Transcriptomic Analysis Reveals Multilayer Regulation Underlying Development and Pregnancy in the Malayan Pangolin

ABSTRACT

The critically endangered Malayan pangolin ( Manis javanica ) exhibits distinctive physiological adaptations; nevertheless, its molecular mechanisms underlying development and reproduction remain inadequately characterized, thereby impeding effective conservation initiatives. In this study, we conducted a comprehensive transcriptomic analysis of six tissues (intestine, kidney, liver, lung, muscle, and spleen) derived from adult, fetal, and pregnant individuals. The fetal stage demonstrated the greatest abundance of stage‐specific genes, which were predominantly enriched in developmental pathways such as epithelial organization and transmembrane signaling. Differential gene expression (DGE) analysis revealed that genes upregulated in adults were primarily associated with metabolic and immune functions, whereas those upregulated in fetuses were linked to developmental processes and cell cycle regulation. Co‐expression network analysis identified critical hub genes, including GLDC in adults and NCAM1 in fetuses, that are implicated in stage‐specific regulatory mechanisms. Additionally, extensive differential alternative splicing (DAS) events were observed, with exon skipping (SE) representing the most frequent splicing pattern. Integrative analyses suggested that DGE and DAS largely constitute independent regulatory layers; however, a subset of hub genes, such as OBSL1 and TNIP1, exhibited concurrent differential expression and splicing, indicating their potential central roles in coordinating developmental processes. This study presents a detailed transcriptomic atlas of Malayan pangolin development and pregnancy, providing candidate biomarker genes for evaluating reproductive health and developmental status, thereby informing future conservation strategies.

1. Introduction

Malayan pangolins (order Pholidota) represent a distinctive group of mammals characterized by their keratinized scales and specialized myrmecophagous diet (Swart et al. 1999; Pietersen et al. 2016; Yan et al. 2023). Unlike the majority of placental mammals, these animals exhibit large, overlapping keratinous scales and lack dentition, instead utilizing an elongated, muscular tongue to consume ants and termites—adaptations that define a unique dietary niche within mammals (Pietersen et al. 2016). Due to extensive poaching and habitat loss, pangolins are among the most threatened mammalian taxa (Samuel et al. 2023; Kanthaswamy 2024). Despite increasing conservation efforts, captive breeding programs have met with limited success, with only a few documented instances of sustained reproduction. This challenge is likely attributable to a fundamental gap in knowledge concerning pangolin developmental and reproductive biology, including critical processes such as embryogenesis and organ maturation. Therefore, elucidating the gene regulatory mechanisms underlying these processes is imperative to advance conservation and breeding strategies.

Transcriptomic analyses have demonstrated considerable utility in uncovering the molecular mechanisms governing development and reproduction in mammals. For instance, transcriptomic profiling has identified stage‐specific gene clusters associated with maternal recognition of pregnancy and uterine implantation in cattle, highlighting the intricate regulatory networks involved in reproduction (Mamo et al. 2011). Prior transcriptomic investigations in adult pangolins have revealed genes and pathways related to dietary specialization, providing insights into their myrmecophagous adaptations (Mohamed Yusoff et al. 2016; Ma et al. 2017). Furthermore, genomic and transcriptomic studies have uncovered evolutionary adaptations in Malayan pangolins, such as the pseudogenization of genes involved in tooth development and vision, reflecting their edentulous condition and nocturnal habits (Choo et al. 2016).

Beyond differential gene expression (DGE), transcriptomic data also enable the analysis of differential alternative splicing (DAS), a critical post‐transcriptional regulatory mechanism that enhances transcriptomic and proteomic diversity in eukaryotes (Yang et al. 2016; Koch 2017). By producing multiple protein isoforms from a single gene, alternative splicing contributes to tissue specificity, developmental transitions, and physiological adaptations (Lopez 1998; Barbosa‐Morais et al. 2012). Accumulating evidence underscores the importance of alternative splicing in embryonic development, organogenesis, and reproductive processes such as placentation and maternal‐fetal communication (Xing et al. 2020; Shen et al. 2021). However, the role of alternative splicing in pangolins remains largely unexplored, particularly across different developmental and physiological stages. Integrating analyses of DGE and DAS can thus provide a more comprehensive understanding of transcriptional regulation by revealing regulatory layers that might otherwise be overlooked.

In light of the limited molecular understanding of pangolin development and reproduction, coupled with the demonstrated efficacy of transcriptomic approaches in elucidating molecular mechanisms, the present study undertook a comparative transcriptomic analysis of fetal, adult, and pregnant individuals to identify associated molecular signatures. This integrative investigation advances our comprehension of pangolin biology and furnishes genomic resources that may facilitate future conservation and breeding initiatives.

2. Materials and Methods

2.1. Data Collection

This study utilized a total of 24 adult Malayan pangolins, including six pregnant specimens. These animals were confiscated by the Customs and Forestry Department of Guangdong Province between March 2019 and April 2020 and had been previously incorporated in our previous investigations (Xiao et al. 2020; Cui et al. 2023; Huang et al. 2023; Liang et al. 2023). Despite rescue efforts, all individuals succumbed. Tissue samples comprising lung, liver, spleen, skeletal muscle, kidney, and intestine were harvested and preserved at −80°C. At the time of sample collection, the prospective utility of these specimens for transcriptomic analysis was not fully appreciated. As a result, tissue collection was not conducted in a systematic manner, resulting in incomplete sample availability across individual specimens. The secondary use of these tissues received approval from the Guangzhou Wildlife Research Center (Permit No. GZZOO20190827C). In total, 101 tissue samples underwent transcriptome sequencing. A portion of these datasets had been published previously, albeit generated for different research aims. For the present study, these prior datasets were integrated with newly sequenced samples to enable a comprehensive analysis (Table S1).

2.2. RNA Sequencing

Total RNA was extracted employing TRIzol reagent (Invitrogen), followed by treatment with DNase I to eliminate contaminating genomic DNA. Ribosomal RNA (rRNA) was depleted prior to library preparation. Strand‐specific complementary DNA (cDNA) libraries were constructed using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina (NEB, USA) in accordance with the manufacturer's protocol, with unique index codes assigned to each sample. Sequencing was performed on an Illumina NovaSeq 6000 platform, generating 150 base pair paired‐end reads. Each sample produced approximately 12 gigabases of raw sequencing data (Table S1). All RNA‐seq libraries analyzed in this study were prepared and sequenced concurrently within a single batch, employing uniform protocols. A subset of the data (PRJNA901878 and PRJNA939083) has been previously published in our earlier works (Cui et al. 2023; Liang et al. 2023), whereas the novel data generated for the current investigation have been deposited under accession number PRJNA1362967.

2.3. Bioinformatic Analysis

Adaptor sequences and low‐quality reads were removed using fastp (version 0.19.7) (Chen et al. 2018) with the following parameters: –q 20, −f 8, −t 1, −F 8, −T 1, −5, −l 75. The Malayan pangolin reference genome (YNU_ManJav_2.0) was served as the reference for RNA‐seq read alignment. Sequence mapping was conducted using Hisat2 (Kim et al. 2019) with the parameter—rna‐strandness RF. Gene‐level quantification was performed with featureCounts (Liao et al. 2014), part of the Subread package version 2.0.0, with parameters –s 2, −t exon, −g gene_id, −p. Principal component analysis (PCA) (Jolliffe 2002) and permutational multivariate analysis of variance (PERMANOVA) (Anderson 2017) were applied to variance‐stabilizing transformed (VST) count data to evaluate sample clustering and to partition variance attributable to tissue type, developmental stage, and sequencing batch. Genes were considered expressed if they exhibited transcripts per million (TPM) values ≥ 1 in at least 50% of replicates per developmental stage (with the exception of adult intestine samples, for which 100% of replicates [n = 2] was required). The robustness of stage‐specific gene expression classifications was further validated through sensitivity analyses employing a range of TPM thresholds (0.5–5) and varying sample proportion cutoffs (50%–100%).

Differential gene expression (DGE) analysis across various tissues and comparative groups was carried out using the DESeq2 R package (version 1.10.1) (Anders and Huber 2010). In the comparison between pregnant and fetal samples, the paired nature of the data, reflecting mother–offspring relationships, was explicitly incorporated into the model. Specifically, a paired DESeq2 model was employed, incorporating mother identity as a fixed effect (design = ~mother_id + condition). Only pairs for which both maternal and fetal spleen samples were available were included; fetal samples lacking a corresponding maternal counterpart were excluded from this paired analysis. To account for multiple hypothesis testing, p‐values were adjusted using the Benjamini–Hochberg method to control the false discovery rate (FDR). Genes exhibiting an absolute log₂ fold change greater than 2 and an adjusted p‐value below 0.05 were classified as differentially expressed.

Co‐expression networks were constructed using the WGCNA R package. For each tissue and comparison, an appropriate soft‐thresholding power was selected to approximate a scale‐free network topology. Subsequently, a signed adjacency matrix was computed, and the topological overlap measure (TOM) was employed to detect co‐expression modules through dynamic tree cutting. Within each module, module membership (kME) was calculated to quantify the correlation between individual gene expression profiles and the module eigengene. Hub genes were defined as the top 20 genes exhibiting the highest kME values within their respective modules.

Alternative splicing (AS) events were identified using the rMATS software (Shen et al. 2014) with the parameters set to –t paired—readLength 150. Normalized percent spliced‐in (PSI) values were computed, and AS events with an absolute ΔPSI greater than 0.1 and FDR less than 0.05 were considered statistically significant.

Functional annotation of protein sequences was performed via BLASTp (version 2.7.1+) alignment against the Swiss‐Prot database with default parameters. Gene Ontology (GO) enrichment analyses for both DEGs and DAS genes were conducted using the clusterProfiler R package (Yu et al. 2012). Visualization of gene expression patterns was achieved through heatmaps generated with pheatmap (Kolde 2025), while dot plots were created using ggplot2 (Wickham 2016).

3. Results

3.1. Quality Control and Criteria for Defining Expressed Genes

All RNA sequencing (RNA‐seq) libraries were prepared and sequenced simultaneously within a single batch, employing uniform protocols. The three BioProject accession numbers (PRJNA901878, PRJNA939083, and the newly generated dataset PRJNA1362967) correspond to distinct submissions associated with previous publications (Cui et al. 2023; Liang et al. 2023); however, all raw sequencing data were produced concurrently under identical experimental conditions. To examine the relationships among samples and to detect potential technical variability, PCA was conducted. As illustrated in Figure S1, samples predominantly clustered according to developmental stage (fetal, adult, and pregnant) and tissue type (lung, liver, spleen, muscle, kidney, and intestine), with no evident systematic segregation based on BioProject identification. To quantitatively assess group differentiation, PERMANOVA was performed. Tissue identity accounted for 52.0% of the observed variance (R ² = 0.520), developmental stage explained 12.7% (R ² = 0.127), and sequencing batch contributed only 7.7% (R ² = 0.077). This distribution aligns with biological expectations: distinct clustering by tissue type was observed, developmental stages exhibited moderate separation (notably, adults and pregnant individuals showed physiological similarity), and batch effects were minimal.

Establishing a rigorous and reproducible definition of “expressed” genes is critical for the identification of stage‐specific gene expression. Accordingly, a gene was classified as expressed if it demonstrated transcripts per million (TPM) values equal to or exceeding 1 in at least 50% of biological replicates within a given developmental stage. For the adult intestine, where only two replicates were available, a more stringent criterion was applied, requiring expression in 100% of samples to reduce the likelihood of stochastic dropout. Additionally, a sensitivity analysis was conducted across multiple TPM thresholds (0.5, 1, 2, 5) and varying proportions of sample expression requirements (50%, 75%, 100%). As detailed in Figure S2, the relative ranking of stage‐specific gene counts (e.g., fetal > adult, and where applicable fetal > pregnant > adult) remained consistent across all sample proportion thresholds in lung, kidney, liver, and muscle tissues. In spleen and intestine, the order between two closely related categories (fetal vs. pregnant in spleen; fetal vs. adult in intestine) reversed at certain thresholds; however, given the quantitative proximity of these categories, such reversals are unlikely to reflect genuine threshold dependencies. These findings substantiate the robustness of the stage‐specific gene classification criteria employed in this study.

3.2. Stage‐Specific Gene Expression Throughout Development

Genes classified as stage‐specific were those expressed exclusively at a single developmental stage (adult, fetal, or pregnant). The Venn diagrams illustrate both the overlap and the distinctiveness of gene expression profiles among these stages (Figure 1). Notably, the fetal stage demonstrated the highest number of uniquely expressed genes across six tissues examined, indicating pronounced transcriptional activation during organogenesis. For example, in the intestine, 2705 out of 19 073 genes (14.2%) were uniquely expressed in the fetal stage, compared to only 464 genes (2.4%) specific to the adult stage. And 2328 of 19 444 genes (12.0%) were fetal‐specific, whereas 626 genes (3.2%) were adult‐specific in kidney. Similarly, in the liver, 2411 of 17 255 genes (14.0%) were fetal‐specific, whereas 604 genes (3.5%) were adult‐specific. For lung, muscle, and spleen tissues, samples from adult, fetal, and pregnant stages were available. In the lung, 55 of 20 592 genes (0.3%) were adult‐specific, 2691 genes (13.1%) were fetal‐specific, and 282 genes (1.4%) were pregnant‐specific. In muscle tissue, 100 of 18 222 genes (0.5%) were adult‐specific, 2824 genes (15.5%) fetal‐specific, and 130 genes (0.7%) pregnant‐specific. In the spleen, 53 genes (0.3%) were adult‐specific, 1082 genes (5.6%) fetal‐specific, and 879 genes (4.6%) pregnant‐specific.

FIGURE 1.

Venn diagram depicting the number of genes expressed at different developmental stages across various tissues. The red, blue, and green circles correspond to adult, fetal, and pregnant samples, respectively.

To elucidate molecular functions conserved across developmental stages in the pangolin, we identified genes expressed consistently across all stages in multiple tissues (Figure 2A). Functional enrichment analysis revealed that these shared genes are significantly associated with fundamental biological processes critical for basic cellular function. The most consistently enriched GO terms encompassed RNA processing, protein synthesis, and post‐translational modification, mitochondrial function, vesicle‐mediated transport, and structural components of the spliceosome and ribosome (Figure 2A). These findings suggest that a core set of genes involved in essential cellular machinery is stably maintained throughout development.

FIGURE 2.

GO term enrichment analysis of overlapping and stage‐specific gene expression across six tissues. (A) Overlapping expressed genes. (B) Stage‐specific expressed genes. GO terms are classified into three categories: Biological process, cellular component, and molecular function. Owing to space limitations, only the top 20 GO terms with the lowest adjusted p‐values are presented for each category. The x‐axis denotes the six tissues examined: Intestine, kidney, liver, lung, muscle, and spleen, while the y‐axis lists the enriched GO terms. The intensity of the color corresponds to the ‐log10 of the adjusted p‐value, and the size of each circle reflects the number of genes associated with the respective GO term.

Conversely, stage‐specific genes displayed distinct functional enrichments reflective of their respective developmental phases (Figure 2B). Adult‐specific genes were predominantly enriched in immune regulatory processes, such as T cell activation and B cell proliferation (Figure 2B). Fetal‐specific genes were primarily enriched in developmental and morphogenetic processes (Figure 2B). Genes specific to pregnancy did not exhibit statistically significant functional enrichment.

3.3. Differential Gene Expression Alterations Across Developmental Stages

To elucidate transcriptional dynamics throughout development, DGE analyses were performed across various tissue types and developmental stages in pangolins. GO enrichment analysis indicated extensive shifts in functional pathways. In the comparison between adult and fetal stages, genes upregulated in adults were primarily enriched in metabolic, immune, and homeostatic processes (Figure 3A), whereas genes upregulated in fetuses were predominantly associated with developmental, cell cycle, and neural functions (Figure 3B). Similarly, in the comparison between pregnant and fetal stages, genes upregulated in pregnant individuals were enriched for immune activation and homeostasis (Figure 3C), while those upregulated in fetuses were linked to developmental and neural processes (Figure 3D). The comparison between adult and pregnant stages revealed a limited number of DEGs, precluding detailed functional analysis (Figure S3).

FIGURE 3.

GO term enrichment analysis of genes upregulated in adult, pregnant, and fetal samples. (A) The enrichment of genes upregulated in adults relative to fetuses. (B) The enrichment of genes upregulated in fetuses compared to adults. (C) The enrichment of genes upregulated in pregnant individuals relative to fetuses. (D) The enrichment of genes upregulated in fetuses compared to pregnant individuals. GO terms are classified into three categories: Biological process, cellular component, and molecular function. Due to limitations in space, only the top 20 GO terms with the most significant adjusted p‐values are presented for each category. The x‐axis corresponds to six tissue types—intestine, kidney, liver, lung, muscle, and spleen—while the y‐axis enumerates the enriched GO terms. The color intensity represents the negative logarithm (base 10) of the adjusted p‐value, and the size of each circle indicates the number of genes associated with the respective GO term.

Through co‐expression network analysis of DEGs across adult, fetal, and pregnant stages, key hub genes were identified (Figure 4). In the comparison between adult and fetal stages, prominent adult hub genes included ISG20, associated with immune regulation; UACA, an immune‐related cytoskeletal protein; and MAP2K7, a critical kinase within the JNK signaling pathway. Conversely, fetal hub genes comprised CLSTN3, involved in neural cell adhesion and synaptic function; RUVBL2, a factor in chromatin remodeling; and FEN1, which plays a role in DNA replication and repair (Figure 4A). In the pregnant versus fetal comparison, IL2RB, implicated in immune regulation, emerged as a central hub within the pregnant network, while NCAM1, related to neural cell adhesion, was highly connected in the fetal network (Figure 4B). In the adult versus pregnant comparison, FGFRL1, which contributes to skeletal development, was identified as a prominent hub gene in the adult network, whereas ERCC1, a DNA repair enzyme, served as a key node in the pregnant network (Figure 4C). These stage‐specific hub genes, alongside others involved in developmental regulation, are predicted to play pivotal regulatory roles during critical transitions in pangolin development.

FIGURE 4.

Heatmap illustrating the hub genes. Panel (A) compares adult and fetal samples, panel (B) contrasts adult and pregnant samples, and panel (C) examines fetal versus pregnant samples. The color intensity corresponds to the log₂ fold change (log₂FC) of each gene in the respective comparison.

3.4. Differential Alternative Splicing During Development

The frequency of DAS events varied among tissues and developmental stage comparisons (Table 1). In the comparison between adult and fetal stages, the intestine exhibited the highest number of DAS events, comprising 3730 out of a total of 15 731 events (23.7%). In the fetal versus pregnant comparison, muscle demonstrated the greatest number of DAS events, with 3291 out of 7249 total events (45.4%). Meanwhile, in the adult versus pregnant comparison, the spleen showed the largest proportion of DAS events, accounting for 2050 out of 3833 total events (53.5%). These findings suggest that no single tissue consistently manifests the highest number of DAS events across different developmental stage comparisons.

TABLE 1.

Distribution of DAS events among various tissues.

Group	Intestine	Kidney	Liver	Lung	Muscle	Spleen	Total
Adult versus fetus	3730 (23.7%)	2160 (13.7%)	1301 (8.3%)	2997 (19.1%)	2930 (18.6%)	2613 (16.6%)	15 731
Pregnant versus fetus	/	/	/	2317 (32.0%)	3291 (45.4%)	1641 (22.6%)	7249
Adult versus pregnant	/	/	/	981 (25.6%)	802 (20.9%)	2050 (53.5%)	3833

DAS events were categorized into five principal types: exon skipping, alternative 3′ splice site, alternative 5′ splice site, mutually exclusive exons, and retained intron. Among these, exon skipping was the most predominant across all comparisons, representing more than 55.6% of total events in the majority of tissues (Table 2). This predominance likely reflects the critical role of exon skipping in modulating protein isoform diversity during developmental processes and physiological transitions.

TABLE 2.

Five types of DAS among various tissues.

Group	Tissue	A3SS	A5SS	MXE	RI	SE	Total
Adult versus fetus	Intestine	287 (7.7%)	206 (5.5%)	614 (16.5%)	40 (1.1%)	2583 (69.2%)	3730
	Kidney	163 (7.5%)	89 (4.1%)	611 (28.3%)	24 (1.1%)	1273 (58.9%)	2160
	Liver	98 (7.5%)	74 (5.7%)	371 (28.5%)	13 (1.0%)	745 (57.3%)	1301
	Lung	184 (4.9%)	108 (3.6%)	1004 (33.5%)	34 (1.1%)	1667 (55.6%)	2997
	Muscle	173 (5.9%)	131 (4.5%)	965 (32.9%)	31 (1.1%)	1630 (55.6%)	2930
	Spleen	139 (5.3%)	78 (3.0%)	920 (35.2%)	9 (0.3%)	1467 (56.1%)	2613
Adult versus pregnant	Lung	100 (2.7%)	43 (4.4%)	245 (25.0%)	12 (1.2%)	581 (59.2%)	981
	Muscle	57 (7.1%)	38 (4.7%)	188 (23.4%)	13 (1.6%)	506 (63.1%)	802
	Spleen	126 (6.1%)	81 (4.0%)	676 (33.0%)	20 (1.0%)	1147 (56.0%)	2050
Fetus versus pregnant	Lung	164 (4.4%)	110 (4.7%)	594 (25.6%)	32 (1.4%)	1417 (61.2%)	2317
	Muscle	231 (7.0%)	153 (4.6%)	768 (23.3%)	19 (0.6%)	2120 (64.4%)	3291
	Spleen	130 (7.9%)	66 (4.0%)	396 (24.1%)	13 (0.8%)	1036 (63.1%)	1641

Abbreviations: A3SS, alternative 3′ splice site; A5SS, alternative 5′ splice site; MXE, mutually exclusive exons; RI, retained intron; SE, exon skipping.

GO enrichment analysis of genes undergoing differential alternative splicing in the adult versus fetal comparison revealed significant enrichment in biological processes associated with cell cycle regulation, ATPase activity, cytoskeletal organization, and protein regulatory networks across multiple tissues, including intestine, kidney, liver, lung, and muscle (Figure 5A). Similarly, GO enrichment analysis of DAS genes in the pregnant versus fetal comparison identified overrepresented processes related to cell cycle regulation, cytoskeletal reorganization, and regulatory interactions involving RNA and proteins within lung, spleen, and muscle tissues (Figure 5B). Notably, no significantly enriched GO terms were observed in the adult versus pregnant comparison.

FIGURE 5.

GO term enrichment analysis for genes exhibiting alternative splicing across various tissues. Panel (A) compares GO term enrichment between adult and fetal samples, while panel (B) contrasts fetal and pregnant samples. In each comparison, GO terms are classified into three categories: biological process, cellular component, and molecular function. Due to limitations in figure space, only the top 20 GO terms with the lowest adjusted p‐values are shown for each category. The x‐axis denotes the tissues analyzed, including intestine, kidney, liver, lung, muscle, and spleen, whereas the y‐axis lists the enriched GO terms. The intensity of the color corresponds to the negative logarithm (base 10) of the adjusted p‐value, and the size of each circle reflects the number of genes associated with the respective GO term. Notably, no significantly enriched GO terms were identified in the comparison between adult and pregnant samples.

3.5. Integrated Analysis Uncovers Complex Multi‐Layered Gene Regulatory Mechanisms During Development

To investigate the interactions among various layers of gene regulation throughout developmental stages, comprehensive integrative analyses were conducted on stage‐specific genes, DEGs, and DAS genes, with particular emphasis on their overlaps across distinct regulatory modalities.

The integrative analysis identified distinct patterns of overlap between stage‐specific genes and DEGs across developmental transitions (Table S2). Notably, fetal stages demonstrated a substantial degree of overlap with DEGs, suggesting that numerous stage‐specific transcriptional programs are actively engaged and dynamically regulated during early development (Table S2). In contrast, adult and pregnant stages exhibited markedly lower levels of overlap. Furthermore, the comparison between adult and pregnant stages revealed minimal shared genes across all tissues analyzed (Table S2). This pattern of coordinated stage‐specific expression coupled with heightened gene activity highlights the likely importance of these genes in meeting the functional demands of fetal development.

A restricted subset of genes demonstrated simultaneous differential expression and DAS across various developmental stages (Table S3). For instance, OBSL1 in the fetal intestine, SOD2 in the fetal lung, and USP13 in fetal muscle exhibited concurrent changes in both transcriptional activity and splicing patterns when comparing adult and fetal samples (Figure 6A). Likewise, PPM1N (adult versus pregnant, Figure 6B) and NCAM1 (fetus vs. pregnant, Figure 6C) showed dual regulatory modifications. Additionally, weighted gene co‐expression network analysis (WGCNA) identified highly interconnected hub genes that also experienced significant alternative splicing events. Prominently, NCAPD1 in the fetal liver, RAP1B in the fetal spleen, and SPHK1 in fetal muscle were identified as central nodes within these networks, characterized by marked splicing alterations (Figure 6D). These hub genes, subject to differential alternative splicing, are likely integral to the regulation of developmental processes. The intersection of these regulatory mechanisms indicates a complex, multi‐layered system orchestrating pangolin development.

FIGURE 6.

Sashimi plots illustrating genes that exhibit both differential expression or serve as hub genes, as well as differential alternative splicing. Panels (A–C) depict the overlap of differentially expressed and differentially alternatively spliced (DAS) genes across the adult versus fetal, adult versus pregnant, and fetal versus pregnant comparisons, respectively. Panel (D) depict the overlap of hub genes and DAS genes. The plots display the relative abundance of RNA‐sequencing reads on the y‐axis across genomic positions on the x‐axis. Splice junctions are represented by arcs annotated with supporting read counts. Color‐coded tracks correspond to sample groups: Red for adult, blue for fetal, and green for pregnant samples.

4. Discussion

The regulation of gene expression throughout development and reproduction is essential for species survival (Clarke 2012; Schall et al. 2019); however, it remains inadequately characterized in endangered pangolins. Advances in high‐throughput sequencing technologies have facilitated comprehensive transcriptomic profiling, enabling systematic characterization of gene expression patterns and alternative splicing events across developmental stages (Bo et al. 2025). Such methodologies are particularly advantageous for non‐model and endangered species, where molecular data are scarce but conservation efforts are imperative (Patnaik et al. 2015; Banerjee et al. 2021). This study presents an extensive transcriptomic analysis encompassing critical developmental and physiological stages—fetal, adult, and pregnant—in the critically endangered Malayan pangolin. By integrating stage‐specific gene expression profiling, DGE, DAS, and WGCNA, we identified complex regulatory mechanisms governing pangolin development and reproduction. These findings illuminate fundamental biological processes unique to this species.

The fetal stage demonstrated the highest number of uniquely expressed genes across various tissues (Figure 1), indicative of pronounced transcriptional activation during organogenesis. GO enrichment analysis revealed that these fetal‐specific genes participate in diverse developmental and morphogenetic processes (Figure 2B), consistent with observations in humans and mice (Zhang et al. 2009; He et al. 2020; Wang et al. 2021; Qiu et al. 2022). Conversely, genes uniquely expressed in adults were enriched for immune regulatory functions (e.g., T cell activation, B cell proliferation, cytokine activity) and metabolic processes such as positive regulation of glucose import (Figure 2B), reflecting conserved patterns also reported in human and murine models (Wang et al. 2021; Qiu et al. 2022).

Differential gene expression analysis further delineated stage‐specific functional transitions. Relative to the fetal stage, immune‐related genes were upregulated in adults, including pregnant individuals, reflecting the increased necessity for environmental defense postnatally. This pattern aligns with findings in humans and mice (Wang et al. 2021; Qiu et al. 2022). In contrast, the reduced expression of immune genes in fetuses corresponds to the protected intrauterine environment, where heightened immune activation could compromise maternal‐fetal tolerance (Osburn et al. 1982). The immunological immaturity of neonatal pangolins emphasizes the importance of maternal antibody transfer and minimizing pathogen exposure in captive breeding programs. However, we have to notice that all samples were obtained from confiscated, smuggled pangolins that died despite rescue efforts. These individuals experienced prolonged transport stress, malnutrition, dehydration, and subclinical or active infections prior to sampling. Such conditions should influence the gene expression more or less. The pangolins are endangered protected animal. Study gene expression in various tissues using healthy pangolins is impossible. Our data still provide valuable information about their gene expression pattern. The fetal stage was characterized by elevated expression of genes implicated in neural development and tissue morphogenesis, indicative of active nervous system and organogenesis. This observation is corroborated by single‐cell studies in mice, which demonstrated that over 40% of fetal cell types are associated with the central nervous and hematopoietic systems, highlighting the intensive neural developmental activity during embryogenesis (He et al. 2020). Given the sensitivity of mammalian fetal neurodevelopment to environmental perturbations such as stress and noise, our genetic findings support that keeping pregnant females in stable, low‐stress environments is beneficial for fetal development.

Alternative splicing plays a crucial role in developmental regulation, as evidenced by the identification of numerous DAS events across developmental stages. The predominance of skipped exon (SE) events (Table 2) highlights its significant contribution to proteomic diversity during development (Barbosa‐Morais et al. 2012). The minimal overlap observed between DEGs and DAS genes suggests that transcriptional and post‐transcriptional regulatory mechanisms largely function independently, thereby facilitating precise spatiotemporal control of gene expression.

Importantly, a subset of genes exhibited integration of multiple regulatory modalities. For instance, OBSL1 was upregulated and alternatively spliced in the fetal lung, emphasizing its critical role in pulmonary organogenesis, potentially acting as a cytoskeletal scaffold that coordinates cellular remodeling processes (Geisler et al. 2007). Similarly, PPM1N, alternatively spliced during the pregnant stage, is likely involved in the fine‐tuning of phosphorylation‐dependent signaling pathways essential for gestation (Kamada et al. 2020). RAP1B, identified as a central hub gene within the co‐expression network and exhibiting significant differential alternative splicing, functions as a critical regulator of integrin‐mediated adhesion, cellular migration, and immune cell homing (Chu et al. 2008). This finding implies a developmental transition toward precise regulatory mechanisms and active modulation of inflammatory responses in the mature spleen. These genes exemplify key regulatory nodes where multiple layers of control converge to orchestrate intricate developmental processes.

A comparative analysis was conducted between the DEGs, hub genes, and DAS genes identified in this study and pangolin genes previously reported to be under positive selection or to exhibit lineage‐specific conservation (Choo et al. 2016; Yan et al. 2023). This comparison revealed significant overlaps, including genes such as HRNR and IFNG. HRNR, a gene associated with keratinization and epithelial barrier formation (Yan et al. 2023), was markedly upregulated in the fetal lung relative to adult tissue and demonstrated alternative splicing across developmental stages. Given its established role in cornification, the elevated expression of HRNR in the fetal lung likely supports the initial development of the respiratory epithelial barrier during fetal maturation. Furthermore, the alternative splicing observed in adult tissues suggests that HRNR may undergo isoform switching to modify its protein structure in response to mature physiological requirements.

Previous genomic investigations have shown that although several interferon (IFN) genes are retained in pangolins, the IFNA subfamily contains fewer functional copies compared to humans, indicating a potential reduction in the functional capacity of this gene family (Choo et al. 2016). In the present transcriptomic dataset, IFNG was the sole IFN family member significantly upregulated in the pregnant spleen and lung relative to fetal stages. This reliance on a limited subset of interferon types from fetal to mature stages may reflect an intrinsic immune constraint, which could contribute to the heightened susceptibility to infections and reduced survival rates observed in pangolins.

In summary, our comprehensive transcriptomic analysis of pangolins across three developmental stages revealed three principal regulatory mechanisms: stage‐specific gene expression, differential expression, and alternative splicing. By integrating these regulatory patterns, we characterized the molecular features unique to each developmental stage and identified core regulatory genes involved in multiple control pathways. These findings not only provide candidate genes for further functional investigation but also establish a molecular framework that may inform strategies to enhance health and survival in captive breeding programs for this endangered species.

Author Contributions

YuShan Ding: resources. Wu Chen: supervision, writing – review and editing. Miaomiao Jia: writing – original draft, investigation, methodology, validation. Xianghui Liang: writing – original draft, validation, methodology, investigation. Xuelin Xu: resources. Chengwei Xiang: resources. He Wang: resources. Fanghui Hou: resources. Yingjin Mao: resources. Luman Chen: resources. Xuejuan Shen: writing – review and editing, supervision, resources, project administration, funding acquisition, conceptualization. Wanhe Huang: resources. Yongyi Shen: writing – review and editing, supervision. Guangda Yang: resources. Zhiguang Wang: resources.

Funding

This work was supported by the Forestry Science and Technology Innovation project in Guangdong Province (2025KJCX003), Natural Science Foundation of Guangxi ‐ Young Scientists Fund (2024GXNSFBA010369), and Guangxi Key Laboratory of Special Non‐wood Forests Cultivation and Utilization Fund (JA‐25‐04‐02).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Principal Component Analysis (PCA) of transcriptomic profiles. (A) Samples are distinguished by developmental stage, including fetal, adult, and pregnant categories. (B) Samples are differentiated according to sequencing batch, identified by BioProject accession numbers. (C) Samples are categorized based on tissue type, encompassing intestine, kidney, liver, lung, muscle, and spleen.

Figure S2: Sensitivity analysis of stage‐specific gene counts conducted across a range of expression threshold values.

Figure S3: The comparison of DEGs between adult and pregnant pangolins. (A) the GO enrichment of genes upregulated in adults relative to pregnant individuals. (B) the GO enrichment of genes upregulated in pregnant individuals compared to adults.

Table S1: Summary of pangolin transcriptome data.

Table S2: Genes shared between stage‐specific and differentially expressed categories.

Table S3: Genes shared between differentially expressed and differentially alternatively spliced categories.

Data Availability Statement

The data that support the findings of this study are openly available in GenBank at https://www.ncbi.nlm.nih.gov/, reference number PRJNA1362967.