005. ExPrimer: to design primers
from exon-exon junctions. Bioinformatics. 21(9):2091– Wang M, Liu X, Chang G, Chen Y, An G, Yan L, Gao S,
2092. Xu Y, Cui Y, Dong J, et al. Qiao J, 2018. Single-cell RNA
Saremi A, Esfandiari N, Salehi N, Saremi MR. 2002. The sequencing analysis reveals sequential cell fate transition
first successful pregnancy following injection of testicular during human spermatogenesis. Cell Stem Cell. 23(4):
round spermatid in Iran. Arch Androl. 48(4):315–319. 599–614.e4.
Sassone-Corsi P. 1997. Transcriptional checkpoints deter-
mining the fate of male germ cells. Cell. 88(2):163–166. Weissig H, Narisawa S, Sikstro€m C, Olsson PG, McCarrey
Shami AN, Zheng X, Munyoki SK, Ma Q, Manske GL, JR, Tsonis PA, Del Rio-Tsonis K, Millan JL. 2003. Three
Green CD, Sukhwani M, Orwig KE, Li JZ, Hammoud SS. novel spermatogenesis-specific zinc finger genes. FEBS
2020. Single-cell RNA sequencing of human, macaque, Lett. 547(1–3):61–68. (03)00669-0.
and mouse testes uncovers conserved and divergent fea-
tures of mammalian spermatogenesis. Dev Cell. 54(4): Wilkinson L. 2011. ggplot2: elegant graphics for data ana-

529–547.e12. lysis by WICKHAM, H. Biometrics. 67(2):678–679.
Schmittgen TD, Livak KJ. 2008. Analyzing real-time PCR
data by the comparative C(T) method. Nat Protoc. 3(6): Wu X, Luo C, Hu L, Chen X, Chen Y, Fan J, Cheng CY,
1101–1108. Sun F. 2020. Unraveling epigenomic abnormality in azoo-
Sohni A, Tan K, Song HW, Burow D, de Rooij DG, spermic human males by WGBS, RNA-Seq, and tran-
Laurent L, Hsieh TC, Rabah R, Hammoud SS, Vicini E, scriptome profiling analyses. J Assist Reprod Genet.
et al. 2019. The neonatal and adult human testis defined 37(4):789–802.
at the single-cell level. Cell Rep. 26(6):1501–1517.e4.
Song H, Wang L, Chen D, Li F. 2020. The function of Pre- Xu WM, Chen J, Chen H, Diao RY, Fok KL, Dong JD, Sun
mRNA alternative splicing in mammal spermatogenesis. TT, Chen WY, Yu MK, Zhang XH, et al. 2011. Defective
Int J Biol Sci. 16(1):38–48. CFTR-dependent CREB activation results in impaired
Soraggi S, Riera M, Rajpert-De Meyts E, Schierup MH, spermatogenesis and azoospermia. PLoS One. 6(5):
Almstrup K. 2021. Evaluating genetic causes of azoosper- e19120.
mia: What can we learn from a complex cellular structure
and single-cell transcriptomics of the human testis? Hum Yusuf D, Butland SL, Swanson MI, Bolotin E, Ticoll A,
Genet. 140(1):183–201. Cheung WA, Zhang XY, Dickman CT, Fulton DL, Lim
Stastna M, Van Eyk JE. 2012. Analysis of protein isoforms: JS, et al. Wasserman WW, 2012. The transcription factor
Can we do it better? Proteomics. 12(19–20):2937–2948. encyclopedia. Genome Biol. 13(3):R24.
Syafruddin SE, Ling S, Low TY, Mohtar MA. 2021. More
than meets the eye: revisiting the roles of heat shock fac- Zhao L, Yao C, Xing X, Jing T, Li P, Zhu Z, Yang C, Zhai
tor 4 in health and diseases. Biomolecules. 11(4):523. J, Tian R, Chen H, et al. 2020. Single-cell analysis of
Tanabe M, Sasai N, Nagata K, Liu XD, Liu PC, Thiele DJ, developing and azoospermia human testicles reveals cen-
Nakai A. 1999. The mammalian HSF4 gene generates tral role of Sertoli cells. Nat Commun. 11(1):5683.
both an activator and a repressor of heat shock genes by
alternative splicing. J Biol Chem. 274(39):27845–27856. Zhu ZJ, Yang S, Li Z. 2015. Transcriptome research on
spermatogenic molecular drive in mammals. Asian J
Androl. 17(6):961–971.

Zhu Z, Li C, Yang S, Tian R, Wang J, Yuan Q, Dong H, He
Z, Wang S, Li Z. 2016. Dynamics of the transcriptome

during human spermatogenesis: predicting the potential
key genes regulating male gametes generation. Sci Rep. 6:
19069.

Zhuang X, Li Z, Lin H, Gu L, Lin Q, Lu Z, Tzeng CM.
2015. Integrated miRNA and mRNA expression profiling
to identify mRNA targets of dysregulated miRNAs in
non-obstructive azoospermia. Sci Rep. 5:7922.