Metatranscriptome Analysis of Sheep Rumen Reveals Methane Production Changes Induced by Moringa oleifera as a Dietary Supplement

Abstract

Global warming has become a significant public health concern, with intensive livestock farming as a major contributor. To mitigate greenhouse gas emissions, strategies such as manipulating the ruminal environment with dietary additives are essential. This study evaluated Moringa oleifera, a globally widespread tree with antioxidant, multivitamin, protein-rich, and anti-inflammatory properties, as a feed additive. Rumen fluid was collected from three Pelibuey sheep, homogenized, and subjected to an in vitro fermentation study for 48 h with three alfalfa/moringa ratio treatments: T0 Control (100:0), T1 Low (85:15), and T2 High (70:30). Total RNA was extracted, followed by high-definition sequencing of the metatranscriptome. The sequencing yielded approximately 456 million sequences. A total of 117 phyla were identified and approximately 1300 genera were mapped. Predominant phylum differed by treatment: T0, Firmicutes; T1, Proteobacteria; and T2 with Synergistetes, at least one sample per treatment. Archaea were nearly absent in T1, which explains a statistically significant decrease in methane production. In the Gene Set Enrichment Analysis (GSEA), it was observed that one of the metabolic pathways with a statistically significant difference (p-value < 0.05) was that of methane, specifically in the low moringa treatment (T1) compared to the control (T0). From the functional analysis, differentially expressed enzymes were identified, some of which are involved in the methane metabolic pathway, such as formate dehydrogenase (EC 1.17.1.9) and glycine hydroxymethyltransferase (EC 2.1.2.1), which are intermediates in methane formation. These results suggest that 15% Moringa oleifera supplementation alters ruminal microbiota, reduces archaeal activity, and suppresses methane-related pathways. These findings provide molecular evidence supporting the potential of M. oleifera as a methane mitigation strategy in ruminant nutrition.

1. Introduction

Rumen fermentation significantly contributes to global warming through the emission of greenhouse gases. It ranks second in methane production, with anthropogenic sources occupying the first place. Reducing methane emissions from the livestock sector has become a necessity due to the growing demand for livestock products and the associated ecological impact [1]. The increasing demand for meat and milk by the population is steadily rising, further increasing the carbon footprint annually [2]. To mitigate this issue, alternatives for livestock feeding are being explored to reduce greenhouse gas emissions. Previous studies have shown that the use of Moringa oleifera [3] as a supplement in livestock diets reduces methane and carbon dioxide emissions. Moringa contains secondary metabolites that directly act on the rumen microbiome, affecting the metabolic pathways inherent to ruminal fermentation [4]. Moringa has been characterized to identify metabolites that may impact metabolic pathways [5], as they possess antioxidant properties, leading to the hypothesis that they may influence the free radicals involved in methanogenesis [6], potentially using them as ionic intermediaries to prevent methane formation. Previous studies have demonstrated moringa’s nutritional [5], antioxidant, and anti-inflammatory [7] potential, and in vitro assays have shown that moringa reduces both carbon dioxide and methane production [8].

Amad and Zentek (2023) [9] describe the moringa like is a fast-growing, drought-resistant tree that yields a high amount of biomass rapidly, making it a cost-effective and accessible feed resource, in addition, it contains significant quantities of bioactive compounds such as alkaloids, flavonoids, phenolics, glucosinolates, carotenoids, sterols, saponins, phenolic acids, tannins, and isothiocyanates. These compounds enhance nutrient digestibility and possess anthelmintic activity, and one of the most important features is that the leaves or seeds, when used as a food supplement, reduce ruminal methane emissions due to their secondary metabolites and interaction with the ruminal microbiota, influencing methane metabolic pathways and fermentation patterns. It is also considered a carbon dioxide sink tree [9].

An experiment called Rusitec (Rumen Simulation Technique) system evaluated the effects of moringa on ruminants, using different treatment groups with increasing concentrations of Moringa oleifera seed: 0% (control), 1%, 2%, and 3% of the dry matter intake [10]. The experiment ran for 14 days, including an initial adaptation period. Researchers collected samples to measure gas production, methane emissions, volatile fatty acid (VFA) profiles, and digestibility. The inclusion of Moringa oleifera seed linearly decreased methane production as the concentration of seeds increased in the simulated diet. This suggests a dose-dependent anti-methanogenic effect. The researchers attributed the methane reduction primarily to the high content of unsaturated fatty acids present in Moringa oleifera seeds. These fatty acids can act as alternative hydrogen sinks in the rumen, diverting hydrogen away from methanogenic archaea and, thus, reducing methane formation through the process of biohydrogenation [10].

Metatranscriptomics is one of the most widely used tools for characterizing complex niches, such as the rumen [11]. In addition to taxonomic classification, it allows for the generation of a complete profile of genes expressed under specific environmental conditions in real time [12]. The proposed tool provides a comprehensive analysis of how environmental changes can modify an entire ecosystem. Several studies have tested the effect of moringa on methane mitigation [8,13]. Gómez-Chávez et al. (2025) [14] published results of in vitro fermentation by adding moringa to a base diet of alfalfa in different doses to observe the effects of moringa. The study found that 15% moringa with 85% alfalfa reduced methane production (

Table 1. In vitro fermentation experiment results. CA: Control treatment (100% alfalfa) T0, MB: Low moringa treatment (15% moringa, 85% alfalfa) T1, MA: High moringa treatment (30% moringa, 70% alfalfa) T2 [14].

Variable	CA	MB	MA	p-Value
pH	6.74 a	6.86 a	6.76 a	≥0.05
DM Digestibility (%)	63.88 a	66.65 ab	67.87 b	<0.05
Neutral detergent fiber (%)	24.64 a	30.96 ab	32.09 b	<0.01
Acid detergent fiber (%)	17.80 a	22.53 ab	23.54 b	<0.01
Gas production (mL/g)	49.06 a	51.18 ab	55.93 b	<0.05
Acetic acid (mmol/L)	63.9 a	42.3 b	31.2 c	<0.05
Propionic acid (mmol/L)	30.6 a	21.0 b	11.1 c	<0.05
Butyric acid (mmol/L)	9.3 a	4.6 b	4.2 b	<0.05
TAGV (mmol/L)	103.9 a	67.9 b	46.5 c	<0.05
CO2 (% molar)	51.7 a	48.5 b	51.9 a	<0.05
CH4 (% molar)	28.6 a	26.0 b	32.3 c	<0.05
NH3 (mmol/mL)	14.6 a	14.9 a	15.1 a	>0.05

^a/b/c Values with the same letter in the same row represent means without statistically significant differences.

) [14]. To measure the molecular effects of moringa, metatranscriptomics was proposed to elucidate which metabolic pathways were active under different treatments.

This study aimed to investigate, using metatranscriptomic analysis, the effects of Moringa oleifera and its metabolites on ruminal metabolism and microbial populations, with a particular focus on methane metabolism, when added to an alfalfa-based diet in an in vitro fermentation assay.

2. Materials and Methods

2.1. Bromatological Characterization of Moringa oleifera

The Moringa oleifera supplement was prepared from leaves harvested from a plantation located in the central region of the state of Chihuahua, Mexico. The leaves were shade-dried and ground into a fine powder.

A comprehensive chemical characterization of the powdered M. oleifera leaves was previously reported by Gómez-Chávez et al. (2025) [14] and is summarized here to provide key nutritional information relevant to this study. Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were determined using the method of Van Soest et al. (1991) [15], adapted for the ANKOM 200 Fiber Analyzer (ANKOM Technology, Fairport, NY, USA). Crude protein was analyzed using Kjeldahl equipment (Firgusa, Cuautitlán-Izcalli, Mexico; AOAC 955.04). Dry matter content was measured using a digital mechanical convection oven (Felisa, Zapopan, Mexico; AOAC 930.15), ash using a digital muffle furnace (Felisa, Zapopan, Mexico; AOAC 942.05), and ether extract using a Goldfish extraction equipment (Novatech, Tlaquepaque, Mexico; AOAC 920.39). All parameters were determined following the official methods of the Association of Official Analytical Chemists (AOAC, 1997) [16]. Mineral content was analyzed using atomic absorption spectrophotometry (AAS) (Buck Scientific, Ansonia, CT, USA), and the amino acid profile was determined by gas chromatography coupled with mass spectrometry (GC-MS) (Agilent, Santa Clara, CA, USA).

The nutritional composition of moringa powder, on a dry matter basis, was: dry matter 93.87%, ash 15.68%, ether extract 5.99%, NDF 26.42%, ADF 14.85%, and crude protein 16.97%. Mineral concentrations were calcium 3.23%, phosphorus 0.27%, magnesium 0.80%, and potassium 1.35%. The nutritional composition of alfalfa was: dry matter 89.4%, ash 11%, ether extract 2.08%, NDF 36.1%, ADF 29%, and crude protein 21.2%. Mineral concentrations were calcium 1.4%, phosphorus 0.26%, magnesium 0.32%, and potassium 3.03% (Gómez-Chávez et al., 2025) [14].

2.2. Rumen Fluid Collection and In Vitro Fermentation

The experimental procedures were approved by the Bioethics and Animal Welfare Committee of the Faculty of Zootechnics and Ecology (CBBA/FZyE). The study was conducted in accordance with NOM-062-ZOO-1999, which establishes technical specifications for the production, care, and use of laboratory animals. Additionally, the study adhered to the Internal Regulation of Bioethics and Animal Welfare of the Faculty of Animal Science and Ecology at the Autonomous University of Chihuahua (2018). The study design and analysis comply with the ARRIVE guidelines for animal research. The in vitro fermentation procedure was previously described by Gómez-Chávez et al. (2025) [14]. Briefly, rumen fluid was directly collected from three adult male fistulated Pelibuey sheep weighing approximately 42 kg (±3 kg), that had been pre-adapted with vaccination, vitamin supplementation, and a high-protein diet prior to surgery. After a 10-day recovery period, the animals underwent a 15-day adaptation phase, during which they were fed alfalfa hay twice daily under a controlled feeding schedule and had unrestricted access to water. The alfalfa hay used for adaptation contained approximately 21.2% crude protein, 36.1% NDF, and 29% ADF on a dry matter basis, as analyzed according to AOAC and Van Soest methods [15,16]. Feed was withheld for 24 h prior to rumen fluid collection to standardize fermentation conditions and minimize postprandial variation in microbial activity. Approximately 500 mL of rumen fluid was collected from each animal, filtered through four layers of cheesecloth, pooled, and maintained under anaerobic conditions with CO₂ flushing.

Based on the bromatological composition of Moringa oleifera, three treatments were prepared and tested with two biological replicates each: T0 (100% alfalfa), T1 (15% Moringa oleifera and 85% alfalfa), and T2 (30% Moringa oleifera and 70% alfalfa). These ratios were selected taking into account the antinutritional compounds present in leaves [9]. The substrates were incubated with rumen fluid in ANKOM FN57 filter bags (ANKOM Technology, Fairport, NY, USA) [14]. For gas production measurements, 2 g of substrate were weighed directly into 100 mL glass tubes sealed with butyl stoppers and agraffes, and incubated anaerobically. For the rest of the parameters, 0.5 g of substrate was used in the ANKOM filter bags placed inside 250 mL bottles. The inoculum consisted of 15 mL for gas production tubes and 40 mL for other incubation bottles, all handled under CO₂ flushing to maintain anaerobiosis. Incubations were conducted in a controlled temperature incubator at 39 °C with continuous stirring at 120 rpm. Samples were collected at 6, 12, 24, and 48 h to evaluate pH, digestibility (%), neutral detergent fiber (NDF, %), acid detergent fiber (ADF, %), gas production, acetic acid (mmol/L), propionic acid (mmol/L), butyric acid (mmol/L), total volatile fatty acids (TVFA, mmol/L), carbon dioxide (CO₂, % molar), methane (CH₄, % molar) and ammonia (NH₃, mmol/mL). These in vitro fermentation parameters were previously measured and published in detail by Gómez-Chávez et al. [14].

From the 48 h samples, approximately 10 mL were preserved in RNAlater™ (Thermo Fisher Scientific, Waltham, MA, USA) [17] and stored at −80 °C for subsequent RNA extraction. The metatranscriptomic analysis presented in this study is based on those 48 h samples, which reflect peak microbial activity under in vitro conditions.

In the referenced study, M. oleifera supplementation significantly reduced methane production, particularly at the T1 (15% inclusion level). These findings provided the biological foundation for conducting the current metatranscriptomic analysis, aimed at elucidating the microbial and enzymatic mechanisms underlying this methane mitigation effect.

2.3. RNA Extraction

The vials containing the stored samples were subjected to an RNA extraction process using a commercial PureLink RNA Mini Kit (Waltham, MA, USA). A protocol provided by the supplier was followed, combining the kit with TRIzol [18], a method commonly used for obtaining high-quality genetic material from difficult samples, which includes organic lysis and selective filtering to preserve genetic material. The quantity and quality of RNA were determined using the Qubit 4.0 fluorometer (Thermo Fisher Scientific, Catalogue number Q33238) [19]. Samples with an RNA Integrity Number (RIN) between 4 and 8 were considered suitable for analysis.

2.4. Sequencing, Data Preprocessing, and Assembly

Six samples were selected, two from each treatment group (control, low, and high), and sequenced at the LSCiences laboratory (Houston, TX, USA). Five μg of total RNA were removed to eliminate ribosomal RNA, following the instructions of the Ribo-Zero™ rRNA Removal Kit [20]. The remaining RNA was fragmented using bivalent cations at high temperature. Retrotranscription was then performed to generate complementary DNA (cDNA), and an “A” base was added to the blunt ends to facilitate adapter ligation with indexing. The ligated fragments were amplified by PCR following the manufacturer’s instructions. The cDNA fragment size was 300 bp (±50 bp). Finally, sequencing was performed using the Illumina NovaSeq 6000 Sequencing System (LS science, Houston, TX, USA) [21] with a read length of 150 bp.

Raw data were processed using FastQC [22] and MultiQC [23]. To discard low-quality sequences and trim sequences by their ends and ambiguous bases, FastP [24] was used, which integrates the features of FastQC [22], Cutadapt [25], Trimmomatic [26], and AfterQC [27] in a single program within the Python (v 3.11.0) environment. Sequences were removed if they had a score < Q20 at the 5′ or 3′ ends, and all sequences shorter than 50 bp were discarded. The statistical power of this experimental design, calculated in RNASeqPower, is 0.95 in the Rstudio package [28].

Sequence filtering was performed in the following order: host, diet, and ribosomal RNA, using the Bowtie2 program [29]. Reference genomes were obtained from the NCBI database [30] for sheep (Ovis aries GCF_016772045.1), alfalfa (three representative genomes of the genus Medicago: M. arabica GCA_946800305.1, M. ruthenica GCA_018208015.1, M. truncatula GCF_003473485.1), and the Moringa oleifera supplement (Moringa oleifera GCA_021397835.1). A second review was performed using the ribosomal RNA index [31] and a control for vector formanPhi (DecontPhixHuman38.p13. Human: GCF_000001405.40, Phix: GCF_000819615.1) to prevent contamination. Multiple alignments were performed, discarding sequences by genome, and the resulting sequences were reviewed with MultiQC [23]. These data are available in the NCBI Bioproject #ID: PRJNA1104775 (SRA accession numbers: SRX24393445, SRX24393446, SRX24393447, SRX24393448, SRX24393449, SRX24393450). Around 250,000,000 sequences of 149 bp were obtained from the alignment, with 88% to 97% of the sequences being successfully identified.

2.5. Taxonomic Characterization, Functional Analysis, Differential Analysis, and Enrichment Analysis

The consensus sequences obtained were processed using the SqueezeMeta program [32] in coassembly mode, which allows for greater coverage, reads from all the samples are pooled and make a single assembly. The databases used for functional and taxonomic analysis were NCBI [30], KEGG [33], and PMAF [34]. Transcript identity was determined using the rnaSPADES command [35].

For result visualization, an object was generated, and the Rstudio program (R version 4.2.2 (31 October 2022 ucrt)) [36] was used with the SQMtools package (v3.1.0) [37] for protein prediction, identification of metabolic pathways, and taxon classification. DeSeq2 [38] was employed to estimate the abundance of active genes based on transcripts per million and their differential expression with a Padj < 0.05. Additionally, MicrobiomeProfiler [39] and FGSEA [40] were used to assess enriched metabolic pathways with a Padj < 0.05.

3. Results

3.1. RNA Sequencing, Mapping, and Taxonomic Characterization

A total of 456,247,410 unique sequences were obtained from six samples—two control, two with low treatment, and two with high treatment. Over 90% of the sequences were successfully aligned and classified, except for the low control sample (T1_1), where only 83% of the sequences were mapped.

Table 2. Total Reads per Sample, Mapping, and Identity Percentage.

Sample	Total Reads	Mapped Reads	Mapping Percentage	Total Bases per Sample
T0_1	73,182,222	71,214,334	97.31	10,785,232,234
T0_2	76,371,052	72,968,710	95.54	11,281,335,036
T1_1	61,435,340	51,225,045	83.38	8,963,537,084
T1_2	76,655,916	73,071,611	95.32	11,286,729,809
T2_1	86,043,460	83,726,582	97.31	12,555,046,099
T2_2	82,559,420	78,956,090	95.64	12,176,189,504
Overall values	456,247,410	431,162,372	Media% = 94.083	67,048,069,766

shows the identity statistics of the samples relative to the total sequences per sample. The first repetition was defined as “_1” and the second repetition of each treatment as “_2” in a standardized manner. For the control (100% alfalfa), the first repetition was defined as T0_1 and the second as T0_2. In the case of the low treatment (85% alfalfa: 15% moringa), the first repetition was named T1_1 and the second T1_2. Finally, for the high treatment (70% alfalfa: 30% moringa), its repetitions were defined as T2_1 for the first and T2_2 for the second.

These reads allowed for the determination of putative taxa for each sample, using the NCBI genomic database [30] as a reference. A total of 117 phyla, 314 classes, 494 orders, 811 families, 1372 genera, and 1910 species were identified across the six samples. The most abundant kingdom was the prokaryote, which is the major contributor to nutrient degradation and the ruminal environment. From the mapping results, it was observed that the kingdom Bacteria had the lowest proportion in the T2 (high) treatment. The proportion of Viruses in the low treatment (T1) was four times greater than in the other treatments. Additionally, the presence of Archaea in the T1 low treatment was nearly absent, correlating with the reduction in methane production, as Archaea are the main producers of methane. The least abundant kingdom in the T0 control treatment was Archaea, which directly correlates with the high abundance of prokaryotes in this treatment. Overall, in the T2 high treatment, most microbial kingdoms were relatively reduced, except for Archaea.

Table A1. Relative abundances and percentage of reads per sample. Total sequences mapped and characterized at kingdom level.

	T0_1		T0_2		T1_1		T1_2		T2_1		T2_2
Kingdom	Abund	%	Abund	%	Abund	%	Abund	%	Abund	%	Abund	%
Archaea	94,929	0.13	213,475	0.28	79,319	0.13	43,708	0.06	12,361	0.01	2,642,305	3.20
Bacteria	52,788,118	72.13	47,159,577	61.75	31,529,695	51.32	61,086,512	79.69	60,309,727	70.09	36,331,623	44.01
Eukaryota	757,736	1.04	914,792	1.20	1,008,417	1.64	948,293	1.24	563,586	0.66	471,579	0.57
No CDS	2,286,794	3.12	4,869,970	6.38	141,358	0.23	995,763	1.30	5,092,198	5.92	13,700,976	16.60
Unclassified	14,873,143	20.32	19,238,680	25.19	14,144,528	23.02	8,238,457	10.75	17,582,487	20.43	24,136,574	29.24
Unmapped	1,967,888	2.69	3,402,342	4.46	10,210,295	16.62	3,584,305	4.68	2,316,878	2.69	3,603,330	4.36
Virus	413,614	0.57	572,216	0.75	4,321,728	7.03	1,758,878	2.29	166,223	0.19	1,673,033	2.03

shows the read abundance for each of the categorized kingdoms per treatment.

Figure 1 shows the most abundant phyla found in each sample. In the T0_1 sample, the most abundant phylum is Firmicutes, within which various lineages such as Staphylococcus, Bacillus, Streptococcus, Ruminococcus, and Lactobacillus can be found. Firmicutes was the predominant phylum in T0, where 100% alfalfa was administered, resulting in an increase in the population of this phylum. Figure 2A shows the proportions of the most abundant phyla in each treatment, by sample. The dominant proportion of Firmicutes in the T0 control treatment is evident. The predominant genus of this phylum in T0 was Streptococcus, a bacterium known for rapidly fermenting starch. Figure 2B shows the proportion of this genus compared to other members of the phylum. In the T1 low treatment, the most abundant phylum, as shown in Figure 1, is Proteobacteria, which is further confirmed in Figure 2C, a comparison of the most prominent phyla across the three treatments. The most abundant genera within the Proteobacteria phylum in T1 were Proteus and Pseudomonas.

In T2 high (Figure 1), the most abundant phylum is Synergistetes. Figure 2D shows the most abundant genus of this phylum, which is Synergistetes bacterium, a relatively less abundant phylum in symbiosis with animals.

Figure A1, Figure A2 and Figure A3 show the taxonomic classification for T0 control, T1 low, and T2 high, respectively, with their two repetitions.

To explore another taxonomic reference point, it was decided to analyze the relative abundances at the family level. In the control sample T0_1, the predominant family is Streptococcecae, accounting for approximately 50%, followed by the Synergistaceae family, which is also present in the second repetition of this treatment and is the most abundant family in treatment T2, as shown in Figure A4. On the other hand, the most abundant family in T1 treatment is Morganellaceae, followed in smaller proportions by the families Nocardiaceae, Pseudomonadaceae, and Enterobacteriaceae, all of which belong to the phylum Proteobacteria.

In the Shannon index analysis, it can be observed that, in general, the samples show low diversity. In treatment T0, the mean was 2.488. The mean index presented in T1 was 3.0275, and in T2 it was 2.6845. The samples with the highest alpha diversity were those from the low treatment (T1), as shown in

Table A2. Shannon index plot per sample.

Sample	Total Reads
T0_1	2.183
T0_2	2.793
T1_1	3.094
T1_2	2.961
T2_1	2.513
T2_2	2.856

and Figure A5.

Differential expression in terms of taxa at the phylum level can be seen in Figure A6 (plotMA), which shows the changes in expression with respect to the mean between the control, called T0, and the treatment T1 [38].

3.2. Functional Analysis

In the SqueezeMeta program [32], with the rnaSPADES assembly parameter [35], a total of 381,192 reads were obtained, which were used as consensus sequences for protein prediction. These counts allowed the identification of 13,313 proteins in the KEGG database [41], 21,194 putative proteins in the eggNOG database [42], and 9600 predicted proteins in the PFAM database [34].

Figure 3 shows the main mapped transcripts and their relative abundances in transcripts per million. Notably, the thermal unstable elongation factor (K02358) was more abundant in the T0 (control) treatment. In the T1 treatment, an mRNA encoding a cold shock protein was found to be greatly increased, while in the other two treatments, the same effect was not observed.

To assess the effect of each treatment more specifically, transcript abundance analysis was performed with respect to the predominant taxon in each treatment. Figure 4A shows the most significant mRNAs from the Firmicutes phylum, where the thermal unstable elongation factor (K02358) remains prominent, suggesting that the genera within this phylum were very active in protein translation. Enzymes such as K01686 enolase (EC 4.2.1.11) and K00134 glyceraldehyde-3-phosphate dehydrogenase (EC 1.2.1.11), both involved in the glycolysis pathway [41], were also observed to be upregulated.

Figure 4B shows the most significant mRNAs from the Proteobacteria phylum, which predominates in the T1 low treatment. Also observed is an abundant mRNA, though not as prominent, for the K04047 DNA-binding protein. In Figure 4C, the most abundant transcripts from the Synergistetes genus are shown, where a general similarity is seen across all three treatments, as confirmed in Figure 2D, where the taxon abundances are also similar. However, it is notable that in the T1 Low treatment, the mRNA for K01952 phosphoribosyl-formylglycinamide synthetase (EC 6.3.5.3) [33], which is involved in purine biosynthesis, is overrepresented.

3.3. Enrichment Analysis

To gain a more general understanding of the transcripts, a sequence overrepresentation analysis was performed. Figure 5 shows the metabolic pathways that were differentially represented across the comparisons of treatments. Overall, the most highly expressed metabolic pathways were two-component systems, carbon metabolism, and cofactor biosynthesis. Cells need to keep these metabolic pathways more active, as they are conditioned by the environment to which they are subjected. In this case, treatment determines metabolism; therefore, the number of transcripts varies depending on the treatment, which is why the specific genes of these metabolic pathways are overexpressed. Methane metabolism was significantly overrepresented in T2 compared to T0 and in T1 compared to T2, with a p-value ≤ 0.00005.

To further strengthen these findings, a Gene Set Enrichment Analysis (GSEA) [43] was performed. This method classifies all the genes in the dataset based on their differential expression and checks whether the genes in a particular set are enriched, taking into account the distribution of gene expression changes in the dataset. In this case, pairwise comparisons were made: T0 vs. T1 (Figure 6A,B), T0 vs. T2 (Figure 7A), and T1 vs. T2 (Figure 7B). The results show the significantly enriched metabolic pathways with a p-value ≤ 0.00041 for each gene set, generated from the evaluated sequences.

In Figure 6A, the methane metabolism pathway is prominently featured, suggesting that the set of genes involved in this metabolism is significantly represented in the T1 low treatment over T0 (control) [40].

3.4. Differential Expression Analysis

A total of 13,313 transcripts were identified in the KEGG database [33], with 2185 being differentially expressed between T0 and T1, 1848 between T0 and T2, and 2929 between T1 and T2. Figure 8 shows the divergence in transcript expression across the different treatment comparisons. Panel a displays the proteins differentially expressed between T0 and T1, panel b shows the enzymes differentially expressed between T0 and T2, and panel c highlights the proteins differentially expressed between T1 and T2. To specifically examine methane metabolism and how moringa supplementation might affect it, a differential analysis was performed on enzymes directly involved in methane metabolism.

Figure 9 illustrates the enzymes that were differentially overexpressed or underexpressed in T1 compared to T0. In the case of methane metabolism, enzymes such as formate dehydrogenase (EC 1.17.1.9) and glycine hydroxymethyltransferase (EC 2.1.2.1) were found to be overexpressed in the control treatment (T0) compared to the low moringa treatment (T1). Conversely, methylamine dehydrogenase (EC 1.4.9.1) and methylglutamate dehydrogenase (EC 1.5.99.5) were underexpressed in the T1 moringa treatment compared to T0.

4. Discussion

In general, it was observed that the most abundant kingdom across all treatments was the prokaryotic kingdom, which is known to play a major role in nutrient degradation and the ruminal environment. Mapping results indicated that the Bacteria kingdom was less abundant in the T2 (high moringa) treatment. Additionally, the proportion of viruses in the T1 (low moringa) treatment was four times higher than in the other treatments. While viruses remain the least studied group of rumen microorganisms, previous studies have shown that bacteriophages may influence bacterial populations and potentially modulate ruminal fermentation patterns [43,44]. In particular, phages from families such as Myoviridae have been reported to target Archaea or bacteria involved in fiber degradation. Although our study does not provide direct evidence of such interactions, these findings highlight the need for further research into the role of viral communities in shaping the rumen microbiome.

Furthermore, in T1, Archaea populations were nearly absent. Archaea are the primary producers of methane in the rumen, as they use hydrogen ions to regulate ruminal pH and prevent acidosis. The decrease in methane production in the low moringa treatment (T1) as described in the in vitro fermentation study can be associated with the reduction in Archaea populations; the results of the in vitro fermentation with Moringa oleifera are shown in Gómez-Chávez et al. (2025) [14].

On the other hand, the least abundant kingdom in T0 (control) was Archaea, which correlates directly with the high abundance of Prokaryotes in this treatment. In general, in T2 (high moringa), most microbial kingdoms were relatively reduced, except for Archaea. Kaplan-Shabtai et al. (2021) provides strong evidence for direct metabolic links and co-dependencies between specific bacterial and archaeal groups that influence the final products of rumen fermentation [44].

Some animal studies have reported toxicity from moringa leaf components, such as terpenes and flavonoids, which require a concentration lower than 5000 mg/kg to avoid acute toxicity [45]. It has also been described that these phytochemicals can interact with the cell membranes of microorganisms such as protozoa and bacteria, which is why dosage is crucial when using moringa as a feed supplement [46], suggesting that dosage is a critical factor when using moringa as a feed supplement. Although our study did not directly evaluate microbial viability or toxicity, the lower abundance of microbial populations observed in T2 may be associated with these effects. Among the treatments, T1 showed a microbial profile that suggests a more balanced modulation, particularly in the Archaea domain.

As previously mentioned, in the sample T0_1, the most abundant phylum was Firmicutes, which is characterized by its high ability to degrade plant cell walls through the enzyme glycoside hydrolase [4], which acts to hydrolyze celodextrins into cellobiose, thus making it available as an energy source from plant tissues. Within this phylum, various lineages can be found, such as Staphylococcus, Bacillus, Streptococcus, Ruminococcus, and Lactobacillus, among others. The adaptability of this phylum to various ecological niches has enabled it to thrive in hostile environments such as the rumen. It has the capacity to encode enzymes that degrade cellulose, lignin, and other components of the plant cell wall [47]. For this reason, we can mention that Firmicutes was the predominant phylum in the T0, where 100% alfalfa was administered, resulting in an increase in the population of this phylum.

In Figure 2A, the predominant percentage of Firmicutes in the T0 control can be observed. The main family of this phylum in T0 was Streptococcaceae (Figure A4), prokaryotes capable of rapidly fermenting starch. Figure 2B shows the proportion of this genus relative to other members of the phylum. In this experiment, no completely mixed diet was administered, only alfalfa and the supplement being tested, which was moringa. The fact that this genus predominates suggests the components of alfalfa, which is highly digestible and rich in protein and nitrogen [48]; this genus is known for its high aminoendopeptidase activity [49]. However, despite being the most abundant genus, other genera capable of fiber degradation are also present for utilization (Figure A1). In addition one of its main functions, in addition to breaking down carbohydrates as a source of energy, is that it contains enzymes to maintain balance within the ruminal environment, such as the CRISPR-CAS system [50]. In this case, two replicates per treatment were taken into account, of which, despite the conditions, sample T0_1 showed a notable predominance of this family, compared to sample T0_2, which presented more variability in terms of populations. Its Shannon index was 2.793 (Table A2 and Figure A5), which is slightly higher than that of T0_1, so although the relative abundance in terms of quantity is quite similar, the populations in each differ considerably from one repetition to another; however, in both, it can be observed that the Streptococcaceae family is one of the most abundant. In Figure 4A, from the functional analysis, it was observed that in sample T0_1, one of the most abundant mRNAs was Elongation Factor Tu (K02358) [33], which is essential for the protein synthesis of enzymes related to the degradation of fiber, starch, and protein, and is characteristic of the genus Streptococcus, which is the most abundant genus in this sample and present under normal dietary conditions [51].

The most abundant phylum in T1 Low was Proteobacteria (Figure 1), which is characterized by its utilization of proteins in the rumen [52] and has been considered as a secondary fermenter; that is, it utilizes metabolites produced from the fermentation of other microorganisms [53]. This proportion is confirmed by Figure 2C, which contrasts the most abundant phyla across the three treatments. In treatment T1 (Low), sample T1_2, which is the second repetition, was found to predominantly contain the Morganellaceae family (Figure A4), while sample T1_1 showed greater variability in terms of populations of families within the Proteobacteria phylum, such as Pseudonomaceae and Nocardiaceae. Of the treatments performed, it is the most distinct in terms of taxonomic composition between replicates. The most abundant genera in T1 within the Proteobacteria phylum were Proteus and Pseudomonas (Figure A2). Both have been studied in various niches, and within the ruminal environment, they are documented to collaborate in acid formation from simple sugars and secrete proteases that assist in protein breakdown [54]. Previous fermentation studies determined that the protein content in the treatments was approximately 20%, which suggests that this concentration of crude protein could stimulate an increase in these genera [14].

In T2 High, the most abundant phylum was Synergistetes (Figure 1), which is considered to utilize antioxidants for detoxifying reactive oxygen species [55]. Moringa contains several metabolites that can function as antioxidants through free radical scavenging, which could explain why this phylum is most abundant in the treatment with the highest moringa concentration [56]. The most abundant genus within this phylum was Synergistetes bacterium (Figure 2D and Figure 5), which is typically the least abundant phylum in animal symbiosis. However, it increases under certain pathogenic conditions [57] or in specific circumstances such as the detoxification effect previously mentioned.

Regarding the divergence between taxa in Treatment T0 and T1 [38], it can be observed that there are lineages at the phylum level that show significant differences in abundance. One of the phyla with significant differences between the T0 control and the T1 treatment was Proteobacteria, as mentioned before, which are mainly present in T1 (Figure A6). In addition to being the most abundant, we must take into account that they are also active in terms of gene expression, this phylum characterized by the degradation of simple and structural carbohydrates for energy generation

One of the advantages of metatranscriptomics is that it allows the identification of current transcripts in real time; in other words, it can precisely describe how an environment is functioning at a specific moment [12]. In Figure 3, the principal mapped transcripts and their relative abundances in transcripts per million can be observed. Notably, the Thermal unstable elongation factor (K02358) is found in higher proportions in T0, which is the control. This factor functions by binding the aminoacyl of transfer RNA, making it essential for protein translation [58]. In this case, the abundance of this mRNA suggests that the thermal unstable elongation factor is required in large quantities during this phase of fermentation under the conditions of T0. Cold shock proteins are those whose domains interact directly with genetic material to modulate its expression, facilitating cell survival or homeostasis [59].

In T1, an mRNA encoding for a cold shock protein is significantly increased by millions in that treatment, while no such effect is observed in the other two treatments. To further examine the specific effects of each treatment, a transcript abundance analysis was performed with respect to the predominant taxon in each treatment. In Figure 4A, the most prominent mRNAs from the Firmicutes phylum are shown, with the Thermal unstable elongation factor (K02358) continuing to stand out, suggesting that genera within this phylum were highly active in protein translation. As mentioned earlier, this phylum is known for its ability to degrade fiber and plant tissue [60], and it can produce fermentative enzymes involved in the production of ionized hydrogen, hydrogenases, and methyl-CoM reductase, which are linked to hydrogenotrophic methanogenesis. Also observed are increased enzymes such as K01686 enolase (EC 4.2.1.11) and K00134 glyceraldehyde-3-phosphate dehydrogenase (EC 1.2.1.11), both involved in the glycolysis pathway [41].

In Figure 4B, the most abundant mRNAs from the Proteobacteria phylum, which is predominant in T1 Low, are displayed. The most abundant mRNA is the cold shock protein, which serves to preserve genetic material. Additionally, there is an abundant, though less evident, mRNA from K04047, a DNA-binding protein involved in starvation response. This protein is translated when the cell harboring it, especially Gram-negative bacteria, behaves as Proteobacteria under stress conditions [61], to protect itself and endure across generations despite the hostile environment. In Figure 4C, the most abundant transcripts from the Synergistetes genus are shown, and in general, similar patterns are observed across all three treatments, which is further supported by Figure 2D, where the taxon abundances also appear very similar. However, it is noteworthy that in T1 Low, the messenger from K01952 phosphoribosyl-formylglycinamide synthase (EC 6.3.5.3), involved in metabolic processes for purine synthesis [33], is overrepresented.

To gain a more general understanding of the transcripts, a sequence overrepresentation analysis (ORA) was conducted [62], which identifies the most represented metabolic pathways during fermentation, specifically those with the highest number of transcripts involved [63]. Figure 5 shows the differentially represented metabolic pathways in each of the treatment pairings. Overall, the most expressed metabolic pathways are those related to two-component systems, carbon metabolism, and cofactor biosynthesis. Methane metabolism is significantly overrepresented in T2 compared to T0, and in T1 compared to T2, with a p-value ≤ 0.05. These metabolic pathways are vital for the survival of organisms, as they are essential to nearly all cellular functions [64].

To strengthen the analysis, a Gene Set Enrichment Analysis (GSEA) was conducted [43], which classifies all genes in the dataset based on their differential expression and tests whether genes within a set are enriched. In other words, it takes into account the distribution of gene expression changes within a dataset. In this case, the pairwise comparisons were performed again: T0 vs. T1 (Figure 6A,B), T0 vs. T2 (Figure 7A), and T1 vs. T2 (Figure 7B), showing significantly enriched metabolic pathways with a p-value ≤ 0.05, based on the gene sets generated from the evaluated sequences. In Figure 6A, the methane metabolism pathway is highlighted, suggesting that the set of genes involved in this metabolism is significantly represented in T1 compared to T0, the control [40]. This is consistent with the results obtained from the in vitro fermentation evaluation [14].

A total of 13,313 transcripts were identified in the KEGG database [33]. Figure 8 shows the divergence in transcript expression across the various treatment pairings. It is worth mentioning that the results from the ruminal fermentation study [14], where it was demonstrated that methane production was significantly reduced in T1 compared to T0, which serves as the control and reference sample. To explore this further, a functional analysis was performed using the methane metabolic pathway to see if any enzymes were significantly differentiated within this specific metabolic pathway. Table 1 shows the results of the proximal fermentation analysis. A statistically significant decrease in methane production was determined in treatment T1 (named MB), and when comparing the genes expressed in these samples using GSEA, there is a statistically significant difference in the methane metabolic pathway between T0 and T1 [14].

To delve into methanogenesis specifically and how the supplementation with moringa might affect it, a differential analysis of enzymes acting directly in the methane metabolic pathway was conducted. Figure 9 displays the enzymes that were differentially overexpressed and underexpressed in T1 compared to T0. In the case of methane metabolism, enzymes such as Formate dehydrogenase (EC 1.17.1.9) and glycine hydroxymethyltransferase (EC 2.1.2.1), involved in the transfer of the methyl group from methyl-hydrofolate, were overexpressed in the control compared to the low moringa treatment (T1). This results in an intermediate compound for methane synthesis. Conversely, enzymes like Methylamine dehydrogenase (EC 1.4.9.1) and Methylglutamate dehydrogenase (EC 1.5.99.5) were found to be underexpressed in the samples treated with T1 Low Moringa compared to T0 Control [65]. Studies show that one strategy for methane mitigation is to use secondary metabolites from plants to inhibit the growth of cellulolytic bacteria and archaea [66], e.g., through molecular docking, pterygospermin showed high affinity for the active site of the enzyme Methyl-Coenzyme M Reductase (EC 2.4.8.1), involved in methane production, suggesting a direct inhibitory effect [67], in this case it does not affect the gene expression, the compound have directly effect. It is important to note that this study was conducted using in vitro fermentation. While this approach allows precise control of experimental conditions and supplementation levels, it does not fully replicate the complex interactions present in a living animal. In this preliminary study, Moringa oleifera was not administered directly to the animals because it is not a commonly available species in our region, although it could potentially be adapted. Therefore, caution should be exercised when extrapolating these results to in vivo scenarios, and further studies directly feeding animals with Moringa oleifera are recommended to confirm the observed effects on methane production and microbial populations.

Overall, these results provide insight into the ruminal microbial and functional shifts induced by Moringa oleifera supplementation. However, further studies, including microbial viability assays and in vivo evaluations, are needed to determine optimal dosing strategies for ruminant diets.

5. Conclusions

Metatranscriptomic analysis provides insights into how environmental factors influence the behavior of the organisms within a niche and allows for the description and characterization of genes expressed under specific conditions. The use of Moringa oleifera as a dietary supplement resulted in significant changes in gene expression and population abundance of ruminal microorganisms compared to the control. Two supplementation levels were evaluated, and differences were observed not only with the use of moringa per se, but also with regard to the dosage, due to changes presented with T1 (15% moringa: 85% alfalfa). It was concluded that this dose is the most appropriate to be implemented as a supplement. One of the detected changes was a reduction in the expression of certain enzymes involved in methane metabolism, suggesting that this supplement, when used at appropriate concentrations, could potentially be used to reduce methane production during ruminal fermentation, while also benefiting the nutrition of ruminants. Incorporating Moringa oleifera as a feed supplement has the potential to enhance ruminant performance and support climate-friendly farming practices by reducing greenhouse gas emissions and promoting carbon sequestration. Although these findings are promising, they are based on in vitro fermentation, and further in vivo evaluations are needed to fully confirm the effects of Moringa oleifera supplementation on rumen function, microbial populations, and methane mitigation. In addition, studies assessing production parameters, such as feed efficiency, weight gain, and milk or meat quality, are required to determine the practical benefits of this supplementation. For future research, increasing the number of biological replicates per treatment is also important to strengthen statistical power, improve the reliability of results, and validate the observed trends.