Evolutionary convergence and niche-specific regulation of core HIF1A pathway genes synergistically drive divergent hypoxia adaptation in mammals

Abstract

Mammals native to hypoxic niches thrive in hypoxic environments, unlike humans and non-adapted species which suffer metabolic dysfunction and disease. Despite progress in understanding hypoxic signaling, the evolutionary strategies enabling niche-specific hypoxia adaptation remain unclear. To address this, we analyzed 21 species representing three hypoxic niches (aquatic deep-diving, terrestrial high-altitude, terrestrial burrowing) and non-hypoxic controls based on identified 25 HIF1A pathway-related core genes. In the coding region, selection pressure analysis initially identified key genes: aquatic deep-diving: CREBBP; terrestrial high-altitude: EGLN1, EGLN2, ELOC and TP53; terrestrial burrowing: EGLN3. Convergent amino acid substitution analysis at the protein level validated these findings and further refined the selection to the most representative key genes for each niche (aquatic deep-diving: CREBBP; terrestrial high-altitude: EGLN1; terrestrial burrowing: EGLN3). Subsequently, non-coding region analysis revealed the specific regulatory patterns of the key genes across the three hypoxic niches: aquatic deep-diving corresponds to acute transient hypoxia adaptation, terrestrial high-altitude exhibits chronic fluctuating hypoxia adaptation, and terrestrial burrowing demonstrates long-term stable hypoxia adaptation. In addition, hypoxia adaptation analysis of the solid tumor in several cancers showed that domain mutations in the core gene CREBBP in the HIF1A signaling pathway contribute to the malignant behavior of cancers. This indicates that distinct hypoxic niches and their key genes have evolved unique regulatory mechanisms for adaptation. The study suggests mammals exploit diverse strategies by modulating key HIF1A pathway gene expression, providing novel insights into hypoxic signaling in evolutionary adaptation to oxygen deprivation.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12862-026-02562-4.

Introduction

Mammals exhibit remarkable adaptability to hypoxic environments, yet the evolutionary and regulatory mechanisms underlying niche-specific hypoxia adaptation remain incompletely understood. The HIF1A signaling pathway, whose elucidation was honored with the 2019 Nobel Prize [1], serves as a central regulator of hypoxia adaptation and related diseases. Physiologically, it governs embryonic development [2] and altitude adaptation [3]; pathologically, its dysregulation drives tumor metastasis, ischemia-reperfusion injury, and renal anemia [4, 5], underscoring its critical roles in metabolism, autophagy, and apoptosis [6–9].

In mammals, hypoxic niches could be classified into three primary types: aquatic deep-diving niches, terrestrial high-altitude niches, and terrestrial burrowing niches [10–13] (Table S1). Aquatic deep-diving niches are characterized by profound hypoxia during prolonged breath-hold dives, where blood oxygen partial pressure (PaO₂) can drop dramatically, driving selection for enhanced hemoglobin-oxygen affinity and oxygen storage capacity [10]. Terrestrial high-altitude niches are defined by chronically low atmospheric oxygen partial pressure coupled with cold temperatures and intense UV radiation, with elevations typically exceeding 2,500 m [11]. Terrestrial burrowing niches feature a unique form of hypoxia caused by restricted air exchange in underground tunnels, leading to concurrent hypoxia and hypercapnia with low oxygen and high carbon dioxide levels [12]. Aquatic deep-diving niches are predominantly exemplified by cetartiodactyl species. The ancestors of these species transitioned from terrestrial to marine habitats during the Eocene epoch, and gradually evolved features adapted to the hypoxic deep-sea environment [14]. Convergent evolution and specific selective pressures have been documented in the HIF1A signaling pathway of cetartiodactyls, which significantly enhances their hypoxic tolerance [15–18]. Upregulation of downstream glycolytic genes enables them to compensate for insufficient oxygen supply via enhanced anaerobic metabolism [19]. Similarly, in yaks, the high expression of HIF1A across multiple organs is closely associated with their hypoxic adaptive capacity [20–23]. Terrestrial burrowing niches are typified by species such as naked mole-rats and Damara mole-rats. The adaptive strategy of these species centers on “oxygen consumption reduction”. They exhibit a substantially reduced HIF1A degradation rate, thereby mitigating cellular injury [24, 25]. These hypoxia-adapted mammal species evolve sophisticated hypoxia adaptations, yet humans and most other mammals suffer metabolic dysregulation or life-threatening pathologies in low-oxygen environments. The cellular oxygen-sensing and hypoxic response are predominantly mediated by the conserved hypoxia-inducible factor 1 (HIF1) signaling cascade. Under normoxic conditions, HIF1A is hydroxylated by HIF1AN/FIH and PHDs, promoting its degradation. Hypoxia suppresses these enzymes, allowing HIF1A to stabilize, translocate to the nucleus, dimerize with HIF1B, and activate downstream target genes via hypoxia response elements (HREs). In species lacking hypoxia tolerance, prolonged or intense hypoxic stimulation leads to dysregulated HIF1A signaling and disrupted metabolic homeostasis, manifesting as a spectrum of pathological phenotypes. Acute mountain sickness (AMS) exemplifies such hypoxia-induced metabolic disorders. Hypoxia can trigger a series of pathological responses in lipid and purine metabolism [26, 27].

Despite progress in HIF1A research, several critical challenges remain. First, most studies have focused on individual hypoxic niches, limiting systematic comparisons across niches [28]. Second, research has emphasized coding-region variations, despite growing evidence that non-coding regions play crucial roles in hypoxia signal transduction [29]. Third, linking evolutionary findings to human hypoxia-related diseases remains an underexplored but promising avenue.

To address these challenges, we integrated evolutionary and regulatory analyses of the HIF1A pathway across three hypoxic niches in mammals. Specifically, we aimed to: (i) identify genes under positive selection and convergent evolution; (ii) characterize niche-specific regulatory patterns in coding and non-coding regions; and (iii) explore links to human hypoxia-related diseases.

Materials and methods

Construction of a dataset of hypoxia-tolerant species and genes

A total of 21 mammalian species from six ecological niches were selected, encompassing three major hypoxic niches (aquatic deep-diving, terrestrial high-altitude, and terrestrial burrowing) and three non-hypoxic niches (aquatic general, semi-aquatic, and terrestrial common) (Table 1). Habitat characteristics of eight mammalian species from the three types of hypoxic niches were primarily retrieved from the existing literature [30–37] (Table S1). According to the integrated data in KEGG [38], Genecards [39], and PathwayCommons [40], the top 25 genes (TOP25 genes) with the strongest interactions with HIF1A were screened as the core genes via protein-protein interaction network analysis using STRING [41] and Cytoscape [42] based on the interaction intensity (score > 0.8). For species with completed genome annotations, the longest transcript sequences of the core genes were downloaded from NCBI and Ensembl databases, others were obtained through homologous alignment using the NCBI online BLAST tool with Mafft and Gblocks [43, 44]. For non-coding region analysis, we focused on the promoter region (1,500 bp upstream to 500 bp downstream of the start codon) to extract them from the selected species.

Table 1.

Summary of species and their associated habitats

Order/suborder	Scientific name	English name	Habitat classification
Cetartiodactyla	Physeter macrocephalus	Sperm whale	Aquatic - Deep-diving
Cetartiodactyla	Monodon monoceros	Narwhal	Aquatic - Deep-diving
Cetartiodactyla	Mesoplodon densirostris	Blainville’s beaked whale	Aquatic - Deep-diving
Cetartiodactyla	Neophocaena	Yangtze finless porpoise	Aquatic - Common
Cetartiodactyla	Lipotes vexillifer	Yangtze river dolphin	Aquatic - Common
Cetartiodactyla	Sousa chinensis	Chinese white dolphin	Aquatic - Common
Carnivora	Enhydra lutris	Northern sea otter	Semi-aquatic - Common
Monotremata	Ornithorhynchus anatinus	Platypus	Semi-aquatic - Common
Sirenia	Trichechus manatus	Florida manatee	Semi-aquatic - Common
Artiodactyla	Bos mutus	Wild yak	Terrestrial - High-altitude
Lagomorpha	Ochotona curzoniae	Black-lipped pika	Terrestrial - High-altitude
Rodentia	Marmota marmota	Alpine marmot	Terrestrial - High-altitude
Rodentia	Heterocephalus glaber	Naked mole-rat	Terrestrial - Burrowing
Rodentia	Fukomys damarensis	Damara mole-rat	Terrestrial - Burrowing
Didelphimorphia	Monodelphis domestica	Gray short-tailed opossum	Terrestrial - Common
Rodentia	Mus musculus	House mouse	Terrestrial - Common
Primates	Homo sapiens	Human	Terrestrial - Common
Diprotodontia	Phascolarctos cinereus	Koala	Terrestrial - Common
Perissodactyla	Equus quagga	Plains zebra	Terrestrial - Common
Carnivora	Panthera tigris	Bengal tiger	Terrestrial - Common
Cingulata	Dasypus novemcinctus	Nine-banded armadillo	Terrestrial - Common

Analysis of evolutionary selective pressure

To decipher evolutionary mechanisms, selection pressure analysis was performed using CODEML: (1) Site-model tests (M0/M3, M1a/M2a, M7/M8, M8a/M8) on HIF1A and TOP25 genes; (2) Branch-site model detection of positive selection across six taxon pairs via ω (dN/dS) comparisons (foreground Ma vs. background Ma0 branches, FDR-corrected 2ΔlnL tests); (3) Branch-model verification of rapidly evolving genes, focusing on hypoxic niche taxa versus non-hypoxic terrestrial taxa to identify key adaptive genes.

Analysis of convergent amino acid substitution

To verify whether the key genes screened at the coding gene level exhibit convergent adaptation at the amino acid level, the DNA2AA function in the FasParser software [45] was utilized to convert nucleotide sequences into amino acid sequences. Four algorithms, including PolyPhen2 [46], Mutation-Taster [47], CADD [48], and DDGun [49], were adopted to predict the deleteriousness of the mutation sites (deleterious mutations often manifest as altered binding energy, functional abnormalities, or pathogenic effects). Ultimately, key adaptive genes at the amino acid level for the three major hypoxic niches were obtained.

Molecular docking

Building on VHL-HIF1A convergent substitutions detected earlier, we analyzed their interaction in hypoxia-adapted species. Using VHL domain sequences [50], and with the aid of HDock [51] and ZDock [52], we performed molecular docking on three hypoxic-niche representatives (the aquatic deep-diving Physeter macrocephalus, the terrestrial high-altitude Bos mutus, and the terrestrial burrowing Heterocephalus glaber) as well as a control species (the terrestrial common Homo sapiens), comparing binding strengths to clarify interaction priorities.

Analysis of gene family expansion and contraction

Six typical representatives from each ecological niche were selected from 21 species, including the aquatic deep-diving Physeter macrocephalus, the aquatic general Neophocaena asiaeorientalis, the semi-aquatic general Trichechus manatus, the terrestrial high-altitude Bos mutus, the terrestrial burrowing Fukomys damarensis, and the terrestrial common Homo sapiens. Orthofinder (v2.5.5) was employed to perform orthologous clustering analysis on the PHDs family genes and the 25 core genes. An orthologous gene table was obtained and then integrated into supergene sequences for subsequent estimation of species divergence times [53]. This study focused on a key event in the divergence of the Artiodactyla clade, the secondary aquatic reentry event. First, the precisely reported fossil-calibrated time of 57.8 million years ago (Mya) from the literature was queried through the Timetree database. Then, the mcmctree program (in PAML v4.10) [54] was used to calculate the precise divergence times of the PHDs family genes and the 25 core genes, and to verify whether their 95% confidence intervals were consistent with the fossil-calibrated time [55]. Next, CAFE5 (v5.0.0) was used to obtain the results of gene family expansion and contraction, and the CafePlotter tool was used for visualization [56]. Meanwhile, to investigate the impact of hypoxic niches on gene evolution, the phylogenetic least-squares method was adopted. The average ω value from the ancestral node to the terminal (root-to-tip ω) was used to depict the gene evolutionary history [57].

Analysis of GC concentrations

We calculated promoter GC content averages per group. We tested for phylogenetic signal and compared inter-group differences via phylogenetic generalized least squares (PGLS) regression (GC Content ~ Group) with Benjamini-Hochberg correction. Results integrated functional groupings, GC means, phylogenetic signals, model parameters, and significance levels.

Analysis of CpG island content

This study conducted CpG island prediction analysis to explore their potential regulatory roles in hypoxia adaptation. The cpgplot tool within the EMBOSS software package [58] was employed for CpG island prediction. Only merged regions with a total length ≥ 200 bp were retained and defined as CpG islands, with the results further outputted.

Analysis of the quantity and positions of TTSs and QSFMs

For each group of promoters, the number of promoters containing triplex target sites (TTSs) and quadruplex-forming sequence motifs (QSFMs) was counted separately. The analysis encompassed both DNA strands and the core promoter research region. TTSs were identified using standardized parameters established in previous studies, requiring a sequence length of at least 15 nucleotides, a guanine content of no less than 50%, and allowing a maximum of one pyrimidine residue. QSFMs were identified through screening using regular expression patterns [59].

DNA stability analysis

DNA double-strand free energy (ΔG) was calculated using a custom script with a 20 bp sliding window (10 bp step), based on nearest-neighbor thermodynamic parameters of 16 dinucleotide steps (derived from 10 unique dinucleotides) [60].

DNA structural property analysis

Promoter DNA shape features (MGW, HelT, ProT, EP) were calculated using DNAshapeR [61] with default parameters.

Visualization analysis of promoter regions

To visualize regulatory element distributions in hypoxic niche species promoters, we used a Python/Matplotlib script [62]. After extracting CpG islands, TTSs, and QSFMs from preprocessed CSV files, we grouped species by niche (deep-diving, high-altitude, fossorial, control).

Expression profile, pathway activity and survival analysis in cancers

We evaluated pathway activity differences using Reverse Phase Protein Array data from The Cancer Proteome Atlas and The Cancer Genome Atlas (TCGA), calculating PAS for 10 cancer pathways across 32 tumor types. PAS differences were assessed via t-test with FDR correction (FDR ≤ 0.05 significant), inferring gene regulatory roles. Survival analysis was performed using R package survival.

Statistical analysis

Group comparisons used Fisher’s exact test (motif frequencies) or Student’s t-test (continuous data), with multiple testing correction (Bonferroni or FDR). Statistical significance was set at α = 0.05. Analyses were performed in Python (SciPy) or R.

Results

Overview of key genes in HIF1A pathway for hypoxia tolerance adaptation

To systematically decipher the molecular mechanisms underlying hypoxia tolerance adaptation in mammals inhabiting multiple ecological niches, we screened 8 species from 3 hypoxic ecological niches and 13 species from 3 non-hypoxic ecological niches (Fig. 1). The top 25 core genes with the highest interaction intensity with HIF1A (> 0.8) were identified based on protein-protein interaction network, and their regulatory networks were analyzed (Fig. 2). These 25 genes are specifically marked in the network diagram of Fig. 2 to visually highlight the Top25 hub genes with the strongest HIF1A interaction. Among them, EGLN1/2/3, VHL, HIF1AN, ARNT, CREBBP and EP300 are recognized as canonical key functional regulators of the HIF1A hypoxia response cascade, which govern the hydroxylation modification, nuclear translocation and transcriptional activation of HIF1A under normoxic and hypoxic conditions respectively; other remaining Top25 genes participate in ubiquitination modification, molecular chaperone regulation and signal crosstalk to jointly mediate mammalian hypoxia adaptation.

Fig. 1.

The flowchart of this study (by Figdraw)

Fig. 2.

The regulatory network of the 25 screened genes within the HIF1A signaling pathway

Analysis of selective pressure on the coding regions of key genes

To clarify the characteristics of selective pressure on key genes, a site-model global analysis was conducted. By comparing four pairs of evolutionary models (M0 vs. M3, M1a vs. M2a, M7 vs. M8, M8a vs. M8) at the whole-clade level, a total of 10 genes exhibited significant signals of positive selection, including EGLN1, EGLN2, ELOC, HDAC1, HDAC4, HIF1A, HSPA4, MDM2, TP53, and UBC (Table S2). Between the aquatic and terrestrial taxonomic groups, only the CREBBP gene showed a significant signal of positive selection (p_adj < 0.05), and two positively selected sites were detected (Table S3). In the aquatic deep-diving vs. terrestrial common taxonomic groups, both CREBBP and HDAC4 were significant (p_adj < 0.05), but only CREBBP had three positively selected sites. In the terrestrial high-altitude vs. terrestrial common taxonomic groups, EGLN1, EGLN2, ELOC, and TP53 showed significant signals of positive selection (p_adj < 0.01), and positively selected sites were detected in all of them, indicating that hypoxia adaptation in the terrestrial high-altitude ecological niche involves multiple key genes (Table S3). Among the terrestrial burrowing vs. terrestrial common taxonomic groups, only EGLN3 exhibited an extremely significant positive selection effect (p_adj < 0.001), suggesting its specific role in hypoxia adaptation in terrestrial burrowing animals. In the hypoxia-tolerant vs. hypoxia-intolerant taxonomic groups, positively selected sites were detected in EGLN1, EGLN2, ELOC, HIF1A, HSPA4, and TP53, all of which showed high significance (p_adj < 0.01), indicating that the adaptation process of hypoxia-tolerant species in different hypoxic ecological niches is driven by positive selection of different key genes. In the aquatic vs. semi-aquatic taxonomic groups, only TP53 had a significant signal of positive selection (p_adj < 0.05). Based on these results, six potential key genes for the three major hypoxic ecological niches were finally identified, including CREBBP, EGLN1, EGLN2, ELOC, TP53, and EGLN3.

To further validate the evolutionary patterns of these potential key genes, three sets of comparative analyses were conducted using branch models. The results revealed that CREBBP gene exhibited a significant signal of accelerated evolution (positive selection effect) in the aquatic deep-diving ecological niche in the Cetartiodactyla vs. Primates taxonomic group, consistent with the findings from the branch-site models, confirming its central role. The TP53 gene showed accelerated evolution signals (positive selection effect) in both ecological niches, with significant signals observed in the Cetartiodactyla population (Table S4). The EGLN3 gene displayed a significant signal of accelerated evolution associated with purifying selection in the terrestrial burrowing ecological niche in the Rodentia vs. Primates taxonomic group, consistent with the branch-site model results, demonstrating the dominant role of EGLN3 in hypoxia adaptation in the terrestrial burrowing ecological niche (Table S4). Overall, six genes (CREBBP, EGLN1, EGLN2, ELOC, TP53, and EGLN3) showed significant signals of positive selection across different ecological niche comparisons (Table 2). To prioritize these into core key genes for each niche, we applied an integrative decision framework requiring genes to meet three criteria: (i) significant positive selection signal in niche-specific comparisons (branch-site or branch model); (ii) detection of convergent amino acid substitutions in protein domain regions; and (iii) predicted deleterious effects on protein stability or function based on multiple algorithms (PolyPhen2, CADD, DDGun). Using this framework, three genes—CREBBP, EGLN1, and EGLN3—met all three criteria. EGLN2, ELOC, and TP53 showed positive selection but did not exhibit convergent substitutions in domain regions (EGLN2 and ELOC) or failed to meet the deleteriousness prediction threshold (TP53) in the niche-specific context, and were therefore classified as auxiliary regulatory genes rather than core drivers of niche-specific adaptation.

Table 2.

Key genes in different ecological niche taxa

Habitat classification	Significant gene
Aquatic/Terrestrial (6/12)	CREBBP (2)
Aquatic deep-diving/Terrestrial common (3/7)	CREBBP (3)
Terrestrial high-altitude/Terrestrial common (3/7)	EGLN1 (4); EGLN2 (1); ELOC (3); TP53 (2)
Terrestrial burrowing/Terrestrial common (2/7)	EGLN3 (13)
Hypoxia tolerant/intolerant (8/10)	EGLN1 (8); EGLN2 (3); ELOC (4); HIF1A (7); HSPA4 (2); TP53 (11)
Aquatic/Semi-aquatic (6/3)	TP53 (4)

Analysis of convergent amino acid evolution in the coding regions of key genes

Convergent amino acid substitutions were analyzed based on representative species (Table S5). Mutation sites located in the protein domain regions were counted, and their mutation effects were analyzed (Table S6). In the aquatic vs. terrestrial taxonomic group, convergent amino acid substitutions were detected in three genes, including EGLN1, EGLN3, and HDAC4 (Table S7). In the aquatic deep-diving vs. terrestrial common taxonomic group, four genes with convergent amino acid substitutions were identified: EGLN1, EGLN3, HDAC4, and VHL. In the terrestrial high-altitude vs. terrestrial common taxonomic group, convergent amino acid substitutions were only detected in the HIF1AN gene. No pseudogene annotations were found, and the functional domain structures were intact. This confirmed that both were functional genes, and their substitutions were due to gene mutation effects (Table S8). The aquatic vs. terrestrial taxonomic group contained three key genes corresponding to three domain region sites, and the aquatic deep-diving vs. terrestrial common taxonomic group contained four key genes corresponding to five domain region sites (Table 3).

Table 3.

Key genes screened at the amino acid level across taxonomic groups in different ecological niches

Habitat classification	Number	Significant gene
Aquatic/Terrestrial (6/12)	3	EGLN1; EGLN3; HDAC4
Aquatic deep-diving/Terrestrial common (3/7)	5	EGLN1; EGLN3; HDAC4; VHL
Terrestrial high-altitude/Terrestrial common (3/7)	1	HIF1AN
Terrestrial burrowing/Terrestrial common (2/7)	4	HIF1AN; CREBBP; HIF1A; SMAD3
Hypoxia tolerant/intolerant (8/10)	0	/
Aquatic/Semi-aquatic (6/3)	8	EGLN1; EGLN3; HDAC4; VHL; TP53

Note: Number indicates number of domain sites

To clarify the potential impacts of the mutation sites in the aforementioned key genes on protein function, multiple algorithms were employed for deleteriousness prediction analysis (Table 4). DDGun stability analysis revealed that mutations at the target sites of four genes—EGLN1, EGLN3, CREBBP, and TP53—significantly affect protein stability (ΔΔG > 0, indicating destabilizing mutations), leading to increased protein unfolding and reduced binding capacity, suggesting that these mutations are deleterious. CADD predictions consistently supported the deleterious nature of the mutations in CREBBP, EGLN1, EGLN3, and TP53. Integrating these findings, we identified a hierarchical set of genes for each hypoxic ecological niche. CREBBP, EGLN1, and EGLN3 met all three decision criteria (positive selection, convergent amino acid substitution in domain regions, and predicted deleteriousness) and were designated as core regulatory genes for the aquatic deep-diving, terrestrial high-altitude, and terrestrial burrowing niches, respectively (Table 5). TP53, despite showing positive selection and deleteriousness predictions, did not exhibit niche-specific convergent amino acid substitutions; it was therefore classified as a common auxiliary gene that may play a broader regulatory role in the hypoxic response across multiple niches. EGLN2 and ELOC, although showing positive selection in the terrestrial high-altitude niche, lacked convergent substitutions in domain regions and were thus considered secondary modulators rather than core drivers of niche-specific adaptation. Further, a phylogenetic tree confirmed that five genes (EGLN1, EGLN3, CREBBP, HIF1A, and TP53) harbored convergent amino acid substitutions in functionally important domain regions (Fig. 3A). Among these, CREBBP, EGLN1, and EGLN3 met all three core criteria (see above), while HIF1A and TP53 showed convergent substitutions but were not uniquely associated with a single hypoxic niche. The visual atlas displayed the protein domain regions where the mutation sites of these five key genes are located, the species assignments of the foreground and background branches, as well as specific mutation site information (Fig. 3B). Furthermore, molecular docking analysis indicated that two functional domains of the VHL gene that have been validated for convergent amino acid substitutions were selected: von Hippel-Lindau disease tumour suppressor beta (10-91aa) and von Hippel-Lindau disease tumour suppressor alpha (103-151aa). The results showed that HIF1A exhibited strong interactions with both domains of VHL (binding energy < -200, with the formation of multiple hydrogen bonds and other interactions) (Fig. 3C). Within each species, the 10-91aa domain (beta) demonstrated lower binding energy than the alpha domain, which is consistent with the established role of the beta domain as the primary direct binding interface for HIF1A [63]. When comparing hypoxic-adapted mammals to humans, the alpha domain exhibited a larger increase in binding affinity (i.e., greater cross-species divergence) than the beta domain, with the most pronounced effect observed in the aquatic deep-diving species Physeter macrocephalus (Δα = -50.19 vs. Δβ = -0.40).

Table 4.

Predicted harmful effects at the protein level for taxonomic groups in different ecological niches

Gene	Site (Hsa)	Poly-Phen2 (0.446)		Mutation Taster			CADD v1.7	DDGun
		Div	Var	PhyloP (-14–6)	PhastCons(0–1)	Tree vote- Prediction	GENOMIC	ΔΔG
EGLN1	359	0.004	0.008	-5.014	1	84%-Deleterious	21.2-quite likely deleterious	0.6 ↑
EGLN3	140	0.484	0.126	-8.017	1	71%-Deleterious	21.7-quite likely deleterious	0.1 ↑
HDAC4	836	0.016	0.056	-1.111	1	85%-Deleterious	/	-1.1 ↓
VHL	68	0	0.003	0.764	0.488	47%-Benign	/	0.0 =
	181	0.021	0.134	-7.779	0.993	81%-Deleterious	/	-0.3 ↓
HIF1AN	213	0.031	0.128	-9.325	1	68%-Deleterious	13.5-likely benign	-0.6 ↓
CREBBP	1575	0	0.001	-2.656	1	56%- Deleterious	18.5- quite likely deleterious	0.1 ↑
SMAD3	59	0.001	0.012	-2.167	1	90%-Deleterious	/	0.0 =
TP53	261	0	0.001	-6.672	1	54%-Deleterious	20.0- quite likely deleterious	0.2 ↑
HIF1A	145	0.052	0.630	-3.438	1	64%-Deleterious	/	-0.9 ↓
	589	0.004	0.003	1.142	0.993	23%-Benign	15.5-potentially deleterious	0.0 =
	641	0.024	0.051	2.145	1	39%-Benign	22.1- quite likely deleterious	0.0 =
	707	0.015	0.014	0.945	1	43%-Benign	/	0.0 =

Table 5.

Decision framework for core gene prioritization

Gene	Positive selection (niche-specific)	Convergent substitution (domain)	Deleteriousness predicted	Classification
CREBBP	✓	✓	✓	Core (Aquatic deep-diving)
EGLN1	✓	✓	✓	Core (Terrestrial high-altitude)
EGLN3	✓	✓	✓	Core (Terrestrial burrowing)
TP53	✓	✗	✓	Auxiliary
EGLN2	✓	✗	✗	Secondary
ELOC	✓	✗	✗	Secondary
HIF1A	✓	✗	✗	Secondary

Fig. 3.

Phylogenetic tree and in-depth analysis of key genes. A. Constructed phylogenetic tree and results for each ecological niche as well as convergent amino acid substitutions. Taxon labels colors indicate ecological niches: blue = Aquatic deep-diving, yellow = Terrestrial high-altitude, red = Terrestrial burrowing, green = Terrestrial common. B. Visualization results of the domain regions for the five mutation sites. C. Interaction intensity of VHL-HIF1A in the domain regions among representative species. Aquatic deep-diving, representative species was Physeter macrocephalus, Terrestrial high-altitude, representative species was Bos mutus, Terrestrial burrowing, representative species was Heterocephalus glaber, and Terrestrial common, representative species was Homo sapiens. The highlighted content in red represents the amino acid residues in the interaction region between two molecules

Correlation between the expansion and contraction of key genes

To further uncover the evolutionary dynamic characteristics, we conducted gene family expansion and contraction analysis. Based on the PHDs family (EGLN1, EGLN2, EGLN3) and the top 25 core genes, orthologous groups were classified (Figure S1A). The divergence time of the secondary aquatic re-entry event in the Cetartiodactyla clade was 57.8 million years ago (Mya) [10]. Herein, the estimated divergence time for the PHDs family and the TOP25 core genes was 56.11 Mya (53.04–59.03 Mya), a range that highly aligns with the fossil-calibrated time, which confirmed that the selected genes are closely related to the key hypoxia adaptation event of secondary aquatic re-entry. Additionally, tracing the divergence time within the Cetartiodactyla clade revealed a result of 30.09 Mya (12.34–48.56 Mya) (Figure S1A). Among the PHDs family, EGLN1 exhibited a significant expansion trend in representative species of Cetartiodactyla (Figure S1B-C) (p = 0.002), while the overall expansion and contraction of the family were relatively modest. For the top 25 core genes, HDAC4, ARNT, and TP53 showed statistically significant expansion or contraction signals (p < 0.05). Notably, HDAC4 showed contraction in the aquatic deep-diving niche (e.g., Physeter macrocephalus), whereas ARNT and TP53 exhibited expansion in the terrestrial general niche species represented by Homo sapiens (Figure S1C). Further, PGLS analysis detected no statistically significant differences in the evolutionary rates of the three core genes after accounting for phylogenetic relatedness among species (Figure S2A-C).

Analysis of the non-coding regions of key genes

For the CREBBP gene, the aquatic deep-diving group exhibited the highest GC content (73.33%) based on the analysis of the non-coding regions of key genes, while the terrestrial burrowing group had the lowest GC content (63.15%), with the other two groups showing similar levels (Fig. 4A). This suggests that CREBBP may be under differential selective pressure in aquatic deep-diving mammals, potentially contributing to their hypoxia adaptation, whereas it may experience intense adaptive evolution in terrestrial burrowing mammals, leading to fundamental changes in its sequence composition. For the EGLN1 gene, the average GC contents of the terrestrial burrowing group (59.45%) and the terrestrial high-altitude group (58.27%) were significantly lower than that of the terrestrial common group (66.01%). This is consistent with the possibility that EGLN1, functioning as an oxygen sensor, may employ reduced GC content as part of its regulatory response to hypoxic environments, though functional validation is needed. The EGLN3 gene showed highly similar average GC contents across niches, suggesting that this gene is more evolutionarily conserved and less influenced by environmental selective pressures. In the aquatic deep-diving group, the CREBBP gene exhibited a marked increase in CpG islands. In striking contrast, the terrestrial burrowing group exhibited the most extreme and significant changes: the CREBBP gene in the Damara mole-rat completely lacked CpG islands, and other burrowing species typically had shorter or fragmented CpG islands, suggesting a substantial reduction in CpG island content. Although EGLN1 and EGLN3 generally contained CpG islands in burrowing species, their structures, numbers, and positions differed from those in the control group (e.g., shorter lengths, altered positions) (Table S9). Analysis of the TTS and QSFM motifs revealed that the number of TTS in CREBBP ranged from 1 to 5, with the terrestrial common group showing relatively stable counts (3–4) (Table S10). The aquatic deep-diving group exhibited polymorphism (2–5 TTS), while the terrestrial high-altitude group showed counts comparable to the terrestrial common group (3–5 TTS), and the burrowing group showed a marked reduction (0–2 TTS). For the EGLN1 gene, the number of TTS ranged from 1 to 10, with the terrestrial common group showing relatively stable counts (1–2). In the EGLN3 gene, the number of TTS ranged from 1 to 9, with the terrestrial common group showing relatively stable counts (2–3). The aquatic deep-diving group had relatively fewer TTS (1–3). The terrestrial high-altitude group showed TTS counts with wider variation (range 1–9). For the EGLN3 gene, the terrestrial burrowing group exhibited TTS counts comparable to the terrestrial common group (Table S10). The number of QSFM motifs ranged from 0 to 2, with the majority in the terrestrial common and terrestrial high-altitude groups showing none (0 in 4/5 species in both groups). Notably, all individuals in the aquatic deep-diving group retained QSFM motifs (1–2 per species), and the burrowing group also had 1–2 QSFM motifs per species (Table S10).

Fig. 4.

Distinct characteristics of promoter regions of key genes. A. Analysis of GC content. B. DNA stability (ΔG) curve. C. Four major positional characteristics of promoter regions (MGW, HelT, ProT and EP)

DNA stability (ΔG) analysis indicated that the critical regions of CREBBP gene in the aquatic deep-diving group exhibited low ΔG values (Fig. 4B), with prominent low ΔG peaks in the core regulatory regions and sharp fluctuations in the curves. In the EGLN1 gene, the low ΔG value intervals in the high-altitude group were relatively concentrated, with low ΔG values in the core regulatory regions and moderate fluctuations in the curves. In terms of DNA structural characteristics, it was found that for the CREBBP gene (Fig. 4C), the aquatic deep-diving group displayed structural features that have been previously linked to enhanced binding and transcriptional activation through a parameter combination of “narrow minor groove width (MGW) + low helical twist (HelT) + low propeller twist (ProT) + high electrostatic potential (EP)” [61]. In contrast, the terrestrial burrowing group exhibited structural properties that may limit binding and favor transcriptional repression, potentially reflecting an energy-conserving strategy via “wide MGW + high HelT + high ProT + low EP”, adhering to a long-term energy-saving strategy. The terrestrial high-altitude and terrestrial common groups, meanwhile, maintained structural conservatism. The aquatic deep-diving and terrestrial burrowing groups, on the other hand, demonstrated functional secondary importance or concessionality of EGLN1 through conservative or inhibitory structures, while the terrestrial common group retained basic structural features.

The characteristics of CREBBP regulatory elements suggest that the aquatic deep-diving niche is rich in and flexible with regulatory elements, highly consistent with its adaptation pattern of acute and transient hypoxia and the core gene status of CREBBP (Figure S3). For the EGLN1 gene, the regulatory elements in the terrestrial high-altitude group are specifically enriched, significantly enhancing the flexibility of the hypoxia response, consistent with the chronic and fluctuating hypoxic pattern of this niche and the key gene function of EGLN1. In contrast, the elements in the aquatic deep-diving and terrestrial burrowing groups are in conservative or inhibitory states, adapting to the acute hypoxia and stable hypoxia patterns, respectively (Figure S4). Based on the element characteristics of the EGLN3 gene, the regulatory elements in the terrestrial burrowing group are specialized and synergistic, capable of supporting the continuous functional expression of the gene, in line with the long-term stable hypoxia pattern and the core regulatory status of EGLN3. The elements in the aquatic deep-diving and terrestrial high-altitude groups are in conservative or inhibitory states, corresponding to the acute hypoxia and fluctuating hypoxia patterns, respectively (Figure S5).

In-depth analysis estimates the potential clinical relevance of the core genes

To validate the pathogenicity and domain-relatedness of convergent amino acid substitution sites, we further analyzed the mutation distribution and clinical relevance within the functional domains of the CREBBP, EGLN1 and EGLN3 genes. Specifically, the CBP/p300-type HAT domain of CREBBP harbored R1446C/H/L mutations (detected in 15 patients) (Fig. 5A). The Fe²⁺/2-oxoglutarate (Fe2OG) dioxygenase domain of EGLN1 exhibited an R344Q mutation (detected in 4 patients) (Fig. 5B), which coincided with the convergent substitution site regions identified in taxonomic groups such as terrestrial high-altitude vs. terrestrial common for EGLN1. Similarly, the Fe2OG dioxygenase domain of EGLN3 displayed an R192C mutation (detected in 2 patients) (Fig. 5C), overlapping with the convergent substitution site regions identified in the taxonomic group of terrestrial burrowing vs. terrestrial common. This suggests that, in addition to its structural basis for hypoxia adaptation, this mutation is also involved in pathogenic regulation. Amino acid substitutions-induced mutations in the CREBBP gene in cancer patients can reduce the stability of complex binding. When comparing hypoxic niches with hypoxia-intolerant niches, it is found that the complex binding in hypoxic niches exhibited higher and more stable binding capacity (Fig. 5D). There were conserved structural interactions between the HAT (histone acetyltransferase) domains of the CBP-p300 complex. Moreover, to further understand the potential clinical roles of the three genes in tumorigenesis, we found that they showed diverse expression patterns across different cancer types (Figure S6A), and CREBBP was found with higher mutation frequency in patients (Figure S6B). The molecular changes may show the potential clinical values, and further analysis showed the activation or inhibition roles in some cancers (Figure S6C), indicating the contributions in tumorigenesis. Then, they showed suggestive associations with prognosis, and abnormal higher or lower expression patterns may show better survival advantages (Figure S6D).

Fig. 5.

The potential associations of the core genes with human complex diseases. A. The mutation profile of CREBBP in human cancers. The red boxes indicate the amino acid substitution sites, and the blue color represents the amino acid sites after the mutation. B. The mutation profile of EGLN1 in human cancers. The red boxes indicate the amino acid substitution sites, and the blue color represents the amino acid sites after the mutation. C. The mutation profile of EGLN3 in human cancers. The red boxes indicate the amino acid substitution sites, and the blue color represents the amino acid sites after the mutation. D. Interaction intensity of the core genes in the domain regions, including between different species of Monodon monoceros and Homo sapiens, and between human normal samples and tumor samples

Discussion

Hypoxia adaptation represents a prevalent adaptive evolutionary phenomenon across the biological kingdom. The elucidation of its molecular mechanisms is not only a core research topic in evolutionary biology but also an important theoretical foundation for the clinical management of human hypoxia-related diseases. This study, centered on the HIF1A signaling pathway, integrated multi-dimensional evolutionary analyses of both coding and non-coding regions to systematically investigate the molecular adaptive strategies of species inhabiting distinct hypoxic niches.

As a crucial transcriptional co-activator, CREBBP regulates the expression of hypoxia-responsive genes by interacting with transcription factors such as HIF1A. It serves as one of the core regulatory molecules in hypoxia adaptation and plays a pivotal role in energy metabolism and cardiac function regulation [64]. Herein, selection pressure analysis of CREBBP demonstrated its universality, and subsequently, significant signals of accelerated evolution in Cetartiodactyla were identified, which refined the previous important conclusion regarding key genes in the aquatic deep-diving niche. In non-coding region analysis, CREBBP in the aquatic deep-diving niche exhibited characteristics such as high GC content, an increased number of CpG islands, an abundance of TTS and QSFM, significant low ΔG peaks in core regulatory regions, and specific curve properties. These findings are consistent with a role for CREBBP in the acute and transient hypoxia adaptation observed in aquatic deep-diving mammals, although direct functional evidence remains to be established. EGLN1 serves as the core oxygen sensor in the HIF1A signaling pathway. Functional variations in this gene are closely associated with high-altitude adaptation capacity and can modulate erythropoiesis by regulating the degradation efficiency of HIF1A [3]. Genomic studies on wild yaks have also confirmed that EGLN1 is a key gene for high-altitude adaptation [65]. Selection pressure analysis of EGLN1 demonstrated its universality, and its mutations were identified as deleterious, reflecting its crucial role as an oxygen sensor. In non-coding region analysis, the regulatory features of EGLN1 in the terrestrial high-altitude niche, including its GC content and TTS counts, support its role in adapting to chronic and fluctuating hypoxia, consistent with the key gene function of EGLN1. EGLN3, as a prolyl hydroxylase, participates in the coordinated regulation of HIF1A degradation efficiency under normoxic conditions [66]. Under hypoxic conditions, EGLN3 can interact with pyruvate kinase M2 (PKM2) and inhibit its nuclear translocation, thereby indirectly enhancing the transcriptional activity of HIF1A. Simultaneously, by regulating the expression of glycolytic genes downstream of HIF1A, EGLN3 mediates metabolic reprogramming to facilitate adaptation to the hypoxic environment [67]. Our study demonstrated the critical role of EGLN3 at the coding gene level from multiple perspectives and also confirmed the strategy of the terrestrial burrowing niche to adapt to long-term stable hypoxia through the EGLN3 gene. Nevertheless, we believe the observation that the same functional domains are targeted by both adaptive and pathogenic mutations provides a foundation for future mechanistic studies.

Despite a series of valuable findings achieved in this study, there are still some limitations that also point out directions for future research. Firstly, the scope of genes and pathways covered in the study is limited. This research focuses on the top 25 core genes that directly interact with HIF1A, potentially overlooking the contributions of peripheral genes in the pathway or other synergistic pathways [68]. Secondly, the representativeness of the species samples can be further enhanced. Although this study includes 21 species from three major hypoxic niches, the number of species within each niche is relatively limited. Thirdly, our gene family expansion/contraction and PGLS analyses yielded weak or inconclusive signals. While some copy-number changes reached statistical significance, the overall magnitudes were modest and their biological relevance to hypoxia adaptation remains uncertain. Similarly, the PGLS analysis did not detect significant phylogenetic signals.

Conclusion

This study reveals coordinated adaptations in the HIF1A pathway across hypoxic niches: aquatic deep-divers optimize hypoxia response via CREBBP domain substitutions, high-altitude species enhance oxygen sensing through EGLN1 tuning, and fossorial taxa maintain hypoxic homeostasis via EGLN3.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AMS

Acute mountain sickness

EP

Electrostatic Potential

Fe2OG

Fe²⁺/2-oxoglutarate

HAT

Histone acetyltransferase

HelT

Helical twist

HIF1

Hypoxia-inducible factor 1

HREs

Hypoxia response elementshypoxia response elements

MGW

Minor groove width

PAS

Pathway activity score

PHD

Prolyl 4-Hydroxylase Domain proteins

PGLS

Phylogenetic generalized least squares

PKM2

Pyruvate kinase M2

ProT

Propeller twist

QSFMs

Quadruplex-forming sequence motifs

TCGA

The Cancer Genome Atlas

TTSs

Triplex target sites

Author contributions

G.Y., and L.G. conceptualized and guided the study. T.L., H.G., G.J., J.Z., C.G., J.S., and J.Z. devised the methodology and managed the resources. T.L., H.G., G.J., J.Z., C.G., J.S., and J.Z. carried out the investigation. H.G., G.J., J.Z., C.G., J.S., and J.Z. carried out the data analyses. T.L., H.G. and L.G. wrote the original manuscript draft. G.Y., and L.G. revised the manuscript. All authors reviewed and edited the manuscript draft.

Funding

This work was supported by National Natural Science Foundation of China (Nos. 62571264 and 62171236), the key project of social development in Jiangsu Province (No. BE2022799), the key projects of Natural Science Research in Universities of Jiangsu Province (No. 22KJA180006), Postgraduate Research & Practice Innovation Program of Jiangsu Province (No. SJCX25_0737), Qinglan Project of Jiangsu Universities in Jiangsu Province.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Ethics statement and patient consent

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.