Transcriptome-guided breeding for Paspalum notatum: producing apomictic hybrids with enhanced omega-3 content Authors: Lara Marino, Silvia Altabe, Carolina Marta Colono, Maricel Podio, Juan Pablo Amelio Ortiz, David Balaban, Juliana Stein, Nicolás Spoto, Carlos Acuña, Lorena Adelina Siena, José Gerde, Emidio Albertini, Silvina Claudia Pessino Published: 2024 ABSTRACT AbstractKey messageTranscriptomics- and FAME-GC-MS-assisted apomixis breeding generated Paspalum notatum hybrids with clonal reproduction and increased α-linolenic acid content, offering the potential to enhance livestock product's nutritional quality and reduce methane emissions AbstractA low omega-6:omega-3 fatty acid ratio is considered an indicator of the nutritional impact of milk fat on human health. In ruminants, major long-chain fatty acids, such as linoleic acid (18:2, omega-6) and α-linolenic acid (18:3, omega-3), originate from dietary sources and reach the milk via the bloodstream. Since forages are the primary source of long-chain fatty acids for such animals, they are potential targets for improving milk lipid composition. Moreover, a high 18:3 content in their diet is associated with reduced methane emissions during grazing. This work aimed to develop genotypes of the forage grass Paspalum notatum with high leaf 18:3 content and the ability for clonal reproduction via seeds (apomixis). We assembled diploid and polyploid Paspalum notatum leaf transcriptomes and recovered sequences of two metabolism genes associated with the establishment of lipid profiles, namely SUGAR-DEPENDENT 1 (SDP1) and PEROXISOMAL ABC TRANSPORTER 1 (PXA1). Primers were designed to amplify all expressed paralogs in leaves. qPCR was used to analyse SDP1 and PXA1 expression in seven divergent genotypes. Reduced levels of SDP1 and PXA1 were found in the polyploid sexual genotype Q4188. Fatty acid methyl esters/gas chromatography/mass spectrometry (FAME/GC/MS) assays confirmed an increased percentage of 18:3 in this genotype. Crosses between Q4188 and the obligate apomictic pollen donor Q4117 resulted in two apomictic F1 hybrids (JS9 and JS71) with reduced SDP1 and PXA1 levels, increased 18:3 content, and clonal maternal reproduction. These materials could enhance milk and meat quality while reducing greenhouse gas emissions during grazing.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00122-024-04788-6. Introduction Livestock production can convert feed crops, such as forages, into human food, with sustainable intensification achieved when system inputs and outputs are balanced (Derner et al. 2017). Although consumer trends are expected to shift towards plant-based food products, animal-derived items remain important in a healthy diet due to their nutritional density and cultural significance (Cramer et al. 2017). This is particularly relevant in developing countries, where undernourishment affects approximately 13% of the population, and nearly one billion people are underfed (Tebeje et al. 2017). Additionally, the social, cultural, and historical importance of animal farming cannot be overlooked (Thornton 2010). Polyunsaturated fatty acids (PUFAs) are characterized by a backbone with two or more carbon–carbon double bonds. Omega-3 and omega-6 fatty acids are families of PUFAs that both feature a final carbon–carbon double bond at the n-3 and n-6 positions, respectively—meaning the third or sixth bond counting from the methyl end. A typical Western diet currently provides an omega-6:omega-3 ratio of about 20:1 in favour of omega-6, which can lead to supraphysiologic inflammatory responses and perpetuate chronic low-grade inflammation, impacting several health conditions (DiNicolantonio and O'Keefe 2021). Omega-3 PUFAs are essential fats; the human body cannot synthesize and must obtain from food. Although common sources of omega-3 include certain fish and seafood, some vegetable oils, nuts, flax seeds, and leafy vegetables, ruminant products such as milk, cream, and cheese can also contribute significant amounts to the human diet (Dewhurst et al. 2003). In this context, it is important to note that the dairy sector has substantial potential to impact early childhood nutrition, particularly in developing countries (Headey 2023). The density of various macro and micronutrients in dairy products and their taste, texture, and consistency make them ideal for infants and young children (Headey 2023). Even though forages contain relatively small amounts of lipids, they are often the primary source of fatty acids in ruminant diets (Dewhurst et al. 2006). Therefore, diet is a key factor that can be targeted to manipulate the lipid composition and nutritional quality of bovine milk (Dewhurst et al. 2006). In addition to diet, milk composition is influenced by several factors, including animal breed, age, health status, and stage of lactation, among others (O'Callaghan et al. 2019). Long-chain fatty acids typically derive from dietary sources and reach the milk via the bloodstream. In contrast, short- and medium-chain fatty acids are produced in the mammary gland through de novo synthesis from precursors such as acetate and butyrate (Alothman et al. 2019). High proportions of α-linolenic acid (18:3, omega-3) are found in fresh pastures and certain oilseeds (e.g., linseed), whereas linoleic acid (18:2, omega-6) predominates in corn silage and cereals (Alothman et al. 2019). Previous studies have shown that the omega-3 content of forage and dairy cows’ milk fat is positively correlated (Hebeisen et al. 1993; Alothman et al. 2019; Techeira et al. 2023). The correlation between the percentage of grass intake and the relative concentration of 18:3 in milk (%) was R2 = 0.7705 (p-value < 0.0001), according to Techeira et al. (2023). Since 18:3 and 18:2 are the most abundant omega-3 and omega-6 fatty acids in milk, respectively, the ratio between their concentrations is considered an indicator of the nutritional impact of milk fat on human health, with lower omega-6:omega-3 ratios being more beneficial (Schwendel et al. 2015). Moreover, recent reports indicate that increased grass intake in cows' diets decreases the relative amount of 16:0 (palmitic acid) in milk, while increasing 18:1 (oleic acid), 18:2, 18:3, and conjugated linoleic acid (CLA) (Techeira et al. 2023). Therefore, grasses are valuable dietary components for increasing UFA content in milk and should be considered viable target crops for improving PUFA concentrations. Beyond nutritional improvements, increased omega-3 content has been reported to enhance butter unctuousness (Thomson and Van der Poel 2000) and even impart a pleasant flavour to meats (Wood et al. 1999). Additionally, increasing the PUFA content in animal diets has a positive effect on digestion. Supplementing ruminant diets with oil has been reported to reduce greenhouse gas emissions (Beauchemin et al. 2008; Moate et al. 2011; Bayat et al. 2018) and minimize economic losses associated with inefficiencies in rumen fermentation (reviewed by Sun et al. 2022). For every 1% absolute increase in oil in a ruminant’s feed, methane (CH4) emissions decrease by 3.5–5.6% (Beauchemin et al. 2008; Moate et al. 2011). Since lipid supplementation of animal diets involves substantial economic costs, using forages with increased endogenous levels of lipids, particularly PUFAs, represents an appealing low-cost option for reducing methane emissions and enhancing the nutritional quality of dairy products and meats. Suppression of SUGAR-DEPENDENT 1 (SDP1) and PEROXISOMAL ABC TRANSPORTER 1 (PXA1) has been shown to lead to lipid accumulation in plant vegetative tissues due to decreased breakdown of triacylglycerols (TAG) or fatty acids (FA) (Kelly et al. 2013; Slocombe et al. 2009). Wijekoon et al. (2020) reported that reducing the activity of these two carbon metabolism genes results in a significant increase in total lipid content and the proportion of 18:3 (18:3 Δ9,12,15), as well as a decrease in stearic acid (18:0) and palmitic acid (16:0) levels in Medicago truncatula seedlings. Therefore, both SDP1 and PXA1 have been proposed as ideal targets for enhancing PUFA content in forages (Wijekoon et al. 2020). Paspalum is one of the largest genera in the Poaceae family, encompassing about 370 grass species with a wide range of morphological and ecological adaptations (Ortiz et al. 2013). The genus’s evolutionary success is attributed to its multiple reproductive strategies, including cross-pollination, pseudogamous gametophytic apomixis (asexual seed reproduction), and vegetative propagation (Bashaw et al. 1970). Several Paspalum species are key components of native South American grasslands, where their natural populations are extensively used for livestock management (Ortiz et al. 2013). The ability of Paspalum spp. to propagate via apomixis has facilitated rapid breeding programmes, currently generating new cultivars and hybrids with desirable traits for livestock farming (Acuña et al. 2019). Apomictic plants can produce non-reduced female gametophytes by bypassing meiosis, develop embryos with a maternal genotype without fertilization (i.e., parthenogenesis), and simultaneously produce viable reduced pollen (Ortiz et al. 2013). Combined with sexuality, apomixis provides a unique reproductive capacity, allowing hybrid genotypes to be fixed and cloned through seeds. This natural process can be harnessed in breeding programmes (Acuña et al. 2019), where sexual mother plants are crossed with apomictic pollen donors. Colchicine-induced sexual tetraploids are self-incompatible, which makes them suitable as maternal parents in large-scale hybrid production. In contrast, sexual polyploid hybrids derived from crosses between sexual and apomictic plants are self-compatible and require emasculation before crossing (Acuña et al. 2019). Our objective here was to generate apomictic genotypes of Paspalum notatum with increased leaf 18:3 content, to enhance feed resources for producing dairy and meat products with improved nutritional quality and to provide a potential means of reducing greenhouse gas emissions. Although studies on the regulation of SDP1 and PXA1 in grasses have not yet been conducted, we hypothesized that reduced expression levels of these genes would alter the quantitative and qualitative profile of PUFAs, as observed in M. truncatula (Wijekoon et al. 2020) and Arabidopsis thaliana (Slocombe et al. 2009; Kelly et al. 2013). This hypothesis is based on the conserved role of these genes in TAG breakdown and the promotion of β-oxidation pathways (Zolman et al. 2001; Footitt et al. 2002; Eastmond 2006). We identified SDP1 and PXA1 homologous sequences expressed in the leaf transcriptomes of several divergent P. notatum genotypes, designed specific qPCR primers to amplify all paralogs expressed in leaves, evaluated their relative expression levels, and characterized leaf fatty acid (FA) profiles using fatty acid methyl ester/gas chromatography/mass spectrometry (FAME/GC/MS). Finally, we crossed selected plant materials with low SDP1 and PXA1 expression and high levels of 18:3, to obtain F1 hybrids with an improved lipid profile and capable of clonal reproduction via seeds. Materials and methods Plant material We used the following P. notatum genotypes: Q4117, apomictic, tetraploid (2n = 4x = 40), collected from a natural population in Southern Brazil, (Ortiz et al. 1997); C4-4x, fully sexual artificially-generated autotetraploid (2n = 4x = 40), obtained by chromosome duplication of a sexual diploid plant using colchicine treatment (Quarin et al. 2001); R1, sexual, diploid (2n = 2x = 20), collected from a natural population near Cayastá city, Santa Fe Province, Argentina (Quarin et al. 2001; Vega et al. 2024); Q3664, facultative apomictic tetraploid (2n = 4x = 40), with a high level (> 70%) of sexual reproduction (Quarin et al. 1984, 2003); Q4188, experimentally obtained sexual tetraploid accession (2n = 4x = 40) derived from the cross Q3664 × Q3853 (Quarin et al. 2003); R8, sexual, diploid (2n = 2x = 20), collected from a natural population near Cayastá city, Santa Fe Province, Argentina (Quarin et al. 2001); Boyero-UNNE, apomictic hybrid cultivar derived from crosses between Q3664 and Q4117, tetraploid (2n = 4x = 40) (Urbani et al. 2017a, b); JS9, JS40, JS71, and JS130, apomictic, tetraploid F1 hybrids derived from the cross Q4188 × Q4117 (Stein et al. 2004). These genotypes belong to the Paspalum living germplasm collection of Instituto de Botánica del Nordeste (IBONE), CONICET-UNNE, Corrientes, Argentina. Vegetative replicates of each plant were grown in pots at Instituto de Investigaciones en Ciencias Agrarias de Rosario (IICAR, CONICET-UNR), Zavalla, Argentina. Rhizome propagules were later transferred to neighbouring plots to ensure identical growing environmental conditions. Paspalum notatum leaf transcriptome assembly We used the raw data generated by de Oliveira et al. (2020) (NCBI SRA repository SRP150615), obtained after transcriptome sequencing of diploid sexual, tetraploid sexual, and tetraploid apomictic P. notatum plants. Raw reads were demultiplexed and quality-checked using FastQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Duplicate sequences, ambiguous reads, and low-quality reads were removed using Trim Galore (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). High-quality reads (QC > 30) were assembled reference-based using the available genome of P. notatum (Vega et al. 2024; NCBI accession number GCA_036689595.1) and the software Trinity v2.0.2 (Grabherr et al. 2011). The Transcriptome Shotgun Assembly project has been deposited at DDBJ/EMBL/GenBank under accession number DAWXEG000000000. The version described in this report is the first one, DAWXEG010000000. Alignments, phylogenetic analysis, and in silico mapping Protein sequences of Setaria italica triacylglycerol lipase SDP1 (XP_004970049.1) and S. italica ABC transporter D family member 1 (PXA1) (XP_012701758.1) were used as queries in TBLASTN analyses to retrieve homologous sequences within the 2 × and 4 × P. notatum assembled transcriptomes. Transcript sequences showing the most significant nucleotide and/or amino acid sequence homology were aligned using Clustal Omega (https://www.ebi.ac.uk/jdispatcher/msa/clustalo) (Madeira et al. 2022). Multiple sequence alignments of proteins were performed with MEGA11 (Tamura et al. 2021) using the ClustalW algorithm. Phylogenetic trees were constructed with MEGA11 using the neighbour-joining (NJ) method (Saitou and Nei 1987). Evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling 1965) and expressed as the number of amino acid substitutions per site. The rate variation among sites was modelled with a gamma distribution (shape parameter = 1). Analysis of conserved protein domains was conducted using the R package RAPG with the hmm option (R Core Team: The R Project for Statistical Computing, https://www.r-project.org.). Primer design and qPCR analysis Specific primers for the amplification of P. notatum leaf homologs of the SDP1 and PXA1 gene transcripts were designed using Primer3Plus (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) (Untergasser et al. 2007). Primer names, sequences, lengths, product sizes, and predicted melting temperatures (Tm) are shown in Online Resource 1. Specific amplification was verified by in silico PCR (http://insilico.ehu.es/PCR/) (Bikandi et al. 2004) using the diploid and tetraploid transcriptomes as templates and confirmed by qualitative PCR. For qPCR, total RNA was extracted from fresh young leaves at 9:00 am using the SV Total RNA Isolation Kit (PROMEGA, Madison, WI, USA). cDNAs were synthesized with ImProm-II Reverse Transcriptase (Promega) following the manufacturer’s recommendations. Quantitative PCR reactions (final volume: 20 µL) were performed in a Rotor-Gene Q thermocycler (QIAGEN, Hilden, Germany). They included 0.5 µM gene-specific primers (Online Resource 1), 1 × Real Mix qPCR (BIODYNAMICS, Buenos Aires, Argentina), and 5.5 ng of cDNA. The constitutive gene β-TUBULIN was used as a reference (Online Resource 1), and reactions without template (cDNA) were included for each primer combination as negative controls. Samples were run in two biological and three technical replicates. The cycling programme consisted of an initial denaturation at 94 °C for two min, followed by 45 cycles of 15 s at 94 °C, 30 s at the annealing temperature, and 40 s at 72 °C. The annealing temperature was set at least two °C lower than the Tm of the corresponding primer (Online Resource 1). The REST9 V2.0.13 software was used to determine the relative expression of SDP1 and PXA1, using the expression level of one of the genotypes (R1 or Q4117, depending on the experiment) as the control (relative expression level: 1). Principal component analysis (PCA) was performed using R Studio, version 4.4.1 (R Core Team: The R Project for Statistical Computing, https://www.r-project.org) with the factoextra (Kassambara and Mundt 2020) and FactoMineR (Lê et al. 2008) packages. Lipid profiling The lipid profiles were analysed using gas chromatography coupled with mass spectrometry (GC–MS). Lipid extraction was performed following the method published by Bligh and Dyer (1959). Samples were processed in triplicate. Briefly, 200 mg of leaves were harvested, grounded with a mortar and pestle, and transferred to a glass tube with 3 mL of chloroform:methanol (1:2) mixture. Dibutylhydroxytoluene (BHT) was added as an antioxidant at a final 5 µg/mL concentration. Samples were incubated for 12 h or more at − 20 °C, centrifuged, and the supernatants were transferred to a new tube. One mL of KCl (1 M) and 1 mL of chloroform were added, the samples were mixed, and then centrifuged for 2 min at 2000 g. Finally, the organic phases were transferred to new tubes and dried under N₂ flow. For methyl ester preparation, lipid samples were incubated with 1–2% concentrated sulfuric acid in methanol for 2 h at 80 °C. Fatty acid methyl esters (FAMEs) were extracted with hexane (2 mL), evaporated to dryness, and dissolved in hexane. The GC–MS analysis was performed on a Shimadzu GC-2010 Plus. The column used was a Supelco WAX-10 (Sigma-Aldrich) with a 100% polyethylene glycol phase. The helium flux was set at 1 mL/min, and the heating programme started at 180 °C (0 min to 32 min), followed by a gradient increase of 3 °C/min from 180 °C to 240 °C. The split ratio was 1:30, and the ionization voltage was 70 eV, with an ionic range from 50 to 600 Da. The ion spectra were recorded as relative abundance in a mass/charge (m/z) function. Peak assignment was done using a mixture of fatty acid methyl ester standards (AGs Supelco, Sigma-Aldrich). The quantification of UFAs, PUFAs, and 18:3 was expressed as a percentage of the total fatty acids. Data are presented as means ± SD of three replicate samples. Data are presented as means ± SD of three replicate samples. Outliers were systematically identified and removed using the following approach: the values for the three replicates were organized in an Excel file, and means and standard deviations were calculated using Excel’s statistical functions. In cases where the SD exceeded 10%, outliers were detected and manually removed to reduce variability. For all samples, at least two replicates with similar values were retained. Total lipid content was assessed gravimetrically using the modified Folch method following Christie and Han (2010) after biomass freeze-drying (Rificor, Argentina), starting from an aliquot of 1.000 g (dried biomass). The first 10 cm of leaves were collected from the base. The significance of the differential levels of 18:3 and total lipid content was assessed using the LSD Fisher test (Fisher 1935), using the Infostat programme (https://www.infostat.com.ar/). Correlation analysis was conducted for two sets of variables: percentage of fatty acids and transcript expression levels. Pearson's correlation between traits was estimated using R statistical software (R Core Team: The R Project for Statistical Computing, https://www.r-project.org), with the Corrplot package (Wei and Simko 2021) and the function “corrplot" to generate the matrix. Cytoembryological classification of F1 plants To determine the mode of reproduction of the experimental plants, embryo sac development and morphology were characterized through cytoembryological observations of cleared ovules, following the protocol described by Young et al. (1979) with modifications as outlined below: spikelets at anthesis were fixed in FAA (70% ethanol:formaldehyde:acetic acid, 18:1:1) for 24–48 h and then transferred to 70% ethanol. Ovaries were subsequently dissected and treated with 3% H₂O₂ for 2 h before dehydration in an ethanol series (50%, 70%, 95%, and two 100% steps; 30 min each step). Finally, dissected ovules were cleared using a series of methyl salicylate/ethanol (v/v) solutions (1:1, 3:1, 5.6:1; 30 min each step), incubated in methyl salicylate for at least 12 h, and examined using a Leica DM2500 microscope equipped with Nomarski differential interference contrast (DIC) optics. Plants presenting at least one ovule with one or more aposporous (non-reduced) embryo sacs (evidenced by the absence of antipodal cells at the pre-anthesis/anthesis stage) were classified as apomicts. The expressivity of apospory was calculated as the proportion of ovules with supernumerary embryo sacs relative to the total number of fertile ovules scored. Results Assembly, recovery, and characterization of SDP1 and PXA1 leaf transcript sequences The public availability of leaf cDNA raw reads from several P. notatum genotypes, including different ploidies and reproductive modes (de Oliveira et al. 2020, NCBI SRA repository, accession number SRP150615), enabled us to construct leaf transcriptome assemblies for two sexual diploids (#22 and #306), one sexual tetraploid (#216), and three apomictic tetraploids (#115, #34, and #30), using the annotated P. notatum genome (Vega et al. 2024) as a reference. The diploid sexual transcriptome produced 76,682 transcripts with a GC content of 46.69% and an N50 length of 1,545 bp; the tetraploid sexual transcriptome yielded 69,969 transcripts with a GC content of 46.75% and an N50 length of 1,507 bp; and the tetraploid apomictic transcriptome generated 70,668 transcripts with a GC content of 46.88% and an N50 length of 1,361 bp. Next, we used the protein sequence of S. italica triacylglycerol lipase SDP1 (XP_004970049.1) as query in TBLASTN surveys, to identify homologs in the P. notatum diploid and polyploid transcriptomes. Overall, considering the three plant groups (2x sexual, 4x sexual, and 4x apomictic), a total of six sequences encoding SDP1 homologs were detected (TRGpnl4×sex54841, TRGpnl2×sex10993, TRGpnl4xapo48133, TRGpnl4×sex25540, TRGpnl4xapo22391, TRGpnl2×sex50945). Phylogenetic analyses involving the S. italica SDP1 sequence XP_004970049.1, two predicted proteins derived from the P. notatum triacylglycerol lipase SDP1 genomic sequences (Pn02t519100.01 and Pn04t133300.01) and the six transcripts expressed in the leaf transcriptomes, revealed two clusters (Online Resource 2). Transcripts from the first cluster (SDP1.1), including TRGpnl4×sex54841, TRGpnl2×sex10993, and TRGpnl4xapo48133, grouped with the Setaria SiSDP XP_004970049.1 and the Paspalum Pn04t133300.01 sequence (Online Resource 2). Transcripts from the second cluster (SDP1.2), including TRGpnl4×sex25540, TRGpnl4xapo22391, and TRGpnl2×sex50945, grouped with the Paspalum Pn02t519100.01 sequence (Online Resource 2). As expected, mapping assays showed that transcripts of cluster SDP1.1 colocalized with locus Pn04t133300, at a distal position on the long arm of chromosome 4 (10,225,745–10,233,097), while those of cluster SDP1.2 mapped to locus Pn02t519100, situated in a sub-telomeric region on the short arm of chromosome 2 (54,637,680–54,641,090) (Fig. 1a). These results indicate that SDP1.1 and SDP1.2 transcripts do not represent allelic or splice variants but originate from distinct gene copies located on different chromosomes. Additionally, the transcriptome heatmaps derived from the six normalized libraries exhibited a highly heterogeneous pattern in the different experimental plants (Fig. 1b), indicating expression variation. The predicted protein domains were PF11815.11 (domain of uncharacterized function DUF3336) and PF01734.25 (patatin-like phospholipase domain). Furthermore, in the Paspalum Pn02t519100.01 sequence, two transmembrane (TM) motifs were detected at the 5’ end (Fig. 1b). Characterization of SDP1 and PXA1 transcripts expressed in Paspalum notatum leaves. a Mapping of transcripts from the four phylogram groups (SDP1.1, SDP1.2, PXA1.1, and PXA1.2) onto the Paspalum notatum reference genome. Each cluster was mapped to distinct genomic regions, indicating they originate from different gene copies. b Top: heatmap of normalized transcriptomes showing expression variation among different genotypes for SDP1 transcripts; bottom: conserved protein domains and features detected in the predicted sequences derived from SDP1 leaf transcripts: PF11815.11 (uncharacterized function, DUF3336); PF01734.25 (patatin-like phospholipase); TM (transmembrane motif). c Left: heatmap of normalized transcriptomes showing expression variation among different genotypes for PXA1 transcripts; right: conserved protein domains and features detected in the predicted sequences derived from PXA1 leaf transcripts: PF00005.30 (ABC transporter); PF06472.18 (ABC transporter transmembrane region 2); PF00675.23 (insulinase, peptidase family M16); PF05193.24 (peptidase M16 inactive domain); TM (transmembrane motif) A similar analysis was conducted for PXA1 transcripts. We used the S. italica ABC transporter D family member 1 (XP_012701758.1) to identify homologous sequences in P. notatum. Among the three plant groups (2 × sexual, 4 × apomictic, 4 × sexual), nine sequences were detected (TRGpnl4xapo45921, TRGpnl4×sex45515, TRGpnl4×sex45517, TRGpnl2×sex12680, TRGpnl4×sex45514, TRGpnl4×sex45516, TRGpnl4xapo14198, TRGpnl4×sex13230, TRGpnl2×sex39853). Phylogenetic analyses revealed two distinct clusters (Online Resource 2). Transcripts in the first cluster (PXA1.1), including TRGpnl4xapo45921, TRGpnl4×sex45515, TRGpnl4×sex45517, TRGpnl2×sex12680, TRGpnl4×sex45514, and TRGpnl4×sex45516, grouped with the Setaria SiPXA XP_012701758.1 and the Paspalum Pn04t006900.01 sequences (Online Resource 2). In contrast, transcripts in the second cluster (PXA1.2), including TRGpnl4xapo14198, TRGpnl4×sex13230, and TRGpnl2×sex39853, grouped with the Paspalum Pn10t005100.02 sequence (Online Resource 2). In silico mapping onto the Paspalum reference genome (Vega et al. 2024) revealed that PXA1.1 sequences colocalized with locus Pn04t006900 on chromosome 4, while PXA1.2 sequences mapped to locus Pn10t005100 on chromosome 10 (Fig. 1a). These results indicate that PXA1.1 and PXA1.2 transcripts do not represent allelic or splice variants but originate from distinct gene copies located on different chromosomes. The transcriptome heatmaps derived from the six normalized libraries exhibited a highly heterogeneous pattern among the different experimental plants (Fig. 1b), suggesting expression variation. The detected protein domains included PF00005.30 (ABC transporter), PF06472.18 (ABC transporter transmembrane region 2), PF00675.23 (Insulinase, peptidase family M16), and PF05193.24 (Peptidase M16 inactive domain). Additionally, transmembrane (TM) features were detected in all family members (Fig. 1c). Natural variability in SDP1 and PXA1 transcript expression revealed by qRT-PCR We analysed seven divergent P. notatum genotypes from our living germplasm collection to evaluate natural SDP1 and PXA1 expression variability (Q4117, C44x, R8, R1, Q4188, Q3664, and Boyero). To amplify each family member, we identified unique sequence regions that distinguish transcripts belonging to different clusters and designed qPCR primers accordingly (Online Resource 1). In silico PCR assays (http://insilico.ehu.es/PCR/) (Bikandi et al. 2004) were used to confirm that amplifications were specific to the intended clusters. Leaf RNA samples were extracted from the P. notatum genotypes, all maintained under identical growing conditions and representing various ploidy levels and/or reproductive modes. Consistent with transcriptome analysis results, qRT-PCR assays revealed natural variability in the expression of SDP1 and PXA1 transcripts (Fig. 2a–e; Online Resource 3). The sexual genotype Q4188 exhibited low expression levels for all isoforms and was identified as a good candidate for high PUFA content (Fig. 2a–d, Online Resource 3). The genotype Q3664 produced both legitimate and alternative amplicons in the PXA1.1 and PXA1.2 reactions, leading to its exclusion from further analysis (data not shown). Principal component analysis (PCA) demonstrated that the four transcript groups (SDP1.1, SDP1.2, PXA1.1, and PXA1.2) contributed to the expression variation among genotypes (Fig. 2f). Relative expression of SDP1 and PXA1 transcripts in a representative collection of Paspalum notatum accessions. a–d Relative expression levels of SDP1.1 (a), SDP1.2 (b), PXA1.1 (c), and PXA1.2 (d) transcripts. Genotype R1 was used as the control (expression value set to 1). Asterisks denote statistical significance for differential expression compared to the control genotype R1 (**: p < 0.05; ***: p < 0.01). Genotype Q4188 exhibited low expression levels for all transcript isoforms (green arrows). e qPCR amplicons for SDP1.1, SDP1.2, PXA1.1, PXA1.2, and β-tubulin controls visualized on 2% agarose gels. Note that the gel bands may not reflect the relative proportions calculated from the qPCR data, as reactions reached their plateau after 45 cycles. f Principal component analysis (PCA) showing expression variation and illustrating the strength and direction of correlations among variables 18:3 levels are higher in genotype Q4188 FAME/GC/MS experiments were conducted to quantify PUFAs in P. notatum leaves (Online Resource 4). Fatty acids ranging from 14 to 26 carbons were detected, with mono-unsaturation occurring only in 16-carbon and 18-carbon members, and poly-unsaturation observed in 18-carbon members (18:2 and 18:3) (Online Resource 4). Consistent with gene expression predictions, the genotype Q4188 exhibited the highest level of 18:3; however, it was not significantly different (alpha = 0.05) from that of Q4117 and R1, as determined by the LSD Fisher test (Table 1). The levels of PUFAs and total UFAs were also higher in Q4188 than in the other genotypes (Table 1). The increase in total PUFA and UFA content was attributed to the higher 18:3 fraction, which came at the expense of the SFA fraction (Table 1). Table 1Percentages of the different fatty acid types in leaf samples of several representative Paspalum notatum genotypesFAQ4117C44XR1Q4188R8BoyeroSFA25.56 ± 2.0237.19 ± 2.7825.06 ± 1.1822.76 ± 1.1625.36 ± 3.0841.24 ± 5.10MUFA5.62 ± 0.972.54 ± 0.312.97 ± 0.054.10 ± 0.653.98 ± 0.433.58 ± 0.6218:212.12 ± 0.9615.72 ± 0.8215.17 ± 0.4313.43 ± 0.0817.21 ± 0.3311.53 ± 0.9718:356.70 ± 1.23b.c44.56 ± 1.77a56.8 ± 0.5b,c59.72 ± 0.01c53.51 ± 0.24b43.64 ± 3.00aPUFA68.82 ± 2.1960.28 ± 2.5971.97 ± 0.9073.15 ± 0.0970.72 ± 0.5755.18 ± 3.96UFA74.44 ± 3.1662.81 ± 2.9074.94 ± 0.9677.24 ± 0.7374.70 ± 1.0058.76 ± 4.58FA, fatty acid category; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; 18:2, linoleic acid; 18:3, linolenic acid; PUFA, poly-unsaturated fatty acids (18:2 + 18:3); UFA, total unsaturated fatty acids (MUFA + 18:2 + 18:3); Letters (a, b, c) indicate the different categories in an LSD Fisher test (alpha = 0.05; LSD = 3.48) Percentages of the different fatty acid types in leaf samples of several representative Paspalum notatum genotypes FA, fatty acid category; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; 18:2, linoleic acid; 18:3, linolenic acid; PUFA, poly-unsaturated fatty acids (18:2 + 18:3); UFA, total unsaturated fatty acids (MUFA + 18:2 + 18:3); Letters (a, b, c) indicate the different categories in an LSD Fisher test (alpha = 0.05; LSD = 3.48) Development of apomictic F1 hybrids with elevated α-linolenic acid levels According to our data, tetraploid genotypes Q4188 and Q4117 were included in the LSD Fisher group showing the highest 18:3 levels. Q4188 is a fully sexual genotype (Quarin et al. 2003), while Q4117 is a well-characterized obligate apomict (Ortiz et al. 1997). Our laboratory previously used these genotypes to produce a perennial mapping population segregating for apomixis (Stein et al. 2004, 2007). Consequently, F1 plants resulting from the cross between Q4188 (sexual mother plant) and Q4117 (apomictic pollen donor) were available. This cross is sexual, since the mother plant forms reduced megagametophytes, while the pollen donor, being apomictic, forms reduced microgametophytes. Both Q4188 and Q4117 are highly heterozygous polyploid genotypes, resulting in the segregation of various traits, including apomixis, in the F1 progeny. From the available F1 population, we selected three individuals previously classified as obligate apomicts (JS9, JS40, JS71) (Stein et al. 2004, 2007) and re-evaluated their reproductive mode, as well as that of the parental plants, using DIC cytoembryology (Fig. 3a, b, Online Resource 5). Additionally, we studied other plants from the F1 population and selected one (JS130) that exhibited a high degree of apospory (Online Resource 5). Our results, consistent with those of Stein et al. (2007), show that the chosen F1 hybrids exhibit high apospory expressivity, ranging from 73.3 to 86.29%. These percentages support the classification of the plants as obligate apomicts, as the observed range of residual sexuality does not usually result in significant formation of variable progeny (Reutemann et al. 2022). Reproductive mode and omega-6:omega-3 content of selected F1 hybrids. a–b Cytoembryological analysis of Paspalum notatum ovules showing meiotic and aposporous embryo sacs (a), and multiple (at least four) aposporous embryo sacs (b). Dashed lines outline the different embryo sacs; black bars represent 100 μm. Abbreviations: pn, polar nuclei; ac, antipodal cells. c Box plot showing the median omega-6:omega-3 ratios for each genotype, based on experiments conducted across three seasons qPCR analyses were used to quantify SDP1 and PXA1 expression levels in parental and F1 plants (Online Resources 3 and 6). The pollen donor genotype Q4177 was used as a control. The apomictic F1 hybrid JS9 showed an expression level significantly lower than those observed in Q4117 for the SDP1.1, SDP1.2, and PXA1.1 isoforms, and values that did not differ significantly from the mother plant Q4188 for all alleles. The same situation was observed for hybrids JS71, JS40, and JS130 regarding the PXA1.1 expression level, but not for the rest of the isoforms (Online Resources 3 and 6). Next, we quantified the fatty acid profiles of both parental plants, Q4188 and Q4117, and all selected F1 hybrids, using FAME/GC/MS experiments (Online Resources 4 and 7). Since these assays were conducted at the end of the southern hemisphere summer, we anticipated a decrease in PUFA levels compared to the initial experiments, due to the documented seasonal fluctuations in 18:3 levels driven by temperature (Thomson and Van der Poel 2000; Falcone et al. 2004; Meľuchová et al. 2008). As expected, the average level of 18:3 decreased from 52.48% to 34.87% in our experiments, which aligns with the data reported by Meľuchová et al. (2008). Despite the general decrease in PUFA levels across all plants due to seasonal variation, the 18:3 content remained significantly higher (p < 0.05) for three of the four tested F1 hybrids (JS9, JS71, JS130; LSD Fisher test group c) compared to both parental plants (Q4188; groups a, b; Q4117; group a) and the JS40 hybrid (group b) (Online Resources 4 and 7). Stability of 18:3 levels in selected apomictic F1 hybrids We conducted three additional experiments to compare the 18:3 content in the JS9, JS71, and JS130 hybrids with that of a control genotype (C4-4x) across different seasons: June 2023, September 2023, and March 2024 (Table 2, Online Resource 4). Hybrid JS130 exhibited variable 18:3 levels compared to the C4-4 × control, showing significantly higher, lower, or similar levels depending on the season. In contrast, hybrids JS9 and JS71 consistently demonstrated higher 18:3 levels in all experiments. Consequently, JS9 and JS71 were selected as improved materials. The dynamics of 18:3 representation for the selected hybrids (JS9/JS71) and the control are illustrated in Online Resource 8. As anticipated, higher 18:3 levels were observed at the end of the cold season (i.e., in the September experiment). Hybrids JS9 and JS71 showed an increase in 18:3 levels at the expense of reduced MUFA and 18:2 levels, resulting in increases of 12.7% and 28.4% in average 18:3 content, respectively. Table 2Percentages of the different fatty acid types in the leaf samples of selected apomictic F1 hybrids and a control of standard expressionExperimentsFAC44xJS9JS71JS130First experiment (June 2023)SFA21.17 ± 0.4723.93 ± 1.3224.08 ± 0.9723.95 ± 0.51MUFA6.01 ± 0.385.18 ± 1.323.87 ± 0.763.64 ± 0.3618:233.56 ± 0.2823.80 ± 0.9021.69 ± 0.1929.33 ± 0.3018:339.38 ± 0.41 a46.92 ± 1.32 c50.48 ± 0.87 d43.09 ± 0.36 bPUFA72.94 ± 0.6970.72 ± 2.2272.16 ± 1.0572.42 ± 0.66UFA78.95 ± 1.0775.90 ± 3.5476.03 ± 1.8176.06 ± 1.02Second experiment (September 2023)SFA21.01 ± 1.0624.03 ± 1.1523.10 ± 1.2024.98 ± 1.36MUFA4.58 ± 0.575.48 ± 0.624.49 ± 0.568.20 ± 2.4018:227.20 ± 0.6720.55 ± 0.6018.71 ± 0.3325.60 ± 0.8118:346.77 ± 0.92 b49.16 ± 1.06 c53.89 ± 0.27 d42.77 ± 0.17 aPUFA73.97 ± 1.5969.71 ± 1.6772.60 ± 0.6068.37 ± 0.98UFA78.55 ± 2.1675.19 ± 2.2977.09 ± 1.1676.57 ± 3.38Third experiment (March 2024)SFA27.52 ± 1.5828.96 ± 1.9127.32 ± 1.3935.67 ± 5.78MUFA7.77 ± 0.207.56 ± 1.174.24 ± 0.767.46 ± 0.6618:234.74 ± 0.4228.69 ± 1.1723.70 ± 0.3629.66 ± 2.8118:329.96 ± 1.03 a34.82 ± 1.43 b44.74 ± 0.92 c27.21 ± 2.67 aPUFA64.71 ± 1.4563.49 ± 2.6568.44 ± 1.2856.87 ± 5.48UFA72.48 ± 1.6571.04 ± 3.8172.68 ± 2.0364.33 ± 6.14FA, fatty acid category; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; 18:2, linoleic acid; 18:3, linolenic acid; PUFA, poly-unsaturated fatty acids (18:2 + 18:3); UFA, total unsaturated fatty acids (MUFA + 18:2 + 18:3); Letters (a, b, c) indicate the different categories in an LSD Fisher test (alpha = 0.05; LSD = 2.19, 1.89, and 3.00, for the first, second, and third experiment, respectively) Percentages of the different fatty acid types in the leaf samples of selected apomictic F1 hybrids and a control of standard expression FA, fatty acid category; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; 18:2, linoleic acid; 18:3, linolenic acid; PUFA, poly-unsaturated fatty acids (18:2 + 18:3); UFA, total unsaturated fatty acids (MUFA + 18:2 + 18:3); Letters (a, b, c) indicate the different categories in an LSD Fisher test (alpha = 0.05; LSD = 2.19, 1.89, and 3.00, for the first, second, and third experiment, respectively) Considering that the 18:2 fraction predominantly represents the omega-6 series, the omega-6:omega-3 average ratio was 0.86 ± 0.23 for the control plant C44x, 0.43 ± 0.07 for JS71, and 0.58 ± 0.17 for JS9. An ANOVA test indicated that means of omega-6:omega-3 ratios varied significantly among the genotypes (p < 2 e−16). According to LSD Fisher's tests, in all experiments, the omega-6:omega-3 ratio of plant C4-4 × differed significantly from that of the hybrids (LSD: 0.043, Online Resource 9). Altogether, our results indicate that genotypes JS9 and JS71 display statistically significant higher levels of 18:3 and lower omega-6:omega-3 ratios across all seasons. A box plot showing the 6:omega-3 ratios medians for each genotype is presented in Fig. 3c. Assessment of total lipid content in the leaves of selected apomictic F1 hybrids Next, we measured total lipid levels in JS9, JS71, and C4-4x (control) leaves, using a gravimetric method (Christie and Han 2010). The average lipid content was 2.28%, 2.85%, and 1.94% for C4-4x, JS9, and JS71, respectively. An LSD Fisher test revealed significant differences for the three genotypes (Categories: C4-4x: b; JS9: c; JS71: a; LSD 0.07795). Considering these and previous results, we concluded that JS9 exhibits an increase in both total lipid content and 18:3 percentage (25% and 12.7% higher than the control, respectively) and a decrease in the omega-6:omega-3 ratio (32.55% lower than the control). Instead, JS71 shows reduced total lipid content, but a significantly improved 18:3 content (14.9% lower and 28.4% higher than the control, respectively), and a decrease in the omega-6:omega-3 ratio (50.0% lower than the control). Correlation among fatty acid classes and SDP1/PXA1 expression levels Correlation studies were conducted among variables reflecting fatty acid abundance by type and SDP1/PXA1 expression levels for the original genotype collection (Fig. 4a, b, p-values in Online Resource 10). As expected, a strong positive correlation was observed among levels of 18:3, PUFA, and UFA. A less pronounced but still positive correlation was found between MUFA and the 18:3/UFA variables, indicating that some unsaturation patterns tend to increase together (Fig. 4a, Online Resource 10). Conversely, SFA exhibited a strong negative correlation with 18:3/PUFA/UFA and a lower but still statistically significant correlation with MUFA (Fig. 4a, Online Resource 10). Expression levels of PXA1.1, PXA1.2, and SDP1.1 showed negative correlations with the 18:3/PUFA/UFA contents, but only the PXA1.1 correlation was statistically significant. Meanwhile, SDP1.2 displayed a low but positive correlation, which was not statistically significant, with these variables (Fig. 4a, b, Online Resource 10). Notably, a negative correlation between SDP1.2 and 18:2 levels was detected (with an associated p-value of 0.05492307), along with a strong positive correlation with the MUFA content (Fig. 4a, Online Resource 10). This suggests that a decrease in SDP1.2 could lead to increased 18:2 levels at the expense of 18:3 and MUFA fractions. Our results indicate that PXA1.1 is the gene/isoform exhibiting the strongest negative correlation with increased 18:3 levels. Correlation analysis among lipid fraction abundance and SDP1/PXA1 expression variables. a Correlation plot showing positive (blue) and negative (brown) correlations between pairs of analysed variables. The size of the circles represents the absolute value of the corresponding correlation coefficients. b Principal component analysis (PCA) illustrating the contribution of each variable. The increase in 18:3/PUFA/UFA was inversely correlated with SFA content (i.e., higher 18:3 levels were associated with decreased SFA levels). Levels of PXA1.1, PXA1.2, and SDP1.1 were negatively correlated with 18:3 increments, while SDP1.2 showed a weak positive correlation. PXA1.1 levels exhibited the strongest negative correlation with 18:3 content Discussion An increase in the dietary omega-6:omega-3 ratio has been observed from the Palaeolithic period ( < 4) to the present (20 in a typical Western diet), which may have adverse consequences for human health (DiNicolantonio and O’Keefe 2021). Despite their considerable proportion of saturated fatty acids (SFA), ruminant products can be a valuable source of beneficial PUFAs, potentially improving the nutritional quality of current human diets. Meats generally have a low fat content, and the omega-6:omega-3 ratio in beef is relatively low (Choi et al. 2000; Scollan et al. 2001). Additionally, milk fat typically contains around 4% PUFA (Jensen 2002). The 18:3 (omega-3) content in meat and milk can be increased by feeding animals specific diets. For example, grasses tend to have a higher 18:3 content than leguminous plants (Whetsell and Rayburn 2022). The fatty acid profile of milk changes rapidly with variations in grass intake, with grass feeding enhancing the nutritional and technological quality of milk (Techeira et al. 2023). Although grasses usually contribute less than 0.05% of total fatty acids, a significant proportion is 18:3 (omega-3), which has a substantial impact on the fatty acid profiles of ruminant products (Dewhurst et al. 2003). Inefficiencies in rumen fermentation of plant materials result in the emission of significant amounts of CH₄, a potent greenhouse gas with a climate change impact 25 times greater than CO₂ (Solomon et al. 2007). Given the complex challenges of global warming, improving forages to reduce CH₄ production has become a priority for both breeders and biotechnologists. Plant oil supplements have been reported to reduce methane emissions in dairy cows fed grass silage-based diets, sparking interest in exploring methods to enhance high-quality lipid intake (Sun et al. 2022). Ruminal microorganisms typically hydrogenate C18 PUFAs and produce conjugated linoleic acids (CLAs), which can influence the competition for metabolic hydrogen between propionate and methane production pathways (Sun et al. 2022). While lipid supplementation of feed shows promise for reducing CH₄ emissions, it involves substantial financial investment. Therefore, utilizing forages with increased shoot lipid content presents a practical alternative (Wijekoon et al. 2020). SDP1 and PXA1 are prominent candidates affecting fatty acid profiles in leaves. SDP1, initially characterized in Arabidopsis thaliana as the dominant lipase involved in triacylglycerol (TAG) turnover (Eastmond 2006), is important but not essential for germination (Kelly et al. 2011; Azeez et al. 2022; Aznar-Moreno et al. 2022). PXA1, also known as COMATOSE (CTS), is a peroxisomal fatty acid ABC transporter responsible for mobilizing triacylglycerol to sucrose during seed germination (Zolman et al. 2001; Footitt et al. 2002). Recent studies using virus-induced gene silencing (VIGS) in M. truncatula have shown that silencing SDP1 or PXA1 leads to significantly increased α-linolenic acid and decreased stearic acid (18:0) levels in total acyl lipid extracts (Wijekoon et al. 2020). Based on these findings, we focused on SDP1 and PXA1 as candidate genes for selecting P. notatum genotypes with potentially elevated 18:3 levels. We chose to explore natural variability within a molecular-assisted breeding framework, rather than employing genetic engineering, because our laboratory houses a large collection of highly heterozygous P. notatum elite material with diverse origins, ploidy levels, and reproductive modes, serving as a rich reservoir of genetic variation. Moreover, detecting natural variability and stabilizing it through apomixis avoids the regulatory procedures typically associated with genetically engineered plants, as the variation results from the crossing of naturally occurring genotypes. Paspalum grasses have been selected and improved through both classical and molecular methods for over 80 years (Ortiz et al. 2020). Since 1940s, 29 apomictic cultivars have been collected in South America, evaluated, registered, and utilized in various parts of the world, including Australia, Japan, Thailand, the USA, Brazil, and Argentina (Acuña et al. 2019). Additionally, an apomictic hybrid cultivar of P. notatum (cv. Boyero-UNNE), derived from polycrosses between a highly sexual mother plant and an apomictic pollen donor and characterized by its upright and rapid growth habit, was released a few years ago (Urbani et al. 2017a, b). The primary goal of hybridizing apomictic species is to widen the genetic variability contained in apomictic races for breeding purposes and to combine heterosis for agronomically desirable traits with high levels of clonal reproduction via seeds. Specifically for P. notatum, the identification of molecular markers (MMs) co-segregating with apomixis (Stein et al. 2007), the generation of artificial sexual polyploids through colchicine duplication (Quarin et al. 2001), the construction of transcriptome databases (Ortiz et al. 2017, 2019; de Oliveira et al. 2020; Podio et al. 2021), and the availability of full genome assemblies (Yan et al. 2022; Vega et al. 2024) facilitate the rapid application of molecular markers and biotechnological tools. These advancements enable the identification, evaluation, and modification of candidate genes, and the examination of apomixis occurrence. Furthermore, the extensive diversity observed in agronomic traits resulting from crossing sexual and apomictic genotypes (Ortiz et al. 2020), along with the establishment of biolistic transformation platforms (Mancini et al. 2014) and editing technologies (May et al. 2023), presents innovative prospects for exploiting both reproductive and non-reproductive traits of interest. In this work, we utilized publicly available cDNA raw reads of P. notatum leaves (de Oliveira et al. 2020) to assemble the transcriptomes of several genotypes and identify homologous transcripts of SDP1 and PXA1. We detected leaf transcripts originating from two different paralogs for each gene. Transcriptome quantitative analysis revealed expression variability among different genotypes. Specific primers were designed to amplify transcripts from different paralogs, to investigate the detected diversity. We explored and confirmed the expression variability in seven genetically divergent genotypes from our P. notatum collection. After determining the isoform-specific expression levels of SDP1 and PXA1 by qPCR and characterizing the lipid fraction content using FAME/GC/MS, we observed a negative correlation between the expression levels of PXA1.1, PXA1.2, and SDP1.1 with the 18:3 content, but only the PXA1.1 correlation resulted statistically significant. Surprisingly, SDP1.2 expression levels were positively correlated with both 18:3 and MUFAs, though the correlation with 18:3 was weak. Additionally, SDP1.2 showed a negative correlation with the 18:2 fractions, suggesting that this particular SDP1 paralog might be involved in increasing 18:2 at the expense of MUFAs and 18:3, possibly indicating an alternative physiological function. Among the candidate transcripts analysed, PXA1.1 exhibited the strongest negative correlation with 18:3 levels, consistent with the findings of Wijekoon et al. (2020). Based on qPCR and FAME/GC/MS experiments, genotype Q4188 emerged as the most promising candidate for further breeding projects. Crosses between Q4188 (sexual mother plant) and Q4117 (obligate apomict pollen donor) produced two hybrid F1 genotypes (JS9 and JS71) with high α-linolenic acid (18:3) content in their leaves and the ability for clonal reproduction via seeds. The level of apospory detected by cytoembryology in these hybrids supports their classification as obligate apomicts. However, in the next few years, the (epi)genetic stability of the successive generations of JS9 and JS71 should be assessed using molecular marker-assisted progeny tests, such as those based on MCSeEd (Marconi et al. 2019). The increased 18:3 levels in these hybrids were consistently confirmed through a series of experiments conducted over the period 2023/2024. This increase originated from converting MUFAs and 18:2 fatty acids into 18:3, while SFA levels were slightly elevated in these plants. This result was unexpected, as the mother plant Q4188 showed an increase in 18:3 at the expense of the SFA fraction, similar to findings by Wijekoon et al. (2020) in M. truncatula plants with reduced SDP1 or PXA1 levels. On average, the hybrid apomictic plants exhibited omega-6:omega-3 ratios that were 32.5% to 50.0% lower than those of the control, making them promising candidates for grazing trials aimed at improving nutritional quality. Additionally, the JS9 line showed a 25% increase in total lipid content in leaves, resulting in higher absolute levels of 18:3 available for hydrogen capture, which could potentially reduce methane (CH₄) emissions. However, this increase was lower than that reported in M. truncatula, where genetic engineering approaches led to an approximately 40% increase in total lipid content (Wijekoon et al. 2020). Further selection and/or crossing would be necessary to achieve a higher value. Regarding 18:3 content, while Wijekoon et al. (2020) reported increases of 21.61% and 23.36% compared to empty vector controls, the JS9 and JS71 genotypes in this study exhibited values of 12.7% and 28.4%, respectively. Considering the estimates reported by Moate et al. (2011), the 25% increase in total lipids detected in JS9 should result in a potential reduction of methane emissions of around 87.5%. Our findings suggest that transcriptional markers can be effectively used for rapid qPCR detection of P. notatum plants with high 18:3 levels. In particular, low expression of PXA1.1 paralogs correlates with higher omega-3 content at a statistically significant level. We hope that this work contributes to the development of superior forage grasses specifically adapted to the North-East rangelands of Argentina and other temperate regions, and to the enhancement of the nutritional quality of ruminant-derived products. We also aim to inspire other researchers in the apomixis field to establish similar breeding schemes. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (PDF 950 kb) Acknowledgements This research was funded by the European Union’s Horizon 2020 Research and Innovation Program under the Marie Skłodowska-Curie Grant Agreements 872417 (MAD) and 101007438 (POLYPLOID). LM, JPAO, LAS, MP, and SCP were supported by the National Agency for the Promotion of Research, Technological Development, and Innovation, Argentina (PICT-2017-01956 and PICT 2019-03414, PICT 2019-02153); CONICET, Argentina (Projects PUE 22920160100043CO and PIP 11220200101680CO); and National University of Rosario, Argentina (Project 80020190300021UR). We are grateful for the collaboration of Guillermo L. Marcuzzi and Gustavo E. Millán in providing GCMS services in the Mass Spectrometry Unit, UEM-CCT Rosario-CONICET. SA, LAS, MP, CA, JG, SCP, and JPAO are research staff members of CONICET, Argentina. Author Contribution statement LM, CMC, CA, and SCP contributed to the conception and design of the study. Material preparation, primer design, and qPCR analysis were performed by LM, CMC, DB, LAS, EA, and SCP. FAME/GC/MS was performed by SA. Cytoembryological analysis was conducted by JS and NS. LM and JG performed a total lipid content analysis. JPAO did data collection and statistical analysis. SCP wrote the first draft of the manuscript, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This research was funded by the European Union’s Horizon 2020 Research and Innovation Program under the Marie Skłodowska-Curie Grant Agreements 872417 (MAD) and 101007438 (POLYPLOID). LM, JPAO, LAS, MP, and SCP were supported by the National Agency for the Promotion of Research, Technological Development and Innovation, Argentina (PICT-2017–01956 and PICT 2019–03414, PICT 2019–02153); CONICET, Argentina (Projects PUE 22920160100043CO and PIP 11220200101680CO); and National University of Rosario, Argentina (Project 80020190300021UR). SA, LAS, MP, CA, JG, SCP, and JPAO are research staff members of CONICET, Argentina. Data availability The Transcriptome Shotgun Assembly project has been deposited at DDBJ/EMBL/GenBank under the accession DAWXEG000000000. The version described in this report is the first one, DAWXEG010000000. The plant material used in this study is available from Herbarium Carmen L. Cristobal, IBONE-CONICET-UNNE under the following voucher numbers: Q4117: CTES0541626; C4-4x: CTES0541627; R1: CTES0553130; Q3664: CTES0489592; Q4188: CTES0542168; R8: CTES0553130. The genotype Boyero-UNNE is available at request from the US National Plant Germplasm System Reg. No. CV-5, PI 676021. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Conflict of interests The authors have no relevant financial or non-financial interests to disclose. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary file1 (PDF 950 kb) Data Availability Statement The Transcriptome Shotgun Assembly project has been deposited at DDBJ/EMBL/GenBank under the accession DAWXEG000000000. The version described in this report is the first one, DAWXEG010000000. The plant material used in this study is available from Herbarium Carmen L. Cristobal, IBONE-CONICET-UNNE under the following voucher numbers: Q4117: CTES0541626; C4-4x: CTES0541627; R1: CTES0553130; Q3664: CTES0489592; Q4188: CTES0542168; R8: CTES0553130. The genotype Boyero-UNNE is available at request from the US National Plant Germplasm System Reg. No. CV-5, PI 676021. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.