Microdroplet-based high-throughput screening for antagonistic bacteria targeting penaeid shrimp pathogenic Vibrio harveyi Authors: Hajime Yuasa, Yuri Katayama, Kayo Konishi, Miho Furukawa, Reiko Nozaki, Hidehiro Kondo, Ikuo Hirono, Keiichiro Koiwai Published: 2026 ABSTRACT AbstractAntagonistic bacteria that suppress the growth of specific bacteria have attracted attention as an antibiotics-independent strategy for infectious disease control in aquaculture. Microfluidics-based water-in-oil droplets (microdroplets) enable high-throughput screening of antagonistic bacteria in the field of medicine or agriculture. However, the use of this screening system in aquaculture has not yet been reported. In particular, penaeid shrimp aquaculture, one of the largest sectors of global aquaculture, has a strong demand for alternative disease control strategies because vaccination is ineffective. Here, we demonstrated a proof-of-concept study of microdroplet-based high-throughput screening system for antagonistic bacteria targeting penaeid shrimp pathogenic Vibrio harveyi. Using this screening system, we successfully isolated 18 bacterial candidates with potential growth-inhibitory activity, representing three genera (Pseudoalteromonas, Shewanella, and Tenacibaculum), from the bacterial community of kuruma shrimp Penaeus japonicus rearing water. 16S rRNA gene-based bacterial community analysis revealed that these isolates included several low-abundance, rare taxa. Although these isolates showed no inhibitory activity on agar plates, one out of four tested strains showed a trend toward improved survival during co-infection tests using kuruma shrimp. Overall, our study highlights both the potential and limitation of microdroplet-based antagonistic bacterial screening to accelerate the development of biological control strategies in shrimp aquaculture.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-68242-x.Keywords: Microdroplet, High-throughput screening, Antagonistic bacteria, Penaeid shrimp, Aquaculture, VibrioSubject terms: Biotechnology, Microbiology Introduction Control of infectious disease is important for stable food production. Antibiotics have been highly effective in the control of bacterial infectious diseases. However, the use of antibiotics is highly restricted due to the risk of antibiotic-resistant bacteria1. As an antibiotics-independent strategy, antagonistic bacteria that can suppress the growth of specific bacteria are a typical target for effective utilization. These bacteria exhibit antagonistic activity against other bacteria through diverse mechanisms, including the production of antimicrobial substances, resource competition, and environmental modification2. These bacteria have been utilized to suppress bacterial diseases in diverse fields such as medicine3, agriculture4 and aquaculture5. Aquaculture has a strong demand for the practical implementation of antagonistic bacteria, because antibiotics are easily dispersed into the surrounding environment through water. Especially, penaeid shrimp, such as whiteleg shrimp Penaeus vannamei, black tiger shrimp P. monodon and kuruma shrimp P. japonicus are the most widely farmed crustacean species in the world6 and frequently experience outbreaks of infectious diseases caused by microorganisms, resulting in significant production losses7,8. Penaeid shrimp aquaculture, which cannot utilize vaccines due to the lack of adaptive immunity, has a strong demand for the practical implementation of biological control9,10. For over a century, agar plate-based bacterial culture methods have been widely employed for isolating antagonistic bacteria from natural environments. However, recent advances in culture-independent analysis have revealed that numerous bacteria in natural environments remain uncultivable on agar plates, including potential antagonistic agents yet to be explored11,12. Moreover, agar plate-based bacterial isolation requires large amounts of space, materials, and manual labor. Therefore, there is a strong demand for the development of next-generation screening methods for beneficial bacteria. Microfluidics-based water-in-oil droplets (microdroplets) offer a promising alternative for high-throughput screening of antagonistic bacteria. This technology can easily and rapidly generate uniform and dispersed pico- to nano-liter droplets13. These microdroplets exhibit high stability and biocompatibility, owing to the use of fluorinated oil. Therefore, these are utilized as independent reactors for cell growth, differentiation, and chemical or enzymatic reaction14. In recent years, fluorescence-based microdroplet sorting techniques have been reported and applied to screening for antagonistic bacteria15–17. Compared to conventional agar plate-based methods, which are often limited by the overgrowth of fast-growing dominant strains, this microdroplet strategy offers distinct advantages in both selection efficiency and throughput. The physical compartmentalization within microdroplets effectively suppresses interbacterial competition, thereby facilitating the isolation of slow-growing and rare bacterial species that would otherwise be outcompeted on agar plates18–20. Furthermore, their compactness and high multiplicity enable. the rapid screening of massive bacterial populations, significantly accelerating the discovery process. The industrial application of this technology is still in its infancy, as recent studies have been conducted using hand-made microfluidic devices. Although this technique has been applied to human15,16 and plant pathogenic bacteria21, its application to aquaculture pathogens has not yet been reported. In this study, we developed microdroplet-based high-throughput screening platform for antagonistic bacteria against penaeid shrimp pathogenic V. harveyi (Fig. 1). This strain was isolated from a kuruma shrimp that originated from a shrimp farm on Kumejima, Okinawa Prefecture, Japan, in 201922. This method provided a user-friendly and reproducible screening system based on commercially available microfluidic devices from On-chip Biotechnologies Co., Ltd. We co-cultivated V. harveyi with bacteria from kuruma shrimp rearing water bacteria at a 1:10 ratio. Using our screening platform, we isolated 18 bacterial candidates with potential growth-inhibitory activity, which were clustered into four operational taxonomic units (OTUs) based on 99% sequence similarity. Here, we present a proof-of-concept for microdroplet-based high-throughput screening platform of antagonistic bacteria aquaculture pathogens. Schematic of the workflow for microdroplet-based high-throughput screening for antagonistic bacteria against penaeid shrimp Vibrio harveyi. As the screening inputs, bacterial communities cultivated from kuruma shrimp rearing water and enhanced green fluorescent protein (eGFP) -expressing V. harveyi were prepared (step 1). These were co-encapsulated into microdroplets of approximately 30 μm in diameter using On-chip Droplet Generator and statically co-cultured at 25 °C (step 2). Microdroplets exhibiting reduced GFP fluorescence - indicative of V. harveyi growth inhibition by environmental antagonistic bacteria - were selectively recovered through high-throughput fluorescence-based sorting and subsequently re-cultured in 96-well plates (step 3). The re-cultured candidate antagonistic strains were analyzed by 16S rRNA gene sequencing and their antagonistic activity was further evaluated by co-infection assays with V. harveyi in kuruma shrimp (step 4). Results Generation of enhanced green fluorescent protein (eGFP) expressingV. harveyi In this screening system, enhanced green fluorescent protein (eGFP) was employed as a marker to evaluate the growth inhibition of V. harveyi in microdroplets. To obtain high stability upon subculturing for many generations, eGFP-expressing gene module was inserted into the middle region of 16S rRNA gene of V. harveyi (Fig. 2A). This module consists of three genetic elements: eGFP gene, a self-cleaving ribozyme (RiboJ), and a bicistronic design element (BCD). RiboJ is the sequence of ribozyme that cleaves the 5’ untranslated region (5’-UTR) of the mRNA to minimize fluctuations in translation efficiency and achieve stable gene expression levels23. BCD is based on a constitutive promoter and tuning by using two ribosomal binding sites RBSs that are translationally coupled24. To insert this eGFP-expressing genetic module into the 16S rRNA gene region of V. harveyi, we employed homologous recombination-based gene integration method (Fig. 2A, Methods). This method has been reported to enable foreign gene integration into a wide range of Gram-negative bacteria, including the genus Vibrio25,26. The generated fluorescent strain was observed using a fluorescence microscope (Fig. 2B). Whole-genome resequencing of our eGFP-expressing V. harveyi using Oxford Nanopore Technologies (ONT) flowcell R9.4.1 confirmed that our genetic module was inserted into a single copy 16S rRNA gene located on chromosome 2 (Fig. 2C). As a result, we successfully generated genetically modified V. harveyi expected to exhibit stable and constitutive expression of eGFP through chromosomal integration. Generation of enhanced green fluorescence protein expressing Vibrio harveyi. (A) Schematic representation of the homologous recombination system introducing a genetic module consisting of RiboJ, BCD, and eGFP into the 16S rRNA gene of V. harveyi. (B) Fluorescence microscopy image of eGFP-expressing V. harveyi. (C) Assembly graph of the eGFP-expressing V. harveyi genome resequenced and visualized using Bandage v.0.8.1. Isolation of 18 candidate antagonistic bacteria by microdroplet-based high-throughput screening Vibrio species are ubiquitous in penaeid shrimp rearing environments, so it is plausible that bacterial species possessing antagonistic activity against them exist in these environments because of niche adaptation. In addition, isolates from the penaeid shrimp rearing environments are expected to possess optimal temperature, pH, and metabolic profiles suitable for practical application in aquaculture settings27,28. Therefore, in this study, we selected the bacterial community present in the laboratory-maintained water of Penaeus japonicus as a source for screening antagonistic bacteria. The bacterial community scaled up by cultivation in Marine Broth, a medium commonly used for the culture of marine bacteria, or 10-fold diluted Matine Broth were co-encapsulated and co-cultured with eGFP-expressing V. harveyi in microdroplets with a diameter of approximately 30 μm. For microdroplet encapsulation, the bacterial concentration was adjusted to a 1:10 ratios of environmental bacteria to eGFP-expressing V. harveyi, allowing isolation of only strongly antagonistic strains. To verify the growth of environmental bacteria, On-chip MiMe Stain (On-chip Biotechnologies, Japan) that broadly labels bacteria through membrane staining was added in the suspension. After three days of cultivation, several microdroplets showing loss of fluorescence from eGFP-expressing V. harveyi together with fluorescence from On-chip MiMe Stain were observed (Fig. 3A). Following the observation of microdroplets, fluorescence-based microdroplet sorting was performed using On-chip Droplet Selector (On-chip Biotechnologies, Japan). Sorting gates were set to isolate microdroplets in which fluorescence derived from eGFP-expressing V. harveyi was low and fluorescence derived from MiMe Stain was high (P1 and P3), as well as microdroplets in which both eGFP-expressing V. harveyi-derived and MiMe Stain-derived fluorescence were low (P2 and P4) (Fig. 3B). P2 and P4 were designed to capture antagonistic bacteria capable of inhibiting V. harveyi growth, even without proliferating within the microdroplets. In the gate of P1, a total of 319,784 microdroplets were analyzed, with 81 were dispensed into a 96-well plate, and bacterial growth was confirmed in 15 wells. In the gate of P2, a total of 92,124 microdroplets were analyzed, and 48 were isolated into a 96-well plate. However, no bacterial growth was detected in any well. In the gate of P3, a total of 381,634 microdroplets were analyzed, with 76 dispensed into a 96-well plate, and bacterial growth was confirmed in 2 wells. In the gate of P4, a total of 106,281 microdroplets were analyzed, with 48 dispensed into a 96-well plate, and bacterial growth was confirmed in 3 wells (Supplementary Table 6). Isolation of antagonistic bacterial candidates via microdroplet-based screening. (A) Microdroplets after three days of co-cultivation in Marine Broth or 10-fold diluted Marine Broth. (B) Scatter plot of microdroplet fluorescence intensities detected by the On-chip Droplet Selector. The x-axis represents eGFP fluorescence intensity, and the y-axis represents MiMe Stain fluorescence intensity. The On-chip Droplet Selector was set to isolate microdroplets falling within gates P1, P2, P3, and P4. (C) Phylogenetic tree of 18 isolated candidate antagonistic strains. The tree was constructed based on ~ 1,400 bp full-length 16S rRNA gene sequences using MAFFT with 1,000 bootstrap replications. P1, P3, and P4 correspond to the gates set by the On-chip Droplet Selector. The isolated bacteria conducted species identification based on the full-length 16S rRNA gene, revealing that the isolates consisted of fifteen strains of the genus Pseudoalteromonas, one strains of the genus Shewanella, and two strains of the genus Tenacibaculum (Fig. 3C, Supplementary Table 2). Interestingly, the genus Pseudoalteromonas were isolated only from P1 (except for Candidate 18 from P4), where Marine Broth was used for bacterial cultivation, whereas the genus Tenacibaculum and the genus Shewanella were isolated from P3 and P4, respectively, where 10-fold diluted Marine Broth was used (Fig, 3 C). Clustering of the full-length 16S rRNA gene sequences with 99% sequence similarity grouped the isolates into a total of four operational taxonomic units (OTUs) (Fig. 3C). Rarity of the isolated bacteria One of the advantages of microdroplet-based screening is that it allows rare bacteria present in the environment to be targeted for screening9–11. To determine the relative abundance of the antagonistic bacterial candidates isolated in this study, we performed a bacterial community analysis based on V4 regions of the 16S rRNA gene sequences (Fig. 4). Kuruma shrimp rearing water used as the screening source (Shrimp rearing water), Marine Broth scaled-up culture (1 x Marine Broth), and 10-fold diluted Marine Broth scale-up culture (0.1 x Marine Broth) were used to extract the environmental DNA. These bacterial cultivates were directly used for antagonistic bacteria screening of this study. A total of 88,828 reads from three samples (mean 29,609.3 per sample) were obtained by the amplicon sequencing of the V4 region of the 16S rRNA gene. After running the DADA2 program, 76,308 high-quality reads remained (mean 25,436 per sample). A total of 275 Amplicon Sequence Variants (ASVs) were detected from the kuruma shrimp rearing water, 20 ASVs from the 1 x Marine Broth sample, and 16 ASVs from the 0.1 x Marine Broth sample (Fig. 4A). In the kuruma shrimp rearing water, Vibrio-associated ASVs constituted approximately 25% of the total community and represented the dominant bacterial group (Fig. 4A). Their relative abundance further increased through cultivation in Marine Broth, reaching approximately 85% of the total ASVs in the 0.1 x Marine Broth sample (Fig. 4A). Although it is generally expected that dilution of the medium enhances the cultivation of a more diverse range of bacteria, in the present study the undiluted Marine Broth culture exhibited higher diversity (Fig. 4A). Relative abundance of screening sources. (A) Stacked bar plot of bacterial communities at the genus level based on the V4 region of the 16S rRNA gene. From left to right: bacterial communities in kuruma shrimp rearing water before cultivation, after cultivation in Marine Broth, and after cultivation in 10-fold diluted Marine Broth. (B) Relative abundances of ASVs corresponding to the three genera (Pseudoalteromonas, Shewanella, and Tenacibaculum) that were isolated in this screening. The candidate antagonistic bacteria isolated in this study exhibited low relative abundances within their respective source communities (Fig. 4B). In the bacterial communities cultured in 10-fold diluted Marine Broth, from which Shewanella and Tenacibaculum strains were isolated, the relative abundances of Shewanella- and Tenacibaculum-associated ASVs were 1.13% and 0.26%, respectively (Fig. 4B). In contrast, Pseudoalteromonas genus showed a relatively high abundance in the Marine Broth cultured community, from which this genus was isolated, with Pseudoalteromonas-associated ASVs accounting for 11.79% of the total community (Fig. 4B). Infection control activity againstV. harveyiinP. japonicus To evaluate the effectiveness of the isolates in this screening, co-cultivation assays on agar plates or artificial seawater. and co-infection experiments via intramuscular injection into P. japonicus were conducted. In the cross-streak assays, none of the isolates exhibited inhibitory activity against V. harveyi (Fig. 5A). For the infection experiments, one representative strain from each of the four OTUs classified by 16S rRNA gene analysis was selected. The bacterial concentration was set at 8.0 × 105 bacteria/shrimp—the dose at which V. harveyi infection was confirmed to cause mortality based on our preliminary infection trials—and both V. harveyi and the isolates were simultaneously injected intramuscularly. The infection experiments revealed that one out of four tested strains showed a trend toward improved survival (Fig. 5B). Notably, Shewanella spp. (classified as OTU2 based on 16S rRNA gene identity) maintained a 30% mortality rate in kuruma shrimp under V. harveyi co-injection conditions. However, shrimp mortality was also observed in the control group, which was treated with artificial seawater only. Effectiveness of the isolated antagonistic bacteria candidates. (A) Evaluation of antagonistic activity using the cross-streaking method. The horizontal line is eGFP-expressing V. harveyi, and the vertical lines are each isolated bacterium. (B) Co-infection test of wild-type V. harveyi and isolated bacteria via intramuscular injection (N = 10). Kuruma shrimp with an average body weight of 6.0 g were used, and the infection concentration of each bacteria was 8.0 × 105 bacteria per shrimp. Statistical analysis was performed using the log-rank test, relative to the positive control group. Discussions This study represents the first attempt to apply a microdroplet-based screening platform for the isolation of antagonistic bacteria against shrimp pathogenic bacteria. Traditionally, the isolation and screening of antagonistic bacteria have predominantly relied on the cultivation-based method introduced by Robert Koch using agar plate media29. This method has also been widely employed for pathogenic bacteria in aquaculture28,30, but it remains a major bottleneck in responding to recurrent and emerging infectious diseases. In this study, we selected V. harveyi, a major pathogenic bacterium of kuruma shrimp, as a model organism and successfully screened 899,823 microdroplets in four culture conditions, leading to the isolation of 18 bacterial candidates with potential growth-inhibitory activity. The fluorinated oil used to generate microdroplets is known to have high oxygen solubility31,32, which was also confirmed in the present work by the high growth rates of bacterial cells within microdroplets. Because microdroplets with a diameter of approximately 30 μm were used as co-culture compartments, the entire cultivation procedure could be achieved within 1.5-mL microtubes, providing a space-efficient workflow. Antagonistic interactions were identified by fluorescence quenching of green fluorescent protein expressed by the pathogenic strain within microdroplets. High-throughput screening of droplets with low fluorescence intensity was achieved by using the On-chip Droplet Selector, a fluorescence-activated droplet sorting (FADS) system developed by On-chip Biotechnologies. We isolated 15 strains of Pseudoalteromonas, two strains of Tenacibaculum, and one strain of Shewanella. Based on full-length 16S rRNA gene sequences determined by Sanger sequencing, the isolates clustered into four OTUs, each sharing ≥ 99% sequence similarity. In general, 16S rRNA gene sequence similarities ≥ 98–99% are widely accepted as indicative of the same bacterial species33. However, because the 16S rRNA gene represents only a very small fraction of the bacterial genome, species-level identification cannot always be concluded solely from high 16S similarity34,35, and therefore these OTU groupings should be regarded as provisional. All three genera isolated in this study have been reported as beneficial bacteria in aquaculture36,37. In particular, Shewanella spp. have been shown to colonize shrimp intestines and are associated with resistance to pathogenic Vibrio infection38. Notably, 16S rRNA gene sequencing analyses indicated that these isolates were originally rare taxa in the environmental samples used for screening. This result supports previous reports demonstrating that microdroplet-based cultivation can selectively recover rare environmental bacteria that are typically missed by conventional cultivation approaches15–17. We acknowledge that the reliance on 16S rRNA gene sequencing is a limitation of this study, as it does not allow for the comparison of genomic differences beyond this specific locus. Since whole-genome sequencing was not performed, we could not evaluate variations in other genes that might contribute to the observed antagonistic phenotypes or functional diversity. Furthermore, the observed dependency of the isolates on specific cultivation media—with Pseudoalteromonas emerging only in full-strength broth and Shewanella and Tenacibaculum in diluted broth—highlights the role of nutrient availability and niche occupancy in this screening system. While this could suggest a selection for rapid growth under specific nutrient regimes, resource competition is widely recognized as a fundamental and effective mechanism of bacterial antagonism. Given that the average encapsulation density was approximately 5 environmental cells to 50 V. harveyi cells per droplet, the successful suppression of the pathogen likely required high ‘functional dominance’ through efficient nutrient utilization and rapid niche occupation. Therefore, our platform provides the versatility to capture diverse functional taxa by ‘tuning’ environmental parameters, such as nutrient concentration, to recover candidates capable of outcompeting pathogens under varied conditions. Despite their successful isolation, the antagonistic effects of these strains against V. harveyi in infection challenges were limited. Interestingly, cross-streak assays on Marine Agar failed to reveal growth inhibition of V. harveyi by any of the isolates. A similar phenomenon was reported by Mahler et al., who documented bacterial taxa that proliferate and express functional phenotypes inside droplets but exhibit poor growth or unstable maintenance under conventional agar cultivation17. These findings suggest that antagonistic traits detected in microdroplets may not be directly reproducible under plate culture conditions, highlighting notable context-dependent differences in bacterial behavior between the two environments. In addition, because bacterial encapsulation into droplets follows Poisson distribution39, it is possible that some fluorescently negative droplets simply contained higher numbers of environmental bacteria relative to the pathogenic reporter, potentially resulting in apparent antagonistic signals. Also, the initial bacterial concentrations were not investigated in this study; therefore, it is necessary to evaluate how variations in concentration ratios influence antagonistic activity in future research. Furthermore, agar plates represent a specialized solid-surface environment that differs significantly from the liquid phase within microdroplets. Since our screening was conducted in a liquid nutrient regime, it is likely that the antagonistic traits observed in the droplets are better evaluated through liquid-based co-culture assays rather than conventional agar-based methods. Therefore, future studies should incorporate liquid-phase validations to more accurately recapitulate the environmental conditions of the initial screening platform. However, in vivo co-infection tests using intramuscular injections in kuruma shrimp demonstrated that one out of four tested strains showed a trend toward improved survival. If bacterial strains clustered within the same OTU exhibit similar phenotypes, one representative strain from each of the four OTUs was selected for infection trials. Among these four strains, the Shewanella isolate grouped into OTU2 maintained shrimp survival at 70%. Because mortality was also observed in shrimp injected with individual isolates in the absence of pathogens, our findings highlight a potential limitation of this screening strategy: the failure to consider host sensitivity toward candidate isolates. Furthermore, the bacterial inoculum used in this study was higher than that used in previous studies40, raising the possibility that the mortality observed may have resulted from endotoxin shock due to excessive lipopolysaccharide exposure41. Also, mortality was also observed in kuruma shrimp injected only with artificial seawater, suggesting that the physical handling and the injection process itself exert significant stress on the shrimps. Future studies should explore more natural infection methods that impose less physiological burden, such as immersion assays, to better evaluate the protective efficacy of candidate strains in a real-world aquaculture environment. Overall, this study provides fundamental insights into the application of microdroplet technology for discovering bacterial candidates with potential growth-inhibitory activity from aquatic environment. While this approach demonstrated the potential to efficiently isolate candidate antagonistic bacteria, including rare environmental bacteria, it also revealed a critical limitation in that host susceptibility cannot be assessed during screening. Furthermore, antagonistic activity detected within microdroplets did not necessarily correlate with growth inhibition on agar plates or disease suppression in shrimp infection assays, indicating that the expression of antagonistic traits is highly environment dependent. Moreover, because the present in vivo assays were limited by a small sample size (10 shrimp per group) and baseline mortality in the control groups, these infection trials remain a preliminary proof-of-concept. Future work should focus on establishing a multi-step evaluation framework that integrates host safety, colonization ability, metabolic characteristics, and the genetic basis of antimicrobial compound production. Methods Generation of enhanced green fluorescent protein expressing V. harveyi A constitutive eGFP expression cassette (eGFP–RiboJ–BCD) was synthesized commercially (Integrated DNA Technologies, Japan). The full-length 16S rRNA gene of V. harveyi TUMSAT-2019 was amplified by PCR and cloned into pGEM-T Easy (Promega, USA). To introduce the eGFP cassette into the central region of the 16S rRNA gene, the pGEM-16S construct was linearized by inverse PCR, and the eGFP module was inserted using the In-Fusion HD cloning system (Takara Bio, Japan), generating a modified 16S::eGFP allele. This recombinant 16S::eGFP fragment was subcloned into the suicide vector pRE112, which carries the R6K origin, sacB counterselection marker, and RP4 mob region42. The resulting plasmid (pRE112-16S::eGFP) was propagated in E. coli SM10 λpir, which supports R6K-dependent replication. Donor (E. coli SM10 λpir) and recipient (Rifampicin-resistant V. harveyi) cultures were spotted onto a 0.22 μm filter on Bacto Heart Infusion (HI) agar and incubated for 12–16 h at 25 °C to allow conjugation. Cells were recovered and plated on Luria-Bertani (LB) agar to obtain single-crossover integrants. Resulting colonies were streaked on LB agar with 5% sucrose to select double-crossover recombinants, in which the sacB-bearing plasmid backbone is excised. Whole genome resequencing of eGFP-expressing V. harveyi eGFP-expressing V. harveyi was cultured in LB medium supplemented with 2.0% (wt/vol) NaCl. An aliquot (4 mL) of the exponentially growing culture was harvested by centrifugation (5,000 × g), and genomic DNA was extracted using a NucleoBond AXG 20 column together with the NucleoBond Buffer Set III (Macherey-Nagel, Germany), following the manufacturer’s instructions. Purified genomic DNA was dissolved in 0.1× Tris-EDTA (TE) buffer and quantified using a Qubit 2.0 fluorometer (Invitrogen, USA). For whole-genome resequencing, a sequencing library was prepared with the Rapid Barcoding Sequencing Kit (Oxford Nanopore Technologies, UK) and run on an R9.4.1 flow cell. Basecalling was performed with Guppy v7.1.4 (Oxford Nanopore Technologies, UK). Long reads were de novo assembled using Flye v2.9.243. The assembly graph of the eGFP-expressing V. harveyi genome was visualized using Bandage44. Scaled-up cultivation from kuruma shrimp rearing water sample Water was collected from a kuruma shrimp rearing tank at the Tokyo University of Marine Science and Technology (TUMSAT), Japan, in March 2024. To remove large particles and debris, the sample was pre-filtered sequentially through 500 μm, 50 μm, and 5 μm nylon mesh filters. A 20-mL aliquot of the pre-filtered water was subsequently passed through a 47-mm-diameter membrane filter with a pore size of 0.22 μm using a Sterifil Aseptic System and Filter Holder (Merck, Germany). The membranes retaining the microbial fraction were placed into 5.0 mL of Marine Broth or 10-fold diluted Marine Broth, prepared using 2% (w/v) NaCl solution to adjust the salinity. Cultures were incubated at 25 °C under static conditions. These cultures were used for antagonistic bacteria screening of this study. Co-encapsulation and co-cultivation in microdroplets eGFP-expressing V. harveyi was cultured overnight at 25 °C in Marine Broth with shaking, and the cell density was adjusted to 7.0 × 109 cells/mL by centrifugation. In parallel, the bacterial community obtained from kuruma shrimp rearing water and enriched in Marine Broth was adjusted to 7.0 × 108 cells/mL. The two suspensions were mixed at a 1:1 ratio and used for microdroplet generation. Water-in-oil microdroplets were produced using an On-chip Droplet Generator equipped with a 2D Chip-800DG (On-chip Biotechnologies, Japan). The fluorinated oil phase consisted of 2% 008-FluoroSurfactant, which was prepared as a mixture of 008-FluoroSurfactant-5wtH (On-chip Biotechnologies, Japan) and 008-FluoroSurfactant-0.1wtH (On-chip Biotechnologies, Japan) at a ratio of 2:3. This oil and aqueous phases were supplied at 30 and 35 kPa, respectively, yielding microdroplets approximately 30 μm in diameter. The droplets were collected into 1.5 mL tubes and incubated statically at 25 °C. Microdroplets were examined using an ECLIPSE Ni fluorescence microscope (Nikon, Japan). Microdroplet dispensing via fluorescence intensity Target microdroplets were dispensed into a 96-well plate using the On-chip Droplet Selector (On-chip Biotechnologies, Japan). Dispensing was performed with a 2D Chip-SD1000-w150 tip, with the sheath reservoir filled with 2% 008-FluoroSurfactant. Prior to sorting, fluorescence intensities of several thousand microdroplets were measured to generate fluorescence distribution plots, and thresholds defining the target population were established. To identify droplets containing potential antagonistic strains, thresholds defining the target population were established based on the reduction of eGFP fluorescence, which indicated the growth inhibition of the pathogenic strain. After threshold determination, fluorescence intensities were re-measured, and droplets falling within the defined range were sorted and isolated. Following dispensing, empty microdroplets containing 20 µL of Marine Broth were added to each well, and two rounds of electrical stimulation were applied using the On-chip Merge system (On-chip Biotechnologies, Japan) to induce droplet fusion. Subsequently, 100 µL of Marine Broth was added, and cultures were incubated at 25 °C with shaking, without centrifugation. 16S rRNA gene sequencing and analysis of isolates Genomic DNA of the isolates in this screening were obtained by heat-shock lysis. Briefly, 1 µL of each glycerol stock was mixed with 20 µL of distilled water and heated at 95 °C for 5 min. The resulting lysate was used as the DNA template for PCR amplification of the full-length 16S rRNA gene using universal primers 27 F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492R (5′-GGYTACCTTGTTACGACTT-3′). PCR products were purified with ExoSAP-IT™ Express PCR Product Cleanup Reagent (Thermo Fisher Scientific, USA). Purified amplicons were then subjected to cycle sequencing using the BigDye™ Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, USA). Sequencing reaction products were cleaned by ethanol precipitation and analyzed on an ABI 3130xl DNA Analyzer (Applied Biosystems, USA) using the Sanger sequencing method. When the obtained reads were insufficient to construct the full-length 16S rRNA gene sequence, additional sequencing was performed using the internal primer 515 F (5′- GTGYCAGCMGCCGCGGTAA-3′). Approximately 1,400 bp full-length 16S rRNA gene sequences were obtained. The sequences were compared against the NCBI nucleotide database using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi), and genus-level taxonomic affiliations were assigned based on the best matches. Multiple sequence alignments were performed using MAFFT v7 (https://mafft.cbrc.jp/alignment/server/index.html), and phylogenetic trees were constructed with 1,000 bootstrap replications45. The sequences were performed OTU clustering at 99% similarity using VSEARCH v2.29.1 software46. 16S rRNA gene amplicon sequence analysis Bacterial community analyses were performed for kuruma shrimp rearing water before cultivation in Marine Broth, after cultivation in Marine Broth, and after cultivation in 10-fold diluted Marine Broth. For the pre-cultivation sample, 20 mL of rearing water was sequentially passed through 500 μm, 50 μm, and 5 μm membrane filters, followed by a 0.22 μm pore-size membrane filter. The final filter was cut into small pieces and DNA was extracted using the ZymoBIOMICS™ DNA Microprep Kit (Zymo Research, USA). For post-cultivation samples, DNA was extracted from 100 µL of bacterial suspension using the same kit. DNA concentration and quality were assessed with a Qubit 2.0 fluorometer. The V4 region of the 16S rRNA gene was amplified following the 16S Metagenomic Sequencing Library Preparation Protocol for MiSeq (Illumina, USA) using primers Miseq515F (5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGCCAGCMGCCGCGGTAA-3′) and Miseq806R (5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGACTACHVGGGTWTCTAAT-3′). PCR was performed with KOD One PCR Master Mix (Toyobo, Japan) under standard cycling conditions, followed by index PCR and purification. Libraries were quantified with Qubit and the Agilent D1000 ScreenTape System (Agilent Technologies, Germany), diluted to 10 nM, pooled, and sequenced on the Illumina MiSeq platform (2 × 250 bp paired-end reads). Sequence data in FASTQ format were analyzed using Quantitative Insights into Microbial Ecology 2 (QIIME2) (https://qiime2.org/)47. Reads were denoised and quality-filtered with the DADA2 plugin, and taxonomic classification was performed against the SILVA 138 reference database. ASVs assigned to mitochondria or chloroplasts were removed prior to downstream analyses. Taxonomic profiles at the genus level were visualized as relative abundance bar plots in QIIME2 View (https://view.qiime2.org/). Cross-streaking test to check the antagonism The antagonistic activity of the 18 isolated candidate strains was evaluated using a cross-streak assay on Marine Agar plates. Each isolated strain and V. harveyi were cultured in Marine Broth at 25 °C with shaking overnight, and the culture broth was cross-streaked on Marine Agar plates. After static incubation at 25 °C, the inhibition of V. harveyi growth was examined at the intersection of the streaks. The determination of no activity on agar plates was based on visual observation of the presence or absence of an inhibition zone. Co-infection test using kuruma shrimp To evaluate the antagonistic ability of the isolates to suppress V. harveyi infection in P. japonicus, shrimp were intramuscularly injected with a mixture of V. harveyi and each isolated bacterium. 10 shrimp were placed in each 10 L seawater tank. One representative strain was selected from each of OTUs 1, 2, 3, and 4; therefore, differences in disease-suppressive ability among strains within the same OTU were not examined in this study. Both V. harveyi and the isolates were cultured in Marine Broth at 25 °C with shaking overnight. The bacterial suspensions were adjusted to a final concentration of 8.0 × 105 cells per shrimp, and cell densities were determined using an Erma bacterial counting plate. A total volume of either 50 µL or 100 µL of the mixed bacterial suspension was injected into the shrimp muscle from the ventral side using a 32G needle. Negative control was injected with artificial seawater using same needle. Shrimp survival was monitored over time following injection. The average shrimp weight used in this assay was 6.0 g. Survival was analyzed using the Kaplan-Meier method and difference was evaluated by log-rank test, relative to the control group. According to the Regulations for Animal Experiments at Tokyo University of Marine Science and Technology, formal ethical approval was not required for this study as it exclusively utilized an invertebrate species. Nevertheless, all animals were handled with care to ensure proper husbandry management and welfare. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We thank Masato Ishikawa (Nagahama Institute of Bio-Science and Technology) for his valuable advice on genome editing. Author contributions H.Y.: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Visualization, Funding acquisition, Writing—original draft. Y.K.: Methodology, Validation, Investigation, Resources, Writing—review & editing. K.Konishi: Investigation, Writing—review & editing. M.F.: Formal analysis, Writing—review & editing. R.N.: Investigation, Writing—review & editing. H.K.: Project administration, Writing—review & editing. I.H.: Project administration, Funding acquisition, Writing—review &editing. K.Koiwai: Conceptualization, Supervision, Project administration, Funding acquisition, Writing—reviewing & editing. Funding This work was supported by Japan Science and Technology Agency (JST) ACT-X (grant number JPMJAX21B5), Noda Institute for Scientific Research Grant, and JSPS KAKENHI (grant number 24H00519) to Keiichiro Koiwai; JSPS KAKENHI (grant number 25H00939), and JSPS Fund for the Promotion of Joint International Research (International Collaborative Research, grant number 24KK0126) to Ikuo Hirono; and the Sasakawa Scientific Research Grant from The Japan Science Society to Hajime Yuasa (grant number 2025–4015). Data availability The 16S rRNA gene sequences of the isolates determined in this study were deposited in the DDBJ/EMBL/GenBank database under accession numbers LC920162-LC920179. The eGFP-expressing V. harveyi genome sequence was deposited in the NCBI database under BioProject accession number PRJDB42897. The 16S rRNA amplicon sequencing data were deposited in the NCBI database under BioProject accession number PRJDB42879. Declarations Competing interests The microfluidic device utilized in this study was manufactured by On-chip Biotechnologies Co., Ltd. Yuri Katayama is an employee of On-chip Biotechnologies Co., Ltd. and holds stock options in the company. The other authors declare no competing interests. Ethics approval and consent to participate No ethics approval or consent to participate was required. Consent for publication Not applicable. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The 16S rRNA gene sequences of the isolates determined in this study were deposited in the DDBJ/EMBL/GenBank database under accession numbers LC920162-LC920179. The eGFP-expressing V. harveyi genome sequence was deposited in the NCBI database under BioProject accession number PRJDB42897. The 16S rRNA amplicon sequencing data were deposited in the NCBI database under BioProject accession number PRJDB42879.