Multi-dimensional epigenomic dynamics converge on H3K4-mediated regulation of low-CO₂ adaptation in Nannochloropsis oceanica

Abstract

Despite their ecological and biotechnological importance, the extent to which microalgae are regulated by epigenetic mechanisms has remained poorly understood. In the model industrial microalga Nannochloropsis oceanica, by comprehensive, multi-dimensional epigenomic analyses, this study uncovers an epigenetic regulatory network responsive to CO₂ levels. This network involves intricate interactions among DNA methylation, histone modifications, dynamic nucleosome positioning, and three-dimensional chromatin organization during adaptation to low-CO₂ conditions. Although DNA methylation is minimal, histone modifications—such as lysine acetylation, crotonylation, and methylation—are associated with active chromatin states and linked to 43.1% of differentially expressed genes. Notably, histone H3K4 di-methylation (H3K4me2) displays a distinct dual-peak profile around the transcription start site and is correlated with chromatin compartment dynamics. Knockout of NO24G02310, a putative H3K4 methyltransferase gene, caused genome-wide shifts in H3K4me2 peaks and decreased H3K4me1 levels, accompanied by direct or indirect downregulation of NoHINT and NoPMA2 expression, slower algal growth, and reduced photosynthetic efficiency (indicated by Fv/Fm), specifically under low-CO₂ conditions. Deletion and overexpression of genes encoding the histidine triad nucleotide-binding protein NoHINT and the plasma membrane H⁺-ATPase NoPMA2 confirmed their roles in growth and photosynthetic efficiency under low CO₂; NoHINT influences growth and NoPMA2 affects photosynthesis. As a previously unrecognized low-CO₂ adaptation mechanism, NO24G02310 likely coordinates the regulation of NoHINT and NoPMA2 through H3K4 modifications. These findings provide a foundation for enhancing microalgal productivity through targeted epigenetic engineering.

This study characterizes the multi-dimensional epigenetic dynamics underlying low-CO₂ adaptation in the model industrial microalga Nannochloropsis oceanica_{. H3K4 histone modification, regulated by NO24G02310, modulates growth and photosynthesis under low-CO2}conditions by controlling the expression of NoHINT_{andNoPMA2, with NoHINT governing growth dynamics and NoPMA2 regulating photosynthetic efficiency.}

Introduction

Marine microalgae play a critical role in mitigating atmospheric CO₂ levels through carbon assimilation, thereby substantially contributing to the reduction of global warming and ocean acidification (Hansen et al., 2006; Doney, 2010). This carbon-assimilation process is catalyzed by the bifunctional enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), which initiates the carbon-fixing Calvin cycle under high-CO₂ conditions but also triggers wasteful photorespiration when the O₂-to-CO₂ concentration ratio is relatively high (Hayer-Hartl and Hartl, 2020). Because photorespiration competes with carboxylation, it is essential for oxygenic photosynthetic organisms to increase the CO₂ concentration around RuBisCO to optimize carbon fixation. This mechanism, known as the CO₂-concentrating mechanism (CCM), is especially important for aquatic plants and microalgae, given that the concentration of dissolved CO₂ in water is limited by its low solubility and slow diffusion rate. To adapt to low-CO₂ environments, microalgal CCMs efficiently convert and concentrate CO₂ near RuBisCO to sustain effective carbon fixation (Bauwe et al., 2010; Hagemann and Bauwe, 2016). However, these processes are energetically demanding and require coordinated interactions among various organelles and metabolic pathways (Burlacot et al., 2022; Arend et al., 2023; Burlacot and Peltier, 2023; Ruiz-Sola et al., 2023), thus confounding efforts to elucidate the regulatory mechanisms that control carbon fixation under low-CO₂ conditions.

Understanding how microalgae regulate metabolic pathways to adapt to low-CO₂ stress is vital for both ecological and biotechnological applications. Under CO₂ limitation, genes associated with CCMs are upregulated to enhance carbon uptake and utilization (Hennon et al., 2015; Huang et al., 2017; Wei et al., 2019; Koh et al., 2023). Additionally, signaling molecules such as cyclic adenosine monophosphate and inositol polyphosphates have been shown to modulate CCMs and photosynthetic capacity in response to low CO₂ levels (Hennon et al., 2015; Morales-Pineda et al., 2023). Physiological adjustments to CO₂ limitation, including alterations in membrane fatty acid composition, help maintain cellular homeostasis and optimize the photosynthetic machinery for improved efficiency (Solovchenko and Khozin-Goldberg, 2013). Therefore, adaptation to low CO₂ involves multiple organelles and metabolic pathways, indicating a complex regulatory network that governs carbon fixation and microalgal growth. Nonetheless, this comprehensive regulatory network remains poorly defined, in part due to the limited understanding of epigenomic regulation in microalgal carbon assimilation—a mechanism known to play essential roles in the regulation of C₄ and crassulacean acid metabolism (CAM) photosynthesis in terrestrial higher plants (Heimann et al., 2013; Shi et al., 2021).

Epigenomic factors—including histone modifications, nuclear positioning, and the 3D conformation of chromatin—act in concert to regulate genomic activity without altering the DNA sequence. In diverse oxygenic photosynthetic organisms, chromatin states (CSs) dynamically respond to environmental fluctuations and play essential roles in gene regulation and adaptation (Probst and Mittelsten Scheid, 2015; Gao et al., 2016; Bacova et al., 2020; Bhadouriya et al., 2021; Ferrari et al., 2023). In Chlamydomonas reinhardtii, promoter-associated histone modifications and DNA 6mA methylation reflect transcriptional status, enabling the identification of regulatory switch genes under environmental stress (Fu et al., 2015; Ngan et al., 2015). However, the establishment of causal relationships between epigenetic variation, gene expression, and resulting phenotypes remains challenging and may be confounded by secondary responses. For example, initial findings in rice suggested that phosphate starvation induces stable hypermethylation near active genes—a counterintuitive observation, given that hypermethylation typically represses transcription. Subsequent analysis revealed that this hypermethylation serves a protective function by silencing nearby transposable elements—elevated local transcriptional activity may otherwise trigger their reactivation (Secco et al., 2015). Therefore, comprehensive genome-wide epigenetic profiling and experimental validation are needed to determine the causal implications of epigenomic changes in response to stress.

Nannochloropsis spp. are unicellular oleaginous microalgae known for their high photoautotrophic biomass productivity and lipid content, as well as suitability for industrial cultivation (Radakovits et al., 2012). These species exhibit a unique CCM that lacks microcompartments such as carboxysomes or pyrenoids, which typically serve as sites for RuBisCO and CCM localization in cyanobacteria and green algae (Rae et al., 2013; Wang et al., 2015; Gee and Niyogi, 2017). We previously showed that 32% of genes in Nannochloropsis oceanica are transcriptionally regulated under low-CO₂ stress (Wei et al., 2019), and variations in several histone marks have also been observed (Wei and Xu, 2018; Liu and Wei, 2023). However, progress in elucidating potential roles of epigenetic regulation has been hindered by the absence of a comprehensive epigenomic map.

Here, we generated an integrated epigenomic roadmap encompassing histone modifications, nucleosome occupancy, and 3D chromatin conformation in N. oceanica before and after transition to low-CO₂ conditions; we also proposed a comprehensive model of epigenetic regulation in N. oceanica IMET1. We found that the histone modification H3K4me2, mediated by NO24G02310, regulates carbon assimilation under low CO₂ by targeting key genes such as NoHINT and NoPMA2. Functional validation through gene knockout and overexpression revealed their distinct but substantial effects on growth and photosynthetic efficiency. These results highlight the intricate epigenetic regulation of carbon metabolism in N. oceanica and establish a foundation for enhancing microalgal productivity through targeted epigenetic modification.

Results

Comprehensive profiling of the epigenomic landscape under low-CO₂ stress

We recently showed that N. oceanica adapts to low-CO₂ conditions by activating its CCM while modulating global metabolic networks (Wei et al., 2019). However, the chromatin dynamics driving differential gene expression—and thus facilitating low-CO₂ adaptation—remain unknown. To investigate the regulatory mechanisms underlying this process, we constructed a multi-dimensional epigenomic atlas using cultures of N. oceanica IMET1 shifted from high CO₂ (HC; 5% CO₂) to low CO₂ (LC; 0.01% CO₂) (Supplemental Figure 1). Microalgal cells were harvested 24 h after the shift to LC to minimize confounding effects from circadian rhythms, given that CCM activation occurs within 3 h (Methods section). This comprehensive atlas integrates genome-wide DNA methylation (5-methylcytosine [5mC]), histone modification marks, nucleosome occupancy, and 3D chromatin architecture (Supplemental Figure 1), providing a detailed landscape of epigenomic remodeling under HC and LC conditions. We explored correlations of these epigenetic changes with gene expression patterns and delineated interrelationships among individual epigenetic features. In total, 73 datasets comprising 504 gigabases of sequencing data (Supplemental Table 1) were generated and deposited in the Sequence Read Archive (accessions PRJNA769792, PRJNA769749, PRJNA769619, and PRJNA513887). These datasets have been integrated into the NanDeSyn database (http://nandesyn.org) (Gong et al., 2020), providing a valuable resource for future microalgal epigenetic research.

DNA 5mC methylomes revealed that N. oceanica exhibits exceptionally low levels of DNA 5mC methylation (0.13% of cytosines) compared with green algae (Supplemental Table 2; supplemental results). Comparative analyses between HC and LC conditions showed minimal 5mC variation across genomic regions and no significant association with transcriptional changes (Supplemental Figure 2; supplemental results), suggesting that DNA 5mC methylation does not contribute to LC adaptation in N. oceanica. This finding highlights the need to focus on other epigenomic layers to elucidate the regulatory mechanisms underlying low-CO₂ adaptation.

Histone modifications orchestrate transcriptional regulation under low CO₂

To investigate the diversity of histone modifications in N. oceanica, histone post-translational modifications (PTMs) were characterized by mass spectrometry of enzymatically digested histone extracts (Methods section). In total, 87 PTMs associated with core and variant histones were identified in N. oceanica (Figure 1A), exceeding those reported in other stramenopiles (Veluchamy et al., 2015; Bourdareau et al., 2021). Consistent with findings in Ectocarpus, N. oceanica also lacks H3K27 tri-methylation and H2AK119 ubiquitination. These histone modifications are mediated by the polycomb repressive complexes PRC1 and PRC2, whose absence corresponds to the loss of these complexes in both algal species. This observation further supports the accuracy and robustness of our PTM-detection workflow. Overall, the expanded spectrum of histone PTMs in N. oceanica suggests broader and potentially more complex biological functions.

Figure 1.

Dynamics of histone modification marks under HC and LC conditions in N. oceanica.

(A) PTMs of Nannochloropsis histones identified by mass spectrometry. Acetylation (Ac), methylation (Me/Me2/Me3), ubiquitylation (Ub), lactylation (La), crotonylation (Cr), and phosphorylation (Ph) were detected in linker, core, and variant histones. Colored boxes denote amino acids covered by peptides identified from mass spectra; amino acids in gray indicate heterogeneity among multiple histone copies; and modified amino acids are highlighted in red.

(B) Enrichment profiles of H3K4me2, H3K9ac, Kcr, and H3K27ac across genes under HC (gray) and LC (red) conditions.

(C) Dynamics of histone modifications around the carbonate dehydratase gene NO20G00630 (NoCA5). Green boxes denote regions with significant peak changes.

(D) Scatterplots illustrating relationships between histone mark changes and gene expression levels.

(E) GO term enrichment for DHMGs. The number of DHM-associated genes corresponding to each GO term is shown on the right.

To clarify the role of histone modifications in transcriptional regulation during low-CO₂ adaptation, we performed chromatin immunoprecipitation sequencing (ChIP-seq) to map seven histone marks genome-wide under HC and LC conditions. Among these, H3K4me2, H3K9ac, Kcr, and H3K27ac were selected for further analysis based on their strong enrichment efficiency and high reproducibility between biological replicates (Supplemental Figure 3; Supplemental Table 3; supplemental results). The distribution patterns revealed significant enrichment of H3K9ac, H3K27ac, and Kcr precisely at transcription start sites (TSSs) (Figure 1B; Supplemental Figures 4A–4C; supplemental results), consistent with patterns reported in land plants and metazoans (Prakash and Fournier, 2018; Zhou et al., 2011). In contrast, H3K4me2 displayed a characteristic dual-peak profile flanking the TSS, with reduced signal intensity immediately adjacent to it (Figure 1B), similar to observations in animals (Hu et al., 2013; Liu et al., 2011; Rickels et al., 2016). Quantitative analysis identified 5587–6180 peaks for H3K9ac, H3K27ac, and Kcr, whereas H3K4me2 exhibited 10 105 peaks due to its dual-peak pattern (Supplemental Table 4). Intriguingly, we observed extensive overlap among these histone marks, with significant co-occurrence across numerous gene loci (Supplemental Figures 4D and 4E). More than 96% of the ChIP-seq peaks for H3K9ac, H3K27ac, and Kcr overlapped (Supplemental Figure 4D); 4932 genes were simultaneously marked by all four types of histone modifications (Supplemental Figure 4E). Genes carrying these combined modifications exhibited higher expression levels relative to those lacking such modifications (Supplemental Figure 5; supplemental results). For instance, the NO20G00630 gene (NoCA5), a key component of the N. oceanica CCM (Gee and Niyogi, 2017), exhibited overlapping but dynamically regulated marks of H3K9ac, H3K27ac, and Kcr, which corresponded to its strong transcriptional upregulation (sixfold increase) under LC conditions (Figure 1C).

Given the strong positive correlation observed between histone marks and gene expression, we next examined differential histone modifications between HC and LC conditions to better characterize transcriptional reprogramming. We identified 1823–2452 differential peaks (differential histone marks [DHMs]) for H3K9ac, H3K27ac, and Kcr but only 263 for H3K4me2 (Supplemental Table 5). DHM-associated genes (DHMGs) overlapped with 43.1% of differentially expressed genes (DEGs). Linear regression analysis of DHM peak abundance versus gene expression changes revealed positive correlations across all histone marks, although the correlation was weaker for H3K4me2, likely due to its limited number of DHMs (Figure 1D). Functional annotation of DHMGs indicated enrichment in molecular functions and cellular components related to ribosomal biogenesis and organelle function. DHMGs enriched in H3K27ac and Kcr were primarily associated with photosynthetic processes; those enriched in H3K4me2 were correlated with diverse biological functions, including gene expression, translation, and biosynthesis of peptides, organonitrogen compounds, and macromolecules (Figure 1E). Based on these DHMGs (Supplemental Data 1), several genes of importance in low-CO₂ adaptation were identified. For instance: (1) NO20G00630 (NoCA5) is epigenetically regulated by H3K9ac, H3K27ac, and Kcr (Figure 1C), enhancing CCM activity under low-CO₂ conditions. (2) Two transporter genes, NO20G00710 (mitochondrial transporter family) and NO12G00380 (plasma membrane ATPase), are regulated by both H3K27ac and Kcr, indicating potential roles as candidate H⁺/HCO₃⁻ transporters. (3) Twenty-three photosynthesis-related genes (Gene Ontology [GO]:0015979) are regulated by either H3K27ac or Kcr; notably, PsbU (NO14G01170), which modulates PSII light sensitivity (Abasova et al., 2011), is regulated by both marks, suggesting active photosynthetic adaptation to low-CO₂ stress. (4) A long-chain acyl–coenzyme A ligase gene (NO25G00730) is regulated by H3K27ac and Kcr, emphasizing the importance of acyl–coenzyme A supply during CO₂ limitation. Intriguingly, these genes exhibit similar transcriptional expression patterns under LC stress (clusters K9 and K12; Supplemental Figure 6). In cluster K9, genes such as NoCA5 show dual activation phases, occurring both early and after 12 h, reflecting complex regulatory dynamics influenced by multiple factors. Collectively, these findings underscore the prominent role of histone modifications in orchestrating gene expression reprogramming in N. oceanica under LC stress and provide new insights into the underlying epigenetic mechanisms of low-CO₂ adaptation.

Nucleosome dynamics underlie altered gene expression in response to low CO₂

The distinct effects of two epigenetic layers—minimal 5mC DNA methylation and the pronounced regulatory roles of histone modifications—revealed a unique epigenomic regulatory system in N. oceanica that differs from systems within other microalgal models, such as C. reinhardtii. Thus, we next examined whether nucleosome positioning, another critical determinant of gene regulation, contributes to adaptation under LC conditions by performing micrococcal nuclease sequencing (MNase-seq) experiments under both HC and LC conditions. High reproducibility among biological replicates was achieved, as shown by a Pearson correlation coefficient of 0.93 for genome-wide read coverage (Supplemental Figure 7A). We identified 81 031 nucleosomes under HC and 87 845 under LC (Supplemental Table 6; Supplemental Figure 7B), with a consistent nucleosome repeat length peak of 178 bp across both conditions, indicating a stable nucleosome arrangement (Supplemental Figure 7C). Regions flanking the TSS and transcription termination site (TTS) showed substantial nucleosome depletion, emphasizing their importance in transcriptional regulation (Figure 2A).

Figure 2.

Dynamics of nucleosome occupancy under HC and LC conditions in N. oceanica.

(A) Nucleosome distribution across genes under HC and LC conditions.

(B) GO term enrichment among downregulated genes linked to dynamic nucleosomes within promoter regions. The number of DPN-associated downregulated genes for each GO term is shown on the right.

(C) Loss of a nucleosome leads to upregulation of the adjacent, promoter-overlapping gene pair NO18G02670 and NO18G02680 . The latter encodes a Zn₂Cys₆ fungal-type transcription factor.

A comparison of LC and HC conditions identified 4496 differentially positioned nucleosome (DPN) peaks under LC stress (false discovery rate [FDR] < 0.01; Supplemental Figure 7D), most of which were located in gene promoters. Motif analysis of DPNs (and DHM-associated peaks) revealed several enriched motifs likely associated with plant-specific GT transcription factors (Supplemental Figure 8; supplemental results). These DPNs mapped to 1892 genes, 35.1% of which exhibited differential expression (Supplemental Data 2). This percentage was slightly higher than the genome-wide expression change (32.4%; Z-test, p < 0.05), indicating that DPNs partially represent transcriptional regulation under LC stress. The enrichment of DEGs among genes containing DPNs (Fisher’s exact test, p < 0.01) underscores the pivotal role of nucleosome positioning in gene regulation during LC adaptation in N. oceanica. Further analysis showed balanced distributions of upregulated (17.5% vs. 17.8% genome-wide; hypergeometric test, p = 0.77) and downregulated genes (17.5% vs. 14.6% genome-wide; hypergeometric test, p < 0.01) (Supplemental Figure 7E). GO enrichment analysis revealed that downregulated genes were predominantly involved in ribosomal components, suggesting widespread transcriptional reprogramming toward reduced metabolic activity under LC stress (Figure 2B; Supplemental Table 7). Notably, the Zn₂Cys₆ fungal-type transcription factor NO18G02680—homologous to Naga_100262g2 in Nannochloropsis gaditana and a likely regulator of LC stress adaptation—was significantly upregulated (Figure 2C). In N. gaditana, another Zn₂Cys₆ transcription factor (Naga_100104g18; homologous to NO09G01030 in N. oceanica IMET1) regulates lipid metabolism without affecting growth (Ajjawi et al., 2017). Additionally, DPNs were linked to the bicarbonate transporter NoBCT2 (NO01G02900), a critical component of the biophysical CCM. Genes such as argininosuccinate synthase (NO05G00910) and ornithine decarboxylase (NO05G00760) within the ornithine–urea cycle also exhibited associations with DPNs, suggesting involvement in the adaptive response to LC conditions.

Chromatin-state alteration in response to low CO₂ adaptation

Because both histone modifications and nucleosome positioning significantly affect gene expression (Figures 1 and 2), we next integrated these datasets to examine CS alterations under LC stress. Five distinct CSs were identified, each defined by a unique combination of nucleosome occupancy and histone marks (Figure 3A). CS1 lacked all four histone marks, whereas CS2 exhibited higher nucleosome occupancy than CS1. CS3 was mainly characterized by H3K4me2, and CS4 was enriched in all histone modifications; CS5 displayed enrichment of H3K9ac, H3K27ac, and Kcr (Figure 3A). The unique distribution of H3K4me2, represented by CS3, suggests that this modification does not consistently overlap with the other three marks and occupies distinct genomic regions. Additionally, the various CSs exhibited preferential enrichment across different gene regions under both HC and LC conditions (Figure 3B). For instance, CS1 was the predominant state, encompassing 36% and 42% of the N. oceanica genome under HC and LC conditions, respectively. In contrast, CS5—representing approximately 10% of the genome—was strongly concentrated near TSSs.

Figure 3.

Dynamics of chromatin states under HC and LC conditions in N. oceanica.

(A) Emission parameters derived from the hidden Markov model constructed through ChromHMM analysis represent the characteristics of five distinct CSs. For both HC and LC samples, the CS of each genomic fragment was defined by the peak occurrence of H3K4me2, H3K9ac, Kcr, H3K27ac, and nucleosome occupancy. Darker red shading indicates a higher probability of detecting a given epigenetic mark in that state.

(B) Fold enrichment of various genomic regions overlapping with the five CSs in HC and LC samples. Relative enrichment values were calculated by subtracting the column minimum and dividing by the column maximum. White indicates the minimum value, whereas the darkest shade represents the maximum value observed in the heatmap.

(C) Transition matrix showing CS shifts of genes, where each gene’s CS was defined at the TSS.

(D) Percentages of DEGs showing CS alterations. Genes were grouped according to their CS under HC conditions, and p values were determined by perturbation testing (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

(E) Dynamics of epigenetic signals and CSs around the malic enzyme gene NO26G00480. Purple rectangles and circles indicate differences between HC and LC samples. Blocks at the bottom represent CSs, where 1–5 correspond to CS1–CS5.

(F–H) Identification of driver genes reveals biologically important genes beyond DEGs. Genes were categorized into four groups: Sig_DEG (driver DEGs), Sig_nDEG (driver genes that are not DEGs), nSig_DEG (non-driver DEGs), and nSig_nDEG (non-driver, non-DEGs). (F) Percentages of genes epigenetically regulated in the four groups across epigenomic datasets. (G) GO terms enriched according to gene set enrichment analysis; the number of associated genes is indicated for each group. (H) Non-DEG driver genes show minimal transcriptional variation but substantially contribute to enrichment analyses, highlighting their biological significance.

Our integrated analysis further demonstrated the dynamic nature of chromatin-state transitions during the shift from HC to LC conditions. To facilitate consistent interpretation, the CS of each gene was defined based on the state at its TSS, and thus only a subset of DPNs was included in the analysis. A substantial proportion of genes exhibited CS shifts at the TSS, reflecting active epigenetic reprogramming (Figure 3C). These changes primarily involved transitions between CS1 and CS2, CS3 and CS4, and CS4 and CS5. Approximately 26.8% of DEGs underwent CS alterations, indicating that transcriptional reprogramming during LC adaptation is largely epigenetically mediated. Mechanistically, transitions between specific CSs were closely associated with differential expression patterns (Figure 3D). Transitions from CS3 to CS1, CS4 to CS3, and CS5 to CS4 corresponded to significant gene downregulation, likely resulting from the loss of active histone modifications near the TSS. Conversely, transitions from CS3 to CS4 and CS4 to CS5 were linked to gene upregulation, indicating a bidirectional regulatory role in transcriptional control.

The functional relevance of these CS alterations was further supported by GO enrichment analysis. For instance, genes transitioning from CS5 to CS4 were enriched for ribosomal components, including those associated with the ribosome (GO:0005840), implying broad transcriptional reprogramming toward reduced metabolic activity in response to LC stress. This finding was consistent with transcriptomic data showing downregulation of these ribosomal genes under LC conditions (Supplemental Table 8). Additionally, key genes such as NoCA5, which is critical for the CCM, and others involved in the ornithine–urea cycle (e.g., ornithine decarboxylase) exhibited CS shifts, further emphasizing the complex interplay between chromatin dynamics and gene regulation under varying CO₂ conditions. Specifically, malic enzyme 1, which catalyzes the oxidation of malate to pyruvate with CO₂ release, showed pronounced CS alterations under LC stress (Figure 3E). This gene was downregulated, showing a correlation with decreased levels of H3K27ac and Kcr. Additionally, an H3K4me2 peak near malic enzyme 1 was repositioned closer to the TSS, whereas the adjacent nucleosome became less accessible (Figure 3E), suggesting multiple layers of transcriptional regulation. Collectively, these multi-dimensional analyses of N. oceanica reveal both immediate chromatin-state responses to LC and provide deeper insight into the epigenetic mechanisms that support adaptive gene regulation.

To enhance the robustness and depth of our analysis, we utilized iClusterBayes (Mo et al., 2017) to integrate the RNA sequencing (RNA-seq), ChIP-seq (for H3K4me2, H3K9ac, H3K27ac, and Kcr), and MNase-seq datasets. iClusterBayes is a machine-learning framework that performs integrative clustering using a Bayesian latent-variable model. Crucially, it enables the identification of key genomic features that drive the formation of distinct clusters, thereby offering valuable mechanistic insights. In the transcriptomic dataset, a substantial proportion of non-differentially expressed genes (non-DEGs; 44.8% of driver genes) contributed to the separation of HC and LC samples (Supplemental Figure 9A). Moreover, the mRNA-seq and MNase-seq datasets shared approximately 55% of driver genes, indicating strong overlap in the regulatory mechanisms captured by these methods (Supplemental Figure 9B). In the ChIP-seq datasets, driver genes exhibited enriched GO terms similar to those of DHMGs, despite representing only a small subset of them, underscoring the biological significance of these driver genes as functionally relevant gene sets (Supplemental Figures 9A, 9C, and 9D). To explore these relationships further, genes were categorized into four groups: Sig_DEG (driver DEGs), Sig_nDEG (driver non-DEGs), nSig_DEG (non-driver DEGs), and nSig_nDEG (non-driver, non-DEGs). The likelihood of epigenetic regulation in the Sig_DEG group was approximately 10% higher than in the other groups (53.1% vs. 44.4%, 38.4%, and 43.3%; Figure 3F). Gene set enrichment analysis revealed that nearly all core enrichment genes were driver genes, even when they were not DEGs (Figure 3G). Additionally, non-DEG driver genes, despite their limited transcriptional variation, played biologically important roles (Figure 3H). Among these non-DEG driver genes, the basic helix-loop-helix (bHLH) transcription factor NO14G02230 displayed epigenetic regulation through H3K27ac and Kcr modifications under LC stress. According to PlantTFDB v5.0 (Tian et al., 2020), NO14G02230 is closely related to Arabidopsis bHLH34 (AT3G23210), a key regulator of glucose and iron homeostasis through signal transduction (Li et al., 2016; Min et al., 2017) and an essential factor in cellular adaptation to CO₂ limitation. Transcriptional profiling in N. oceanica revealed a strong inverse correlation (Pearson’s R = −0.83) between NO14G02230 and NoCA5 expression across HC and LC time-course experiments. These findings suggest an underappreciated but crucial regulatory paradigm in which epigenetically modified transcription factors coordinate low-CO₂ adaptation by modulating signal amplification within interconnected signaling networks. Moreover, these results highlight the importance of considering both DEGs and non-DEG driver genes to fully understand the regulatory mechanisms governing HC and LC adaptation, providing a more comprehensive view of transcriptional and epigenetic interplay.

Reorganization of 3D chromosome architecture in response to low-CO₂ adaptation

Building upon the fine-scale CS alterations encompassing 5mC DNA methylation, histone modifications, and nucleosome positioning, we further explored the role of higher-order 3D chromatin organization in LC adaptation by generating high-resolution Hi-C interaction maps through Hi-C sequencing of N. oceanica under HC and LC conditions (Supplemental Table 9; Supplemental Figures 10A–10C; Methods section). The resulting 3D genome architecture (Supplemental Figure 10D) revealed that each chromosome folds in half, with telomeres colocalizing and centromeric regions exhibiting fusion—a configuration resembling the lattice-like structure of haploid yeast (Shao et al., 2018)—and suggesting susceptibility to large-scale chromosomal rearrangements. In eukaryotes, each chromosome can be partitioned into multiple compartments through principal component analysis of genome-wide contact matrices. The first eigenvector distinguishes chromatin regions into compartment A (positive values) and compartment B (negative values) (Lieberman-Aiden et al., 2009). Compartment A represents open, transcriptionally active chromatin that is gene-dense and enriched in activating histone marks, whereas compartment B corresponds to closed, transcriptionally repressed chromatin that is gene-poor and associated with repressive modifications. In N. oceanica, chromosomal compartmentation was relatively simple (Figure 4A; Supplemental Figure 10D), likely due to the short length of its chromosomes (each < 2 Mbp). Nevertheless, these compartmental distinctions were biologically meaningful: compartment A was positioned toward central chromosomal regions and associated with open chromatin and elevated gene expression, whereas compartment B localized near chromosomal termini, corresponding to closed chromatin and lower transcriptional activity (Wilcoxon test, p < 0.0001; Figure 4B).

Figure 4.

Dynamics of 3D chromatin architecture under HC and LC conditions in N. oceanica.

(A) Chromosome-specific A/B compartment profiles of N. oceanica under HC and LC conditions at 40-kb resolution. Red and blue represent compartments A and B, respectively.

(B) Comparison of gene expression levels between compartments A and B under both HC and LC conditions.

(C) Difference interaction matrix of chromosome 14 at 10-kb resolution, showing differences between HC and LC conditions for replicate 1.

(D) Correlations between local chromatin compartments and differential gene expression in N. oceanica. “A2B” indicates genomic intervals transitioning from compartment A to B upon the shift from HC to LC, whereas “B2A” represents the reverse transition. “A2A” and “B2B” denote regions with stable compartment profiles.

(E) Correlations between local compartment dynamics and changes in histone-modification in N. oceanica. Statistical significance was determined via the Wilcoxon test (ns, p > 0.05; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001).

A comparison of the HC and LC contact maps at 10-kb resolution revealed that, although the higher-order chromatin architecture was largely conserved (Supplemental Figure 10E), extensive local chromatin remodeling occurred (Figure 4C). Intriguingly, these structural dynamics aligned with gene expression changes: transitions from compartment A to B were associated with a higher frequency of downregulated genes and a pronounced decrease in overall expression (A2B vs. A2A; Wilcoxon test, p < 0.001). In contrast, B-to-A transitions were correlated with an increased proportion of upregulated genes, although the overall change in expression levels was not statistically significant (B2A vs. B2B; Wilcoxon test, p > 0.05) (Figure 4D). Prior studies have indicated potential crosstalk between chromatin compartments and histone modifications (Wang et al., 2019; Zheng et al., 2024). To explore this relationship, we compared the dynamics of histone marks between regions with and without compartment shifts (Figure 4E; A2B vs. A2A, B2A vs. B2B). Among all histone marks analyzed, compartment transitions were most prominently reflected by changes in H3K4me2 levels (Figure 4E). Specifically, a significant reduction in H3K4me2 variation was observed in the A2B transition group compared with A2A (Wilcoxon test, p < 0.0001), whereas a substantial increase was noted in the B2A transition group relative to B2B (Wilcoxon test, p < 0.0001). Because H3K4me2 is also recognized as an active enhancer mark (Wang et al., 2008; Pekowska et al., 2011; Chepelev et al., 2012; Ong and Corces, 2012), we investigated whether distal regulatory interactions—such as promoter–enhancer or promoter–promoter loops—contribute to transcriptional control (Grubert et al., 2020; Hamamoto and Fukaya, 2022; Zagirova et al., 2024). Several significant “mid-range” chromatin contacts were identified under HC (10 contacts) and LC (24 contacts) conditions (∼4–835 kb, FDR < 0.05); 11 of the 21 genes involved (52.4%) showed significant differential expression. Furthermore, these chromatin contacts were associated with the promoter regions of 10 genes (Supplemental Figure 10F), indicating that enhancer–promoter and promoter–promoter interactions may participate in LC adaptation in N. oceanica. These findings suggest that alterations in histone modifications—particularly H3K4me2—underlie chromatin compartment transitions, modulating gene expression dynamics in response to LC stress.

Functional validation of epigenetic regulators in carbon assimilation

Given the intricate epigenetic landscape revealed through comprehensive profiling, we next examined carbon-assimilation mechanisms in N. oceanica by focusing on a specific histone mark. H3K4me2 emerged as a prime candidate: (1) it exhibits a distinctive distribution pattern predominantly located within genomic regions of active CSs (Figures 1B and 3A); (2) it is a key epigenetic feature in our 3D chromatin compartment analysis, where its distribution is correlated with spatial genome organization and transcriptional dynamics (Figure 4E); and (3) it displays a unique dual-peak profile around the TSS (Figure 1B), distinct from those of other microalgae and plants. Together, these findings suggest a specialized and complex role for H3K4me2 in transcriptional regulation in N. oceanica.

To validate the function of H3K4me2, two putative H3K4 methyltransferases (NO17G01960 and NO24G02310) were identified from the NanDeSyn database (Gong et al., 2020). NO17G01960 exhibited stable expression under both HC and LC conditions, whereas NO24G02310 was differentially expressed under LC stress (VLC_24h vs. HC_24h; log₂FC = −1.16). NO24G02310 is a member of the protein family HISTONE-LYSINE N-METHYLTRANSFERASE ATXR3 (PTHR46655:SF1); it shares 32% sequence identity with Arabidopsis ATXR3/SDG2 across 557 residues (E-value = 1e−70, bit score = 251) and 32% sequence identity with Oryza sativa SDG701 across 603 residues (E-value = 1e−72, bit score = 250). ATXR3/SDG2 has been validated in vitro as an H3K4me1/2/3 methyltransferase (Guo et al., 2010), and SDG701 protein functions as an H3K4me2/3 methyltransferase (Liu et al., 2017; Liu et al., 2019). Accordingly, NO24G02310 was disrupted via CRISPR-Cas9, generating two frameshift mutants (M4 and M6) with truncations occurring upstream of the catalytic domains (Figure 5A). H3K4me2 ChIP-seq profiling revealed largely similar yet distinct distribution patterns between mutants and wild type (WT), with numerous differential peaks detected (Supplemental Figures 11A–11C; Figure 5B). Notably, H3K4me2 peaks in both mutants shifted closer to TSSs genome-wide, indicating that this methyltransferase significantly influences H3K4me2 positioning (Supplemental Figures 11A and 11D). Western blot analysis also showed a significant reduction in H3K4me1 abundance in the mutants relative to WT (Supplemental Figures 12A and 12B), strongly suggesting that NO24G02310 also acts as an H3K4me1 methyltransferase in vivo. Under LC conditions, mutants M4 and M6 exhibited significant reductions in growth (20.0% and 21.6%, respectively) and biomass productivity (14.7% and 10.1%, respectively), along with decreased Fv/Fm values (5.7% and 11.3%, respectively) (analysis of variance [ANOVA], p < 0.05; Figures 5C–5E). However, these phenotypic differences were absent under HC conditions; there were no significant changes in growth, biomass, or photosynthetic efficiency compared with WT, aside from a slight increase in Fv/Fm in M4 at 2-5 days and a 2.6% reduction in M6 (ANOVA, p > 0.05; Supplemental Figures 12C–12E). These results suggest that H3K4 methyltransferase knockout impairs carbon assimilation and emphasize the essential role of H3K4 methylation in LC adaptation in N. oceanica.

Figure 5.

NO24G02310 knockout reveals that H3K4me2 regulates carbon assimilation.

(A) Genome sequence of NO24G02310-knockout mutants at guide RNA target sites, showing truncation before the functional domains of NO24G02310.

(B) Number of significantly altered peaks, indicating that gene knockout induced greater H3K4me2 dynamics than CCM induction.

(C–E) Growth, biomass productivity, and photosynthetic efficiency (Fv/Fm) of the mutants under low CO₂ (0.04%) conditions. Statistical significance was determined via the t-test (∗p < 0.05, ∗∗p < 0.01).

(F) Correlations between histone mark dynamics and gene expression.

(G) Relationship between differential gene expression and positional shifts of H3K4me2 peaks.

(H) Expression profiles of selected genes potentially contributing to the phenotypes of NO24G02310-knockout mutants. Mean transcripts per kilobase per million mapped reads values are shown.

(I) Dynamics of H3K4me2 modifications around the NoHINT (NO14G01650) gene. The green box denotes the region where peaks disappeared.

To further validate the role of NO24G02310, we conducted a genetic complementation assay by overexpressing NO24G02310 in the mutant strains, which generated the complemented lines C1 and C2. After 6 days of cultivation under LC conditions, the expression levels of NO24G02310 in C1 and C2 increased by 4.2-fold and 9.9-fold, respectively, relative to the WT (Supplemental Figure 12F; t-test, p < 0.01), confirming successful overexpression. The complemented strains showed no significant differences in growth (as determined by optical density at 750 nm [OD₇₅₀]) or biomass at day 6 compared with the WT, except for modest OD₇₅₀ fluctuations on days 1, 2, 4, and 5 (in C1) or days 1 and 2 (in C2) (t-test, p < 0.05; Supplemental Figures 12G and 12H). Intriguingly, Fv/Fm values were reduced by 16.3% and 9.7% in C1 and C2, respectively (Supplemental Figure 12I; t-test, p < 0.01 for C1 and p < 0.05 for C2), indicating a complex relationship between NO24G02310 and the photosynthetic machinery. These results demonstrate that the growth defects observed in the knockout mutants specifically result from the loss of NO24G02310 function.

Transcriptomic analysis of the mutants and WT identified 448 DEGs (M4 vs. WT) and 423 DEGs (M6 vs. WT), clearly distinguishing the mutants from the WT (Supplemental Figures 11E and 11F). Linear regression analysis revealed a positive correlation between changes in H3K4me2 peak abundance and gene expression, supporting a role for H3K4me2 in gene activation (p < 0.01; Figure 5F). Furthermore, both the upregulated and downregulated peak groups exhibited significantly higher percentages of DEGs than randomly selected gene sets, reinforcing the influence of H3K4me2 on transcriptional regulation (p < 0.01; Figure 5G). Transcriptomic profiles under LC conditions revealed distinctive trends among DEGs (Figure 5H). For example, TIA1, which encodes a stress-granule-associated protein, was upregulated, indicating an enhanced stress response in the mutants. Additionally, the histidine triad nucleotide-binding protein gene NoHINT (NO14G01650) lacked detectable H3K4me2 peaks and showed substantially reduced expression (log₂FC = −1.49), indicating critical roles in maintaining growth and carbon assimilation (Figure 5I). Examination of published transcriptomic data from the NanDeSyn database (Gong et al., 2020) revealed that NoHINT was specifically downregulated under LC conditions (log₂FC = −1.29) and remained unaffected by other stressors. This selective response underscores the potential involvement of NoHINT in H3K4me2-mediated regulation of LC adaptation in N. oceanica.

Concurrently, two plasma membrane H⁺-ATPases (PMAs)—potentially critical for nutrient uptake, intracellular pH regulation, and CCM activity—were downregulated in the mutants. Intriguingly, whereas NoPMA1 (NO12G00390) expression remained stable under both HC and LC conditions, NoPMA2 (NO17G00440) was consistently downregulated (LC vs. HC, log₂FC = −1.35), mirroring observations in C. reinhardtii, where reduced PMA expression was linked to impaired adaptation to elevated CO₂ levels (Choi et al., 2021).

The convergence of these expression patterns, coupled with the absence of H3K4me2 peaks at the NoHINT locus, strongly implicates both NoHINT and NoPMA2 as key targets of H3K4 methylase activity, potentially exerting major effects on LC adaptation in N. oceanica. To verify these functional roles, we generated two knockout mutants, NoHINTko-1 and NoHINTko-2 (Figure 6A). Under LC conditions, significant differences in OD₇₅₀ were observed for both mutants (ANOVA, p < 0.05), although only NoHINTko-2 displayed pronounced variation in Fv/Fm. These mutants exhibited subtle but significant growth reductions (5.4%–14.0%, p < 0.05) between days 3 and 7 of cultivation compared with WT, despite the absence of substantial differences in final biomass or Fv/Fm after 9 days (Figures 6B–6D). This transient growth impairment, particularly under nutrient-replete conditions, supports a putative role for NoHINT in regulating carbon assimilation, possibly through specific interactions with enzymes involved in growth processes.

Figure 6.

Knockout and overexpression mutants of NoHINT and NoPMA2 reveal epigenetically mediated regulation of carbon assimilation in N. oceanica.

(A) Genotypic validation of NoHINT-knockout strains.

(B–D) Analysis of growth phenotype, biomass productivity, and photosynthetic efficiency (Fv/Fm) in NoHINT-knockout strains.

(E) Genotypic validation of NoPMA2-knockout mutants.

(F–H) Analysis of growth phenotype, biomass productivity, and photosynthetic efficiency (Fv/Fm) in NoPMA2-knockout mutants.

(I) qPCR analysis of expression levels in NoPMA2 overexpression strains.

(J–L) Analysis of growth phenotype, biomass productivity, and photosynthetic efficiency (Fv/Fm) in NoPMA2 overexpression strains. Statistical significance was determined via the t-test (∗p < 0.05, ∗∗p < 0.01).

For NoPMA2, both knockout (NoPMA2ko-1 and NoPMA2ko-2; Figure 6E) and overexpression (NoPMA2ov-1 and NoPMA2ov-2; Figure 6I) mutants were generated. Under LC conditions, knockout mutants displayed no significant changes in growth or biomass, although NoPMA2ko-2 showed a slight increase in Fv/Fm (5.2%) (Figures 6F–6H). Conversely, overexpression lines showed no differences in growth or biomass but exhibited decreased Fv/Fm values (6.0% and 8.4% for NoPMA2ov-1 and NoPMA2ov-2, respectively; Figures 6J–6L). These consistent alterations in photosynthetic efficiency (all mutants showed significant Fv/Fm differences; ANOVA, p < 0.05) suggest that NoPMA2 modulates photosynthesis by influencing pH gradients across chloroplast membranes.

Collectively, our findings provide compelling evidence that H3K4me2 regulates carbon assimilation in N. oceanica under LC conditions by targeting NoHINT and NoPMA2 (Figure 7). The subtle but reproducible phenotypic changes in the mutants underscore the fine-tuned nature of epigenetic regulation in microalgal carbon metabolism and emphasize the potential of targeted epigenetic modifications to enhance microalgal productivity.

Figure 7.

Conceptual model of NO24G02310-mediated low-CO₂ adaptation in N. oceanica.

Discussion

The epigenome, characterized by hierarchical layers of regulation ranging from fine-scale modifications such as DNA methylation, histone modifications, and nucleosome dynamics to complex higher-order chromatin conformations, plays a pivotal role in controlling gene expression (Casas-Mollano et al., 2008; Kouzarides, 2007; Xiong et al., 2016; Zhang et al., 2009). However, in Nannochloropsis spp.—organisms with high potential for industrial CO₂-to-oil conversion (Radakovits et al., 2012) and a unique position among heterokonts due to the absence of chlorophyll c (unlike diatoms and related taxa; Koh et al., 2019)—the full scope of the epigenetic landscape remains largely unexplored. In this study, we comprehensively examined the structural and chemical features of the N. oceanica epigenome, then integrated them with genomic and transcriptomic data after a shift to LC conditions. Our analyses revealed multi dimensional epigenomic profiles encompassing 3D chromatin organization, nucleosome occupancy, DNA methylation, and histone modifications such as acetylation, methylation, and crotonylation, providing new insight into the epigenetic mechanisms that underlie LC adaptation in carbon assimilation. Moreover, through genetic manipulation of a candidate H3K4 methyltransferase and multi-omics analyses, we identified a central role for H3K4me2 in regulating LC adaptation via its effects on key genes, including NoHINT and NoPMA2 (Figure 7). Functional validation of these targets through gene knockout and overexpression confirmed their subtle but important influences on growth and photosynthetic efficiency. Collectively, these findings establish distinctive epigenetic signatures and highlight the active involvement of epigenetic regulation in Nannochloropsis under LC stress.

N. oceanica exhibits unique epigenetic characteristics compared with higher plants and other algae. Specifically, H3K4me2 peaks show substantial enrichment within promoter and gene-body regions of active or poised genes, differing from the promoter-proximal enrichment observed at TSSs in higher plants (Liu et al., 2019). This dual-peak enrichment pattern more closely resembles that observed in animals (Hu et al., 2013; Rickels et al., 2016) and the brown alga Ectocarpus (Bourdareau et al., 2021), but it sharply contrasts with the distribution patterns observed in C. reinhardtii and plants (Ngan et al., 2015; Liu et al., 2019). H3K4 methylation, generally catalyzed by SET domain–containing methyltransferases, has diverse functional roles due to evolutionary divergence. The H3K4 methyltransferase NO24G02310 in N. oceanica contains a SET domain (IPR001214), a zinc-finger domain (IPR011124), and a chromo-like domain (IPR016197). Functionally, H3K4me2 in Nannochloropsis promotes gene transcription, similar to its role in animals. In contrast, its function in plants remains debated—studies in rice have shown an inverse relationship with gene transcription during development (Liu et al., 2019), whereas in cotton (Gossypium hirsutum) it has been implicated in heat stress response (He et al., 2022).

In addition, N. oceanica displays an exceptionally low level of genomic 5mC (∼0.1%), substantially lower than the methylation levels reported in animals, plants, and even closely related diatoms such as Phaeodactylum tricornutum, where methylated regions represent approximately 5% of the genome (Feng et al., 2010; Veluchamy et al., 2013; Lopez et al., 2015; Fan et al., 2020). This deficiency likely results from the absence of key DNA methyltransferases (DNMTs), including DNMT2 and DNMT3; only a candidate DNMT1 (NO12G03180) has been identified, consistent with the low-methylation landscape evident in Saccharomyces (He et al., 2011). The near absence of methylation, potentially exacerbated by incomplete bisulfite conversion, reflects a distinctive epigenetic architecture that distinguishes N. oceanica from both green algae (e.g., Chlamydomonas) and diatoms, underscoring its unique mode of genomic regulation.

At the higher level of chromatin organization, we propose that H3K4me2 marks are associated with the spatial genome configuration of N. oceanica, particularly regarding compartment switching. Previous studies have demonstrated a role for H3K9 methylation in chromatin compartmentalization in animals (Bian et al., 2020; Wang et al., 2019); H3K27ac and H3K36me3 have been identified as key predictors of A/B compartments (Zheng et al., 2024). Our findings emphasize the intricate interplay between histone modifications and 3D chromatin architecture in Nannochloropsis, underscoring the need for further high-resolution structural analyses. At the fine scale of chromatin structure, N. oceanica displays a nucleosome repeat length of 179 bp; open nucleosome configurations facilitate transcription at promoter and transcription termination site (TTS) regions (Figure 2A; Supplemental Figure 7C). In diatoms such as P. tricornutum, nucleosome depletion primarily occurs ∼150 bp upstream of the TSS in highly expressed genes but not at the TTS (Veluchamy et al., 2015). In plants, nucleosome positioning strongly influences gene expression, evolution, and alternative splicing (Jabre et al., 2021; Zhang et al., 2015). Remarkably, nucleosome depletion around the TTS is rare and has primarily been observed in Nannochloropsis spp. and maize (Chen et al., 2017), suggesting a unique mechanism of transcriptional regulation in N. oceanica.

Based on this comprehensive epigenomic map, we analyzed the epigenetic patterns of key CCM components and assessed their contributions to LC adaptation (Supplemental Table 10). Importantly, most key CCM genes with differential expression exhibited at least one form of epigenetic regulation (Supplemental Figure 13). These epigenetic signatures play critical roles in modulating transcriptional efficiency and coordinating carbon-assimilation processes in N. oceanica. Amid the ongoing challenge of establishing direct causal relationships between phenotypic changes and specific chromatin-modifying enzymes, this study highlights the strong potential of epigenetic breeding by confirming the influence of an H3K4 methyltransferase candidate on the carbon-assimilation phenotype of N. oceanica. Additionally, through genetic manipulation of the machinery underlying H3K4me2 regulation, we proposed a comprehensive model of epigenetic control of carbon assimilation under LC conditions, in which NoHINT regulates growth and NoPMA2 modulates photosynthesis (Figure 7). Thus, to enhance carbon fixation in Nannochloropsis, chromatin-modifying enzymes represent key targets for (epi)genetic engineering, and the principles of epigenetic regulation may inform strategies for improving microalgal productivity. It should be noted that, although our findings suggest a regulatory role for NO24G02310 in controlling NoHINT and NoPMA2 (Figure 7), these interactions are inferred from correlative omics data and phenotypic validation of engineered strains. Further mechanistic studies will be necessary to fully elucidate these direct regulatory relationships and their effects on growth and photosynthetic efficiency under LC conditions.

Taken together, this study provides the first panoramic epigenomic map of carbon assimilation under LC stress in a photosynthetic organism and establishes a model for the epigenetic regulation of carbon fixation in N. oceanica. These findings lay a foundation for enhancing carbon assimilation and biomass productivity through targeted epigenetic modification of industrial microalgae.

Methods

Algae cultivation

N. oceanica IMET1 was inoculated into modified f/2 liquid medium prepared with 35 g l⁻¹ sea salt (Realocean, USA), 1 g l⁻¹ NaNO₃, 67 mg l⁻¹ NaH₂PO₄·H₂O, 3.65 mg l⁻¹ FeCl₃·6H₂O, 4.37 mg l⁻¹ Na₂EDTA·2H₂O, a trace metal mixture (0.0196 mg l⁻¹ CuSO₄·5H₂O, 0.0126 mg l⁻¹ NaMoO₄·2H₂O, 0.044 mg l⁻¹ ZnSO₄·7H₂O, 0.01 mg l⁻¹ CoCl₂, and 0.36 mg l⁻¹ MnCl₂·4H₂O), and a vitamin mixture (2.5 μg l⁻¹ vitamin B₁₂, 2.5 μg l⁻¹ biotin, and 0.5 μg l⁻¹ thiamine HCl) (Kang et al., 2015). Cultures were maintained at 25°C under continuous illumination (80 ± 5 μmol m⁻² s⁻¹) in a 1-l column reactor (inner diameter: 5 cm) and bubbled with 5% CO₂. Cells in the logarithmic growth phase (OD₇₅₀ = 3.0) were harvested via centrifugation, then washed with fresh medium before use in subsequent experiments.

In total, six identical column reactors were used, each containing 800 ml of modified f/2 liquid medium supplemented with 10 mM Tris–HCl buffer (pH = 8.2) to maintain stable pH during cultivation. Equal quantities of seed cells were re-inoculated into the six reactors to achieve an initial OD₇₅₀ of 1.5. The light intensity was maintained at 80 ± 5 μmol m⁻² s⁻¹. All six algal cultures were initially aerated with air containing 5% CO₂ (HC conditions) for 1 h. After this pre-adaptation phase, three cultures continued under HC as controls, whereas the remaining three were switched to aeration with 0.01% CO₂ (LC conditions) (Brueggeman et al., 2012; Fang et al., 2012) to induce CCM activity (Supplemental Figure 1). After transfer to the designated conditions, cell aliquots were collected at 0 and 24 h from each column using a syringe for profiling of DNA methylation, histone modification, chromatin accessibility, and chromatin interaction.

Mutant strain cells were cultured in liquid f/2 medium and maintained at 25°C under illumination of 50 μmol photons m⁻² s⁻¹, with aeration by air or 5% CO₂. Growth curves were generated by cell counting with three biological replicates for each strain. After 6 days of cultivation, cells were harvested by centrifugation and stored at −80°C for use in subsequent experiments.

RNA-seq profiling and data analysis

RNA-seq datasets generated from LC experiments, as described in our previous study (Wei et al., 2019), were retrieved from the NCBI GEO database (GEO: GSE55861). For mutant analyses, 2 μg of total RNA per sample was used as input material for library construction, following previously described procedures (Wei et al., 2019). Paired-end sequencing (2 × 150 bp) of each sample was performed on an Illumina HiSeq platform (Illumina, USA).

RNA-seq datasets were primarily analyzed using the nf-core/rnaseq pipeline v1.4.2 (https://github.com/nf-core/rnaseq). In brief, raw reads were quality-checked with Trim Galore (https://github.com/FelixKrueger/TrimGalore). High-quality paired reads were then aligned to the N. oceanica IMET1 genome (Gong et al., 2020) using STAR (Dobin et al., 2013) with modified intron-length parameters (“--alignIntronMin 20 --alignIntronMax 3000”), and duplicates were marked via Picard (http://broadinstitute.github.io/picard/). Because the N. oceanica IMET1 genome is haploid and a high-quality reference genome for this strain is available (Gong et al., 2020), the mapping accuracy and reliability of RNA-seq achieved by the STAR aligner (Dobin et al., 2013) are guaranteed. Gene expression abundance was quantified using featureCounts (Liao et al., 2013) and StringTie (Pertea et al., 2016). Scripts from Trinity (Grabherr et al., 2011) were then used to construct a gene expression matrix with the RSEM2 method and TMM normalization to correct for variation in library size across samples. Transcripts per kilobase per million mapped reads (TPM) values were subsequently averaged among replicates. For the LC experiments, raw counts from HC (0 h) and LC (VLC_24h) samples obtained via featureCounts (Liao et al., 2013) were used for differential gene expression analysis with edgeR (Robinson et al., 2010), applying an FDR ≤ 0.001 and a minimum fold change >2.

Detection of histone PTMs using mass spectrometry

Histone proteins from N. oceanica IMET1 (Supplemental Table 11) were isolated and analyzed at Beijing Biotech Pack Scientific. Histones were first extracted via acid extraction. Cells were resuspended in 1 ml of hypotonic lysis buffer, incubated at 4°C for 30 min on a rotator, and centrifuged (12 000 g, 10 min, 4°C); the pellet was resuspended in 400 μl of 0.4 N H₂SO₄ (0.2 M H₂SO₄) for ≥ 30 min at 4°C. After centrifugation (12 000 g, 10 min, 4°C), the supernatant was transferred to a fresh tube and precipitated with trichloroacetic acid at 4°C for 1.5 h, incubated on ice for 30 min, and centrifuged (12 000 g, 10 min, 4°C); the pellet was washed twice with pre-chilled acetone, air-dried, and resuspended in 30 μl of ultrapure water. Histones were then separated by 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). A 10-μl sample mixed with 2.5 μl of 5× loading buffer was heated at 100°C for 5 min. A 5-μl marker and 12.5-μl sample were loaded onto the gels, separated at 80 V for 15 min and 120 V for 60 min, and visualized by Coomassie Brilliant Blue staining. Target bands were excised into ∼1-mm³ gel pieces, destained with 50% acetonitrile (ACN)–50 mmol/l NH₄HCO₃, dehydrated with 100% ACN, and dried. Gel pieces were rehydrated in 30 μl of 100 mmol/l NH₄HCO₃ and reacted with 200 μl of propionylation reagent (150 μl propionic anhydride + 50 μl ACN) under pH 8–9 (repeated). Samples were then washed, dehydrated, and digested overnight at 37°C with trypsin (0.025 μg/μl; Promega). Peptides were extracted with 5% trifluoroacetic acid–50% ACN–45% water, sonicated, centrifuged, vacuum-dried, desalted (self-packed column), resuspended in 0.1% formic acid–2% ACN, and analyzed on an Orbitrap Exploris 480 mass spectrometer (Thermo Scientific). The mobile phases were A (0.1% formic acid in water) and B (0.1% formic acid in 80% ACN) with a gradient of 0–2 min (4%–8% B), 2–105 min (8%–28% B), 105–120 min (28%–40% B), and 120–131 min (95% B) at 600 nL/min (column: 150 μm i.d. × 170 mm, Reprosil-Pur 120 C18-AQ 1.9 μm). MS parameters were set as follows: MS1—resolution 60 000, AGC 300%, maximum injection time 50 ms, m/z 300–1800; MS2—resolution 15 000, AGC 100%, maximum injection time 22 ms, NCE 30. Spectra were analyzed using Byonic software with the following parameters: protease (trypsin); variable modifications (oxidation, acetylation, propionylation, lactylation, phosphorylation, mono-/di-/trimethylation, GlyGly, crotonylation); maximum missed cleavages (3); and peptide/fragment mass tolerances (20 ppm/0.02 Da). The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository under dataset identifier PXD066363.

ChIP-seq profiling and data analysis

N. oceanica cells (1–2 million cells/ml) from 800-ml cultures were fixed under vacuum for 30 min with 1% formaldehyde (Sigma, USA) at room temperature. After fixation, cells were incubated for 10 min under vacuum with 0.15 M glycine at room temperature. Nuclei were isolated as previously described (Wei and Xu, 2018), and nuclei from 1 g of fixed material were used for each ChIP assay. The isolated nuclei were resuspended in 1 ml of sonication buffer (10 mM potassium phosphate, pH 7.0, 0.1 M NaCl, 0.5% sarkosyl, and 10 mM ethylenediaminetetraacetic acid [EDTA]), and chromatin was sheared by sonication using a Covaris S220 to yield fragments of approximately 300–1000 bp. The sonicated sample was mixed with 100 μl of 10% Triton X-100, and 50 μl of this mixture was reserved as the input control. The remaining sheared chromatin was combined with an equal volume of IP buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 10 μM ZnSO₄, 1% Triton X-100, 0.05% SDS) and incubated with antibodies against H3K9ac (Abcam, UK; ab10812), H3K27ac (Millipore, Germany; 07-360), H3K4me2 (Abcam; ab7766), or crotonyllysine (PTM BioLab, USA; PTM-501). After overnight incubation at 4°C, 50 μl of protein A/G magnetic beads (Millipore, Germany) were added and incubated at 4°C for 2 h. The beads were washed at 4°C in the following sequence: three times with IP buffer, once with IP buffer containing 500 mM NaCl, and once with LiCl buffer (0.25 M LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, and 10 mM Tris, pH 8.0) for 5 min each. Chromatin retained on the beads was eluted in 200 μl of elution buffer (50 mM Tris, pH 8.0, 200 mM NaCl, 1% SDS, and 10 mM EDTA) at 65°C for 30 min, followed by proteinase K digestion at 65°C for 6 h. DNA was purified using a Qiagen kit (Qiagen, Germany); subsequent end-repair, A-tailing, adaptor ligation, and library amplification were performed using a standard Illumina protocol (Illumina). The final libraries were sequenced on an Illumina HiSeq 2000 platform with 2 × 150 bp paired-end reads.

Additionally, three antibodies targeting histone modifications—H3K9me3, H3K27me3, and H2A.Z—were used to profile genome-wide distributions in N. oceanica. Among these, H3K9me3 and H3K27me3 have been reported in P. tricornutum (Veluchamy et al., 2015), and H2A.Z has been detected in Arabidopsis (Gómez-Zambrano et al., 2019). However, these marks were not effectively enriched by ChIP in N. oceanica (ChIP-seq data deposited in the Sequence Read Archive, accessions SRR16288281–SRR16288288), suggesting that the histone modification landscape of N. oceanica differs from that of diatoms and Arabidopsis.

The ChIP-seq datasets were analyzed using the nf-core/chipseq pipeline v1.2.1 (https://github.com/nf-core/chipseq) with minor modifications. In brief, reads were trimmed with Trim Galore (https://github.com/FelixKrueger/TrimGalore) and aligned to the reference genome using BWA (Li and Durbin, 2009). Mapped reads were marked for duplication with Picard (http://broadinstitute.github.io/picard/) and quality-checked via phantompeakqualtools (Landt et al., 2012) and deepTools (Ramírez et al., 2016). Peak calling was performed with MACS2 (Zhang et al., 2008) using the “--narrow_peak” flag and input samples as controls, followed by peak annotation (relative to gene features) via annotatePeaks.pl from HOMER (Heinz et al., 2010). For each histone modification, consensus peaks were merged with BEDTools (Quinlan and Hall, 2010) and quantified using featureCounts (Liao et al., 2013). Differential binding analysis was then carried out with DESeq2 (Love et al., 2014), using an FDR threshold of < 0.01. Promoters were defined as genomic regions extending 1 kbp upstream from each TSS.

MNase-seq library construction and sequencing data analysis

Nannochloropsis nuclei were isolated and digested according to the method of Dai (Dai et al., 2017). Briefly, N. oceanica cells (1–2 million cells/ml) from 800-ml cultures were fixed under vacuum for 30 min with 1% formaldehyde (Sigma) at room temperature. After fixation, cells were incubated under vacuum for 10 min with 0.15 M glycine (Wei and Xu, 2018). Fixed cells were collected by centrifugation at 2500 g and ground under liquid nitrogen with a mortar and pestle. The resulting powder was resuspended in hypertonic buffer A (50 mM HEPES [pH 7.5], 1 mM EDTA [pH 8.0], 150 mM NaCl, 1% Triton X-100, and 1× protease inhibitor cocktail [Roche, Switzerland]) and shaken for 30 min at 4°C. The suspension was filtered through Miracloth (Calbiochem, Germany) into fresh 50-ml tubes. Nuclei were collected by centrifugation at 4°C for 20 min at 4000 g, washed twice in buffer A, and resuspended in buffer D (10% sucrose, 50 mM Tris–HCl [pH 7.5], 25 mM MgCl₂, and 1 mM CaCl₂). Chromatin was digested for 10 min at 37°C with two units of micrococcal nuclease (Sigma), and reactions were stopped by adding 5 mM EDTA. Mononucleosome-sized fragments were gel-purified (Petesch and Lis, 2008), and the nucleosomal population was subjected to paired-end high-throughput sequencing.

The MNase-seq datasets were analyzed using the nf-core/mnaseseq pipeline v1.0 (https://github.com/nf-core/mnaseseq). In brief, raw MNase-seq reads were quality-checked with Trim Galore (https://github.com/FelixKrueger/TrimGalore), and high-quality reads were aligned to the reference genome (IMET1v2) using Bowtie2 (Langmead and Salzberg, 2012). Unpaired and discordant alignments were removed; duplicates were marked with Picard (http://broadinstitute.github.io/picard/). Alignments from biological replicates were merged and assessed for quality using deepTools (Ramírez et al., 2016). Nucleosome positions, occupancy levels, and differential occupancies between LC and HC samples were determined using DANPOS2 (Chen et al., 2013).

Chromatin state analysis

ChromHMM (Ernst and Kellis, 2017) was used to integrate MNase-seq data with ChIP-seq data for H3K9ac, H3K27ac, Kcr, and H3K4me2 to identify major recurrent combinatorial and spatial patterns of histone marks. Multiple models were trained (“LearnModel” mode) in parallel with CS numbers ranging from three to nine; the optimal model was selected using the “CompareModels” mode of ChromHMM (Ernst and Kellis, 2017). The model with five CSs was chosen, and the CS of each gene was defined as the CS at its TSS. Transitions in CSs from HC to LC conditions were summarized using custom scripts. GO and pathway enrichment analyses of gene sets with distinct transition types were performed using tools provided by the NanDeSyn database (Gong et al., 2020).

Generation of targeted knockout N. oceanica mutants via CRISPR

A modular CRISPR–Cas9 toolbox system (pNOC-ARS-CRISPR; Poliner et al., 2018) was used to construct CRISPR plasmids. Briefly, guide RNAs were designed via the CHOPCHOP platform (http://chopchop.cbu.uib.no). The CRISPR–Cas9 episome was assembled as described by Poliner et al. (2018) and introduced into N. oceanica via electroporation (Wang et al., 2016). Target regions in mutants and WT were amplified by genomic PCR, and PCR products were verified through Sanger sequencing.

Photosynthesis parameter monitoring

Chlorophyll fluorescence in mutant and WT strains (control) was measured using a pulse-amplitude–modulated fluorometer (Image PAM, Walz, Effeltrich, Germany) after 20 min of dark adaptation. The PSII maximum quantum yield (Fv/Fm) was determined according to established protocols (Maxwell and Johnson, 2000; Wei et al., 2017).

Western blot analysis of H3K4me1, H3K4me2, and H3K4me3

Western blotting was performed at Sinogene Biotech. For total protein extraction, approximately 10 mg of algal culture was ground in liquid nitrogen and transferred to a 1.5-ml centrifuge tube. In total, 150 μl of radio immunoprecipitation assay (RIPA) lysis buffer containing protease inhibitors was added, followed by sonication for 30 s. The mixture was centrifuged at maximum speed (4°C, 15 min), and the supernatant was collected as the soluble protein fraction. Protein concentration was quantified using the Bradford assay. For SDS–PAGE, a 4% stacking gel and 12% separating gel were used with a protein marker (PM2510, SMOBIO). Thirty micrograms of total protein were loaded per lane. Electrophoresis was conducted at 90 V for 30 min and 120 V for 2 h. Proteins were transferred to a polyvinylidene fluoride membrane at a constant current of 300 mA for 2 h using a Bio-Rad transfer system, in accordance with the manufacturer’s instructions. Nonspecific binding sites were blocked with Fast Protein-Free Block Buffer (#W1028, Sinogene Biotech, China) for 3–5 min at room temperature. The membrane was incubated with primary antibodies against H3K4me1 (#91289), H3K4me2 (#91321), or H3K4me3 (#91263) (all from Active Motif) at a 1:1000 dilution for 1 h at room temperature, followed by three 10-min washes in Tris-buffered saline plus Tween (TBST). The membrane was then incubated with a horseradish peroxidase–conjugated secondary antibody (1:2000 dilution) for 1 h at room temperature; it was washed twice with TBST and once with TBS (10 min each). For detection, the membrane was treated with ECL chemiluminescent substrate (29050, Engreen, China) on plastic wrap and exposed to X-ray film in a darkroom. Each film was scanned and saved as a JPEG file for densitometric analysis. After target protein detection, the membrane was stripped using stripping buffer to remove primary antibodies and reprobed with a β-actin antibody as an internal control. Densitometric values of target proteins were normalized to β-actin to adjust for experimental variation, representing relative expression levels in each sample.

ChIP-seq sequencing

For the crosslinking reaction, mutant and WT cells (as controls) were incubated with 37% formaldehyde as previously described (Wei and Xu, 2018). Cells were then incubated with an anti-H3K4me2 antibody (Cell Signaling Technology, USA; #9725). ChIP-seq library preparation was performed by Novogene Corporation (Beijing, China), and library quality was evaluated using an Agilent Bioanalyzer 2100 system. Paired-end sequencing (2 × 150 bp) of each sample was subsequently conducted on an Illumina platform (Illumina).

Overexpression of NoPMA2 and NO24G02310

To construct the NoPMA2 overexpression vector, the NoPMA2 coding sequence from N. oceanica IMET1 was amplified from cDNA by PCR using primer F (5′-acctctttttaaattggtaccatggttgttgtcgacgagtcc-3′) and primer R (5′-tctcttccggtaggggtacctcagaccataatcgagatatcagat-3′). The amplified fragment was inserted into the KpnI-digested vector pXJ53-1 (GenBank: OR338766.1). NoPMA2 expression was driven by the endogenous VCP1 promoter, and the bleR gene served as the selection marker. Transformants were cultured in triangular flasks at 25°C under illumination (50 μmol m⁻² s⁻¹) with shaking at 200 rpm.

To construct the NO24G02310 overexpression vector for complementation experiments, the NO24G02310 coding sequence was amplified from N. oceanica IMET1 cDNA by nested PCR using three primer pairs: F1 (5′-acctctttttaaattggtatgaagcatccacccagcc-3′) with R1 (5′-ttggaggagccgtagttctgtt-3′), F2 (5′-aacagaactacggctcctcca-3′) with R2 (5′-gccatcttcctcaccttcatc-3′), and F3 (5′-ggatgaaggtgaggaagatggcaagcgtc-3′) with R3 (5′-tcactccctcctcctggagctcggtaccctatgccagtatgaccacgtcct-3′). The final PCR product was cloned into the KpnI-digested pXJ199H-GFP1 expression vector (GenBank: PV943961) using an infusion-based seamless cloning strategy. In this construct, transgene expression was controlled by the endogenous VCP1 promoter with the hygromycin phosphotransferase (HygR) gene serving as the selectable marker. Transformants of NO24G02310-knockout strains were cultured in triangular flasks at 25°C under 50 μmol m⁻² s⁻¹ light and shaken at 200 rpm.

Hi-C sequencing and data analysis

Hi-C libraries were constructed from N. oceanica cells under LC and HC conditions following the standard protocol (Belton et al., 2012) with minor modifications. Briefly, microalgal cells were crosslinked with 2% formaldehyde at room temperature under vacuum for 30 min. The crosslinking reaction was quenched with 2.5 M glycine for 10 min at room temperature. Cells were then ground in liquid nitrogen and resuspended in 25 ml of extraction buffer I (0.4 M sucrose, 10 mM Tris–HCl, pH 8, 10 mM MgCl₂, 5 mM β-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride [PMSF], and protease inhibitor cocktail, 13 units). The suspension was centrifuged at 4000 rpm for 20 min at 4°C. The resulting pellet was resuspended in extraction buffer II (0.25 M sucrose, 10 mM Tris–HCl, pH 8, 10 mM MgCl₂, 1% Triton X-100, 5 mM β-mercaptoethanol, 0.1 mM PMSF, and protease inhibitor, 13 units) and centrifuged at 14 000 rpm for 10 min at 4°C. The pellet was then resuspended in extraction buffer III (1.7 M sucrose, 10 mM Tris–HCl, pH 8, 0.15% Triton X-100, 2 mM MgCl₂, 5 mM β-mercaptoethanol, 0.1 mM PMSF, and protease inhibitor, 13 units) and layered onto an equal volume of clean extraction buffer III before centrifugation at 14 000 rpm for 10 min. The pellet was washed twice with 500 μl of ice-cold 1× CutSmart buffer and centrifuged at 2500 g for 5 min. Nuclei were washed once with 0.5 ml of restriction enzyme buffer and solubilized with dilute SDS, then incubated at 65°C for 10 min. SDS was quenched with Triton X-100, and chromatin digestion was performed overnight at 37°C on a rocking platform using the four-cutter restriction enzyme MboI (400 units).

DNA ends were subsequently labeled with biotin-14-dCTP and subjected to blunt-end ligation of crosslinked fragments. Proximal chromatin fragments were re-ligated using a ligation enzyme. Nuclear complexes were reverse crosslinked by incubation with proteinase K at 65°C, and DNA was purified via phenol–chloroform extraction. Biotin residues on non-ligated ends were removed using T4 DNA polymerase. The sonicated DNA fragments (200–600 bp) were end-repaired with T4 DNA polymerase, T4 polynucleotide kinase, and Klenow DNA polymerase. Biotin-labeled Hi-C DNA was enriched using streptavidin C1 magnetic beads. After A-tailing and ligation of Illumina paired-end adapters, Hi-C libraries were amplified by PCR (10–15 cycles) and sequenced on an Illumina HiSeq 2500 platform (2 × 125 bp).

The Hi-C sequencing datasets were processed via the HiC-Pro pipeline v2.9.0 (Servant et al., 2015), which utilizes Bowtie2 (Langmead and Salzberg, 2012) for read alignment to the reference genome, performs quality filtering, and normalizes contact maps using ICE normalization (https://github.com/hiclib/iced). Interaction maps were visualized and compared with HiCPlotter (Akdemir and Chin, 2015). The ultra-deep sequencing depth, averaging 1910×, provided sufficient coverage for robust interaction analysis. Data quality was assessed through multiple metrics, including an average mapping rate of 94.44% (67.45% uniquely mapped), an average valid interaction rate of 94.98%, and high concordance between replicates (Pearson’s R = 0.81 for HC samples and 0.85 for LC samples). ICE normalization effectively reduced variance in interaction frequencies by 10.38%–12.91% (mean 11.93%) at 10-kb resolution, mitigating technical bias. To capture chromatin architecture across multiple scales, Hi-C data were analyzed at various resolutions to detect chromosome compartments (10 and 40 kb), chromatin loops (5 kb), and significant chromatin contacts (1 kb and fragment levels). Using the 10-kb resolution maps, the 3D genome structures were reconstructed via 3DMax v1.0 (Oluwadare et al., 2018), which applies a maximum-likelihood algorithm and visualizes the models with NGLVieweR v1.3.1 (https://github.com/nvelden/NGLVieweR/). Chromosomal compartments were identified using cooltools v0.4.0 (https://github.com/open2c/cooltools). Significant chromatin contacts were identified via FitHiC v1.1.3 (Ay et al., 2014), significant interaction differences were analyzed via HiCcompare v1.8.0 (Stansfield et al., 2018), and chromatin loops were called via Juicer v1.5.6 (Durand et al., 2016). Only interactions supported by both biological replicates were retained for further analysis.

Whole-genome bisulfite sequencing (WGBS) and data analysis

In total, 5.2 μg of genomic DNA was fragmented into 200- to 300-bp fragments by sonication (Covaris S220, USA), then processed via end repair and adenylation. Cytosine-methylated barcodes were ligated to the sonicated DNA, and fragments were subjected twice to bisulfite conversion using the EZ DNA Methylation-Gold Kit (Zymo Research, USA). The resulting single-stranded DNA was PCR-amplified using KAPA HiFi HotStart Uracil+ ReadyMix (2×). Libraries were quantified with a Qubit 2.0 fluorometer (Life Technologies, USA) and quantitative reverse transcription PCR; insert sizes were evaluated using an Agilent Bioanalyzer 2100 system. Indexed samples were clustered on a cBot Cluster Generation System using the TruSeq PE Cluster Kit v3-cBot-HS (Illumina, USA) in accordance with the manufacturer’s instructions. Sequencing was performed on an Illumina HiSeq 2500 platform, generating 125-bp paired-end reads. Image analysis and base calling were conducted via the Illumina CASAVA pipeline.

Bismark software (version 0.16.1) (Krueger and Andrews, 2011) was used to align bisulfite-treated reads to the reference genome under default parameters. Methylation extractor results were converted into bigWig format for visualization using the IGV browser (Thorvaldsdóttir et al., 2012). Only cytosines supported by sequencing reads were considered valid, and each valid cytosine was treated equally in subsequent analyses. Bisulfite conversion efficiency was assessed by spiking unmethylated lambda phage DNA into each sample before bisulfite treatment. Sequencing reads were aligned to the lambda phage reference genome (accession J02459.1) using Bismark v0.19.0 (Krueger and Andrews, 2011), and cytosine conversion rates were calculated. The mean conversion efficiency across all samples was 99.87%; the minimum efficiency was 99.85%. High reproducibility among biological replicates, reflected by a standard deviation of 0.0093%, confirmed the reliability of methylation detection. Differentially methylated regions were detected via the R/Bioconductor package DMRcaller v1.26.0 (Catoni et al., 2018) with default parameters, integrating multiple methods (“noise_filter”, “neighbourhood” and “bins”). The methylation level of each valid cytosine (MLmc) was calculated as the ratio of methylated cytosine reads (mCs) mapped to a given locus relative to the total cytosine reads at that locus. The methylation level of a genomic fragment (MLgf) was determined as the mean methylation of all cytosines (and complementary guanines) within the fragment. For genomic components, the methylation level (MLgc) was defined as the average MLmc across all cytosines within that component. Whole-genome methylation (MLwg) was calculated similarly; to correct for technical artifacts, noise, or batch effects, MLgc values were normalized against MLwg, yielding the normalized methylation level (NMLgc) (Xiang et al., 2010).

Data and code availability

Raw sequencing data for WT N. oceanica are available in the Sequence Read Archive under accession numbers PRJNA769792 (ChIP-seq), PRJNA769749 (MNase-seq), PRJNA769619 (WGBS), and PRJNA513887 (Hi-C). RNA-seq and ChIP-seq data for mutant strains are deposited under PRJNA769958. All 73 datasets have also been integrated into the NanDeSyn database (Gong et al., 2020) and can be accessed through its genome browser (http://nandesyn.single-cell.cn/browser) for comparative analyses with other Nannochloropsis omics data. To promote transparency and reproducibility, all scripts and pipelines used for data processing and visualization have been deposited in the GitHub repository (https://github.com/gongyh/NannoEpiCode). Key datasets and outputs from this study are also available in Figshare (https://doi.org/10.6084/m9.figshare.c.7620293). For additional information or assistance with the analysis code, please contact the corresponding author.

Funding

This work was supported by the Synthetic Biology Program of the Ministry of Science and Technology of China (2021YFA0909700), the National Natural Science Foundation of China (grant numbers 32200448, 32370097, 31741005, and 31900071), the Shandong Provincial Natural Science Foundation (ZR2022QC020), the Biological Carbon Sequestration Program (ZDRW-ZS-2016-3 and KSZD-EW-Z-017) of the Chinese Academy of Sciences, and grant I201908 from the Qingdao Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences.

Acknowledgments

No conflict of interest is declared.

Author contributions

Y.G., Q.W., L.S., and J.X. conveived and designed the research and wrote the manuscript. L. Wei and Q.W. performed the experiments and acquired the data. Q.W., L. Wang, N.L., and X.D. generated the mutants and performed phenotypic profiling. Y.G. performed the bioinformatic analyses. Y.G., Q.W., C.S., Y.X., L.S., and J.X. analyzed and interpreted the data. All authors read and approved the final manuscript.

Published: September 25, 2025