Blood-derived gene expression profiles associated with dietary microalgae oil intake and methane emission variation in lambs

Abstract

Background

Minimising methane (CH₄) emissions from livestock production is a global priority, and feed modifications, such as supplementing diets with microalgae, have previously been shown to help reducing enteric CH₄ production. This study explored blood-derived host gene expression profiles from twenty lambs supplemented with increasing levels of microalgae oil to investigate their transcriptional responses associated with varying microalgae oil levels while also exploring the host systemic responses towards varied CH₄ productions.

Results

Findings revealed no significant changes in CH₄ production with increasing levels of microalgae oil intake through phenotypic analysis (P = 0.18). However inter-individual variations in CH₄ production ranged from 27.02 to 47.86 g/day throughout the study period. Blood RNA-Sequencing identified 64 significant genes including DHCR7, DHCR24, HMGCS1, INSIG1, LSS, MSMO1, and SQLE, which were involved in lipid metabolism, and steroid biosynthesis that became enriched alongside increasing microalgae oil intake levels thereby contributing to a positive impact on lambs’ metabolic functions. Additionally, seven significant blood-expressed host genes (NME4, MARCHF3, PLXNB3, LOC132657460, LOC121819234, LOC105603087, LOC101116551) functionally enriched in nucleotide metabolic pathways and immune responses were identified to have significant positive associations with increasing CH₄ production. Importantly, this study found no overlap between genes associated with microalgae oil intake and those linked to CH₄ emissions.

Conclusions

Findings suggest that microalgae oil intake and inter-individual variations in CH₄ production are associated with distinct blood-derived transcriptional responses. Although such signals should be interpreted as proxies for systemic host responses rather than direct measures of rumen-specific processes, these results emphasise the importance of considering host-associated molecular variations alongside dietary CH₄-mitigation strategies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12864-026-13110-1.

Background

The persistent rise in greenhouse gas (GHG) emissions poses a considerable challenge to climate change mitigation efforts, with methane (CH₄) from livestock production systems being a particularly significant contributor. Sheep produce CH₄ as a byproduct of ruminal fermentation, a process involving breakdown of feed by rumen microbes, leading to volatile fatty acids (VFA) production including propionate, acetate, and butyrate along with hydrogen (H₂) release, which is utilized by methanogens to produce CH₄ [1]. This process accounts for approximately 15% of overall feed energy intake [2] necessitating increased feed consumption to compensate for energy loss and contributing to further GHG generation during feed production, increasing costs and reducing economic efficiency. Consequently, several diet-based strategies have been explored aiming to reduce and suppress livestock CH₄ production [3–5].

Microalgae (Schizochytrium sp.) has been identified as one of the promising supplements [6] to reduce CH₄ emissions in vitro [7] and in vivo [8, 9]. However, as reported in sheep and cows [10], excessive microalgae use can negatively affect dry matter intake (DMI) and feed palatability. This underscores the importance of balanced supplementation, which may require long-term trials to assess both effects on animals and the potential of rumen microbes to adapt to such diet manipulation strategies limiting effectiveness. Consequently, it is equally crucial to explore additional approaches alongside dietary modifications, which can further contribute to CH₄ emission lowering.

The adoption of genetic or genomic selection methods to identify animals with inherently low CH₄ production offers complimentary avenues for long-term mitigation of CH₄, given the moderate trait heritability (heritability, h² = 0.29 ± 0.05) [11]. Transcriptomics-based ribonucleic acid sequencing (RNA-Seq) can support these methods by identifying host genes and pathways associated with variation in CH₄ productions. However, only few RNA-Seq studies [12, 13] have explored CH₄ production in sheep particularly using rumen samples, and blood gene expression profiles associated with CH₄ variations still remain unexplored in sheep and whole livestock, apart from a comparable study in beef-on-dairy cattle [14]. In addition, little is known about host blood transcriptomic responses to varying levels of dietary microalgae oil.

Blood samples can therefore be an important proxy for CH₄ studies, as CH₄ production is not only influenced by rumen fermentation, but also by host physiological and metabolic status. While CH₄ is produced in the rumen, host-associated factors such as genetics, feed intake patterns or metabolism can influence CH₄ production [15] and subsequently partially reflected in blood RNA profiles [16]. For instance, immune system activities or metabolic stress can indirectly influence CH₄ production by modifying patterns of feed intake, partitioning of nutrients, rate of digestion as well as functioning of rumen [16]. A recent study on Japanese black cattle has shown correlations between circulating blood metabolites including beta hydroxy butyric acid, and CH₄ production, supporting the relevance of blood-based molecular phenotypes for CH₄-related traits [17]. In contrast, rumen sampling is an invasive procedure requiring surgical fistulation or tissue collection post-slaughter, making it less practical for studies involving live animals [18]. While blood metabolomics is useful to detect downstream biochemical markers associated with CH₄ production, blood RNA-Seq may provide additional insights into upstream pathways such as host immune, and physiological responses, indirectly contributing to alterations in CH₄ production, and thus can serve as a proxy for systemic host responses associated with inter-individual CH₄ variation and increasing intake levels of microalgae oil.

Therefore, this study used whole blood RNA-Seq in lambs to investigate host transcriptional responses towards graded levels of microalgae oil (Schizochytrium sp.) supplementation and inter-individual variations in CH₄ production. CH₄ production was chosen as the primary phenotype other than CH₄ yield or intensity because CH₄ production represents the total amount of CH₄ produced by an animal. However, CH₄ yield or intensity are CH₄ efficiency traits which are ratio-based and incorporate other phenotypic variables (like feed intake) into calculation. Consequently, variations in CH₄ yield or intensity may arise due to changes in either CH₄ emissions or denominator trait, making biological interpretation complex [19]. We hypothesize that dietary microalgae oil induces distinct molecular responses in the host and that inter-individual variations in CH₄ production are associated with specific transcriptional pathways. By linking blood gene expression profiles in lambs with dietary microalgae oil and CH₄ measurements, this study aimed to characterize host-level transcriptional signatures that may complement nutritional and genetic strategies for CH₄ mitigation while maintaining animal productivity.

Materials and methods

Experimental design and animal management

Samples in this study were obtained from a previously conducted experimental trial [20, 21]. Briefly, twenty male weaned lambs (crosses of Texel and mule (Bluefaced Leiscester x Scottish Blackface breeds), with average age of 13 ± 1.20 weeks (mean ± standard deviation), average weight of 30 ± 3.6 kg) were acquired from Agri-Food and Biosciences Institute (AFBI), Loughgall flock which were then housed and managed at the AFBI Hillsborough research facility. Lambs were assigned to one of the four dietary groups, which consisted of a total mixed ratio (TMR) diet introduced as 50:50 ratio of grass silage and concentrate at dry matter (DM) basis (see Additional file 1). Diets were formulated to be isoenergetic and isoproteic, providing approximately 11.72 MJ/kg DM metabolizable energy and 13.66 gN/kg DM metabolizable protein, supporting a target live weight gain of 0.20 kg/day. Dietary groups differed only in the level of microalgae oil (Schizochytrium sp.) included in the concentrate component, specifically: 0.00% (Control, n = 6), 0.54% (Low, n = 4) 1.08% (Medium, n = 5) and 1.62% (High, n = 5) per kg DM along with ad libitum access to water. To balance the microalgae oil intake across the four dietary groups, soya oil was used at inverse rations: 1.62% (Control, n = 6), 1.08% (Low, n = 4) 0.54% (Medium, n = 5) and 0.00% (High, n = 5)). The TMR diet was provided once per day, between 09:00 and 10:00 h. The study was conducted over nine weeks, including a prior three‑week adaptation period during which lambs were housed in individual pens and offered ad libitum grass silage. Feed intake was recorded daily, and body weight was measured weekly for each lamb. Dry matter intake (DMI) and average daily live weight gain (ADG) were individually calculated as follows:

1. , where,

   

2. 

After nine weeks, lambs were moved to indirect open-circuit respiratory chambers for three days following 1 day of adaptation. CH₄ production (g/day) for each animal was measured over the last 48-hour period of the CH₄ production measurements.

Sample acquisition

Blood was collected in the morning following a 3-hour period of fasting and post CH₄ measurement (on the last day of chamber trial) using PAXgene blood RNA tubes (BD, Franklin Lakes, New Jersey, USA; and QIAGEN, Venlo, Netherlands), centrifuged (2,750 g,10 min at 4 °C) and plasma was isolated. The animal study work was conducted at the Agri-Food and Biosciences Institute (AFBI) Hillsborough, under the regulations of the Department of Health for Northern Ireland, according to the United Kingdom (UK) Animals (Scientific Procedures) Act 1986 requirements and AFBI (Hillsborough) Animal Welfare and Ethical Review Body approval.

RNA extraction, sequencing and data analysis

Total RNA was isolated from acquired plasma using PAXgene Blood RNA Kits following manufacturer’s protocols. Subsequently, RNA samples underwent library preparation using the TruSeq3 RNA Library Prep Kit v2 (Illumina, San Diego, California, USA). Resulting libraries underwent quality control (QC) to ensure the expected size distribution of cDNA fragments and adapter-ligated libraries were sequenced on the Illumina NovaSeq Platform employing 150 bp (bp) paired-end reads according to the manufacturer’s protocol. To ensure data quality, the sequenced data underwent sample QC using FastQC (v0.12.1) [22] followed by pre-processing steps, such as adapter sequence trimming and filtering for low-quality reads, using Trimmomatic (v0.39) [23]. The trimmed reads were aligned to sheep (Ovis aries) reference genome (Genome assembly, ARS-UI_Ramb_v3.01) available in the National Centre for Biotechnology Information (NCBI) using the HISAT2 (v2.1.0) [24] software. A count matrix was generated using featureCounts (v2.0.1) [25] representing the number of mapped reads for each gene. Normalization of the count matrix was performed using DESeq2 [26] which uses the median-of-ratios method to correct for sequencing depth variation. To filter out the genes with low expression, as genes with low counts often reflect non-meaningful associations [27], a strict filtering criterion was applied where genes with counts less than 10 were first converted to zero and subsequently removing all rows with zero counts in more than 70% samples. Total consumption of microalgae oil was adjusted for animals’ body weight (g/kg body weight (BW)) to account for individual differences in body size and feed intake. DGE analyses were conducted using DESeq2 in R (v 4.3.3) [28], incorporating microalgae oil intake (g/kg BW) and CH₄ production (g/day) as continuous traits, thereby allowing gene expression levels to vary as a linear function of microalgae oil intake and CH₄ production which were analysed as two separate analyses within the DESeq2 design. Modelling these traits as continuous phenotypes improves statistical power by leveraging the full spectrum of phenotypic variation rather than relying on discrete group contrasts. A primary cut-off threshold was established for the adjusted P-value (P_adjusted) from the Wald test using the Benjamini-Hochberg (BH) method and fold change, specifically with P_adjusted < 0.05 and an absolute Log₂ fold change (henceforth |Log₂ fold change|) ≥ 2 (4-fold change) to identify significant differentially expressed genes (DEGs) [29]. To further investigate the relationship between CH₄-associated gene expression and CH₄ production, a linear regression analysis was carried out in R utilizing the normalized count matrix obtained from variance stabilising transformation (VST) method in DESeq2. Model outputs generated from the linear regression analyses such as regression coefficient (β), coefficient of determination (R²), and associated P were extracted. P values were then subjected to multiple testing correction using the BH method. Scatter plots were generated to better visualize each gene’s association.

Functional enrichment analysis of DEGs

To infer biological processes associated with varying levels of microalgae oil and CH₄ production involving significant DEGs, functional overrepresentation analysis (ORA) were conducted with the R package clusterProfiler [30] based on gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) databases. As input gene list for DEGs, the list of DEGs ranked for decreasing absolute Log₂ fold change was used. Analyses were conducted with a BH corrected P cutoff of 0.05 applied to identify significant GO terms and KEGG pathways. The AnnotationHub (v 3.12.0) (hub: AH114633) [31] was used to annotate the GO entry for sheep.

Construction of Protein-Protein Interaction (PPI) network on DEGs associated with microalgae oil inclusion in the diet

For constructing the PPI network from DEGs associated with increased microalgae oil intake, STRING database [32] was used. A cut-off criterion of a combined score greater than 0.4 (medium confidence) and greater than 0.7 (high confidence) was applied to ensure the reliability of the interactions. Subsequently, the outcomes were visualized using CytoScape software v.3.10.1[33]. Highly interconnected (hub) genes in the PPI network were determined by employing CytoHubba, another CytoScape plug-in, based on their degree of association with other genes in the PPI network. Maximal clique centrality (MCC) algorithm in CytoHubba was employed to evaluate and select the top hub genes.

Gene semantic similarity analysis

Microalgae oil intake-associated and CH₄-associated DEGs were mapped to their corresponding human orthologs to retrieve their Entrez Gene identifiers (IDs). GO semantic similarity analysis was performed using the GOSemSim R package [34] by employing mgeneSim() function in two separate analyses for each trait-specific gene set utilizing the Wang method and Best-Match Average parameter. Similarity score computation was carried across the three GO categories: Biological Process (BP), Molecular Function (MF), and Cellular Component (CC). For exploratory purposes on CH₄-associated gene set, a less stringent threshold was applied on the normalized gene set from DESeq2 where CH₄-associated genes with P_adjusted < 0.05 and |Log₂ fold change| > 0 were extracted and gene semantic similarity analyses were carried out. Results were visualized using pheatmap R package [35]. Furthermore, to observe any semantic similarities between microalgae oil-associated gene cluster and CH₄-associated gene cluster, clusterSim() function was utilised.

Statistical analysis

Statistical analysis of the data was performed using R software v4.4.2 [28]. To assess normality, the Shapiro-Wilk test was employed. Equality of variances among the groups was evaluated using Bartlett’s test, followed by a One-Way Analysis of Variance (ANOVA) to assess whether different inclusion levels of microalgae oil influenced body weight, DMI or CH₄ emissions. A cut-off of P < 0.05 was used to determine statistically significant effects. Results were presented as means ± standard error of mean (SEM), providing an indication of the variability within the data.

Results

Phenotypic responses to microalgae oil inclusion

Table 1 presents a summary of the mean growth performance, DMI, and CH₄ production in twenty lambs included in this study fed on increasing levels of microalgae oil over a 9-week period. The final body weight of lambs receiving no microalgae oil was slightly higher than that of lambs supplemented with 5.4–16.2 g/kg DM of microalgae oil to replace the same levels of soya oil; however, this difference was not statistically significant (Initial bodyweight: P = 0.66, F = 0.54; Final bodyweight: P = 0.81, F = 0.31). Additionally, supplementation of microalgae oil to replace soya oil did not lead to a significant effect on DMI (P = 0.62, F = 0.60). Although mean CH₄ production across the four dietary groups varied among studied lambs, ranging from 40.87 ± 2.09 g/day (control) to 34.11 ± 1.59 g/day (high) (mean ± SEM) (Table 1), these variations were also not statistically significant (P = 0.18, F = 1.86). However, a significant negative correlation between CH₄ production and actual intake levels of microalgae oil was observed (R² = -0.45, P = 0.04), while no significant correlation between CH₄ production and DMI (R² = 0.33, P = 0.15), or final body weight (R² = 0.44, P = 0.05).

Table 1.

Phenotypic effects of microalgae oil inclusion. Mean growth performance, DMI and CH₄ production in twenty lambs supplemented various levels of microalgae oil to replace the same amount of soya oil are tabulated. The data is presented as mean ± SEM. No significant differences were observed in the phenotypes studied with increasing levels of microalgae oil

Phenotype	0 g/kg DM (control) (n = 6) (mean ± SEM)	5.4 g/kg DM (low) (n = 4) (mean ± SEM)	10.8 g/kg DM (medium) (n = 5) (mean ± SEM)	16.2 g/kg DM (high) (n = 5) (mean ± SEM)	P
Initial body weight (kg)	35.91 ± 2.09	34.37 ± 2.11	35 ± 1.36	33 ± 0.82	0.66
Final body weight (kg)	50.41 ± 3.22	47.62 ± 3.33	48.20 ± 2.00	47.10 ± 1.91	0.81
DMI (kg/day)	1.30 ± 0.05	1.21 ± 0.07	1.29 ± 0.06	1.23 ± 0.03	0.62
ADG (g/day)	187 ± 25.58	199 ± 12.69	159 ± 13.55	180 ± 16.32	0.57
CH4 production (g/day)	40.87 ± 2.09	37.38 ± 2.15	36.24 ± 2.66	34.11 ± 1.59	0.18

RNA-sequencing

On average, 83.4% (24,668,185 reads) of the total reads aligned uniquely with the sheep reference genome obtained from the NCBI database (see Additional file 2). Furthermore, 72% (44,533,410 reads) of the aligned reads were successfully assigned to gene symbols corresponding to the reference genome build (see Additional file 2), resulting in the identification of 31,304 genes. Subsequently, 8,875 genes were excluded due to low expression levels (gene expression counts < 10 in over 70% of samples) which left a total of 22,429 genes for further downstream analyses, including differential gene expression (DGE) analyses and ORA.

Increasing levels of microalgae oil influenced the downregulation of genes involved mainly in steroid biosynthesis

DGE analysis revealed a total of 64 DEGs, comprising 12 upregulated and 52 downregulated genes that were statistically significant with increased microalgae oil intake (P_adjusted < 0.05 and |Log₂ fold change| ≥ 2) (Fig. 1, see Additional file 3). The top five upregulated and downregulated genes are listed in Table 2. ORA was conducted on these DEGs, revealing that they were enriched in GO terms in the BP category, with “sterol metabolism”, “steroid biosynthesis”, and “lipid biosynthesis” identified as the top enriched terms (Fig. 2, see Additional file 4). The KEGG pathway enrichment analysis indicated “steroid biosynthesis” as the only enriched pathway, involving five downregulated genes: 24-dehydrocholesterol reductase (DHCR24), 7-dehydrocholesterol reductase (DHCR7), methyl sterol monooxygenase 1 (MSMO1), lanosterol synthase (LSS), and squalene epoxidase (SQLE) (see Additional file 4).

Fig. 1.

DEGs associated with microalgae oil replacement rates for soya oil (g/kg BW) in lambs. The volcano plot represents Log₂ fold change on the x-axis and -Log₁₀ P on the y-axis. Red colour dots represent genes that passed both P_adjusted and |Log₂ fold change| cutoff. Green dots represent genes that only satisfy |Log₂ fold change| cutoff, and blue dots represent genes that only passed P_adjusted cutoff. Grey dots represent non-significant (NS) genes

Table 2.

Top DEGs associated with microalgae oil replacement rates for soya oil (g/kg BW) in lambs. The significant threshold was determined as P_adjusted < 0.05, and absolute |Log₂ fold change| = 2. Microalgae oil intake was treated as a continuous phenotype. Hence Log₂ fold change here corresponds to fold change per unit increase in microalgae oil intake. The P_adjusted was calculated by adjusting P-value from Wald test using BH method

Gene ID	Gene name	Log2 fold change	P adjusted
LOC101114495_1	A-kinase anchor protein 17 A	-7.128	0.0244
SCD	Stearoyl-CoA desaturase	-4.824	1.24E-16
LDLR	Low density lipoprotein receptor	-4.493	8.76E-07
RAMP2	Receptor activity modifying protein 2	-3.766	0.022
INSIG1	Insulin induced gene 1	-3.730	2.43E-19
LOC132657867	Large ribosomal subunit protein uL16-like	2.476	0.031
LOC132659401	Uncharacterized	2.558	0.026
LOC114110105	Uncharacterized	2.618	0.041
LOC132659352	Uncharacterized	3.588	0.003
CAPN13	Calpain 13	5.244	0.004

Fig. 2.

Top enriched pathways of DEGs of microalgae oil replacement rates for soya oil (g/kg BW). The x-axis indicates number of genes (in ratios) associated with each pathway, and the y-axis represents -log₁₀ (P). Each circle is color-coded according to -log₁₀ (P), with red indicating the highest significance

PPI network of microalgae oil-intake associated DEGs

In lambs fed varying replacement levels of microalgae oil for soya oil, the PPI network identified a significantly enriched interaction landscape (PPI enrichment P: 1.0e-16), showing a strong connectivity (interaction score greater than 0.4) among the DEGs (Fig. 3a). A stringent interaction score cutoff of greater than 0.7 was also applied as a sensitivity analysis and is presented in Fig S1. Key downregulated genes, including 3-hydroxy-3-methylglutaryl-CoA synthase (HMGCS1), SQLE, MSMO1, insulin-induced gene 1 (INSIG1), LSS, DHCR7, and DHCR24, were central to the network (also in both medium and high-confidence interaction score cutoffs), and among the top hub genes (MCC score > 200), as identified using the CytoHubba plug-in in Cytoscape software (Fig. 3b). These genes were found to primarily interact with other genes involved in steroid biosynthesis.

Fig. 3.

PPI Network of DEGs associated with increasing levels of microalgae oil (g/kg BW). a The network constructed from STRING database consisting of 44 nodes and 46 edges, highlighting significant interactions (PPI enrichment P: 1.0e-16). b Top connected genes in the network as identified by the CytoHubba plugin based on the MCC algorithm. The colour codes represent the scale of importance with red being most important gene and yellow being the least important gene

Host gene expression signatures and pathways associated with increased CH₄ production

From the DGE analysis with CH₄ production (g/day) treated as a continuous phenotype, no genes passed the strict |Log2 fold change| cut off of 2. Therefore, to examine subtle transcriptional responses associated with CH₄ production, a less-strict cut off of 0.584, corresponding to a 1.5-fold change [36] was applied resulting in seven DEGs- four upregulated (LOC121819234, NME4, MARCHF3, and PLXNB3) and three downregulated (LOC105603087, LOC132657460, and LOC101116551 (Fig. 4; Table 3). GO enrichment analysis revealed NME4 overrepresentation in multiple GO terms including purine nucleoside triphosphate biosynthetic process (GO:0009145), ribonucleoside triphosphate biosynthetic process (GO:0009201), and pyrimidine-containing compound metabolic process (GO:0072527) (see Additional file 5). Furthermore, upregulation of NME4 gene significantly enriched five KEGG pathways: “pyrimidine metabolism”, “drug metabolism”, “nucleotide metabolism”, “purine metabolism”, and “biosynthesis of cofactors”, while PLXNB3 upregulation enriched the “axon guidance pathway” (see Additional file 5). The remaining five genes showed no significant pathway enrichment due to the uncharacterized or incompletely annotated functions rather than the absence of biological relevance.

Fig. 4.

DEGs associated with lambs’ CH₄ production (g/day). The volcano plot shows Log₂ fold change on the x-axis and -Log₁₀ P on the y-axis. Red colour dots represent genes that passed both P_adjusted and |Log₂ fold change| cutoff. Green dots represent genes that only satisfy Log2 fold change cutoff, and blue dots represent genes that passed P_adjusted cutoff. Grey dots represent NS genes

Table 3.

DEGs associated with CH₄ production identified from DGE analysis. The significant threshold was determined as P_adjusted < 0.05, and absolute |Log₂ fold change| = 0.584 (1.5-fold change) where P_adjusted was calculated by adjusting P from Wald test using BH method. CH₄ production was treated as a continuous phenotype. Hence Log₂ fold change here corresponds to fold change per unit increase in CH₄ production

Gene ID	Gene name	Gene Function	Log2 fold change	P adjusted
LOC132657460	Uncharacterized		-1.072	0.003
NME4	NME/NM23 nucleoside diphosphate kinase 4	cellular differentiation, proliferation, apoptosis [37], and catabolism of short-chain fatty acids (SCFA) [38]	0.957	0.003
MARCHF3	Membrane associated ring-CH-type finger 3	regulates immune responses [39]	1.065	0.009
LOC105603087	Myeloid-associated differentiation marker-like		-1.502	0.033
PLXNB3	Plexin B3	semaphorin signalling pathway, axonal guidance [40], regulating morphology and motility of various cell types [41]	1.253	0.036
LOC121819234	Uncharacterized		0.669	0.038
LOC101116551	GTPase IMAP family member 1-like		-0.679	0.047

Linear regression analysis confirmed five out of seven DEGs with significant CH₄ production-associated effects (P_adjusted < 0.05) (Table 4, Fig S2). Among these genes, NME4 displayed the strongest significant effects (R² of 0.58, β = 2.38). Moderate to strong significant effects were also observed for LOC121819234 and LOC101116551, with a negative gene expression-CH₄ relationship for the latter gene. Comparatively weak effects were observed for LOC132657460 and PLXNB3, however these effects were still statistically significant. MARCHF3 and LOC105603087 gene expression displayed non-significant effects with CH₄ production.

Table 4.

Association between gene expression and CH₄ production for the seven DEGs. The table shows regression coefficients (β), coefficient of determination (R²), and BH-corrected P (P_adjusted) for the linear models which evaluated the association between CH₄ production and normalized expression of the seven CH₄‑associated genes identified from DESeq2

Gene	Slope (β)	Coefficient of determination (R2)	P adjusted
LOC132657460	-2.58	0.38	0.006
NME4	2.38	0.58	0.0006
MARCHF3	7.19	0.14	0.09
LOC105603087	-34.12	0.17	0.08
PLXNB3	1.40	0.29	0.02
LOC121819234	1.05	0.51	0.001
LOC101116551	-2.16	0.47	0.001

Functional clustering of DEGs using GO semantic similarity

Distinct gene semantic similarity patterns were observed for microalgae oil intake-associated DEGs across the three GO categories: CC, MF and BP. The heatmap generated displayed functional relationship between genes with red colour representing GO annotations that are closely related (high semantic similarity), orange or light orange with moderate similarities and blue with low similarities (Fig S3). Several gene pairs with moderate similarities were observed for CC category (Fig S3a) while high similarities between gene-pairs were found for the MF category (Fig S3b). The BP ontology displayed greatest heterogeneity with most of the genes showing lower semantic similarities (Fig S3c). Importantly, seven hub genes identified from PPI network (DHCR7, DHCR24, HMGCS1, INSIG1, LSS, MSMO1, and SQLE) displayed moderate to high semantic similarities, while four hub genes- ACSL3, CEBPB, SREBF1, and SCD exhibited low similarity scores (Fig S3).

Three (NME4, MARCHF3, PLXNB3) out of seven CH₄-associated DEGs were successfully mapped to their corresponding human orthologs. Heatmap showing semantic similarity scores between these three genes using CC, MF and BP ontologies are represented in Fig S4. PLXNB3 displayed low similarity with both NME4 and MARCHF3, while NME4 and MARCHF3 exhibited moderate similarity for the CC category (Fig S4a). Almost similar patterns were observed for MF and BP category with low similarities between all three genes (Fig S4b, c). Incorporating a relaxed threshold of genes (P_adjusted < 0.05 and |Log₂ fold change| > 0) resulted in 33 genes with all three ontologies displaying substantial similarities particularly for MF ontology (Fig S5). Furthermore, we could not find any gene cluster semantic similarities between the microalgae oil intake-associated DEGs and CH₄-associated DEGs.

Discussion

The implications of CH₄ production from ruminants raise serious concerns about both the environmental sustainability and the energy efficiency of food production practices associated with sheep farming. To address this issue, while genetic selection of animals with low CH₄ production would be a cost-efficient, and permanent strategy across generations [42, 43], dietary interventions remain an important alternative approach. In this study, feed supplementation with varying levels of microalgae oil over a nine-week period did not reveal any statistically significant differences in CH₄ production, suggesting that under the tested conditions, group-level inclusion to diet was not sufficient to alter CH₄ production.

No significant changes in bodyweight, total DMI or ADG in response to dietary microalgae oil were observed suggesting no changes in feed intake patterns or animal growth performance. These findings are consistent with previous studies which observed no significant growth performance effects from microalgae oil consumption in swine, lambs, and calves [44–46]. These observations are particularly relevant as dietary interventions using microalgae oil that aim to reduce CH₄ production may not compromise or negatively affect other production traits including intake or overall growth. Therefore, any changes observed in CH₄ production levels in the animals in this study are unlikely to be due to differences in microalgae oil or differences in feed intake.

To uncover the gene expression profiles associated with varying levels of microalgae oil, DGE analysis was conducted using actual microalgae oil intake treated as continuous exposure rather than group-level inclusion. Treating microalgae oil intake as continuous variable improves statistical power and helps in identifying genes with even subtle expression changes that might get overlooked through a group-level inclusion analysis. Dichotomizing actual microalgae oil intake levels into discrete groups may result in significant information loss as a result of collapsing individual intake variations into different groups. However, preserving the complete numerical intake values of microalgae oil (as they represent real biological measurement scale) enables continuous statistical models to utilise even minute differences between intake levels, increasing sensitivity and precision [47].

Six key genes out of total 64 DEGs- HMGCS1, INSIG1, sterol regulatory element binding transcription factor 1 (SREBF1), DHCR24, DHCR7, and LSS- were consistently downregulated across several enriched biological terms (sterol metabolism, steroid biosynthesis, and lipid biosynthesis) and were also among the hub genes in the PPI network pointing to a collective transcriptional response towards varying levels of dietary microalgae oil. Although the exact drivers of this expression pattern cannot be confirmed, downregulation of these six genes are consistent with known regulatory effects of long-chain n-3 polyunsaturated fatty acids (LC-n3 PUFA), particularly eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which are abundant in microalgae oil [48]. Similar findings were reported in a study where male hamsters fed with diets enriched in PUFA displayed approximately 60% in vivo sterol synthesis reduction [49]. These six hub genes were also reported to have critical roles in cholesterol and sterol biosynthetic pathways, with HMGCS1 catalysing the conversion of acetyl-CoA and acetoacetyl-CoA into HMG-CoA, a compound that plays a pivotal role in cholesterol synthesis [50]. INSIG1 is a lipogenic gene activated by SREBF1 [51] that plays a role in regulating cholesterol synthetic genes [52]. DHCR24, DHCR7, and LSS are crucial genes at different stages of the steroid biosynthetic pathway: LSS catalyses the conversion of (S)-2,3-oxidosqualene to lanosterol [53], and DHCR7 and DHCR24 reduce the delta-7 and delta-24 double bonds in sterol intermediates [54, 55]. Therefore, a combined downregulation of these genes’ points to an altered biosynthesis of cholesterol in lambs, likely as a consequence of increased microalgae oil within ingested feed. Such downregulation may suggest either a protective or a compensatory response, with genes involved in cholesterol biosynthesis displaying an initial increase in expression within rumen epithelial cells of dairy cattle during the first week of a high-grain diet, and subsequently downregulated by week 3 [56]. Elevated cellular levels of cholesterol can lead to inflammation, oxidative stress, altered cell growth, and changes in membrane function [56]. Therefore, downregulation of cholesterol biosynthetic genes observed in this study suggest that cholesterol biosynthesis is reduced with increasing microalgae oil consumption. RNA-Seq was carried out on blood samples collected post-feeding, without paired pre-feeding transcriptomic profiles as a baseline reference. Hence, although the enrichment of lipid metabolism, cholesterol biosynthesis and steroid biosynthetic pathways supports a biologically plausible association with varying microalgae oil, this analysis cannot fully distinguish microalgae oil effects from pre-existing inter-individual variations. These genes should therefore be interpreted as post-feeding blood-derived transcriptional signatures associated with microalgae oil intake. Nevertheless, observed downregulation may reflect a compensatory mechanism to prevent the overproduction of sterols, which, at elevated concentrations may otherwise potentially induce inflammation, oxidative stress, cell proliferation, and membrane permeability changes, leading to tissue damage [50].

CH₄ production was observed to vary inter-individually across the study lambs with a negative correlation between CH₄ production and actual intake of microalgae oil observed. We used soya oil to balance the microalgae oil, thus minimizing the effect of oil itself on CH₄ production. Evidence indicates that dietary supplementation with any type of oil can influence rumen fermentation and reduce CH₄ emissions [57, 58]. Therefore, the present design could minimize the effect of oil itself but explore the effect of anti-methanogenesis function of microalgae oil. It is often assumed that CH₄ production varies with feed consumption or growth rate; but lack of significant correlation between CH₄ production and DMI or final body weight of lambs in this study suggests that CH₄ production may be influenced by other factors (biological or genetic factors) beyond total feed intake or growth such as host metabolic factors, digestive efficiency, or rumen fermentation patterns. As such, CH₄ production, when analyzed as continuous phenotype, enabled the identification of host transcriptional responses associated with inter-individual CH₄ differences, with seven significantly associated genes. Notably, none of these genes overlapped with the set of genes (n = 64 DEGs) identified in response to microalgae oil, suggesting different regulatory or functional roles for these two sets of genes, potentially pointing to non-overlapping biological mechanisms driving inherent CH₄ production and dietary responses to microalgae oil. This was further confirmed through the gene cluster semantic similarity analysis where we could not find any semantic similarities between the two gene clusters suggesting no functional overlap. Genes including NME4, LOC121819234, MARCHF3, and PLXNB3 were found to be upregulated in lambs with higher CH₄ production. NME4 encodes mitochondrial nucleoside di-phosphate kinase (NDPK) and its upregulation suggests a possible connection between mitochondrial functioning and energy metabolism which are associated with CH₄ production, as reduced efficiency of energy use in animals may result in greater availability of hydrogen for methanogenesis in the rumen. LOC121819234 is an uncharacterized gene, whereas MARCHF3, normally expressed in immune cells such as B cells, and T cells, reflects host immune responses towards microbial activities in the rumen [39]. PLXNB3 regulates the morphology and motility of several immune cells [41]. Few studies have reported PLXNB3 gene and its associations with growth [59], stature [60], and carcass quality [61], however no studies were found reporting its association with CH₄ production. Whilst the role of PLXNB3 in CH₄ production is unknown, the association identified in the current study may suggest a broad regulatory role of the PLXNB3 gene in connecting host physiology to CH₄ production which is correlated to efficient fermentation. Furthermore, immune system associated roles of MARCHF3 and PLXNB3 suggest that upregulation of these genes may disturb the immune system functions in lambs with inherently increased CH₄ production. Uncharacterized LOC132657460, LOC105603087, and LOC101116551 were the three downregulated DEGs associated with increasing CH₄ production. LOC105603087, annotated as MYADM-like gene is expressed in hematopoietic cells with putative links to cell migration, myeloid differentiation and lamb weight traits [62, 63]. Knockdown of MYADM has been associated with an inflammatory-like phenotype and altered barrier function [64]. Similarly, LOC101116551 belongs to GIMAP1 proteins with roles in regulating lymphocyte survival and maintaining homeostasis. Altered GIMAP1 expressions were associated with inflammatory and autoimmune diseases [65], particularly, its downregulation was observed in cattle with Johne’s disease [66] confirming an immune resilience role. However, it should be noted that any functional interpretation of uncharacterized genes (i.e. LOC132657460, LOC105603087, LOC101116551) should remain cautious due to limited characterization in sheep. Nevertheless, based on their predicted functions, the observed downregulation of these two genes may suggest a weak functioning of myeloid and lymphocytes in lambs that emit more CH₄, but does not establish a causal role in CH₄ production. Therefore, further studies on functionally annotating such uncharacterized genes are critical in sheep research.

Importantly, MARCHF3 and LOC105603087, among the seven CH₄-associated DEGs did not retain significance in the linear regression analyses. This difference possibly reflects the different statistical methods of the two analyses, as DGE analysis find genes with significant changes in expression across samples whereas linear regression analyses investigate the association strength between the continuous CH₄ measurement and gene expression. Despite these differences, consistent identification of five CH₄-associated genes from both approaches provides strong evidence for the involvement of these genes in CH₄‑related physiological processes. While these candidate genes contribute new insight into potential regulatory mechanisms influencing CH₄ production, RT-qPCR validation as well as replication in an independent or larger population were beyond the scope of the present work. These steps will be incorporated into future studies to further support and refine the gene-CH₄ associations identified here.

Blood transcriptomics in this study should be interpreted as a proxy for systemic gene expression responses in the host, rather than as a direct representation of transcriptional activities in the rumen. However, the current study provides host-associated molecular information complementary to approaches focused on rumen sampling methods, although integrative analyses utilizing blood, rumen tissues, microbiome, and metabolomics will be needed to fully resolve tissue-specific mechanisms underlying CH₄ variation.

Conclusions

The present study has reported alterations in the gene expression profiles of lambs supplemented with varying levels of microalgae oil, while also identifying host genes linked to CH₄ production. Different levels of microalgae oil (group-allocation) replacement were shown to not significantly influence CH₄ production or growth performance, but to have a positive impact on the metabolic functioning of lambs by altering blood gene expression leading to downregulation of genes (HMGCS1, SQLE, MSMO1, INSIG1, LSS, DHCR7, DHCR24) associated with steroid biosynthesis and other lipid metabolic pathways. In addition, various DEGs (including NME4, MARCHF3, and PLXNB3) associated with CH₄ production were identified that participate in key biological processes including immune regulation and cellular differentiation. Whilst exact roles in CH₄ production remains to be elucidated, involvement in immune-related activities highlights the need for further investigations to examine how external factors such as infection status or overall health might influence these genes and their connection to CH₄ production. Findings also signify that CH₄ production is governed by inter-individual differences in part and underscores the importance of incorporating host-level mechanisms alongside dietary strategies for CH₄ mitigation programmes, thereby promoting more sustainable and environmentally responsible agricultural practices.

Abbreviations

DEG

Differentially expressed gene

DGE

Differential gene expression

DHCR7

7-dehydrocholesterol reductase

DHCR24

24-dehydrocholesterol reductase

DMI

Dry matter intake

GO

Gene ontology

HMGCS1

3-hydroxy-3-methylglutaryl-CoA synthase

INSIG1

Insulin-induced gene 1

KEGG

Kyoto encyclopaedia of genes and genomes

LSS

Lanosterol synthase

LOC105603087

Myeloid-associated differentiation marker-like

LOC101116551

GTPase IMAP family member 1-like

MARCHF3

Membrane associated ring-CH-type finger 3

MSMO1

Methyl sterol monooxygenase 1

NME4

NME/NM23 nucleoside diphosphate kinase 4

ORA

Overrepresentation analysis

PLXNB3

Plexin B3

SQLE

Squalene epoxidase

Authors’ contributions

MM and MS secured UKRI funding for the project (supervision). SM and SH secured Horizon Europe funding. OC and AA participated in protocol development for dietary treatments, experimental design and data collection. SK analysed, interpreted the dataset and prepared the first draft of the manuscript. MM, AA, OC, TY, FR, RH, SM, SH and MS were involved in editing and reviewing of manuscript. All authors read and approved the final manuscript.

Funding

The first author acknowledges funding support from the UK Research and Innovation (UKRI) doctoral training (Grant no: BB/T008776/1) UKRI Biotechnology and Biological Sciences Research Council (BBSRC) Doctoral Training Partnership (DTP) FoodBioSystems under grant no BB/T008776/1, and CASE studentship funding from Agri-food Bioscience Institute (AFBI) through Northern Ireland Farm Animal Biobank (NIFAB) (project no: 21/5/01 funded by DAERA). Authors also acknowledge H2020 that financed the project.

Data availability

The datasets generated and analysed during the current study are deposited in the NCBI sequence read archive (SRA) repository, BioProject ID: PRJNA1451055, which can be accessed at -https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Feu-west-1.protection.sophos.com%2F%3Fd%3Dnih.gov%26u%3DaHR0cDovL3d3dy5uY2JpLm5sbS5uaWguZ292L2Jpb3Byb2plY3QvMTQ1MTA1NQ%3D%3D%26i%3DNjU1NTMyNDY4MjQyNDMyZTkzMTQwMTIw%26t%3DR2RTY09UNzZOVjgzZkpoazNaQlBMRVVPWjMxZ2lTTjQ3aS9kNGh2OTBCcz0%3D%26h%3D55b6147562614c178ba4051f841b8b41%26s%3DAVNPUEhUT0NFTkNSWVBUSVaGtG3OaKxlTcq0fhGjKKcMkmp_laJ8K5UdFPgZa4vZPQ&data=05%7C02%7Cschackokaitholil01%40qub.ac.uk%7Ca6d62488b2ef4efb57d008de962b3e51%7Ceaab77eab4a549e3a1e8d6dd23a1f286%7C0%7C0%7C639113309646423665%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=qMbH0rAt20%2BuuIJvx2wVfhjvAf3%2B92pOC8kjor5hcPI%3D&reserved=0.

Declarations

Ethics approval and consent to participate

The animal study work was conducted at the Agri-Food and Biosciences Institute (AFBI) Hillsborough, under the regulations of the Department of Health for Northern Ireland, according to the UK Animals (Scientific Procedures) Act 1986 requirements and AFBI (Hillsborough) Animal Welfare and Ethical Review Body approval.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.