# Fileset

[Takano Sotaro ISMEj 2025.pdf](https://mdr.nims.go.jp/filesets/e102a355-6658-405f-8ae3-578d30deb7f5/download)

## Creator

[Sotaro Takano](https://orcid.org/0000-0003-2858-6587), Satoshi Takenawa, Divya Naradasu, Kangmin Yan, Xinxin Wen, Tomoko Maehara, Nobuhiko Nomura, Nozomu Obana, Masanori Toyofuku, Michihiko Usui, Wataru Ariyoshi, [Akihiro Okamoto](https://orcid.org/0000-0002-8102-4316)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Enrichment of horizontally transferred gene clusters in bacterial extracellular vesicles via non lytic mechanisms](https://mdr.nims.go.jp/datasets/eaf84ac0-5f6f-4a1d-924d-ac00039e1011)

## Fulltext

Enrichment of horizontally transferred gene clusters in bacterial extracellular vesicles via non lytic mechanismsReceived: 07 April 2025. Revised: 01 August 2025. Accepted: 25 August 2025© The Author(s) 2025. Published by Oxford University Press on behalf of the International Society for Microbial Ecology.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), whichpermits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.The ISME Journal, 2025, 19(1), wraf193https://doi.org/10.1093/ismejo/wraf193Advance access publication: 29 August 2025Original ArticleEnrichment of horizontally transferred gene clusters inbacterial extracellular vesicles via non lytic mechanismsSotaro Takano 1,2, Satoshi Takenawa1, Divya Naradasu1, Kangmin Yan1, Xinxin Wen1, Tomoko Maehara1, Nobuhiko Nomura3,4,Nozomu Obana4,5, Masanori Toyofuku3,4, Michihiko Usui6, Wataru Ariyoshi7, Akihiro Okamoto1,3,8,9,*1Research Center for Macromolecules and Biomaterials, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan2Integrated Bioresource Information Division, BioResource Research Center, RIKEN, 3-1-1 Koyadai, Tsukuba, Ibaraki 305-0074, Japan3Graduate School of Life and Environmental Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577, Japan4Microbiology Research Center for Sustainability, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577, Japan5Transborder Medical Research Center, Faculty of Medicine, University of Tsukuba, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577, Japan6Division of Periodontology, Department of Oral Function, Kyushu Dental University, 2 Chome-6-1 Manazuru, Kokurakita Ward, Kitakyushu, Fukuoka 803-8580,Japan7Division of Infection and Molecular Biology, Department of Health Promotion, Kyushu Dental University, 2 Chome-6-1 Manazuru, Kokurakita Ward, Kitakyushu,Fukuoka 803-8580, Japan8Graduate School of Chemical Sciences and Engineering, Hokkaido University, North 13 West 8, Kita-ku, Sapporo, Hokkaido 060-8628, Japan9Research Center for Autonomous Systems Materialogy, Institute of Innovative Research, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku,Yokohama, Kanagawa 226-8501, Japan*Corresponding author. Research Center for Macromolecules and Biomaterials, National Institute of Materials Science (NIMS). 1-1 Namiki, Tsukuba, Ibaraki305-0044, Japan. E-mail: OKAMOTO.Akihiro@nims.go.jp.AbstractBacterial extracellular vesicles are emerging as key mediators of horizontal gene transfer, enhancing microbial adaptability. A criticalfactor determining the effectiveness of horizontal gene transfer is the fraction of vesicles containing specific functional genes. However,the proportion of containing specific DNA fragments has not been adequately determined, which hinders the understanding of theconditions and mechanisms that facilitate the incorporation of specific genes into the vesicles and possible evolutionary roles of vesicle-derived DNA. Here, we demonstrate that enrichment of horizontally transferred genes into bacterial extracellular vesicles is driven bycellular processes by profiling the DNA content of hundreds of individual vesicles using a microdroplet-based sequencing technique.This approach revealed unique DNA profiles in vesicles from the oral pathogen Porphyromonas gingivalis, pinpointing genomic regionsrelated to DNA reorganization such as CRISPR-Cas clusters. Comparative genomic and phylogenetic analyses of Porphyromonas genomesrevealed traces of horizontal gene transfer in vesicle-enriched genes. Modulating vesicle-biogenesis routes, quantitative real-time PCRrevealed that this selective enrichment was driven by blebbing-driven DNA packaging mechanisms rather than stress-induced lysis.Applied to dental plaque-derived bacterial extracellular vesicles, the droplet-based approach reveled O-antigen biosynthetic genes,key for host–bacterial interactions, were prevalent in the vesicles from Alcaligenes faecalis, suggesting the vesicles from this bacteriumcan modulate pathogenicity in oral biofilms through targeted DNA packaging. These findings suggest the prevalence of functionallyrelevant gene clusters in bacterial extracellular vesicles in oral microbiota and their evolutionary roles as DNA cargoes for modulatingphage–bacterial and host–bacterial interactions via horizontal gene transfer.Keywords: membrane vesicles; horizontal gene transfer; oral microbiota; CRISPR-Cas; O-antigen gene cluster; Porphyromonas gingivalisIntroductionHorizontal gene transfer (HGT), a fundamental mechanism ofmicrobial evolution [1], enables bacterial cells to acquire novelphenotypic traits (e.g. antibiotic resistance), fostering rapid adap-tations to environmental challenges [2–4]. Besides ensuring indi-vidual survival, HGT can also contribute to genetic diversifica-tion within microbial populations [5]. This allows the sharingand integration of beneficial genes across species barriers [6, 7],catalyzing the emergence of novel capabilities within bacterialcommunities [2, 8].Bacterial extracellular vesicles (BEVs), nano-vehicles of diversebiomolecules produced by both Gram-positive and Gram-negativebacteria in various habitats [9, 10], have emerged as critical facil-itators of HGT [11, 12]. A recent study focusing on BEVs in marineenvironment revealed the prevalence of mobile genetic elementsin BEV-incorporated DNA, supporting its significance as DNAcargo [13]. The molecules inside BEVs (e.g. DNA) are generallycondensed, secreted, and protected from the risks of degradation[14, 15], increasing the success of gene transfer compared withuptake of extracellular DNA (e.g. natural transformation) [16].BEVs have the potential to transfer genetic material not onlywithin a species but also across phylogenetically distant species[16–18]. These characteristics are distinct from other molecu-lar mechanisms (e.g. conjugation and natural transformation),Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://orcid.org/0000-0003-2858-6587 5801 28398 a 5801 28398a mailto:OKAMOTO.Akihiro@nims.go.jpmailto:OKAMOTO.Akihiro@nims.go.jpmailto:OKAMOTO.Akihiro@nims.go.jpmailto:OKAMOTO.Akihiro@nims.go.jp2 | Takano et al.suggesting the unique role of BEVs for expanding the geneticrepertoire and adaptability of microbial communities.Despite their potential evolutionary significance via HGT, theorigin, and nature of the genetic material within BEVs remainslargely underexplored. For instance, although certain genesappear to be selectively enriched in BEVs and amount or regionsof packaged DNA vary depending on bacterial conditions or BEVtypes [19–24], the mechanisms behind the selective incorporationof specific DNA remain unclear. In contrast, existing modelssuggest that DNA incorporation into BEVs occurs randomly duringbacterial cell death [14], abolishing the idea of selective DNApackaging. Thus, whether BEV-mediated HGT plays a role inbacterial evolution through specific DNA packaging mechanisms,rather than being purely stochastic, remains a subject of debate.A critical factor determining the effectiveness of HGT is thequantity of BEVs containing specific genes [17], underscoringthe need to understand the genes present and the diversity ofthe DNA encapsulated in individual BEVs. However, identifyingand characterizing DNA within BEVs pose significant challenges,as conventional metagenomic methods cannot adequately dif-ferentiate DNA variations among individual vesicles. To addressthese limitations, we applied a single-nanoparticle droplet DNAsequencing (NP-DS) technique, adapted from single-cell genomicmethodologies [25–27]. The genetic content of individual vesicleswas profiled by applying P. gingivalis—and human dental plaque-derived BEVs. Given that this organism and its BEVs are impli-cated in periodontitis and systemic diseases [28, 29], our studyhighlights the specific features of BEV-derived DNA that may playimportant roles in microbial interactions, particularly within oralbiofilms and infections.Materials and methodsIsolation and purification of bacterialextracellular vesicles from P. gingivalisP. gingivalis W83 was grown in Gifu Anaerobic Medium at 37◦C. Tomaintain anaerobic conditions, 20 min of N2/CO2 (80,20 v/v) gassparging was performed prior to culture inoculation. The culturewas grown at 37◦C until the OD600 reached over ≈0.6, followed bycentrifugation for 10 min at 7800 rpm and 4◦C. The supernatantwas passed through 0.22 μm filters to remove any cell debrisand then ultracentrifuged for 2 h at 200,000 × g at 4◦C, and theresulting pellets were resuspended in phosphate-buffered saline(PBS).DNase I treatmentDNase I (13 units (U) / μl) (NIPPON GENE) was added to a finalconcentration of 2 U to treat the external DNA of nanoparticles.The sample was stored at 37◦C for 30 min, followed by DNaseI deactivation by heating the sample at 80◦C for 10 min. Weconfirmed that same amount of DNase I treatment can degradethe control DNA samples (200 ng of pUC19 plasmid) using thesame procedure.DNA isolation and sequencing of bulk bacterialextracellular vesicle samplesDNA from DNase I-treated BEV samples was extracted andpurified using an Isoplant-II DNA extraction kit (NIPPON GENE)according to the manufacturer’s instructions. The total amountof extracted DNA was quantified using a Qubit 1X dsDNA HighSensitivity Assay Kit (Thermo Fisher). Sequencing libraries wereprepared using the QIAseq FX DNA Library Kit (QIAGEN) accordingto the manufacturer’s protocols and sequenced using the NextSeq2000 System (Illumina) with a 2 × 150 bp configuration. Theobtained fastq files were processed using fastp 0.19.5 [30], andthe sequence reads were aligned to the assembly genomes of thehost bacterium (P. gingivalis: GCF_000007585.1) using bowtie2 [31]with option (−sensitive-local).Nanoparticle tracking analysisNTA was performed using a ZetaView (Particle Metrix) to quantifythe nanoparticles. The samples were measured by scanning 11cell positions and capturing images at 30 fps. For measurementsusing fluorescent signals, 488/500 nm and 660/680 nm laser-filterunits were used with the following camera settings: sensitivity:70–80; shutter: 200; minimum trace length: 15. For measure-ments in scatter mode, we used the following settings: sensitivity:65; shutter: 200; minimum trace length: 15. The captured videoimages were further analyzed using the ZetaView Software. Weconfirmed that measuring PBS only with lipid-dye did not harborany detectable signal and the detected particles were stainednanoparticles and not dye-aggregates.Droplet genome sequencingDroplet genome sequencing was performed by bitBiome (Japan),as previously described [32]; the details are described in the Sup-plemental Methods. Briefly, the cell or nanoparticle suspensionswere mixed with 1.5% agarose solutions to yield an expectedpercentage of positive droplets of 40% [25]. For the cell-DS, theencapsulated cells were lysed using lysis solutions in gel beads.The droplets in the cell-DS and NP-DS were processed for multipledisplacement analysis using the REPLI-g Single Cell Kit (QIAGEN)and the beads were stained with 1× SYBR Green (Thermo FisherScientific). Green fluorescence-positive beads were sorted usinga FACSMelody cell sorter (BD Biosciences). The collected positivedroplets proceeded to the second round of whole-genome amplifi-cation using the REPLI-g Single Cell Kit. Whole-genome amplifiedsamples with sufficient DNA were further subjected to whole-genome sequencing analysis using the Nextera XT DNA LibraryPrep Kit (Illumina) according to the manufacturer’s protocols andsequenced using HiSeq System (Illumina).Subjects and dental plaque sample collection fordroplet sequencingPatients were recruited with periodontitis, classified as havinggeneralized stage III-grade C [33]. These patients had not takenantimicrobials within the previous 3 months, were non-smokers,or had diabetes or any other serious systemic disease. The presentstudy was approved by the ethics committee of Kyushu DentalUniversity (ethical approval number: 26–28) and conducted inaccordance with the approved guidelines. All the participantsprovided written informed consent.Dental plaque biofilm samples (n = 3) were collected as follows.Rolled cotton was placed next to the teeth to prevent salivacontamination. Supra- and subgingival plaques were collectedwithout blebbing by a periodontist using a curette. The collecteddental plaque biofilm samples were immediately resuspended in1 ml of PBS. Thereafter, the biofilm was treated with 100 μg/mlProteinase K (QIAGEN) for 1 h at 37◦C with vortexing at 20-min intervals to degrade extracellular matrices, and the treatedsample was centrifuged for 10-min at 6000 × g at room tempera-ture. The resulting pellet was resuspended into OMNIgene-ORALOM-501 (DNA Genotek) and used for cell-DS. The supernatantwas passed through 0.22 μm filters to remove any extracellularsubstrates. The filtered supernatant was ultracentrifuged for 2 hDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025Decoding HGT by single-vesicle genomics | 3at 200,000 × g at 4◦C, and the resulting pellets were re-suspendedin 10 mM HEPES with 0.85% NaCl for NP-DS.Data analysis in NP-DS and cell-DSThe paired-end read sequences from each droplet were first pro-cessed using fastp 0.19.5 and then assembled using SPAdes 3.12.0.Coding sequences (CDSs), rRNAs, and tRNAs were extracted fromthe contigs using Prokka 1.14.6 [34]. Extracted CDSs were groupedand clustered using CD-HIT. Clustered CDSs were first annotatedby a homology search against the National Center for Biotechnol-ogy Information (NCBI) nonredundant (nr) database (downloadedon July 1, 2021) using DIAMOND 2.0.8 [35]. For each CDS, the Gen-Bank accession number of the best-hit protein sequence was usedfor taxonomic annotation with BASTA [36]. The CDSs classifiedas Bacterial at the kingdom-level were further annotated by ahomology search against protein sequences in the genome taxon-omy database [37] (version r202) by DIAMOND stricter taxonomicannotation, with a threshold in a percentage identity ≥80%.The bacterial taxa classified as potential contaminants in low-biomass human samples in previous studies were eliminated [38,39] (see Supplementary Data 4). For droplets determined to beBEV-containing, the CDSs were first grouped by GTDB accessionnumber and the total CDS length assigned to each GTDB taxon-omy was calculated. Thereafter, the most abundantly detectedGTDB taxonomy was designated to the MFT (the most frequentlydetected taxon) for each droplet. For details of the computationalanalysis, see Supplementary Methods.ResultsEnrichment of specific genomic regions in P.gingivalis-derived BEVsWe explored the characteristics of DNA sequences in BEVsproduced by P. gingivalis, a prominent pathogen of periodontitis[40, 41]. Nanoparticle tracking analysis of the isolated samplesdetected >109 particles (/ml) stained with a lipophilic dye(Supplementary Fig. 1A). Transmission electron microscopy (TEM)revealed the presence of spherical particles in the collectedsample (Supplementary Fig. 1C), and no prophage genes includingthose related to lipid enveloped virus was identified in this strain[42], indicating that BEV is dominant nanoparticles in the bacterialcell cultures. To understand the composition of the DNA withinthese vesicles, shotgun sequencing for the bulk BEV samples wasperformed (Fig. 1A), after treatment with DNase I to remove anyexternal DNA (eDNA). The sequence reads from these DNA werewidely mapped to the original bacterial genome; however, severalgenomic regions were over-detected (Fig. 1B), suggesting selectiveDNA packaging within the BEV population.To check the overrepresentation of specific genomic regions ata higher resolution, WGA of these particles was performed using adroplet-based method [25]. The samples were mixed with agarosesuch that ≈40% of the droplets contained a single nanoparti-cle (Table 1, Supplementary Fig. 2, denoted as Rt). This processharbored 4.6% of the droplets contained amplified DNA (stainedwith SYBR-Green-I, Table 1, Supplementary Fig. 2, represented asRo). Given that nearly 20% of the total nanoparticles were lipid-stained (Table 1, Supplementary Fig.1, 2, presented as Rb), wecould estimate that ≈60% of lipid-stained nanoparticles (BEVs)harbored DNA (Table 1). This substantial population underscoresthe prevalent DNA packaging process within the BEVs of P. gin-givalis, substantially surpassing the value estimated by SYBR-Green-I (a dye for nucleic acids) directly on BEVs like a conven-tional method (0.44 ± 0.32%) of positively stained BEVs in NTA(Supplementary Fig. 1). This discrepancy between two methodsis likely attribute to the low-permeability of the dye into thebacterial cytoplasmic space or the presence of too short DNAfragments to be detected by this technique, given fluorescencestaining sensitivity is contingent on DNA length [43].To determine the DNA content of individual BEV particles,whole-genome sequencing of DNA amplified by MDA was per-formed. Of the 96 droplets, most sequence reads were mappedto the P. gingivalis genome in 93 droplets (Supplementary Fig. 3),and wide regions of the host chromosome were detected in atleast one droplet, similar to the profile in bulk-BEV sequencing(Fig. 1B). Although there were variations in the detected genomicregions within the BEV-population from P. gingivalis (Fig. 1B), 10–30kb of genomic regions were detected in ≈70% (65/96) of droplets(Fig. 1C). By calculating the continuous genomic regions mappedby the sequence reads within each droplet, we found that innearly 70% of droplets (68/96), the majority (>80% in length) ofthe detected genomic regions originated from the single longestgenomic region (Supplementary Fig. 4). This suggests that most ofthe DNA fragments in each BEV derived from one single genomiclocus. Some droplets containing DNA from the multiple chromo-somal loci (Supplementary Fig. 4), and such droplet would pack-age a BEV possibly containing the DNA fragments from substan-tially different genomic loci or multiple BEVs with DNA from dif-ferent genomic loci. However, the distribution of the total detectedregions and the longest continuous detected regions exhibitedquite similar trend (Fig. 1C), and such BEVs or droplets were minorin the analyzed population.Although the regions mapped by read sequences from P. gingi-valis-BEVs were not uniformly distributed among the 96 droplets(Fig. 1B), we found a few genomic regions that were significantlyover-detected among the BEV population (binominal test, P < 0.01,see Supplemental Methods) (Fig. 1B). Common enriched regionsexisted between bulk-BEV sequencing and droplet sequencing ofBEVs from P. gingivalis (Fig. 1B), yet numerous frequently mappedregions in bulk-BEV sequencing failed classification as enrichedbased on detected droplet criteria. A deviation between thetwo approaches was also observed in the prevalence of eachprotein CDSs. The mapped sequence reads (FPKM) in bulk-BEV-sequencing exhibited longer-tailed abundance distributionof 1873 CDSs compared to the detected droplets in NP-DS(Fig. 1D). Consequently, the bulk-BEV-sequencing identifiedapproximately seven times more enriched CDSs (i.e. outliers inthe distribution; Supplementary Data 1) than NP-DS (Fig. 1D).This deviation occurred because mapped sequence reads in bulk-BEV sequencing responded significantly to GC content of eachgenomic locus (Supplementary Fig. 5), as reported in previousstudies [44, 45], whereas GC content minimally affected detecteddroplet counts (Supplementary Fig. 5).We validated enrichment of specific genomic regions in BEVsusing a simpler approach; quantitative PCR (qPCR) analysis. Wetargeted two regions: one with significantly enrichment (magentain Fig. 1B, bottom) and another with zero droplet detection (grayin Fig. 1B, bottom). Results showed high detection levels forthe enriched region (0.06 ± 0.03 copies/BEV particle) (Fig. 1F),but the other target site showed minimal detection using qPCR(0.002 ± 0.003 copies/BEV particle) (Fig. 1F), aligning with NP-DSresults.We applied the qPCR analysis to extracellular DNA (eDNA) fromthe P. gingivalis culture supernatant, finding that the DNA purifiedfrom the eDNA fraction contained fragments of DNA from bothBEV enriched and nonenriched genomic regions at a comparablelevel (Supplementary Fig. 6), and this tendency was distinct fromDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-data4 | Takano et al.Figure 1. Enrichment of DNA at specific loci of the host genome in P. gingivalis BEVs. (A) A schematic of the analysis. Using the isolated bacterialextracellular vesicles (BEVs) from the bacterial culture, we performed nanoparticle droplet sequencing (NP-DS) by DNA amplification of encapsulateddroplets or bulk-BEV sequencing by directly extracting DNA from the whole BEV particles for shotgun sequencing. In NP-DS, the collected sequencereads in each droplet were mapped to the original bacterial genome to obtain the mapping profiles of individual droplets. In bulk-BEV sequencing, allsequence reads were collectively mapped to the genome. In both cases, we analyzed BEV samples derived from a single culture medium. (B) Loci ofmapped regions of sequence reads in BEVs on the original bacterial chromosomes. The number of mapped reads in the next-generation sequencing ofthe bulk-BEV sample is displayed in the top panel. In the second top panel, the frequency of positive droplets among 96 droplets across each bacterialgenome is shown. In the third top panel, the detected genomic regions in all 96 droplets are displayed vertically. In each row, the positions on thegenome where the sequence reads were mapped in each droplet are indicated with filled areas. The genomic regions determined as significantlyenriched in bulk-BEVs (more than 75th percentile +2∗IQR (interquartile range)) or droplet sequencing (binomial test, P < 0.01) are indicated with filledareas in the bottom panels. The enriched chromosomal regions in both the droplet and bulk-BEVs sequence analysis are indicated with filled areas (inthe bottom-most panel). (C) The distributions of the total lengths (enclosed by solid lines) or the longest contiguous genomic region (enclosed bydashed lines) mapped by sequence reads in each droplet (n = 96). (D) Abundance distribution of 1873 CDSs in BEVs. In the bulk-BEV sequencing, theabundance was estimated according to the frequency of mapped reads (RPKM). In the case of the NP-DS, the number of droplets detected was used asa metric of the abundance of each CDS. Dashed lines indicate the threshold for enriched CDSs (i.e. outliers, 75 percentile +2∗IQR (interquartile range)).(E) SEM images of P. gingivalis cells treated with artepillin-C or petroselinic acid. BEV productions are indicated by arrows. (F), (G) DNA copy numbers ofthe target genomic regions in the BEV samples quantified using qPCR when the cells were treated with artepillin-C (APC, panel F) or petroselinic acid(PA, panel G). We targeted two representative regions: region 1 was significantly enriched in the NP-DS and region 2 was not detected in any of the 96droplets (those target sites were labelled as bars in Fig. 1B, respectively). Here, the copy numbers per particle of each genomic region under twodifferent conditions are displayed. Error bars indicate the standard deviation in triplicate experiments. Asterisks show statistical significance levelsaccording to the Student’s t-test (∗: P < 0.05, n.s.: not significant).Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025Decoding HGT by single-vesicle genomics | 5Table 1. Estimated proportion of BEVs containing DNA in the analyzed population. The ratio of droplets to particles with DNA (Ro). Thetheoretical ratio of droplets to total particles (Rt) was controlled to a fixed value in our analysis. Ro and Rvb were estimated by NTA andtaxonomic annotation, respectively. The proportion of BEVs with DNA, r, was calculated based on those parameters. For calculationdetails, see Supplementary Methods and Supplementary Fig. 2.P. gingivalis BiofilmParticles with DNA / Droplets (Ro) 0.046 0.035Total particles / Droplets (Rt) 0.36 0.36Viruses / BEVs with DNA (Rvb) 0 3.97Total BEVs / Total nanoparticles (Rb) 0.21 0.053BEVs with DNA / Total BEVs (r) 0.62 0.35that observed in the extracted DNA from equivalent amount ofBEV fraction (Supplementary Fig. 6). This result indicates that theenrichment of specific DNA was only observed in the BEV fractionin our experiments, and thus most of the DNA observed in ouranalysis was derived from BEVs but not eDNA remnants.DNA profiles in BEVs were susceptible to thebiogenesis mechanismAlthough membrane blebbing and explosive lysis followed byrecircularization are reportedly major mechanisms of BEV for-mation [14, 46, 47], the mechanism of DNA packaging into BEVsremains debatable. To address this issue, the selective packag-ing of specific genomic loci into BEVs was compared betweenthese two mechanisms. First, cells were treated with artepillin-C, which promotes BEV formation via membrane blebbing [48](Fig. 1E). The copy numbers of the DNA fragments from the tar-get genomic regions (Fig. 1B, bottom; gray and magenta) werequantified using qPCR. Despite an increase in BEV production(Supplementary Fig. 7B), the same genomic regions were detectedat levels comparable to those of the untreated controls (Fig. 1F),indicating that artepillin-C treatment did not significantly alterthe DNA profile of BEVs.When cells were treated with petrocelinic acid (PA), BEV bio-genesis was induced by explosive cell lysis or bubbling cell death[49] (Fig. 1E), where the genomic regions frequently detected innaturally occurring BEVs (i.e. those produced during normal cellgrowth) were barely detectable in PA-induced BEVs (Fig. 1G). Thissuggests that membrane blebbing rather than cell lysis, is respon-sible for the enrichment of specific DNA fragments under normalgrowth conditions.PA-induced cell lysis harbored both BEV enriched and nonen-riched genomic regions in eDNA fraction at a comparable orhigher level in a BEV fraction (Supplementary Fig. 6), and thusthis process contribute to the increase in eDNA amount similar toa previous study [46]. However, both genomic regions were barelydetected in PA-induced BEVs (Supplementary Fig. 6) and total DNAwas ∼10-fold lower than that extracted from naturally occurringBEVs (Supplementary Fig. 8), suggesting that cell lysis and theaccompanying increase in eDNA do not contribute as effectivelyto the selective DNA packaging into BEVs like blebbing case.Enrichment of functional genes in BEVs fromP. gingivalisWe investigated the functional implications of DNA enrichment inP. gingivalis BEVs. To explore the genes loaded into a large fractionof the BEV population, the over-detected CDSs were statisticallyextracted among 96 droplets in NP-DS, revealing 39 significantlyover-represented CDSs among the 1873 present in the genome(Supplementary Table 1). A homology search of the enrichedCDSs in BEVs against the UniProtKB/Swiss-Prot database [50]enabled GO terms categorization of the statistically screenedgenes (Fig. 2A and Supplementary Table 2). Five categories among497 showed statistical enrichment (hypergeometric test, P <0.05, Supplementary Methods). “DNA transposition (GO:0006313)”emerged significant, indicating the predominance of genome-arrangement functions in the BEV-derived DNA. ∼20% of theenriched CDSs were related to genome arrangements suchas transposase, site-specific integrase, and CRISPR-Cas system(9/39). The transposase and integrase are flanked by the terminalinverted repeats (TIRs) (Supplementary Data 2) and possess thetypical excision signatures in their adjacent region (Fig. 2B).≈15% (13/96) of droplets contained IS5 family transposases(WP_010956028.1 and WP_010955945.1) with the conservedregions in TIR (Fig. 2B), suggesting that specific transposaseswith the specific excision signatures are associated with BEVbiogenesis and DNA-packaging.The CRISPR category, linked to “Maintenance of CRISPRrepeat elements (GO:0004803)” and “defense response to virus(GO:0051607)”, was particularly highlighted by the packaging of anentire CRISPR-Cas gene cluster within an enriched genomic region(Fig. 2D), quite similar tendency as DNA in BEVs from the marinemicrobiota [24]. The genes grouped into “Cobalamin biosyntheticprocess (GO: 0009236)”, another enriched functional category(Fig. 2C), is clustered at a specific locus (Fig. 2C). This cobalaminbiosynthesis cluster is also adjacent to the type VI-B2 CRISPR-Casgene clusters (Fig. 2C), indicating the strong relationship betweenthe CRISPR-Cas gene clusters and selective packaging into BEVsin this strain.Bulk-BEV sequencing produced the bias derived from the GCcontent and also harbored approximately triple the screened cat-egories compared NP-DS in GO term analysis, spanning metabolic,stress response, and enzymatic categories (Supplementary Table 3).Nevertheless, identical or similar GO terms emerged, including“cobalamin biosynthetic process (GO: 0009236)” and “threonine-phosphate decarboxylase activity (GO:0048472)” (SupplementaryTable 3), demonstrating methodological consistency.We characterized the chromosomal location and excision sig-natures of the overrepresented gene clusters. The packaging ofDNA fragments excised by Xer recombinase at dif site into BEVswas previously reported [51], but any concordance of the difsite and the enriched genomic loci in P. gingivalis was found(Supplementary Fig. 9). Then, we focused on TIRs, more commonexcision signatures, as in the case of, and identified the possiblecandidates of TIRs around the enriched genomic regions (Fig. 2Cand D, Supplementary Fig. 10, and Supplementary Table 4). Thoseidentified TIRs are usually ≈14 bp, yet we found the long tandeminverted repeats flanking cobalamin biosynthesis gene cluster(≈600 and ≈50 bp, Supplementary Fig. 10 and Fig. 2C and D),suggesting the high probability of excision in this region, giventhe high excision frequency of the long inverted repeats [52]. TIR-Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-data6 | Takano et al.Figure 2. Functional enrichment of genes in P. gingivalis BEVs. (A) Enriched gene categories in CDSs detected in BEVs of P. gingivalis. Frequently detectedCDSs in BEVs were statistically extracted (hypergeometric test, P < 0.05, see supplementary methods). Gene ontology terms (GO terms) were used forthe classification of the frequently detected CDSs. Heatmap indicates the statistical significance level of enrichment. (B) The identified possible TIRsadjacent to the overrepresented IS elements in the P. gingivalis BEVs. Asterisks indicate the aligned nucleotide pairs between left-arm and right-arm ofthe inverted repeats. Shaded regions are the conserved nucleotide sequences between those two transposases. (C), (D) Clustered functional geneslocated on the enriched genomic regions in P. gingivalis BEVs. Detection profiles in 96 droplets where (C) cobalamin biosynthetic genes and type VI-B2CRISPR-Cas gene cluster or (C) type III-B CRISPR-Cas gene cluster located. In each row, the positions on the genome where the sequence reads weremapped in each droplet are filled. In the bottom, possible TIRs located around each region were plotted. Each TIR pair was colored by its length.flanked regions on the P. gingivalis genome were generally moreprevalent than other regions in the BEVs (Supplementary Fig. 11),implying correlation between the presence of TIRs and the selec-tive packaging in this bacterium genome. However, most of theidentified TIRs adjacent to the targeted gene clusters were notfound in the other loci and unique on the P. gingivalis genome(Supplementary Data 3), which also implies that specific excisionsignature relates to the selective packaging.Enrichment of gene clusters that are possiblehotspots of horizontal transmissionDNA fragments in BEVs are potentially transferred to other bac-teria [16–18], suggesting that enriched regions identified in ouranalysis could be horizontally transmitted among this bacterialgroup during evolution. We investigated the possibility of HGTsfor two representative gene clusters: cobalamin biosynthesis andtype III-B CRISPR-Cas gene clusters (Fig. 2C and D). Analysis of244 high-quality genomes in the Porphyromonas genus revealed theprevalence of both clusters across diverse Porphyromonas species(Fig. 3A and Supplementary Fig. 12), yet type III-B CRISPR-Cascluster appeared only in specific strains within each species,including P. gingivalis (Fig. 3A), indicating limitations of verticaltransmission mechanism in explaining its distribution amongPorphyromonas.We tested the possibility of horizontal transmission ofthe CRISPR-Cas cluster among Porphyromonas species using acomparative genomics approach. Horizontally transferred genesoften show evolutionary history differing from species-levelDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-dataDecoding HGT by single-vesicle genomics | 7Figure 3. Prevalence and possible evolutionary history of the gene-cluster enriched in the BEVs in Porphyromonas group. (A) The prevalence of the sevengenes in the CRISPR-Cas cluster among 244 genomes from the Porphyromonas group is displayed on the phylogenetic (species) tree tips. The presence ofthe genes in each genome is displayed as a heatmap from gene 1 to 7 (the number corresponds to the gene arrows in Fig. 2). The genes existing on thegenome as a cluster are indicated with darker shading. If the gene is present but located on the distant locus from the gene cluster, it is indicated withlighter shading. The major species group is shaded by colors in the phylogenetic tree. (B) Discordance between the overall similarity of the genome andthe similarity of the genes in the CRISPR-Cas cluster across the Porphyromonas species. Pearson’s correlation (r) between two similarity scores isdisplayed. The P-value indicates the probability that the randomly selected gene similarity score exhibited a lower correlation than the observed valuein the permutation test (see supplementary methods). (C) The correlation between the overall similarity of the genome and the similarity of the genesin all analyzed members. The numbers correspond to those in Fig. 2. The statistical significance levels of correlation values are displayed as P values.(D) Comparison of the concatenated gene-cluster tree of CRISPR-Cas cluster and the corresponding species tree. The blue dots on the internal nodes ofthe trees show the branch support values. The red arrows indicate possible the horizontal gene transfer (HGT) events predicted by the treereconciliation approach by RANGER-DTL. The color density indicates the likelihood that the corresponding HGT was observed in RANGER-DTLanalysis (see supplementary methods).phylogenies, manifesting as phylogenetic distances discrepanciesbetween genes and species particularly in comparison to othergenes [53, 54]. Comparison of the similarity of CRISPR-Cas geneswith overall genomic similarities within the Porphyromonas grouprevealed significant deviations between those two parametersin four of the seven cases (Fig. 3B and C), suggesting that HGTevents contributed to the dissemination of this gene cluster.In contrast, cobalamin biosynthesis genes showed no suchDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 20258 | Takano et al.phylogenetic discrepancies (Supplementary Fig. 12), suggestingprimarily vertical transmission.To further corroborate the possibility of horizontal transmis-sion, two phylogenetic trees were directly compared, one based onwhole-genome sequences (general marker genes) and the otherbased on the five genes in the CRISPR-Cas cluster that are widelypresent in Porphyromonas, resulting in considerable discrepanciesbetween those two trees (Fig. 3D). Tree reconciliation analysisidentified four potential species-level HGT events within Porphy-romonas (Fig. 3D). These findings support horizontal transmissionof CRISPR-Cas cluster among Porphyromonas species, emphasizingthe potential role of BEVs in the evolutionary dynamics of thisgene cluster.DNA profiling of nanoparticles from human oralbiofilmWe attempted to characterize DNA content in BEVs producedin the dental plaque biofilm of periodontal patients, the nat-ural habitat of oral pathogenic bacteria. Transmission electronmicrographs of the isolated samples revealed spherical particleswith diameters in the range of ≈100–200 nm (Fig. 4A). NTA alsodetected particles stained with lipophilic and double-strandedDNA dyes (Supplementary Fig. 13A), with DNA-positive particlesexceeding lipid-stained particles (Supplementary Fig. 13B), sug-gesting presence of nanoparticles with DNA that do not consistof a lipid layer, such as viruses. However, the amount of extractedDNA was below the minimum amount required for conventionalmetagenome sequencing (<1 ng per sample).Droplet sequencing was successful for collected BEVs withlow DNA quantities. Following the protocol for P. gingivalis, theDNA-positive gel beads in NP-DS and lipid-stained nanoparti-cles were quantified (Supplementary Fig. 2), and the fractionof nanoparticles containing DNA was estimated as ≈10%(Supplementary Fig. 2B and C), matching the number of DNApositive particles in NTA (Supplementary Fig. 13B). Next 384positive droplets were sequenced (Fig. 4B). The sequence readsfrom each droplet were computationally assembled, dividedinto CDS units, and searched against the nr database (seeMaterials and Methods). This method detected an averageof 52 ± 29 CDSs per droplet (Fig. 4C), predominantly virus(bacteriophage)-derived (vCDS). TEM analysis revealed the pres-ence of particles with the morphology of tailed bacteriophages(Supplementary Fig. 14A), and all the vCDSs were assignedto Caudovirales (Supplementary Fig. 14B). Thus, the origin ofmost viral DNA would be phage particles. However, we alsoobserved several droplets containing both bCDSs and vCDSs;the origin of the viral DNA in such cases will be discussed later.Approximately 20% of droplets (73/384) contained ≥5 bacterialCDSs (bCDSs) (Fig. 4D), and bCDSs / (vCDSs + bCDSs) ratio inthose droplets >0.15 (Supplementary Fig. 15) and regarded as BEV-containing based on a previous comprehensive study of auxiliarymetabolic genes in viral genomes [55]. However, CDSs annotatedas host (human)-derived were barely detected (Fig. 4C, Eukaryote).Considering the fraction of virus-containing droplets in NP-DS, theratio of BEVs with DNA to the total BEVs was estimated as ≈30%(Table 1, Supplementary Fig. 2, presented as r). Thus, a significantfraction of BEVs contained DNA in the oral biofilm.The DNA content of a BEVs was compared to that of a bacterialcell by droplet sequencing of bacterial cells (cell-DS) isolated fromthe same dental plaque samples (Fig. 4B). In cell-DS, an averagelyof 1204 ± 542 CDSs and 790 ± 490 kb of CDS region was detected(Fig. 4C and Supplementary Fig. 16A), and full-length 16S rRNAsequences were observed in >90% of the droplets (Fig. 4E). Incontrast, the CDS regions detected in most BEV-containingdroplets were ≈9 kb in each droplet (8.6 ± 6.5 kb, Supplementary Fig. 16B),corresponding to ≈1% of those detected in cell-DS. In addition,full-length 16S rRNA sequences were detected in 4% (3/68) ofBEV-containing droplets (Fig. 4E).Enrichment of specific genomic regions in BEVsfrom the oral cavityTaxonomic annotation of the detected bCDSs in each BEV-containing droplet through homology search against the GTDB(Supplementary Fig.17) revealed dominance of bCDSs from asingle phylum, family, or genus-group (∼70% –90% of purity,Fig. 5A and Supplementary Fig. 18), suggesting that the mostfrequently detected bacterial taxon (MFT) for each dropletreflected the BEV origin. Mapping of the sequence reads to theassembled genome of MFT (Supplementary Fig. 17) revealed thatBEV-containing droplets covered 0.5%–1% of the genomic regions,whereas droplets in cell-DS reasonably covered 40%–60% of thegenomic regions (Fig. 5B).BEVs from dental plaque biofilms also exhibited enrich-ment of specific genomic regions, similar to the pure culturesystem. Analysis of droplets with an MFT derived from A.faecalis (GCF_002443155.1), the predominant taxon in BEV-containing droplets (Fig. 5C), revealed the presence of 5–30 kbof the continuous genomic regions in ≈70% of those droplets(Supplementary Fig. 19), which were distributed across awide range of host chromosomal regions (Fig. 5C). However,some regions were commonly detected among the droplets(Fig. 5C, binomial test P < 0.01), including a genomic locusconsisting of many lipopolysaccharide (LPS) biosynthesis genes,which has a possible TIR in its adjacent region (Fig. 5D andSupplementary Table 5). A statistical screening of enrichedfunctional categories also identified GO terms associatedwith bacterial antigen (LPS) biosynthesis, such as “O antigenbiosynthetic process (GO:0009243)” and “LPS biosynthetic process(GO:0009103)” (Fig. 5E and Supplementary Table 6). O-antigenbiosynthesis genes appeared in ≈14% of A. faecalis-derived BEVs(≈10% of all analyzed DNA-containing BEVs in the oral biofilm),suggesting the prevalence of these genes in the biofilm BEVs.Origins of BEV-derived DNA in dental plaqueTaxonomic profiles of bCDSs detected in the NP-DS weremarkedly distinct from those detected in the cell-DS. At thephylum level, abundant groups in BEV-containing droplets, suchas Proteobacteria, were barely detected in the cell-DS (Fig. 5A);thus, most of the DNA sequences detected in BEVs were derivedfrom bacterial species with very low abundance in dental plaquebiofilms. Indeed, Actinomyces, Peptidiphaga, and Streptococcus werethe most frequently detected taxa and MFTs in ∼90% of dropletsin the cell-DS, whereas these CDSs were barely detected in theNP-DS (only in six droplets) (Fig. 5A). In contrast, CDSs fromAlcaligenes (Proteobacteria), the most abundant genus detectedin BEV-derived DNA, were quite low in frequency in the cell-DS(Fig. 4A). In addition to Alcaligenes, we also found the bCDSs fromseveral minor bacterial genera in the cell-DS, such as Prevotella,Capnocytophaga, Campylobacter, Leptotrichia, and Pseudomonas_E, inthe NP-DS, although some of these (e.g. Prevotella) were previouslyreported to be prevalent in periodontal oral biofilms [56]. Theseresults suggest that the BEV-derived DNA detected using ourmethod could reveal the distribution of functional genes amongparticles and highlight the taxonomically minor but activeproducer strains of BEVs in the microbiome.Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-dataDecoding HGT by single-vesicle genomics | 9Figure 4. Distinct characteristics of DNA in BEVs compared to bacterial cells. (A) The typical transmission electron microscope (TEM) micrographs ofBEVs isolated from the dental plaque of periodontal patients. (B) Schematic of microbial cell-DS and NP-DS analysis of the dental plaque biofilmsamples from periodontal patients. The separated nanoparticles and bacterial cells in dental biofilms were encapsulated into droplets, and the internalDNA fragments were amplified (see materials and methods). In NP-DS, the positive amplification droplets (enclosed by dashed lines) were classified bythe number of detected bCDSs and vCDSs as BEV-containing, viral DNA-containing, or both. (C) The total length of detected CDSs in NP-DS and cell-DSanalysis. The results of all positive droplets from the biofilm samples were analyzed and are shown as boxplots. Medians and outliers are shown asbold lines and circles, respectively. (D) Kingdom-level classification of detected CDSs in NP-DS and cell-DS. In each droplet, the length of CDSs assignedto bacteria, viruses, eukaryotes, or unknown (i.e. not assigned to any taxon in the database) was normalized by the total length of detected CDSs and isshown as a heatmap. The droplets where less than five CDSs were detected were eliminated. (E) The percentage of BEV- or bacterial cell-containingdroplets in which 16S rRNA sequence was detected. The DNA sample is a mixture of BEVs collected from the biofilms of three patients.DiscussionThe present study revealed that ≈40% of the BEV population ofP. gingivalis and ≈ 20% of the BEVs from the oral biofilm (corre-sponding to ≈70% of the BEVs from A. faecalis) contained 10–30kb or 5–30 kb fragments respectively. These findings indicate thata considerable fraction of BEVs from the oral microbiota containDNA fragments with multiple genes, with certain functional geneclusters being prevalent across the BEV population (Fig. 2, andFig. 5). NP-DS analysis revealed gene enrichment in P. gingivalisBEVs associated with functions more specific than those identi-fied by bulk sequencing (Fig. 1). Bulk sequencing tended to capturea broad array of miscellaneous gene sets (Supplementary Data 1)and was susceptible to amplification bias during library prepara-tion (Supplementary Fig. 5). It is possible that DNA amplificationwithin individual droplets, combined with count-based statisticalanalysis, mitigates these biases, leading to a precise identificationof functionally enriched genes within BEVs. Furthermore, thehigh sensitivity of droplet-based approach for detecting DNA-containing BEVs (Supplementary Fig. 2) has the potential to fur-ther clarify the ubiquity of BEVs as DNA carriers in microbialcommunities in diverse environments.The enrichment profiles, the prevalence of DNA content, andtheir unique evolutionary history strongly suggest the potentialecological impacts of DNA cargoes on the microbial communityvia HGT. Enrichment of a specific group of metabolic or virulence-associated genes was found in both bacterial cases (Fig. 2A andFig. 5E), similar to other mobile genetic elements (e.g. plasmidsand integrative conjugative elements) [57, 58].Our phylogeneticanalysis revealed discordance between the evolutionary historiesof species and genes in the BEV-enriched gene cluster (Fig. 3),suggesting that BEVs contain HGT hotspots. Clustered gene setswere estimated to be present in 3%–12% of BEVs of P. gingivalis,which was a 10–100-fold greater than those reported in the otherbacteria [24]. Although one possible reason for this deviationwould be the difference in bacterial species and growth condition,our results in P. gingivalis using traditional DNA dye stainingtechniques also detected <1% of DNA positive BEVs, and thus itis possible that DNA-packaging BEVs are far more prevalent thanthose estimated by the direct staining.The “O-antigen biosynthesis genes” were highly prevalent inthe oral plaque biofilm-derived BEVs. These gene sets are gen-erally clustered on the genome of Gram-negative bacteria [59,Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-data10 | Takano et al.Figure 5. Enriched DNA sequences in BEV-containing droplets. (A) Genus-level taxonomic classification of bCDSs detected in BEV-containing dropletsin NP-DS and bacterial cell-containing droplets in cell-DS. The length of the CDSs assigned to each bacterial genus group was normalized to the totaldetected bCDSs and are shown as a heatmap. We only display the genera whose maximum frequency of detected CDSs in a single droplet is more than0.4. (B) The percentage of the genomic region covered by the sequence reads in a BEV- or bacterial cell-containing droplet. For each droplet, sequencereads are aligned to the assembled genome of the MFT. Medians and outliers are shown as bold lines and circles in box plots, respectively (C) genomicregions in A. faecalis (GCF_002443155.1) mapped by sequence reads in BEV-containing droplets. The frequency of droplets, including each genomicregion (top). In each row, the positions on the genome where the sequence reads are mapped in each droplet are filled with black (middle). The genomicregions that were significantly enriched in BEV-containing droplets are filled with red (bottom). We analyzed the sequence data of 53 BEV-containingdroplets whose MFT was determined as A. faecalis (GCF_002443155.1). (D) Detection profiles of the lipopolysaccharide (LPS) biosynthesis gene cluster(encircled by magenta in panel C) in 53 droplets. The color of each gene arrow corresponds to the functional group. (E) Screened functional categories(GO terms) that were significantly overrepresented in the enriched genomic region of BEVs from A. faecalis. Similar to Fig. 2, we statistically extractedthe GO terms (hypergeometric test, P < 0.05, see supplementary methods). The statistical significance level is shown as a heatmap.60], and the high sequence diversity and unique GC content ofthis genomic region suggest that this gene cluster is a frequentsite for HGT [61, 62]. Given that divergence in O-antigens canpotentially impact host–bacterial interactions [63], it is possiblethat the enriched genes in BEVs modulate the pathogenicity ofthe oral biofilm via HGT.P. gingivalis BEVs showed enrichment of genes associatedwith DNA integration, including DNA transposition, and CRISPR-Cas system, which are also consistent with the prevalence ofmobile genetic elements (MGEs) in BEVs from marine environ-ment [13]. Insertion sequences (IS) are believed to facilitategenome rearrangements in this species group [64]. Furthermore,comparative genomic analyses revealed discordance betweenspecies and gene similarities in the CRISPR-Cas cluster amongthis bacterial group (Fig. 3), strongly suggesting a criticalrole for HGTs via BEVs. CRISPR spacer sequences are highlyhomologous to IS regions in the P. gingivalis genome, potentiallypreventing IS transposition and recombination [64]. In addition,targets for some spacer sequences correspond to prophagesspread among the Porphyromonas group [42]. The enrichmentof CRISPR-Cas elements and the traces of lateral transferssuggest that BEVs in this bacterium play an important role inthe interspecies recombination of IS and in defense againstprophages through HGT. Although the frequency and extentof gene transfer among bacteria requires further investigation,our study lays the groundwork for identifying transferable genecandidates via BEVs and their potential effects on microbialcommunity.Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025Decoding HGT by single-vesicle genomics | 11The overrepresentation of CRISPR-Cas elements and IS ele-ments in P. gingivalis BEVs also provides significant clues forelucidating the mechanisms of the selective packaging. The BEVDNA is rich in IS elements with typical TIRs (Fig. 2B), suggestingthat these DNA regions tend to be excised and loaded into BEVs viacertain mechanisms. The potential of Cas1-containing elements(so called Casposon) as mobile genetic elements has been positedfrom the aspects of presence of TIRs, the integrase activity of Casprotein, and its potential role in the evolution of CRISPR-Cas [65–67]. Although the identified enriched genomic regions in BEVsin the current study (Fig. 2) and previous one [13] is the typicalCRISPR-Cas cluster rather than Casposons, our genomic analysisalso identified the TIRs possibly works as excision sites (Fig. 2),and the trace of the horizontal transmission in the type III CRISPR-Cas element among Porphyromonas (Fig. 3), suggesting that itspotential as a MGE in BEVs via the excision from the chromosomeand subsequent packaging into BEVs. In contrast, the excisionby Xer family recombinase at dif sequences, one of the widelyobserved recombination system in bacteria [68], would not bethe mechanism of the selective DNA packaging in the targetedbacteria given the no concordance of the dif site and the overrep-resented genomic loci in both bacteria (Supplementary Fig. 9).Our findings highlight the enrichment of specific genomic lociin BEVs and reveal the requirements of certain biogenesis routesfor this enrichment. A biogenesis route for DNA-containing BEVsthat aligns with membrane blebbing was previously proposed [69],where the peptidoglycan layer is locally and transiently weak-ened by autolysins, facilitating the translocation of cytoplasmiccontent, including chromosomal DNA—a process observed inP. gingivalis BEVs [70]. Moreover, outer-inner membrane vesicleformation through blebbing proposed as the major mechanismsof DNA packaging across Gram-negative bacteria [22, 23, 71], sup-porting the prevalence of specific mechanisms for DNA enrich-ment across bacterial species. Excision of IS elements can be alsoinfluenced by environmental stressors or specific genetic factors[72, 73], which may alter the frequency of their incorporation intoBEVs depending on the type of stress [74]. Another mechanismis that BEVs take up free eDNA through a process similar to nat-ural transformation [75]. However, given the biased gene profilederived from BEVs, eDNA contamination was not a major factor(Supplementary Fig. 6). These results suggest that the DNA cargoin P. gingivalis BEVs is not simply a byproduct of cell death (lysis)but rather a result of specific biological mechanisms.Taxonomic classification in NP-DS underscore the rarity of 16SrRNA sequences within BEVs (Fig. 4E), indicating that shotgun-based approach provided a more comprehensive view of thetaxonomic composition of BEV-derived DNA than 16S rRNA-basedapproach. Our analysis revealed substantial taxonomic differ-ences between the profiles obtained from cell-DS and NP-DS(Fig. 5A). This discrepancy suggests that minor members produceBEVs or package DNA into BEVs more actively than major bac-terial groups in human dental plaques. NP-DS identified DNAfrom several other pathogens (e.g. Capnocytophaga and Leptotrichia),which were minor bacterial groups in our cell-DS data (Fig. 4A);however, some of these were prominent pathogens in periodonti-tis [76]. Given the highly organized structure of the oral biofilm[77], it is also possible that bacterial species that tend to formlarge aggregates or become trapped in biofilms are less frequentlydetected by droplet sequencing, whereas BEVs from these bacte-rial taxa are more easily released into the environment becauseof their small size. Assuming that DNA encapsulation into BEVreflects the physiological status of the original cells, we canassume that these enriched DNA regions may serve as biomarkersto assess the physiology of such pathogens. Therefore, the uniquecharacteristics of BEV-derived DNA imply its potential applicabil-ity in biopsy for diagnosis.Our analysis also detected a large amount of phage-derivedDNA in NP-DS. Although the origin of these viral DNA fragmentswould be mostly phages themselves, ∼20% of the dropletscontained both bacterial and viral CDSs (i.e. containing 15%or more bCDS of vCDS). Although such mixed particles wouldpartly result from the aggregations in the purification procedure,incorporation of viral fragments into BEVs by biological processessuch as phage induction and infection in bacterial cells [21,46] or phage infection to BEVs [78] would be possible. Furtherexperimental investigation is necessary to understand thecoexistence of bacterial and phage DNA, yet droplet sequencingcould be a potential tool to explore phage-BEV interactionsin nature.AcknowledgementsWe gratefully thank Prof. Dianne Newman and Prof. YoosephShibu for critical reading of the manuscript. This work wasfinancially supported by JST-PRESTO (JPMJPR19H1 to A.O.), JST-ACT-X (24028593 to So.T.), a JSPS -KAKENHI (grant numbers22H02265 to A.O. and grand numbers 24 K17823 to So.T.),and the Japan Agency for Medical Research and Development(19gm6010002h0004 to A.O. and 21ae0121044h0001 to A.O.and So.T.). This work was supported by Advanced ResearchInfrastructure for Materials and Nanotechnology in Japan of theMinistry of Education, Culture, Sports, Science and Technology(MEXT) (ARIM, PMXP1224NM5397).Author contributionsSo.T. and A.O. designed the research; So.T. developed analysispipelines and analyzed the data; Sa.T., N.D., Y.K., T.M., X.W, M.U.,and A. W. performed the experiments; So.T. and A.O. wrote thefinal version of the paper; All authors contributed to data inter-pretation and the writing of the manuscript.Conflicts of interestThe authors, So.T. and A.O., applied for patents for the single-particle DNA sequencing method. The authors declare no com-peting financial interests.FundingNone declared.Data availabilityAll sequence raw data used in this study were deposited in theDNA Data Bank of Japan (DDBJ) with the accession ID PRJDB17260and PRJDB17266. The intermediate analysis data were alsodeposited in Zenodo (doi:10.5281/zenodo.16662605). Codes for dataanalysis can be found in the GitHub repository: https://github.com/sotarotakano/DropletSequenceAnnotator_2025.References1. Polz MF, Alm EJ, Hanage WP. Horizontal gene transfer and theevolution of bacterial and archaeal population structure. TrendsGenet 2013;29:170–5. https://doi.org/10.1016/j.tig.2012.12.006Downloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-datahttps://academic.oup.com/ismej/article-lookup/doi/10.1093/ismejo/wraf193#supplementary-data10.5281/zenodo.1666260510.5281/zenodo.16662605https://github.com/sotarotakano/DropletSequenceAnnotator_2025https://github.com/sotarotakano/DropletSequenceAnnotator_2025https://github.com/sotarotakano/DropletSequenceAnnotator_2025https://github.com/sotarotakano/DropletSequenceAnnotator_2025https://github.com/sotarotakano/DropletSequenceAnnotator_2025https://doi.org/10.1016/j.tig.2012.12.006https://doi.org/10.1016/j.tig.2012.12.006https://doi.org/10.1016/j.tig.2012.12.006https://doi.org/10.1016/j.tig.2012.12.006https://doi.org/10.1016/j.tig.2012.12.00612 | Takano et al.2. Power JJ, Pinheiro F, Pompei S. et al. Adaptive evolutionof hybrid bacteria by horizontal gene transfer. Proc NatlAcad Sci USA 2021;118:e2007873118. https://doi.org/10.1073/pnas.20078731183. Peabody GL, Li H, Kao KC. Sexual recombination andincreased mutation rate expedite evolution of in varied fitnesslandscapes. Nat Commun 2017;8:2112. https://doi.org/10.1038/s41467-017-02323-44. Cooper TF. Recombination speeds adaptation by reducingcompetition between beneficial mutations in populations ofEscherichia coli. PLoS Biol 2007;5:e225. https://doi.org/10.1371/journal.pbio.00502255. Shapiro BJ, Friedman J, Cordero OX. et al. Population genomics ofearly events in the ecological differentiation of bacteria. Science2012;336:48–51. https://doi.org/10.1126/science.12181986. Smillie CS, Smith MB, Friedman J. et al. Ecology drives a globalnetwork of gene exchange connecting the human microbiome.Nature 2011;480:241–4. https://doi.org/10.1038/nature105717. Schönknecht G, Chen WH, Ternes CM. et al. Gene trans-fer from bacteria and archaea facilitated evolution of anextremophilic eukaryote. Science 2013;339:1207–10. https://doi.org/10.1126/science.12317078. Pang TY, Lercher MJ. Each of 3,323 metabolic innovations in theevolution of arose through the horizontal transfer of a singleDNA segment. Proc Natl Acad Sci USA 2019;116:187–92. https://doi.org/10.1073/pnas.17189971159. Tulkens J, De Wever O, Hendrix A. Analyzing bacterial extracel-lular vesicles in human body fluids by orthogonal biophysicalseparation and biochemical characterization. Nat Protoc 2020;15:40–67. https://doi.org/10.1038/s41596-019-0236-510. Biller SJ, Schubotz F, Roggensack SE. et al. Bacterial vesi-cles in marine ecosystems. Science 2014;343:183–6. https://doi.org/10.1126/science.124345711. Schwechheimer C, Kuehn MJ. Outer-membrane vesicles fromgram-negative bacteria: biogenesis and functions. Nat Rev Micro-biol 2015;13:605–19. https://doi.org/10.1038/nrmicro352512. Brito IL. Examining horizontal gene transfer in microbialcommunities. Nat Rev Microbiol 2021;19:442–53. https://doi.org/10.1038/s41579-021-00534-713. Biller SJ, Ryan MG, Li J. et al. Distinct horizontal gene trans-fer potential of extracellular vesicles versus viral-like parti-cles in marine habitats. Nat Commun 2025;16:2126. https://doi.org/10.1038/s41467-025-57276-w14. Toyofuku M, Nomura N, Eberl L. Types and origins of bacterialmembrane vesicles. Nat Rev Microbiol 2019;17:13–24. https://doi.org/10.1038/s41579-018-0112-215. Toyofuku M, Schild S, Kaparakis-Liaskos M. et al. Compositionand functions of bacterial membrane vesicles. Nat Rev Microbiol2023;21:415–30. https://doi.org/10.1038/s41579-023-00875-516. Fulsundar S, Harms K, Flaten GE. et al. Gene transfer potentialof outer membrane vesicles of Acinetobacter baylyi and effectsof stress on vesiculation. Appl Environ Microbiol 2014;80:3469–83.https://doi.org/10.1128/AEM.04248-1317. Tran F, Boedicker JQ. Genetic cargo and bacterial species setthe rate of vesicle-mediated horizontal gene transfer. Sci Rep2017;7:8813. https://doi.org/10.1038/s41598-017-07447-718. Schaar V, Nordstrom T, Morgelin M. et al. Moraxella catarrhalisouter membrane vesicles carry beta-lactamase and promotesurvival of Streptococcus pneumoniae and haemophilus inf luenzaeby inactivating amoxicillin. Antimicrob Agents Chemother 2011;55:3845–53. https://doi.org/10.1128/AAC.01772-1019. Surve MV, Anil A, Kamath KG. et al. Membrane vesicles of groupb streptococcus disrupt feto-maternal barrier leading to pretermbirth. PLoS Pathog 2016;12:e1005816. https://doi.org/10.1371/journal.ppat.100581620. Bitto NJ, Chapman R, Pidot S. et al. Bacterial membrane vesiclestransport their DNA cargo into host cells. Sci Rep 2017;7:7072.https://doi.org/10.1038/s41598-017-07288-421. Yasuda M, Yamamoto T, Nagakubo T. et al. Phage genesinduce quorum sensing signal release through membrane vesi-cle formation. Microbes Environ 2022;37:ME21067. https://doi.org/10.1264/jsme2.ME2106722. Perez-Cruz C, Carrion O, Delgado L. et al. New type ofouter membrane vesicle produced by the gram-negative bac-terium Shewanella vesiculosa M7T: implications for DNA content.Appl Environ Microbiol 2013;79:1874–81. https://doi.org/10.1128/AEM.03657-1223. Perez-Cruz C, Delgado L, Lopez-Iglesias C. et al. Outer-inner membrane vesicles naturally secreted by gram-negativepathogenic bacteria. PLoS One 2015;10:e0116896. https://doi.org/10.1371/journal.pone.011689624. Biller SJ, McDaniel LD, Breitbart M. et al. Membrane vesiclesin sea water: heterogeneous DNA content and implications forviral abundance estimates. ISME J 2017;11:394–404. https://doi.org/10.1038/ismej.2016.13425. Hosokawa M, Nishikawa Y, Kogawa M. et al. Massively par-allel whole genome amplification for single-cell sequencingusing droplet microfluidics. Sci Rep 2017;7:5199. https://doi.org/10.1038/s41598-017-05436-426. Gole J, Gore A, Richards A. et al. Massively parallel poly-merase cloning and genome sequencing of single cells usingnanoliter microwells. Nat Biotechnol 2013;31:1126–32. https://doi.org/10.1038/nbt.272027. Macosko EZ, Basu A, Satija R. et al. Highly parallel genome-wideexpression profiling of individual cells using nanoliter droplets.Cell 2015;161:1202–14. https://doi.org/10.1016/j.cell.2015.05.00228. Cecil JD, Sirisaengtaksin N, O’Brien-Simpson NM. et al. Outermembrane vesicle-host cell interactions. Microbiol Spectr2019;7:10.1128/microbiolspec.psib-0001-2018. https://doi.org/10.1128/microbiolspec.PSIB-0001-201829. Zhang Z, Liu D, Liu S. et al. The role of Porphyromonas gingivalisouter membrane vesicles in periodontal disease and related sys-temic diseases. Front Cell Infect Microbiol 2020;10:585917. https://doi.org/10.3389/fcimb.2020.58591730. Chen S, Zhou Y, Chen Y. et al. Fastp: an ultra-fast all-in-oneFASTQ preprocessor. Bioinformatics 2018;34:i884–90. https://doi.org/10.1093/bioinformatics/bty56031. Langmead B, Salzberg SL. Fast gapped-read alignment withbowtie 2. Nat Methods 2012;9:357–9. https://doi.org/10.1038/Nmeth.192332. Arikawa K, Ide K, Kogawa M. et al. Recovery of strain-resolvedgenomes from human microbiome through an integrationframework of single-cell genomics and metagenomics. Micro-biome 2021;9:202. https://doi.org/10.1186/s40168-021-01152-433. Caton JG, Armitage G, Berglundh T. et al. A new classifica-tion scheme for periodontal and peri-implant diseases andconditions—introduction and key changes from the 1999 clas-sification. J Clin Periodontol 2018;45 Suppl 20:S1–8. https://doi.org/10.1111/jcpe.1293534. Prokka ST. Rapid prokaryotic genome annotation. Bioinformatics2014;30:2068–9. https://doi.org/10.1093/bioinformatics/btu15335. Buchfink B, Xie C, Huson DH. Fast and sensitive protein align-ment using diamond. Nat Methods 2015;12:59–60. https://doi.org/10.1038/nmeth.317636. Kahlke T, Ralph PJ. Basta—taxonomic classification ofsequences and sequence bins using last common ancestorDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://doi.org/10.1073/pnas.2007873118https://doi.org/10.1073/pnas.2007873118https://doi.org/10.1073/pnas.2007873118https://doi.org/10.1073/pnas.2007873118https://doi.org/10.1038/s41467-017-02323-4https://doi.org/10.1038/s41467-017-02323-4https://doi.org/10.1038/s41467-017-02323-4https://doi.org/10.1038/s41467-017-02323-4https://doi.org/10.1371/journal.pbio.0050225https://doi.org/10.1371/journal.pbio.0050225https://doi.org/10.1371/journal.pbio.0050225https://doi.org/10.1371/journal.pbio.0050225https://doi.org/10.1371/journal.pbio.0050225https://doi.org/10.1126/science.1218198https://doi.org/10.1126/science.1218198https://doi.org/10.1126/science.1218198https://doi.org/10.1126/science.1218198https://doi.org/10.1038/nature10571https://doi.org/10.1038/nature10571https://doi.org/10.1038/nature10571https://doi.org/10.1038/nature10571https://doi.org/10.1126/science.1231707https://doi.org/10.1126/science.1231707https://doi.org/10.1126/science.1231707https://doi.org/10.1126/science.1231707https://doi.org/10.1073/pnas.1718997115https://doi.org/10.1073/pnas.1718997115https://doi.org/10.1073/pnas.1718997115https://doi.org/10.1073/pnas.1718997115https://doi.org/10.1038/s41596-019-0236-5https://doi.org/10.1038/s41596-019-0236-5https://doi.org/10.1038/s41596-019-0236-5https://doi.org/10.1038/s41596-019-0236-5https://doi.org/10.1126/science.1243457https://doi.org/10.1126/science.1243457https://doi.org/10.1126/science.1243457https://doi.org/10.1126/science.1243457https://doi.org/10.1038/nrmicro3525https://doi.org/10.1038/nrmicro3525https://doi.org/10.1038/nrmicro3525https://doi.org/10.1038/nrmicro3525https://doi.org/10.1038/s41579-021-00534-7https://doi.org/10.1038/s41579-021-00534-7https://doi.org/10.1038/s41579-021-00534-7https://doi.org/10.1038/s41579-021-00534-7https://doi.org/10.1038/s41467-025-57276-whttps://doi.org/10.1038/s41467-025-57276-whttps://doi.org/10.1038/s41467-025-57276-whttps://doi.org/10.1038/s41467-025-57276-whttps://doi.org/10.1038/s41467-025-57276-whttps://doi.org/10.1038/s41579-018-0112-2https://doi.org/10.1038/s41579-018-0112-2https://doi.org/10.1038/s41579-018-0112-2https://doi.org/10.1038/s41579-018-0112-2https://doi.org/10.1038/s41579-023-00875-5https://doi.org/10.1038/s41579-023-00875-5https://doi.org/10.1038/s41579-023-00875-5https://doi.org/10.1038/s41579-023-00875-5https://doi.org/10.1128/AEM.04248-13https://doi.org/10.1128/AEM.04248-13https://doi.org/10.1128/AEM.04248-13https://doi.org/10.1128/AEM.04248-13https://doi.org/10.1038/s41598-017-07447-7https://doi.org/10.1038/s41598-017-07447-7https://doi.org/10.1038/s41598-017-07447-7https://doi.org/10.1038/s41598-017-07447-7https://doi.org/10.1128/AAC.01772-10https://doi.org/10.1128/AAC.01772-10https://doi.org/10.1128/AAC.01772-10https://doi.org/10.1128/AAC.01772-10https://doi.org/10.1371/journal.ppat.1005816https://doi.org/10.1371/journal.ppat.1005816https://doi.org/10.1371/journal.ppat.1005816https://doi.org/10.1371/journal.ppat.1005816https://doi.org/10.1371/journal.ppat.1005816https://doi.org/10.1038/s41598-017-07288-4https://doi.org/10.1038/s41598-017-07288-4https://doi.org/10.1038/s41598-017-07288-4https://doi.org/10.1038/s41598-017-07288-4https://doi.org/10.1264/jsme2.ME21067https://doi.org/10.1264/jsme2.ME21067https://doi.org/10.1264/jsme2.ME21067https://doi.org/10.1264/jsme2.ME21067https://doi.org/10.1264/jsme2.ME21067https://doi.org/10.1128/AEM.03657-12https://doi.org/10.1128/AEM.03657-12https://doi.org/10.1128/AEM.03657-12https://doi.org/10.1128/AEM.03657-12https://doi.org/10.1371/journal.pone.0116896https://doi.org/10.1371/journal.pone.0116896https://doi.org/10.1371/journal.pone.0116896https://doi.org/10.1371/journal.pone.0116896https://doi.org/10.1371/journal.pone.0116896https://doi.org/10.1038/ismej.2016.134https://doi.org/10.1038/ismej.2016.134https://doi.org/10.1038/ismej.2016.134https://doi.org/10.1038/ismej.2016.134https://doi.org/10.1038/s41598-017-05436-4https://doi.org/10.1038/s41598-017-05436-4https://doi.org/10.1038/s41598-017-05436-4https://doi.org/10.1038/s41598-017-05436-4https://doi.org/10.1038/nbt.2720https://doi.org/10.1038/nbt.2720https://doi.org/10.1038/nbt.2720https://doi.org/10.1038/nbt.2720https://doi.org/10.1016/j.cell.2015.05.002https://doi.org/10.1016/j.cell.2015.05.002https://doi.org/10.1016/j.cell.2015.05.002https://doi.org/10.1016/j.cell.2015.05.002https://doi.org/10.1016/j.cell.2015.05.002https://doi.org/10.1128/microbiolspec.PSIB-0001-2018https://doi.org/10.1128/microbiolspec.PSIB-0001-2018https://doi.org/10.1128/microbiolspec.PSIB-0001-2018https://doi.org/10.1128/microbiolspec.PSIB-0001-2018https://doi.org/10.1128/microbiolspec.PSIB-0001-2018https://doi.org/10.3389/fcimb.2020.585917https://doi.org/10.3389/fcimb.2020.585917https://doi.org/10.3389/fcimb.2020.585917https://doi.org/10.3389/fcimb.2020.585917https://doi.org/10.1093/bioinformatics/bty560https://doi.org/10.1093/bioinformatics/bty560https://doi.org/10.1093/bioinformatics/bty560https://doi.org/10.1093/bioinformatics/bty560https://doi.org/10.1093/bioinformatics/bty560https://doi.org/10.1038/Nmeth.1923https://doi.org/10.1038/Nmeth.1923https://doi.org/10.1038/Nmeth.1923https://doi.org/10.1038/Nmeth.1923https://doi.org/10.1186/s40168-021-01152-4https://doi.org/10.1186/s40168-021-01152-4https://doi.org/10.1186/s40168-021-01152-4https://doi.org/10.1186/s40168-021-01152-4https://doi.org/10.1111/jcpe.12935https://doi.org/10.1111/jcpe.12935https://doi.org/10.1111/jcpe.12935https://doi.org/10.1111/jcpe.12935https://doi.org/10.1093/bioinformatics/btu153https://doi.org/10.1093/bioinformatics/btu153https://doi.org/10.1093/bioinformatics/btu153https://doi.org/10.1093/bioinformatics/btu153https://doi.org/10.1093/bioinformatics/btu153https://doi.org/10.1038/nmeth.3176https://doi.org/10.1038/nmeth.3176https://doi.org/10.1038/nmeth.3176https://doi.org/10.1038/nmeth.3176Decoding HGT by single-vesicle genomics | 13estimations. Methods Ecol Evol 2019;10:100–3. https://doi.org/10.1111/2041-210x.1309537. Parks DH, Chuvochina M, Waite DW. et al. A standardizedbacterial taxonomy based on genome phylogeny substantiallyrevises the tree of life. Nat Biotechnol 2018;36:996–1004. https://doi.org/10.1038/nbt.422938. Poore GD, Kopylova E, Zhu Q. et al. Microbiome analyses ofblood and tissues suggest cancer diagnostic approach. Nature2020;579:567–74. https://doi.org/10.1038/s41586-020-2095-139. Salter SJ, Cox MJ, Turek EM. et al. Reagent and laboratorycontamination can critically impact sequence-based micro-biome analyses. BMC Biol 2014;12:87. https://doi.org/10.1186/s12915-014-0087-z40. Ho MH, Chen CH, Goodwin JS. et al. Functional advantagesof Porphyromonas gingivalis vesicles. PLoS One 2015;10:e0123448.https://doi.org/10.1371/journal.pone.012344841. Nakao R, Takashiba S, Kosono S. et al. Effect of Porphyromonasgingivalis outer membrane vesicles on gingipain-mediateddetachment of cultured oral epithelial cells and immuneresponses. Microbes Infect 2014;16:6–16. https://doi.org/10.1016/j.micinf.2013.10.00542. Matrishin CB, Haase EM, Dewhirst FE. et al. Phages areunrecognized players in the ecology of the oral pathogenPorphyromonas gingivalis. Microbiome 2023;11:161. https://doi.org/10.1186/s40168-023-01607-w43. Tomaru Y, Nagasaki K. Flow cytometric detection and enu-meration of DNA and RNA viruses infecting marine eukaryoticmicroalgae. J Oceanogr 2007;63:215–21. https://doi.org/10.1007/s10872-007-0023-844. Aird D, Ross MG, Chen WS. et al. Analyzing and minimizing PCRamplification bias in illumina sequencing libraries. Genome Biol2011;12:R18. https://doi.org/10.1186/gb-2011-12-2-r1845. Sato MP, Ogura Y, Nakamura K. et al. Comparison of the sequenc-ing bias of currently available library preparation kits for illu-mina sequencing of bacterial genomes and metagenomes. DNARes 2019;26:391–8. https://doi.org/10.1093/dnares/dsz01746. Turnbull L, Toyofuku M, Hynen AL. et al. Explosive cell lysis asa mechanism for the biogenesis of bacterial membrane vesiclesand biofilms. Nat Commun 2016;7:11220. https://doi.org/10.1038/ncomms1122047. Toyofuku M, Carcamo-Oyarce G, Yamamoto T. et al. Prophage-triggered membrane vesicle formation through peptidoglycandamage in Bacillus subtilis. Nat Commun 2017;8:481. https://doi.org/10.1038/s41467-017-00492-w48. Yoshimasu Y, Ikeda T, Sakai N. et al. Rapid bactericidal actionof propolis against Porphyromonas gingivalis. J Dent Res 2018;97:928–36. https://doi.org/10.1177/002203451875803449. Yoshino N, Ikeda T, Nakao R. Dual inhibitory activity of pet-roselinic acid enriched in fennel against Porphyromonas gin-givalis. Front Microbiol 2022;13:816047. https://doi.org/10.3389/fmicb.2022.81604750. Boutet E, Lieberherr D, Tognolli M. et al. Uniprotkb/swiss-prot. Methods Mol Biol 2007;406:89–112. https://doi.org/10.1007/978-1-59745-535-0_4 Totowa, NJ, Humana Press.51. Wang H, Beier N, Boedeker C. et al. Dinoroseobacter shibae outermembrane vesicles are enriched for the chromosome dimerresolution site dif . mSystems 2021;6:10.1128/msystems.00693-20.https://doi.org/10.1128/mSystems.00693-2052. Shen P, Huang HV. Homologous recombination in Escherichiacoli: dependence on substrate length and homology. Genetics1986;112:441–57. https://doi.org/10.1093/genetics/112.3.44153. Ravenhall M, Skunca N, Lassalle F. et al. Inferring horizontalgene transfer. PLoS Comput Biol 2015;11:e1004095. https://doi.org/10.1371/journal.pcbi.100409554. Douglas GM, Langille MGI. Current and promising approachesto identify horizontal gene transfer events in metagenomes.Genome Biol Evol 2019;11:2750–66. https://doi.org/10.1093/gbe/evz18455. Luo XQ, Wang P, Li JL. et al. Viral community-wide auxiliarymetabolic genes differ by lifestyles, habitats, and hosts. Micro-biome 2022;10:190. https://doi.org/10.1186/s40168-022-01384-y56. Nemoto T, Shiba T, Komatsu K. et al. Discrimination of bac-terial community structures among healthy, gingivitis, andperiodontitis statuses through integrated metatranscriptomicand network analyses. mSystems 2021;6:e0088621. https://doi.org/10.1128/mSystems.00886-2157. Gaillard M, Vallaeys T, Vorholter FJ. et al. The clc elementof pseudomonas sp. strain B13, a genomic island with variouscatabolic properties. J Bacteriol 2006;188:1999–2013. https://doi.org/10.1128/JB.188.5.1999-2013.200658. Durrant MG, Li MM, Siranosian BA. et al. A bioinformatic analy-sis of integrative mobile genetic elements highlights their rolein bacterial adaptation. Cell Host Microbe 2020;27:140–153.e9.https://doi.org/10.1016/j.chom.2019.10.02259. Iguchi A, Iyoda S, Kikuchi T. et al. A complete view of thegenetic diversity of the Escherichia coli O-antigen biosynthesisgene cluster. DNA Res 2015;22:101–7. https://doi.org/10.1093/dnares/dsu04360. Rocchetta HL, Burrows LL, Lam JS. Genetics of O-antigen biosyn-thesis in Pseudomonas aeruginosa. Microbiol Mol Biol Rev 1999;63:523–53. https://doi.org/10.1128/MMBR.63.3.523-553.199961. Reeves P. Evolution of salmonella O antigen variation by interspe-cific gene transfer on a large scale. Trends Genet 1993;9:17–22.https://doi.org/10.1016/0168-9525(93)90067-R62. Hester SE, Park J, Goodfield LL. et al. Horizontally acquireddivergent O-antigen contributes to escape from cross-immunityin the classical bordetellae. BMC Evol Biol 2013;13:209. https://doi.org/10.1186/1471-2148-13-20963. Lerouge I, Vanderleyden J. O-antigen structural variation: mech-anisms and possible roles in animal/plant-microbe interac-tions. FEMS Microbiol Rev 2002;26:17–47. https://doi.org/10.1111/j.1574-6976.2002.tb00597.x64. Watanabe T, Nozawa T, Aikawa C. et al. CRISPR regulationof intraspecies diversification by limiting IS transposition andintercellular recombination. Genome Biol Evol 2013;5:1099–114.https://doi.org/10.1093/gbe/evt07565. Hickman AB, Dyda F. The casposon-encoded Cas1 protein fromAciduliprofundum boonei is a DNA integrase that generates targetsite duplications. Nucleic Acids Res 2015;43:10576–87. https://doi.org/10.1093/nar/gkv118066. Krupovic M, Makarova KS, Forterre P. et al. Casposons: a newsuperfamily of self-synthesizing DNA transposons at the ori-gin of prokaryotic CRISPR-Cas immunity. BMC Biol 2014;12:36.https://doi.org/10.1186/1741-7007-12-3667. Koonin EV, Makarova KS. Mobile genetic elements and evolutionof CRISPR-Cas systems: all the way there and back. Genome BiolEvol 2017;9:2812–25. https://doi.org/10.1093/gbe/evx19268. Kono N, Arakawa K, Tomita M. Comprehensive prediction ofchromosome dimer resolution sites in bacterial genomes. BMCGenomics 2011;12:19. https://doi.org/10.1186/1471-2164-12-1969. Kadurugamuwa JL, Beveridge TJ. Virulence factors are releasedfrom Pseudomonas aeruginosa in association with membranevesicles during normal growth and exposure to gentamicin:a novel mechanism of enzyme secretion. J Bacteriol 1995;177:3998–4008. https://doi.org/10.1128/jb.177.14.3998-4008.199570. Zhou L, Srisatjaluk R, Justus DE. et al. On the origin of membranevesicles in gram-negative bacteria. FEMS Microbiol Lett 1998;163:223–8. https://doi.org/10.1111/j.1574-6968.1998.tb13049.xDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://doi.org/10.1111/2041-210x.13095https://doi.org/10.1111/2041-210x.13095https://doi.org/10.1111/2041-210x.13095https://doi.org/10.1111/2041-210x.13095https://doi.org/10.1038/nbt.4229https://doi.org/10.1038/nbt.4229https://doi.org/10.1038/nbt.4229https://doi.org/10.1038/nbt.4229https://doi.org/10.1038/s41586-020-2095-1https://doi.org/10.1038/s41586-020-2095-1https://doi.org/10.1038/s41586-020-2095-1https://doi.org/10.1038/s41586-020-2095-1https://doi.org/10.1186/s12915-014-0087-zhttps://doi.org/10.1186/s12915-014-0087-zhttps://doi.org/10.1186/s12915-014-0087-zhttps://doi.org/10.1186/s12915-014-0087-zhttps://doi.org/10.1186/s12915-014-0087-zhttps://doi.org/10.1371/journal.pone.0123448https://doi.org/10.1371/journal.pone.0123448https://doi.org/10.1371/journal.pone.0123448https://doi.org/10.1371/journal.pone.0123448https://doi.org/10.1371/journal.pone.0123448https://doi.org/10.1016/j.micinf.2013.10.005https://doi.org/10.1016/j.micinf.2013.10.005https://doi.org/10.1016/j.micinf.2013.10.005https://doi.org/10.1016/j.micinf.2013.10.005https://doi.org/10.1016/j.micinf.2013.10.005https://doi.org/10.1186/s40168-023-01607-whttps://doi.org/10.1186/s40168-023-01607-whttps://doi.org/10.1186/s40168-023-01607-whttps://doi.org/10.1186/s40168-023-01607-whttps://doi.org/10.1186/s40168-023-01607-whttps://doi.org/10.1007/s10872-007-0023-8https://doi.org/10.1007/s10872-007-0023-8https://doi.org/10.1007/s10872-007-0023-8https://doi.org/10.1007/s10872-007-0023-8https://doi.org/10.1186/gb-2011-12-2-r18https://doi.org/10.1186/gb-2011-12-2-r18https://doi.org/10.1186/gb-2011-12-2-r18https://doi.org/10.1186/gb-2011-12-2-r18https://doi.org/10.1186/gb-2011-12-2-r18https://doi.org/10.1093/dnares/dsz017https://doi.org/10.1093/dnares/dsz017https://doi.org/10.1093/dnares/dsz017https://doi.org/10.1093/dnares/dsz017https://doi.org/10.1093/dnares/dsz017https://doi.org/10.1038/ncomms11220https://doi.org/10.1038/ncomms11220https://doi.org/10.1038/ncomms11220https://doi.org/10.1038/ncomms11220https://doi.org/10.1038/s41467-017-00492-whttps://doi.org/10.1038/s41467-017-00492-whttps://doi.org/10.1038/s41467-017-00492-whttps://doi.org/10.1038/s41467-017-00492-whttps://doi.org/10.1038/s41467-017-00492-whttps://doi.org/10.1177/0022034518758034https://doi.org/10.1177/0022034518758034https://doi.org/10.1177/0022034518758034https://doi.org/10.3389/fmicb.2022.816047https://doi.org/10.3389/fmicb.2022.816047https://doi.org/10.3389/fmicb.2022.816047https://doi.org/10.3389/fmicb.2022.816047https://doi.org/10.1007/978-1-59745-535-0_4https://doi.org/10.1007/978-1-59745-535-0_4https://doi.org/10.1007/978-1-59745-535-0_4https://doi.org/10.1128/mSystems.00693-20https://doi.org/10.1128/mSystems.00693-20https://doi.org/10.1128/mSystems.00693-20https://doi.org/10.1128/mSystems.00693-20https://doi.org/10.1093/genetics/112.3.441https://doi.org/10.1093/genetics/112.3.441https://doi.org/10.1093/genetics/112.3.441https://doi.org/10.1093/genetics/112.3.441https://doi.org/10.1371/journal.pcbi.1004095https://doi.org/10.1371/journal.pcbi.1004095https://doi.org/10.1371/journal.pcbi.1004095https://doi.org/10.1371/journal.pcbi.1004095https://doi.org/10.1371/journal.pcbi.1004095https://doi.org/10.1093/gbe/evz184https://doi.org/10.1093/gbe/evz184https://doi.org/10.1093/gbe/evz184https://doi.org/10.1093/gbe/evz184https://doi.org/10.1093/gbe/evz184https://doi.org/10.1186/s40168-022-01384-yhttps://doi.org/10.1186/s40168-022-01384-yhttps://doi.org/10.1186/s40168-022-01384-yhttps://doi.org/10.1186/s40168-022-01384-yhttps://doi.org/10.1186/s40168-022-01384-yhttps://doi.org/10.1128/mSystems.00886-21https://doi.org/10.1128/mSystems.00886-21https://doi.org/10.1128/mSystems.00886-21https://doi.org/10.1128/mSystems.00886-21https://doi.org/10.1128/JB.188.5.1999-2013.2006https://doi.org/10.1128/JB.188.5.1999-2013.2006https://doi.org/10.1128/JB.188.5.1999-2013.2006https://doi.org/10.1128/JB.188.5.1999-2013.2006https://doi.org/10.1016/j.chom.2019.10.022https://doi.org/10.1016/j.chom.2019.10.022https://doi.org/10.1016/j.chom.2019.10.022https://doi.org/10.1016/j.chom.2019.10.022https://doi.org/10.1016/j.chom.2019.10.022https://doi.org/10.1093/dnares/dsu043https://doi.org/10.1093/dnares/dsu043https://doi.org/10.1093/dnares/dsu043https://doi.org/10.1093/dnares/dsu043https://doi.org/10.1093/dnares/dsu043https://doi.org/10.1128/MMBR.63.3.523-553.1999https://doi.org/10.1128/MMBR.63.3.523-553.1999https://doi.org/10.1128/MMBR.63.3.523-553.1999https://doi.org/10.1128/MMBR.63.3.523-553.1999https://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1016/0168-9525(93)90067-Rhttps://doi.org/10.1186/1471-2148-13-209https://doi.org/10.1186/1471-2148-13-209https://doi.org/10.1186/1471-2148-13-209https://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1111/j.1574-6976.2002.tb00597.xhttps://doi.org/10.1093/gbe/evt075https://doi.org/10.1093/gbe/evt075https://doi.org/10.1093/gbe/evt075https://doi.org/10.1093/gbe/evt075https://doi.org/10.1093/gbe/evt075https://doi.org/10.1093/nar/gkv1180https://doi.org/10.1093/nar/gkv1180https://doi.org/10.1093/nar/gkv1180https://doi.org/10.1093/nar/gkv1180https://doi.org/10.1093/nar/gkv1180https://doi.org/10.1186/1741-7007-12-36https://doi.org/10.1186/1741-7007-12-36https://doi.org/10.1186/1741-7007-12-36https://doi.org/10.1093/gbe/evx192https://doi.org/10.1093/gbe/evx192https://doi.org/10.1093/gbe/evx192https://doi.org/10.1093/gbe/evx192https://doi.org/10.1093/gbe/evx192https://doi.org/10.1186/1471-2164-12-19https://doi.org/10.1186/1471-2164-12-19https://doi.org/10.1186/1471-2164-12-19https://doi.org/10.1128/jb.177.14.3998-4008.1995https://doi.org/10.1128/jb.177.14.3998-4008.1995https://doi.org/10.1128/jb.177.14.3998-4008.1995https://doi.org/10.1128/jb.177.14.3998-4008.1995https://doi.org/10.1111/j.1574-6968.1998.tb13049.xhttps://doi.org/10.1111/j.1574-6968.1998.tb13049.xhttps://doi.org/10.1111/j.1574-6968.1998.tb13049.xhttps://doi.org/10.1111/j.1574-6968.1998.tb13049.xhttps://doi.org/10.1111/j.1574-6968.1998.tb13049.xhttps://doi.org/10.1111/j.1574-6968.1998.tb13049.x14 | Takano et al.71. Hagemann S, Stoger L, Kappelmann M. et al. DNA-bearingmembrane vesicles produced by Ahrensia kielensis and Pseudoal-teromonas marina. J Basic Microbiol 2014;54:1062–72. https://doi.org/10.1002/jobm.20130037672. Siguier P, Gourbeyre E, Chandler M. Bacterial insertionsequences: their genomic impact and diversity. FEMS MicrobiolRev 2014;38:865–91. https://doi.org/10.1111/1574-6976.1206773. Kusumoto M, Ooka T, Nishiya Y. et al. Insertion sequence-excision enhancer removes transposable elements from bacte-rial genomes and induces various genomic deletions. Nat Com-mun 2011;2:15210. https://doi.org/10.1038/ncomms115274. Orench-Rivera N, Kuehn MJ. Differential packaging into outermembrane vesicles upon oxidative stress reveals a generalmechanism for cargo selectivity. Front Microbiol 2021;12:561863.https://doi.org/10.3389/fmicb.2021.56186375. Renelli M, Matias V, Lo RY. et al. DNA-containing membrane vesi-cles of Pseudomonas aeruginosa PAO1 and their genetic transfor-mation potential. Microbiology (Reading) 2004;150:2161–9. https://doi.org/10.1099/mic.0.26841-076. Kononen E, Fteita D, Gursoy UK. et al. Prevotella species asoral residents and infectious agents with potential impacton systemic conditions. J Oral Microbiol 2022;14:14. https://doi.org/10.1080/20002297.2022.207981477. Mark Welch JL, Rossetti BJ, Rieken CW. et al. Biogeogra-phy of a human oral microbiome at the micron scale. ProcNatl Acad Sci USA 2016;113:E791–800. https://doi.org/10.1073/pnas.152214911378. Manning AJ, Kuehn MJ. Contribution of bacterial outer mem-brane vesicles to innate bacterial defense. BMC Microbiol2011;11:258. https://doi.org/10.1186/1471-2180-11-258© The Author(s) 2025. Published by Oxford University Press on behalf of the International Society for Microbial Ecology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.The ISME Journal, 2025, 19(1), wraf193https://doi.org/10.1093/ismejo/wraf193Original ArticleDownloaded from https://academic.oup.com/ismej/article/19/1/wraf193/8243901 by guest on 25 December 2025https://doi.org/10.1002/jobm.201300376https://doi.org/10.1002/jobm.201300376https://doi.org/10.1002/jobm.201300376https://doi.org/10.1002/jobm.201300376https://doi.org/10.1111/1574-6976.12067https://doi.org/10.1111/1574-6976.12067https://doi.org/10.1111/1574-6976.12067https://doi.org/10.1038/ncomms1152https://doi.org/10.1038/ncomms1152https://doi.org/10.1038/ncomms1152https://doi.org/10.1038/ncomms1152https://doi.org/10.3389/fmicb.2021.561863https://doi.org/10.3389/fmicb.2021.561863https://doi.org/10.3389/fmicb.2021.561863https://doi.org/10.3389/fmicb.2021.561863https://doi.org/10.1099/mic.0.26841-0https://doi.org/10.1099/mic.0.26841-0https://doi.org/10.1099/mic.0.26841-0https://doi.org/10.1099/mic.0.26841-0https://doi.org/10.1080/20002297.2022.2079814https://doi.org/10.1080/20002297.2022.2079814https://doi.org/10.1080/20002297.2022.2079814https://doi.org/10.1073/pnas.1522149113https://doi.org/10.1073/pnas.1522149113https://doi.org/10.1073/pnas.1522149113https://doi.org/10.1073/pnas.1522149113https://doi.org/10.1186/1471-2180-11-258https://doi.org/10.1186/1471-2180-11-258https://doi.org/10.1186/1471-2180-11-258https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://doi.org/10.1093/ismejo/wraf193  Enrichment of horizontally transferred gene clusters in bacterial extracellular vesicles via non lytic mechanisms Introduction Materials and methods Results Discussion Acknowledgements Author contributions Conflicts of interest Funding Data availability