Single-cell transcriptomics reveals lateral transfers of multiple functional genes from prokaryotes to free-living ciliated protists in detrital food webs Authors: Qianqian Zhang, Eleni Gentekaki, Michelle M Leger, Songbao Zou, Gong-Ao-Te Zhang, Atef Omar, Yingjun Fu, Jun Gong Published: 2026 ABSTRACT AbstractLateral gene transfer (LGT) is a key driver of evolutionary innovation, underlying protists’ lifestyles and interactions in anaerobic environments. Yet, its significance in free-living protists remains underexplored. Here, we address this gap by presenting the first single-cell transcriptomes of Metopus yantaiensis and genome-wide LGT screens across 36 omics datasets from nine anaerobic APM ciliates (classes Armophorea, Muranotrichea, and Parablepharismea)—a group in soil/sediment environments. Through phylogenetic analyses and validation testing, we identified 63 candidate prokaryotic LGT genes preferentially enriched in APM ciliates. Among these, 19 form interconnected pathways for degrading complex organics (polysaccharides, amino sugars); their high diversity and completeness are rarely seen in reported protist LGTs. A rare fused gene (arcC-OTC) and two novel genes (acs, ME2) were exclusively identified in APM ciliates, with their potential as the first evidence of LGT-mediated carbon metabolite retention and ammonia assimilation in phagotrophic protists inferred. Notably, 27 LGTs (including arcC-OTC, acs, and ME2) trace to candidate phyla radiation (CPR) bacteria or described prokaryotes, marking the first CPR-to-eukaryote LGT documentation. Collectively, these 63 LGTs are predicted to enhance nutrient utilization (complex organics, other carbon metabolites, inorganic elements), bioenergetic efficiency, and stress resistance (heavy metals, oxygen), facilitating soil/sediment adaptation. Overall, our results highlight lateral prokaryotic gene acquisition may be key for free-living anaerobic ciliates’ adaptation to new environments, shedding light on protists’ evolutionary dynamics and ecological roles.Supplementary InformationThe online version contains supplementary material available at 10.1007/s42995-026-00382-5.Keywords: Anaerobes, Candidate phyla radiation, Lateral gene transfer, Ciliophora, Biogeochemical cycle Introduction Lateral gene transfer (LGT) is increasingly recognized as a major driver of evolutionary novelty in eukaryotes (Husnik and McCutcheon 2018; Keeling 2009, 2024). LGT from prokaryotes to protists—predominantly single-celled eukaryotes that are neither animals, plants, nor fungi—represents an area of significant interest. This is due to the frequent biotic interactions and direct physical contact between these organisms and prokaryotes, including predation and endosymbiosis (Keeling 2009, 2024). Documented cases of gene transfer between prokaryotes and diverse protistan groups, including extremophiles (e.g., ice-adapted or thermophilic algae) (Dorrell et al. 2023; Raymond and Kim 2012), parasites (e.g., Giardia and Blastocystis) (Alsmark et al. 2013; Eme et al. 2017), and diatoms (Vancaester et al. 2020). While much research has focused on parasitic or symbiotic protists, relatively little is known about the identity and potential roles of LGTs in free-living, heterotrophic protistan species in natural environments (Vancaester et al. 2020; Xu et al. 2016). Soil microbiomes are pivotal in biogeochemical cycling of elements, producing and consuming greenhouse gases, such as CO2, CH4 and N2O, and maintaining the health of plants, animals and humans (Banerjee and van der Heijden 2023; Jansson and Hofmockel 2020). Protists are ubiquitous and abundant in soil, where they play well-recognized roles in bacterivory, food web dynamics, and nutrient remineralization (Geisen et al. 2015, 2018). Single-cell-omics approaches have emerged as powerful tools for elucidating the physiology, ecology, and evolution of protists (Gao et al. 2025; Kolisko et al. 2014; Lyu et al. 2024; Ma et al. 2025; Wang et al. 2026; Zhang et al. 2025b; Zheng et al. 2025). However, research on soil protists has lagged significantly behind that on their marine counterparts (Geisen et al. 2015, 2018). Specifically, transcriptomic data of free-living, heterotrophic soil-dwelling protist species remain scarce (Lax et al. 2018). Ciliates are among the most dominant and diversified protozoa in soil and sediment micro- and mesofauna, playing significant roles in the flow of matter and nutrients (Acosta-Mercado and Lynn 2004; Feng et al. 2025; Liu et al. 2025; Ma et al. 2025). In this study, we generated single-cell transcriptomes of Metopus yantaiensis, a member of a well-known ciliate genus that is abundant and widely distributed in soils (Foissner et al. 2002). Together with genome-wide screening across 36 omics datasets derived from another nine members of anaerobic APM ciliates (classes Armophorea, Muranotrichea, Parablepharismea), we identified 63 candidate LGT genes of prokaryotic/viral origin that are uniquely enriched in the APM ciliates. Our analyses revealed genes of bacterial, archaeal and viral origin involved in nutrient utilization and scavening, energy metabolisms, and stress resistance. These findings highlight the previously unrecognized biogeochemical potential of free-living protists in terrestrial and sedimentary ecosystems. Materials and methods Organism source Metopus yantaiensis was isolated from the soil of a coastal apple orchard in Yantai, Shandong, China (Omar et al. 2017). The upper layer of soil containing a small amount of litter was collected and air-dried. The dried soil sample was examined using the non-flooded Petri dish method described by Foissner et al. (2002). Dried, sterilized grass pieces were added and mixed to boost organic matter content and filtration capacity of the fine-grained soil. A few ciliate cells were picked with a micropipette, washed, and then inoculated into mineral water with sterilized chopped grass to establish initial cultures. Complementary DNA (cDNA) libraries and sequencing, DNA extraction Although Metopus yantaiensis is lab-culturable, isolating sufficient cells from soil media free of other microbial contaminants remains challenging for omics studies. We thus sequenced the transcriptomes of two single cells (hereafter C1 and C2) using a Smart-seq2 protocol for single-cell amplification (Kolisko et al. 2014; Picelli et al. 2014). Two individuals of average cell size were picked using a micropipette, washed three times with 0.22 μm-filtered mineral water, and then starved for 3 h to allow complete digestion of ingested prey. The generation of full-length cDNA and sequencing libraries is detailed in the Supplementary text and Table S1. For three additional washed, starved M. yantaiensis individuals, genomic DNA was extracted using the REDExtract-N-Amp Tissue PCR Kit (Sigma, St. Louis, MO) following Omar et al. (2017). This was to further validate LGT via polymerase chain reactions (PCR) and amplify the 16S rRNA gene of M. yantaiensis-associated bacteria for PacBio sequencing (see below). Assembly and completeness evaluation of the single cell transcriptomes Sequencing adapters were trimmed, and low-quality reads (N content > 10%, low quality bases > 50%) were filtered out using in-house scripts (Novogene, Beijing). The clean datasets (99.5% for C1, 99.08% for C2) were assembled using Trinity v2.12.0 with standard settings. To assess potential biases introduced by SCT and calculate the expression level (FPKM), Bowtie 2 (Langmead et al. 2009) and SAMtools v1.3 (Li et al. 2009) were used to map reads back to the assembly, both with the default settings. Unigenes with read counts lower than 2 were discarded (4993 for C1 and 8602 for C2). To remove potential contaminants and identify preliminary LGT candidate genes, unique genes were annotated via BLASTx against National Centers of Biotechnology Information (NCBI) nonredundant protein (nr) database (December 2024) with an e value cutoff of 1.00E-10. Sequences with high identity (≥ 80%) and coverage (≥ 50%) to non-protist organisms were regarded as contaminants and removed (1867 for C1, 930 for C2) (Kolisko et al. 2014). To evaluate the completeness of the SCTs, two reference datasets comprising 20 tRNA synthetases and 248 eukaryotic core proteins (http://korflab.ucdavis.edu/Datasets/genome_ completeness/) were used (Kolisko et al. 2014). In addition, previously published EST data from cultures of Nyctotherus ovalis, and its annotated mitochondrial protein datasets (Boxma et al. 2005; De Graaf et al. 2011) were used as references, given that M. yantaiensis is closely related to N. ovalis based on 18S rRNA gene phylogenies (Omar et al. 2017). Reciprocal BLASTx searches were used to identify unigenes of M. yantaiensis SCTs against these reference databases with an e value cutoff of 1.00E-10. Furthermore, BUSCO version 4.1 (Simão et al. 2015) was employed to assess the completeness of the SCTs using an Alveolata database (alveolata_odb10). Based on the qualified SCTs, we used a workflow to estimate the proportion of M. yantaiensis unigenes acquired via lateral gene transfer from prokaryotes (Fig. 1). Schematic overview of the analytical pipeline used to identify lateral gene transfers (LGTs) in the Metopus yantaiensis transcriptome. The number of unigenes retained after each filtering step is shown Primary screen for LGTs in M. yantaiensis and homologs in anaerobic ciliates The BLASTx returned 1140 unigenes whose top three hits were exclusively bacterial or archaeal; this set included the 360 unigenes predicted as contaminant from prokaryotes in the earlier qualifying step (Fig. 1). These 1140 unigens were selected as the primary LGT candidates derived from prokaryotes and were translated to proteins. Prior to translation, Augustus 3.4.0 (Stanke and Morgenstern 2005) was run on two M. yantaiensis SCT cDNAs to determine stop-codon usage bias at the end of the transcripts. Both transcripts preferred "UAA" (76.7% and 83.4%, respectively), mirroring the bias reported for Euplotes (Jin et al. 2023). The LGT candidates were, therefore, translated with ExPASy Translate (Gasteiger et al. 2003) using the Euplotes code 10; reading frames were taken from the highest scoring BLASTx alignment. To confirm that transferred genes are fixed within anaerobic ciliates phylogentically related to M. yantaiensis, we screened 36 recently published anaerobic ciliate transcriptomes and genomes (Fig. 1; Supplementary Table S2). This dataset includes nine species of the APM clade (classes Armophorea, Muranotrichea, Parablepharismea) plus three outgroup taxa (Trimyema finlayi, Plagiopyla frontata, and Cyclidium porcatum) (Lewis et al. 2020; Rotterová et al. 2020). Raw reads were re-assembled using Trinity v2.12.0/Metaspades v3.13.1 following the same settings as reported in the original publications (Lewis et al. 2020; Rotterová et al. 2020). Homologous sequences were searched by BLASTx of the M. yantaiensis LGT candidates against the dataset (e value ≤ 1E-10, coverage ≥ 50%). Transcripts were translated with ExPASy (Gasteiger et al. 2003) and genome scaffolds were annotated using Augustus 3.4.0 (Stanke and Morgenstern 2005). Both steps employed the Euplotes nuclear code 10 (Jin et al. 2023) on the basis of stop-codon usage conservation across the APM clade. Candidates lacking homologues in at least one anaerobic ciliate species were discarded; the remaining set was carried forward to phylogenetic analyses (Fig. 1). Phylogenetic analyses and phylogenetic tree interpretation To collect sufficient homologs for phylogenetic analysis, three independent BLASTp searches (December 2024 release) were performed against the NCBI-nr database using each M. yantaiensis LGT candidate protein as a query. For each BLASTp search, taxonomic filters were applied to restrict hits to Eukaryota, Bacteria, and Archaea separately (e value ≤ 1E-5, maximum 2000 hits per search). Additional tBLASTn searches were also performed against the NCBI nonredundant nucleotide (nt) database (December 2024) without taxonomic restrictions (e value ≤ 1E-5, maximum 100 hits). We also screened 21 aerobic ciliate transcriptomes (Supplementary Table S3) derived from the Marine Microbial Eukaryote Transcriptome Sequencing Project (MMETSP; Keeling et al. 2014) for ciliates’ homologs (e value ≤ 1E-5, maximum 1000 hits). Hits with > 50% sequence overlap were retained as reference sequences for all BLAST results. MMETSP transcripts and tBLASTn nucleotide hits were translated to proteins using ExPASy with ciliate code (6) and standard code (1), respectively. For phylogenetic tree construction, datasets were created by combining M. yantaiensis LGT proteins, the anaerobic ciliate homologs, and all the reference datasets. Similar protein sequences were removed to reduce the dataset to approximately 200–500 sequences using cd-hit (Li and Godzik 2006). The curated datasets were aligned with MAFFT version 7, and the output sequences were ordered by the aligning quality (Kuraku et al. 2013). The alignments were subsequently trimmed using BMGE-1.1 with a BLOSUM30 substitution matrix and a block size of 4 to remove ambiguously aligned regions (Criscuolo and Gribaldo 2010). The final alignments longer than 80 amino acids and comprising more than 30 reference sequences were retained for the further phylogenetic analysis. Maximum likelihood analysis was conducted using IQ-tree v1.4.4 (Nguyen et al. 2014) under substitution models LG + C60, LG + C20, or LG4X, the specific model parameters in each case was chosen by ModelFinder Plus according to the Bayesian Information Criterion (Supplementary Table S4). Branch supports were evaluated via ultrafast bootstrap approximation (1000 replicates; Hoang et al. 2018) and the SH-like approximative likelihood ratio test (SH-aLRT; 1000 replicates; Guindon et al. 2010). For each determined LGT, details of alignment and models used in phylogenetic analysis were listed in Supplementary Table S4. Following the IQ-TREE manual, bipartitions with UFboot ≥ 95% and SH-aLRT ≥ 80% were considered well-supported. A M. yantaiensis protein was inferred as a potential LGT from prokaryotes if it formed a robust clade with at least one other APM ciliate species and if: the APM clustered with prokaryotic sequences with high support; the sequences were composed primarily of prokaryotic sequences and sequences from members of the APM clade; and/or sequences from the APM clade clustered with sequences from distant eukaryotic lineages which in turn clustered with prokaryotes with high support (Fig. 1; Eme et al. 2017; Ricard et al. 2006; Xu et al. 2016). Validation testing for the LGTs To rule out residual prokaryotic contamination, we screened for eukaryotic or ciliate features in every M. yantaiensis and APM LGT-encoding transcripts/scaffolds (i.e., PolyA tails, introns, and eukaryote or ciliate affiliations of its flanking CDS) (Eme et al. 2017; Xu et al. 2016) (Fig. 1). Transcripts were scanned using a custom Python script (Additional data); only sequences ending in ≥ 6 consecutive adenines were scored as eukaryotic. Augustus 3.4.0 (Stanke and Morgenstern 2005) with the Euplotes code (Jin et al. 2023) was used to predict splice sites in genomic scaffolds; presence of canonical introns confirmed nuclear origin. Predicted CDS up- and downstream of each candidate were queried against NCBI nr (e value ≤ 1E-10); hits were required to affiliate with eukaryotes or, preferably, ciliates. These checks were applied twice: (1) during initial homologue retrieval to remove contaminants and (2) after phylogenetic confirmation to ensure monophyly of ciliate copies (Fig. 1). All phylogenetically confirmed LGT candidates were also screened for in-frame stop codons (UGA, UAA, and UAG) using ExPASy. Under the ciliate-/Euplotes-specific genetic code (Tables 6, 10), these codens encode amino acids rather than stop signals; thus their presence within open reading frames confirms the sequences represent genuine ciliate transcripts and not prokaryotic contamination. Finally, to exclude tree artefacts generated by mis-aligned non-homologous regions, we computed the Homology-derived Functional Similarity of Proteins (HFSP) score between the M. yantaiensis protein and the sequences forming its sister prokaryotic clade in the ML tree (Fig. 1; Mahlich et al. 2018; Rost 2002). Only candidates whose highest HFSP ≥ 5.0 were accepted as genuine prokaryote-to-ciliate LGTs (Alsmark et al. 2013). In addition, for seven key LGT candidates (acs, arcC-OTC, lacZ, man_1, man_4, ME2, and OTC), the presence of the genes were also validated by performing Polymerase Chain Reaction (PCR) and Sanger sequencing on single-cell genomic DNA of M. yantaiensis (starved and washed cells, see above; Fig. 1). Mapping LGTs onto metabolic pathways Annotations against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for the M. yantaiensis LGT genes were retrieved with BLAST–KOALA (Kanehisa et al. 2016) and KofamKOALA (Aramaki et al. 2020), using the predefined Kofam thresholds and an e value threshold of 0.01 (Fig. 1). The functional assignment was complemented with the annotation for the proteins by InterPro (version 101.0) against the InterPro database consortium (Blum et al. 2025). Cellular localizations of the LGT proteins were predicted using TargetP 1.1 (Emanuelsson et al. 2000). PacBio sequencing for 16S rRNA gene of bacteria associated with M. yantaiensis To compare the LGTs’ prokaryotic donors and the current bacteria associated with M. yantaiensis, we performed single-cell Pac-Bio sequencing on the genomic DNA of 3 ciliate cells. The nearly full-length 16S rRNA gene was amplified and sequenced using using PacBIo II HiFi Circular Consensus Sequencing (CCS). The amplification, sequencing and data processing methods for the 16S rRNA gene are detailed in the Supplementary text. Results General features of the M. yantaiensis single-cell transcriptome datasets After filtering out low-quality and short reads (< 200 bp), a total of 14,095,201 (4.23 Gb) and 16,819,304 reads (5.05 Gb) were retained for C1 and C2, respectively. Assembly using Trinity yielded 26,038 contigs for C1 and 29,484 contigs for C2 (Supplementary Table S5). The two single-cell transcriptomes (SCTs) showed low sequencing bias (Supplementary Fig. S1A), and high reproducibility, with similar distributions of gene lengths and GC contents (Supplementary Fig. S1B, C). They shared 76% of the contigs (Supplementary Fig. S1D), and the successfully annotated transcripts exhibit a high degree of consistency in terms of species classification (Supplementary Fig. S2A) and functional classification (gene ontology and KEGG assignments) (Supplementary Fig. S2B, C; Table S7). Both SCTs exhibited high integrity and low contamination levels (Supplementary Table S6). Gene recovery rates of the SCTs were 100% against the aa-tRNA synthetases dataset, 87% and 88% against the 248 core-eukaryotic-gene set, 79% and 71% against the BUSCO Alveolata core genes, and 90% against the Nyctotherus ovalis annotated proteins (Supplementary Table S6). Collectively, we believe that the data from these two cells can represent the main profile of the species transcriptome. Detailed information on GC content, contig length, contaminant levels and completeness of the two SCTs was provided in the Supplementary text, Tables S5–S7, and Supplementary Figs. S1, S2. The number, distribution in APM taxa, and multi-faceted identification evidence of lateral transferred genes (LGT) A total of 63 genes of M. yantaiensis were identified as high-confidence prokaryote-derived LGT candidates (Table 1; Fig. 2). Of these, 62 out of 63 formed robust clades with prokaryotic/viral sequences in phylogenetic trees, or composed primarily of prokaryotes/viruses and the APM clade (Fig. 2; Supplementary Table S4; Additional data). The remaining gene (lip) exhibited high-confidence BLASTp and HFSP matches to prokaryotic sequences but lacked sufficient homologs (< 20) for phylogenetic analysis. All 63 LGT genes were substantially expressed in M. yantaiensis (8–1232 FPKM; Fig. 1) and seven key LGT candidates (acs, arcC-OTC, lacZ, man_1, man_4, ME2, and OTC) were successfully amplified on the M. yantaiensis single-cell genomes (accession numbers: OR730783–OR730792; Supplementary Table S4). All 63 genes had homologs in ≥ 2 APM species, while 15 were conserved in ciliates outside the APM clade (Fig. 2). Most genes exhibited clear eukaryotic characteristics in M. yantaiensis transcripts or homologous APM transcripts/genome scaffolds: 56 genes contained poly(A) tails, and 60 were either flanked by ciliate/eukaryotic genes or harbored introns (Fig. 2; Supplementary Fig. S3, Table S4). In addition, 52 genes contained in-frame stop codons specific to ciliate or Euplotes (UGA/UAA/UAG) in their corresponding transcripts (Fig. 2; Supplementary Table S4). All genes showed high HFSP similarity (≥ 5) to members of prokaryotic sister groups in the trees (Additional data). These evidences rule out the possibilities that the aforementioned LGT genes were random sequencing artifacts, bacterial contamination, or examples of orthology misassignment. Table 1Enzyme names and abbreviations for the laterally transferred gene candidates identified in Metopus yantaiensis and other APM ciliates. The numbers in the parentheses represent the number of copies discovered for the same geneAnnotated enzyme name/[EC]Gene abbreviationML phylogenyaAnnotated enzyme name/[EC]Gene abbreviationML phylogenyaDegradation of complex carbohydrateCarbon and energy metabolismPolysaccharides degradationglucokinase [EC: 2.7.1.2]glkAddmannan endo-1,4-beta-mannosidase [EC: 3.2.1.78] gene 1 (2)man_1Fig. S4Cacetyl-CoA synthetase [EC: 6.2.1.1] (2)acs, ACSS1_2Fig. 3mannan endo-1,4-beta-mannosidase [EC: 3.2.1.78] gene 2 (2)man_2Add.malate dehydrogenase (oxaloacetate-decarboxylating) [EC: 1.1.1.38]ME2, sfcA, maeAFig. 4mannan endo-1,4-beta-mannosidase [EC: 3.2.1.78] gene 3man_3Adddiphosphate-dependent phosphofructokinase [EC: 2.7.1.90]PFK,pfkAAddmannan endo-1,4-beta-mannosidase [EC: 3.2.1.78] gene 4man_4Fig. S4Dpyruvate, orthophosphate dikinase [EC: 2.7.9.1] (5)ppdKFig 5glycoprotein endo-alpha-1,2-mannosidase [EC: 3.2.1.130]MANEAAddsuccinate-semialdehyde dehydrogenase [EC: 1.2.1.16/24]sadAddbeta-galactosidase [EC: 3.2.1.23] gene 1lacZFig. S4Aciliate FeFe hydrogenase 51 Kda/NuoF-like domainHoxUFig. S9Bbeta-galactosidase [EC: 3.2.1.23] gene 2bgaB, lacAAddciliate FeFe hydrogenase 24 Kda/NuoE-like domain (2)HoxFAddarabinan endo-1,5-alpha-L-arabinosidase [EC: 3.2.1.99]abnAFig. S4BNitrogen metabolism/amino acid metabolismL-fucose/D-arabinose isomerase [EC: 5.3.1.25 5.3.1.3]fucIAddNitrogen assimilation and absorption4-alpha-glucanotransferase [EC: 2.4.1.25]malQAddammonium transporter/Amt familyamt, AMT, MEPAddglucoamylase-like [PF10091]GAMAddcarbamate kinase [EC: 2.7.2.2]-ornithine carbamoyltransferase [EC: 2.1.3.3]arcC-OTCFig. 6DUF4434 domain proteinDUF4434AddIndependent carbamate kinase [EC: 2.7.2.2]arcCFig. 6AAmino sugar degradationindependent ornithine carbamoyl transferase [EC: 2.1.3.3]OTCFig. 6Bserine-type D-Ala-D-Ala carboxypeptidase [EC: 3.4.16.4] (2)pbp4bFig. S5Bhydroxylamine reductase [EC: 1.7.99.1]hcpAddexo-beta-N-acetylmuramidase NamZ [PF07075]NamZAddAmino acid degradationalpha-N-acetylglucosaminidase [EC: 3.2.1.50]NAGLUAdddiaminopimelate decarboxylase [EC: 4.1.1.20]lysAAddanhydro-N-acetylmuramic acid kinase [EC: 2.7.1.170]anmKFig. S8histidine ammonia-lyase [EC: 4.3.1.3] (3)HALAddglucosamine-6-phosphate deaminase [EC: 3.5.99.6]-N-acetylglucosaminylphosphatidylinositol deacetylase [EC: 3.5.1.89] (2)nagB-PIGLFig. S5Aornithine cyclodeaminase [EC: 4.3.1.12]OCDFig. S8N-acetylmuramic acid 6-phosphate etherase [EC: 4.2.1.126]murQAddbeta-ureidopropionase/N-carbamoyl-L-amino-acid hydrolase [EC: 3.5.1.6/87]pydCAddN-acylglucosamine 2-epimerase [EC: 5.1.3.8]RENBPFig. S6Phosphate and sulfur metabolismNucleotidase/ribozyme recyclephospholipase D1/2 [EC: 3.1.4.4]PLD1_2Add5'-nucleotidase/UDP-sugar diphosphatase [EC: 3.1.3.5 3.6.1.45] (2)ushAFig. S8D-cysteine desulfhydrase [EC: 4.4.1.15]dcyDAddYjgF/chorismate_mutase-like [PF14588]YjgFFig. S8Lipid metabolism and cell membrane stabilityEndonuclease/Exonuclease/phosphatase family [PF03372]EndonucleaseAddSqualene-tetrahymanol cyclase [EC: 4.2.1.123]STCAddadenine deaminase [EC: 3.5.4.2]adeAddtriacylglycerol lipase [EC: 3.1.1.3]lip, TGL2Napurine-nucleoside phosphorylase [EC: 2.4.2.1]punAAddphosphatidylethanolamine/phosphatidyl-N-methylethanolamine N-methyltransferase [EC: 2.1.1.17 2.1.1.71]pmtAAddProtease/PeptidaseStress resistance, detoxification and deoxygenL-asparaginase/beta-aspartyl-peptidase [EC: 3.5.1.1/3.4.19.5]iaaA, ASRGL1Addpeptide-methionine (S)-S-oxide reductase [EC: 1.8.4.11]msrAFig. S9Apeptidyl-dipeptidase A [EC: 3.4.15.1] (2)ACEAddsigma intracellular receptor 2TMEM97Addaminopeptidase N [EC: 3.4.11.2]pepNAddRNA ligase (2)Rnl2Addcopper homeostasis proteincutCAddFAD-dependent oxidreductaseoxidreductaseAddarsenite methyltransferase [EC: 2.1.1.137] (2)AS3MTAddlactoylglutathione lyase [EC: 4.4.1.5]GLO1, gloAAddNAD(P)H dehydrogenase (quinone) [EC: 1.6.5.2] gene 1 (2)NQO1_1Addmonothiol glutaredoxingrxD, GLRX5AddNAD(P)H dehydrogenase (quinone) [EC: 1.6.5.2] gene 2 (5)NOQ1_2AddCOMPASS component SWD5WDR5AddNAD(P)H dehydrogenase (quinone) [EC: 1.6.5.2] gene 3 (2)NOQ1_3Addprolycopene isomerase [EC: 5.2.1.13]crtISOAddribosyldihydronicotinamide dehydrogenase (quinone) [EC: 1.10.5.1]NQO2AddaThe ML phylogeny of each gene is shown in the corresponding figure (Fig.) and additional trees figures (Add.) Enzyme names and abbreviations for the laterally transferred gene candidates identified in Metopus yantaiensis and other APM ciliates. The numbers in the parentheses represent the number of copies discovered for the same gene aThe ML phylogeny of each gene is shown in the corresponding figure (Fig.) and additional trees figures (Add.) Functional annotation, gene features and expression of the lateral gene transfer (LGT) candidates in Metopus yantaiensis and their distribution across the Armophorea-Muranotrichea-Parablepharismea (APM) group and other ciliate classes. Only ciliate homologues sharing a single prokaryotic origin with the M. yantaiensis LGT are plotted. MY: Metopus yantaiensis. Gene abbreviations are defined in Table 1. Expression (FPKM) is the higher value observed in the two MY single-cell transcriptomes. Values in parentheses indicate multi-copy genes whose FPKM is the sum of all copies Functional annotation Overview of functional annotation results The 63 identified LGT genes can be categorized into the following functional categories: complex carbohydrate degradation (19 genes), protein/nucleotide hydrolysis and salvage (eight genes), carbon/energy metabolism (eight genes/domains), nitrogen/amino acid metabolism (nine genes), phosphate/sulfur metabolism, lipid metabolism, membrane stability, and detoxification/oxidative stress responses (19 genes) (Table 1). Among these, four genes/domains (grxD, ciliate FeFe hydrogenase domains HoxU and HoxF, and sad) are predicted to be targeted to hydrogenosomes (Supplementary Table S8). Other LGT candidates are considered to encode cytoplasmic proteins due to the absence of amino acid signals for mitochondria and hydrogenosomes. Below, we list the genes that deserve attention in protists and even across the entire eukaryote lineage. Genes related to complex carbohydrate decomposition and proteins/nucleotides salvage A total of 27 prokaryotic LGT genes were predicted to participate in complex carbohydrate degradation (of polysaccharides and amino acid sugars) and protein/nucleotide salvage, with 25 of them widespread more than three APM species (Table 1; Fig. 2). Among these, ten genes (man_1–4, MANEA, lacZ, bgaB, malQ, abnA and GAM) encode glycoside hydrolases involved in polysaccharide degradation, including hemicellulose (mannan), starch and other glycans (Table 1; Supplementary Fig. S4). Seven LGT candidates (pbp4b, NamZ, anmK, NAGLU, nagB-PIGL, murQ, and RENBP) were predicted to participate in amino sugar degradation (Table 1; Supplementary Figs. S5, S6). Particularly, 6 genes showed isozymes (man_1–4, lacZ and bgaB), while 4 genes had multiple copies (man_1, man_2, pbp4b, and nagB-PIGL) (Table 1; Fig. 2). In addition, identified LGT genes included three proteases (iaaA, ACE, and pepN) and five nucleic acid/ribose metabolic genes (ushA, YjgF, endonuclease, ade, and punA) (Table 1). Novel genes related to carbon and energy metabolism Eight laterally acquired genes/domains (acs, ME2, glk, PFK, ppdK, sad, HoxU, and HoxF) are predicted to be involved in carbon and energy metabolism (Table 1; Fig. 2). Among these, two genes—acs and ME2—are novel acquisitions in eukaryotes. To the best of our knowledge, this is the first report of the transfer of this type of prokaryotic AMP-forming ACS gene (EC 6.2.1.1) in eukaryotes (though genes encoding other, distantly paralogous ADP-forming ACS enzymes have been reported to be transferred into other anaerobic eukaryotes Jerlström-Hultqvist et al. 2013; Leger et al. 2017; Sánchez et al. 2000). Transfers of ME2 have only been recorded in the parasites Trichomonas and Entamoeba (Doležal et al. 2004; Field et al. 2000). Phylogenetic analyses revealed distinct evolutionary trajectories of the two genes to the APM group (Figs. 3, 4). The APM acs gene (hereafter shorten as apm-acs) is distinct from the classical eukaryotic acetyl-CoA synthetase and was transferred from a novel prokaryotic clade (full supports; Fig. 3), composed predominantly of environmentally derived archaeal lineages (e.g., Bathyarchaeota, Pacearchaeota, Odinarchaeota, Hadesarchaea, and MSBL1) and CPR taxa (e.g., Parcubacteria, Nealsonbacteria, and Wildermuthbacteria). The APM ME2 gene (hereafter shorten as apm-ME2) formed a clade separate from other eukaryotes (e.g., Trichomonas, Entamoeba, and other ciliates) and was transferred from a newly classified archaeal lineage, Candidatus Woesearchaeota (full ultrafast bootstrap support; Fig. 4). Another LGT—the ppdK gene—is particular notable, because it maintains five homologous copies in M. yantaiensis and specific APM taxa (Fig. 5). The origin of the APM ppdK gene (hereafter shorten as apm-ppdK) is distinct from those of other eukaryotes, and was transferred from novel prokaryotic clades, composed predominantly of environmentally derived archaeal phyla (e.g., Ca. Woesearchaeota, Ca. Aenigmarchaeota and Ca. Nanoarchaeia) and bacterial MAGs (full supports; Fig. 5). Maximum-likelihood phylogenies of the gene encoding AMP-forming acetyl-CoA synthetase (acs) specifically detected in the APM group (apm-acs), constructed under the LG + C60 + F + I + R9 model and based on 477 amino acids, with corresponding genomic scaffold depictions. Numbers in parentheses indicate collapsed homologs; key SH-aLRT/UltraFast bootstrap support values are shown. Filled circles on the nodes mark bipartitions with SH-aLRT > 95% and ultrafast BS ≥ 95%. The full protein tree is in online Additional data. Suffixes "T", "G", and "Amplified" after APM species denote transcriptome, genome, and PCR-amplified sequences, respectively. In scaffold sketches, predicted genes are color-coded; bold lines indicate introns. Letters "C", "E", "B", and "N" above genes denote affiliation to ciliates, eukaryotes, bacteria, and no significant NCBI nr homolog, respectively Maximum-likelihood phylogenies of the gene encoding the small malic enzyme (ME2), based on 357 amino acids and constructed under the LG + C60 + I + R8 model, along with depictions of the corresponding genomic scaffolds. Maximum likelihood topologies include key SH-aLRT bootstrap support values followed by UltraFast bootstrap support values. The legend from Fig. 3 applies Maximum-likelihood phylogenies of the gene encoding pyruvate phosphate dikinase (apm-ppdK) based on 703 amino acids and constructed under the LG + C20 + F + I + R10 model, and depictions of the corresponding genomic scaffolds. Maximum likelihood topologies with key SH-aLRT bootstrap support followed by UltraFast bootstrap support values. The legend from Fig. 3 applies Novel genes related to nitrogen metabolism Nine of the identified LGT genes are associated with nitrogen metabolism (amt, arcC-OTC, arcC, OTC, and hcp) and amino acid degradation (lysA, HAL, OCD, and pydC) (Table 1; Fig. 2). Among these, the arcC-OTC gene, composed of a carbamate kinase (arcC; EC: 2.7.2.2) domain and an ornithine carbamoyltransferase (OTC; EC: 2.1.3.3) domain, is particularly rare across the entire domain of life (Fig. 6C). No significantly similar sequences existed in the NCBI nr database, and similar gene arrangement was only found in the operons of some Enterobacter and Gammaproteobacteria bacteria, i.e., arcC and argF (isozyme of OTC; Fig. 6C). Notably, independent OTC and arcC genes were simultaneously identified among the LGT genes of M. yantaiensis or other APM members (Fig. 2). Phylogenetic analyses on the arcC and OTC fragments were conducted separately to trace the evolutionary origin of this arcC-OTC fusion gene (Fig. 6A, B). The results indicated that the closest neighbors of the arcC and OTC fragments in the fusion gene were not identical: the APM arcC originated from Cloacimonetes bacteria (Fig. 6A; full support), while the OTC clustered with a large proteobacterial clade (Fig. 6B; full support). Neither gene was related to known eukaryotic arcC and OTC genes, nor to the aforementioned arcC/argF genes in the Enterobacter and Gammaproteobacteria operons (Fig. 6A, B). A, B Maximum-likelihood phylogenies of the genes encoding carbamae kinase (arcC) and ornithine carbamoyltransferase (OTC), showing the phylogenetic positions of fused/non-fused arcC and OTC in APM ciliates, with corresponding genomic scaffold depictions. The arcC phylogeny (215 amino acids) was constructed under LG + C60 + R8, and OTC phylogeny (222 amino acids) under LG + C60 + I + R9; key SH-aLRT/UltraFast bootstrap support values are shown. The legend from Fig. 3 applies unless specified otherwise. Purple stars after APM species denote regions from fused genes. Red stars after the bacterial clades indicate the location of the known bacterial arcC and argF (OTC) operon genes. Arrow in (B) indicates the position of Ca. Cloacimonetes. C Sketches of the independent and fused LGT genes encoding APM arcC and OTC, and the known arc operons from Enterobacter species (NCBI accession no. CP077290.1) and Salinivirga cyanobacteriivorans (NCBI accession no. CP013118.1); Arrows indicate putative open reading frames (ORFs); Lines indicate noncoding regions; dotted line represents the presence of other operons between the arcA and arcC. Abbreviations in the bacterial operons: arcA, gene encoding arginine deiminase (ADI); arcC, gene encoding carbamate kinase; arcF, gene encoding ornithine transcarbamoylase (OTC) Genes involved in other pathways Two LGT genes in M. yantaiensis and the APM group (dcyD and PLD1_2) are implicated in sulfur and phosphorus metabolism, respectively (Table 1; Fig. 2). We additionally identified 17 LGT genes associated with diverse cellular processes, including: (i) lipid metabolism and membrane stabilization (lip, STC, and pmtA); (ii) signal transduction and RNA repair (TMEM97 and Rnl2); (iii) redox homeostasis and oxidative stress response (MsrA, NQO1_1–3, NQO2, oxidreductase, GLO1, grxD, crtI and crtISO); and (iv) heavy metal detoxification (AS3MA and cutC) (Table 1; Fig. 2). Notably, a substantial proportion of these genes (especially those related to lipid metabolism and heavy metal resistance) appear to be absent from the gut symbiont Nyctotherus ovalis (Fig. 2). The prokaryotic donors of the LGTs of the APM ciliates Phylogenetic analyses and BLASTp results for lip confirmed prokaryotic/viral origins for the 63 LGT candidates: 33 from common bacteria, 1 from a virus, and 27 from metagenome-assembled genomes (MAGs) of newly classified bacterial/archaeal taxa derived from environmental sequences (Fig. 2; Supplementary Table S4; Additional data). Notably, these MAGs are predominantly affiliated with Candidate Phyla Radiation (CPR) prokaryotes and other recently identified "Microbial Dark Matter" groups—largely uncultured microbial lineages recognized primarily via molecular data (e.g., Figs. 3–6; Supplementary Figs. S8, S9) (Brown et al. 2015; Rinke et al. 2013). The candidate donors of the APM LGTs are taxonomically distinct from dominant associated bacteria (Sphingomonas, Streptococcus) detected in M. yantaiensis single cells (Supplementary Fig. S7; Table S9; Omar et al. 2017) and share no genus-level affiliations with 360 potential contaminant prokaryotic mRNA sequences (Supplementary Fig. S7). This suggests the LGT donors are unlikely to be current endosymbionts of M. yantaiensis, but rather relics of ancestral symbionts (Husnik and McCutcheon 2016) or past prokaryotic prey (Keeling 2024). Discussion Through rigorous phylogenetic and multi-faceted genomic analyses, we identified 63 prokaryote/virus-derived lateral transferred genes (LGTs) in APM ciliates, with several genes and their evolutionary trajectories representing rare cases in eukaryote. If functionally fixed, these LGTs would likely confer significant adaptive advantages for harsh environments or novel roles in detrital ecosystems. Functional validation of LGTs via heterologous protein expression, gene editing, or RNA interference (RNAi) is limited to select model eukaryotic organisms, e.g., oomycetes and insects (Attah et al. 2024; Fang and Tyler 2016; Zhang et al. 2025a). For Metopus yantaiensis and most APM taxa, axenic cultures cannot be established—precluding validation of the LGTs identified here via these approaches, and thus leaving the open possibility of their silent or pseudogenic status. However, their high expression in transcriptomes of M. yantaiensis and other APMs, conservation across APM lineages, and presence as multiple copies (2–5) (Fig. 2; Table 1) strongly suggest that these LGTs are likely functional (or have been functional) in APM ciliates. Under an assumption of their functionality, we therefore infer potential physiological and ecological roles conferred by these LGTs to APM ciliates (Figs. 7, 8), aiming to provide a framework for future functional validation (as technological capabilities advance) and the refinement of ecological models. Predicted cellular processes of complex carbonate degrading (A) and carbon and energy metabolism (B) affected by lateral gene transfer (LGT) in Metopus yantaiensis cells. Cell membranes and internal organelles are delineated with solid lines, while distinct pathways are represented by color blocks. Enzyme names were indicated adjacent to arrows. Gene names and arrows shown in red indicate LGT-derived genes, whereas those in black indicate ancestrally inherited orthologs. Dotted arrows denote multiple-step reactions or processes. Full names of LGT-derived genes are listed in Table 1. Full names of the metabolite abbreviations and additional enzymes: AnhMurNAc, Anhydro-N-acetylmuramic acid; GlcNAc, N-acetylglucosamine; ManNac, N-Acetylmannosamine; HydA: Hydrogenase A; MDH, Malate Dehydrogenase Predicted cellular processes of nitrogen, sulfur, and phosphate metabolism (A) affected by lateral gene transfer (LGT) in Metopus yantaiensis cells and the contribution of LGT on detoxification and deoxygenation (B). Cell membranes are delineated with solid lines, while distinct pathways are represented by color blocks. Enzyme names are indicated adjacent to arrows. Enzymes derived from LGT are labeled in red. Dotted arrows denote multiple-step reactions or processes. Full names of the LGT-derived genes are listed in Table 1. Full names of metabolite abbreviations and additional enzymes: L-DAL, L-2,4-diaminobutyric acid; PC, Phosphatidylcholine; SAH, S-Adenosylhomocysteine; SAM, S-Adenosylmethionine Potential significance of the complex organic matter degradation gene transfers in free-living anaerobic ciliates Of the 63 lateral transferred genes, 19 are involved in the degradation of polysaccharides (hemicellose and other glycans) and amino sugars (Table 1); their high diversity and completeness are rarely seen in reported protist LGTs (Fig. 7A). Several of these genes possess isoenzymes (i.e., man_1–4, lacZ and bgaB) and multiple copy numbers (i.e., man_1, man_2, pbp4b, and nagB-PIGL) (Table 1; Fig. 2), implying the importance of their potential functions. Previously, cellulase/hemicellulase LGTs have primarily been reported in rumen ciliates/saprotrophic protists (Ricard et al. 2006; Savory et al. 2015), with sporadic occurrences in free-living bacterivores (e.g., Mastigamoeba and Pelomyxa) (Treitli et al. 2023; Verner et al. 2021); several of these recipient lineages commonly possess prokaryotic symbionts, suggesting symbiosis as an important source of these LGTs. For the APM taxa, the aforementioned genes may expand the range of carbohydrates that can be utilized in soil/sediments (Fig. 7A), as reported in rumen ciliates (Ricard et al. 2006). The polysaccharide-degrading enzymes may enable the ciliates to decompose polysaccharides (e.g., hemicellulose) in plant residues and algal cell walls (Fig. 7A). Amino sugars are the main components of peptidoglycan, and the amino sugar-degrading enzymes acquired by APM taxa (e.g., NamZ, nagB-PIG-L and murQ) constitute a complete pathway for peptidoglycan degradation (Fig. 7A). The acquisition of these enzymes may enable APM taxa to decompose biological residues, such as bacterial cell walls and fungal cell walls in soils and sediments (Joergensen 2018; Konopka 2012). Amino sugar degrading LGT have only occasionally been reported in protists, mainly in parasites, where they facilitated host invasion or disruption of surface glucosamine antigens (Alsmark et al. 2013; Andersson 2006; Eme et al. 2017). Most interestingly, genes involved in glycan sensing and processing have been implicated in control of algal symbionts in the ciliate Paramecium bursaria; as in our ciliate data, these include several laterally transferred genes of bacterial origin, including NamZ and several beta-mannosidases (Jenkins et al. 2024). Our phylogeny of one amino sugar degrading gene, RENBP, suggests a bacterial symbiont origin for this gene in APM ciliates (Supplementary Fig. S6), raising the possibility that host–symbiont interactions might also have acted as an initial selective pressure in the acquisition of this pathway in this clade. Finally, identified LGT genes included three proteases (iaaA, ACE, and pepN) and five nucleic acid/ribose metabolic genes (ushA, YjgF, endonuclease, ade, punA) (Table 1; Fig. 7A). These genes primarily function in protein catabolism or nucleic acid/ribose salvage—recovering DNA/RNA/ribose components for nucleotide, phosphate, and sugar recycling (Fig. 7A). Synergizing with polysaccharide/amino sugar-degrading LGTs, these elements may collectively enhance nutrient availability for nitrogen, carbon, and nucleotide metabolism (Fig. 7A). Predicted functions of the novel carbon metabolism LGTs in transfer to anaerobic ciliates Among the 63 newly identified lateral gene transfers in APM taxa, we revealed several novel LGT events in eukaryotes. Notably, two carbon metabolism-related genes (apm-acs and apm-ME2) are rare among eukaryotes, while the apm-ppdK gene is distinctive as it exhibits five copies in APM taxa. Functionally, these novel genes may be involved in the recycling of carbon metabolic products and the enhancement of energy production efficiency (Fig. 7B). Enzymes encoded by acs and ME2 can utilize malate, acetate to produce key substrates (acetyl-CoA and NADPH) for lipid biosynthesis (Doležal et al. 2004; Starai and Escalante-Semerena 2004; Stegink and Coon 1968). Malate and acetate are glycolysis and fermentation intermediate products of anaerobic ciliates and are typically excreted to be utilized by environmental prokaryotes (Müller et al. 2012). In M. yantaiensis and other APM members, these ciliates may have acquired the aforementioned genes from prokaryotes to recycle anaerobic respiration metabolites (Fig. 7B). This mechanism could reduce both the loss of carbon sources and the demand for external carbon sources of the anaerobic ciliates. A similar mechanism has been reported in Trichomonas vaginalis, which acquired a ME2 gene in the cytoplasm to utilize malate and produce additional NADPH for biosynthesis (Doležal et al. 2004). In addition, the carbon metabolism-related apm-ppdK gene exhibits five copies (Fig. 5), implying import function. Along with another gene PFK, these gene acquisitions have been reported in other anaerobic protist (oxymonads and parabasalia) to enhance the efficiency of ATP generation during glycolysis, compensating for the low metabolic yield inherent in anaerobic metabolism (Slamovits and Keeling 2006). This mechanism may similarly apply to the anaerobic APM ciliates (Fig. 7B). Potential functions of the arcC-OTC fusion gene and other nitrogen-metabolism-related LGTs The nitrogen-metabolism-related fusion LGT gene arcC-OTC identified in APM ciliates had no homologous sequences across the entire domain of life. Functionally, this rare gene may be involved in the utilization of inorganic nitrogen sources by anaerobic ciliates (Fig. 8A). The fused arcC-OTC gene architecture suggests its involvement in dual-functional metabolic pathways. One such pathway, the arginine deiminase pathway (ADH) —an energy-generating route (Biagini et al. 2003; Linstead and Cranshaw 1983; Schofield et al. 1992) was excluded, because APM ciliates’ omics lack the key gene encoding arginine deiminase (arcA/ADI), the inverted gene order (argF/OTC upstream of arcC; Fig. 6C) and the phylogenetically inferred divergence (Fig. 6A, B). Instead, similarities to the ornithine–urea cycle (OUC) of diatoms were identified (Allen et al. 2011; Armbrust et al. 2004). In the diatom OUC, LGT-derived arcC and OTC (upstream arrangement) utilize ammonia nitrogen and carbonate to synthesize carbamoyl phosphate, a key substrate for the synthesis of nitrogen-containing components (Allen et al. 2011; Fig. 7A). Other prokaryote-derived nitrogen metabolic LGTs (i.e., OCD and hcp) were also identified in both APM ciliates and diatoms (Allen et al. 2011; Armbrust et al. 2004). In this context, APM’s arcC-OTC fusion, along with other LGTs (e.g., OCD, hcp), likely mirrors this OUC-like functionality, facilitate inorganic nitrogen assimilation and carbon–nitrogen homeostasis (Allen et al. 2011) (Fig. 8A). These underscore convergent evolution of prokaryote-derived nitrogen metabolism in disparate eukaryotic lineages. Beyond arcC-OTC, the LGT genes identified in the APM ciliates also include five other genes involved in nitrogen metabolism and amino acid metabolism (Table 1; Fig. 8A). The functions of these genes can be predicted from existing information: 1) ammonia uptake (amt) (Soupene et al. 2002); 2) biosynthesis of essential amino acids (lysA) (Stragier et al. 1983); 3) interconversion of complex amides with ammonia, CO₂, or carbon backbones (HAL); 4) synthesize complex amino acids and polypeptides (e.g., proline and spermine) essential for protein synthesis, bioenergetics, and stress resistance (OCD) (Christgen and Becker 2019); and 5) nitric oxide-resistant and regulating fermentative and nitrogen utilization pathways (hcp) (Wang et al. 2016). These genes, in synergy with arcC-OTC, if functional in the APM group, may assist the ciliates in adapting to nitrogen source deficiency or other environmental stresses through diversifying amino acid metabolic flux (Fig. 8A). Potential contributions of other LGT genes to soil/sedimental environmental adaption Beyond the aforementioned metabolic pathways, the lateral gene transfers (LGTs) identified in APM ciliates also include two genes involved in phosphorate and sulfur metabolism (Table 1; Fig. 8A). The known functions include: utilization of D-sulfur-containing amino acids other than L-amino acids (dcyD) (Soutourina et al. 2001); decomposition of phospholipids to recycle phosphate groups, which supports cell proliferation, cytoskeleton remodeling, and stress resistance (PLD1_2) (Christeller and Tolbert 1978; Jenkins and Frohman 2005). In addition, 19 genes/domains are potentially involved in adaptive processes (Fig. 8B): lipid catabolism and utilization (lip and STC), membrane stability (pmtA), signal transduction (pmtA), and rRNA repair under stress (Rnl2) (Banta et al. 2015; Bexkens et al. 2018; Salsabila and Kim 2024; Shen et al. 2021), oxygen tolerance (HoxU, HoxF, MsrA, NQO1_1-3, NQO2, oxidoreductase, GLO1, grxD, and crtISO) (de Koning et al. 2000; Dobri et al. 2014; Rotterová et al. 2020), and heavy-metal resistance (AS3MA and cutC) (Chen et al. 2017; Ge et al. 2022). Together with the aforementioned novel genes, these genes may enable the APM ciliate group to acquire novel functions for adapting to harsh anaerobic soil and sediment environments, such as nutrient deficiency, heavy metal stress, and oxygen fluctuation (Figs. 7, 8). Notably, most of these genes—particularly those involved in lipid metabolism and heavy-metal resistance—are absent from the gut symbiont Nyctotherus ovalis (Fig. 2). This may reflect secondary gene loss in N. ovalis, driven by reduced oxygen and heavy-metal exposure in gut environments. This study presents the first systematic analysis of LGTs in free-living ciliates, as previous investigations have focused primarily on rumen and parasitic lineages (Ricard et al. 2006; Xiong et al. 2015). Comparative analysis reveals distinct ecological patterns: rumen ciliates have acquired genes for complex carbohydrate metabolism similar to our free-living APM species (Ricard et al. 2006), whereas fish-parasitic ciliates harbor LGTs predominantly involved in heme utilization, signaling, and adhesion—functions associated with host invasion and exploitation (Xiong et al. 2015). Collectively, these findings suggest that LGT repertoires in free-living and commensal ciliates primarily reflect environmental nutrient availability and metabolic pressures, whereas parasitic lineages exhibit distinct signatures of host adaptation. However, given the limited sampling of ciliate and microbial eukaryotic LGTs to date, these patterns remain preliminary. Future comparative analyses across diverse eukaryotic lineages, ecological contexts, and symbiotic states will be necessary to establish general principles governing LGT evolution. CPR bacteria and newly classified prokaryotic groups as gene donors to ciliated protists Our analysis provides the first evidence that protists can acquire genes from Candidate Phyla Radiation (CPR) prokaryotes and other recently discovered "Microbial Dark Matter" groups (Fig. 2; Supplementary Table S4)—largely uncultured microbial lineages known primarily through molecular data (Brown et al. 2015; Liu et al. 2022; Murugkar et al. 2020; Rinke et al. 2013). These genes contributed to diverse potential functions for the APM ciliates: complex carbohydrate degradation (anmK, nagB-PIGL, murQ, RENBP, iaaA, ushA, YjgF, endonuclease), acetate/malate retention and energy efficiency enhancement (apm-acs, ME2, and apm-ppdK), ammonia assimilation and amino acid metabolism (arcC, OCD), and resistance to stress and oxygen (STC, lip, oxidoreductase, HoxU) (Figs. 3–6; Supplementary Figs. S8, S9). Genomic and metagenomic analyses indicate that many "Microbial Dark Matter" groups, including CPR bacteria, lack complete citric acid cycles or respiratory chains, consistent with adaptation to low-oxygen environments (Brown et al. 2015; Rinke et al. 2013). In addition, many CPR members are inferred to be symbionts of diverse organisms, including ciliates (Brown et al. 2015; Gong et al. 2014; Nelson and Stegen 2015). This suggests uncultivated CPR bacteria may have formed associations with free-living anaerobic protists, enabling hosts to access novel carbon sources and adapt to hypoxic soil environments via LGT. Thus, CPR-to-ciliate gene transfer represents a previously unrecognized mechanism for protists to expand metabolic capabilities and ecological niches in anaerobic habitats. Conclusions Symbiotic and predatory interactions between protists and prokaryotes are widespread across diverse natural ecosystems. While significant attention has been devoted to elucidating the metabolic potential of environmental prokaryotes, recent studies are beginning to uncover the genomic changes that these associations may induce in free-living protists, and the impact on their metabolic and ecological potential. Through the first survey of expressed genes of the free-living non-model APM ciliate, we found evidence that many genes in these single-celled eukaryotes were likely acquired from prokaryotes via lateral gene transfer (LGT). These LGT genes are predicted to be involved in the utilization of organic matter (hemicellulose and amino sugar), the recycling of intermediate metabolites (malate, acetate), the enhancement of energy efficiency, anabolism of inorganic nutrients (e.g., nitrogen), and tolerance to hostile environment (e.g., oxidative, toxic, and nutrient-deficient). Future studies will need to employ techniques such as gene editing (or RNAi) and isotope labeling of productions to verify the functions of these genes in ciliates. If proved functional, these genes would play significant roles in the adaptation of the APM ciliates to the lifestyle in the soil/sediment environments. Given the high diversity and abundance of free-living protists and their extensive interactions with bacteria and archaea in natural habitats, protists likely harbor a rich repertoire of prokaryote-originated genes. This finding implies that protists may be underappreciated contributors to biogeochemical activities previously attributed primarily to prokaryotes and fungi. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (PDF 4598 KB) Supplementary file2 (RAR 81 KB) Acknowledgements This work was supported by the Natural Science Foundation of China (grant numbers 42176163, 42376086, and 31970398), the Youth Innovation Promotion Association, CAS (grant numbers 2019216, 2022211). MML was supported by a Marie Skłodowska-Curie Individual Fellowship under the EU Framework Program for Research and Innovation Horizon 2020 (project ID 747789). The computational analyses conducted in this study were additionally supported by the High-Performance Computing Platform of Yantai Institute of Coastal Zone Research, Chinese Academy of Sciences. We would like to thank Drs. Weibo Song, William H. Lewis, T. Martin Embley and Johanna Rotterová for kindly sharing data. We would like to thank Dr. Xiao Chen from Shandong University (Weihai Campus), Dr. Kaihao Tang from South China Sea Institute of Oceanology, CAS and Dr. Hongan Long from Ocean University of China for their help on the genome data processing and MS drafting. We also thank MS. Xiaoxin Zhang, MS. Chenchen Zhao, and MS Ning Xu for their assistant on the illustrations. Author contributions Conception and design of the study: QZ and JG. Data acquisition: QZ, AO, and YF. Data analysis and interpretation: QZ, EG, ML, and SZ. Drafting of the manuscript: QZ and JG. Critical revision of the manuscript: QZ, EG, ML and JG. Editing and polishing of the Fig. plates: SZ and AZ. Study supervision and funding acquisition: QZ and JG. All authors have read and approved the final version of the manuscript. Data availability The transcriptomic sequencing reads generated in this study are deposited in the NCBI BioProject under accession number PRJNA1025929. The accession numbers for the individually amplified lateral transferred gene sequences reported in this paper are GenBank: OR730783-OR730792. The 16S rRNA gene reads from PacBio sequencing are available under NCBI’s GenBank BioProject ID PRJNA1025937. Additional data containing Phylogenies, alignments, LGT protein sequences of M. yantaiensis and its homologs in other APM ciliates, and HFSP values have been deposited to Mendeley Data at https://data.mendeley.com/preview/bj6v5szrk3?a=1dc63412-5f61-4c4d-89a0-199dbd5b17d1. Declarations Conflict of interest All the authors declare that there are no conflicts of interest. Animal and human rights statement This article does not contain any studies with human participants or animals performed by any of the authors. Footnotes Ciliatology: mini-compilation. 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 4598 KB) Supplementary file2 (RAR 81 KB) Data Availability Statement The transcriptomic sequencing reads generated in this study are deposited in the NCBI BioProject under accession number PRJNA1025929. The accession numbers for the individually amplified lateral transferred gene sequences reported in this paper are GenBank: OR730783-OR730792. The 16S rRNA gene reads from PacBio sequencing are available under NCBI’s GenBank BioProject ID PRJNA1025937. Additional data containing Phylogenies, alignments, LGT protein sequences of M. yantaiensis and its homologs in other APM ciliates, and HFSP values have been deposited to Mendeley Data at https://data.mendeley.com/preview/bj6v5szrk3?a=1dc63412-5f61-4c4d-89a0-199dbd5b17d1.