An Aeromonad selectively removes a class of pathogens from shrimp, prevents disease and preserves a healthy commensal microbiome

Abstract

On their own, probiotics do not entirely eliminate pathogens during infection and disease. Instead, they often serve as adjuncts that rely on distinct mechanisms to reduce the presence of harmful bacteria. Our prior research indicates that an isolated Aeromonas dhakensis strain A603 kills pathogenic Vibrio strains through two antibacterial mechanisms: the type VI secretion system (T6SS) and phenazine (AdPhen). Here, we investigate A603 as a standalone probiotic for a shrimp disease model. This study shows that A603 prevents mortality from acute hepatopancreatic necrosis disease (AHPND) in shrimp caused by pathogenic Vibrio spp. that produce the PirAB toxin. AHPND infection alters the shrimp microbiota by increasing pathogen abundance and decreasing beneficial bacterial abundance prior to death. As both a prophylactic and treatment, A603 removes pathogenic Vibrio from shrimp and reverses such alterations in the microbiota using both T6SS and AdPhen. Collectively, our findings show that A603’s antibacterial mechanisms prevent AHPND.

Probiotics are widely used in humans and livestock to promote health and immunity and to prevent disease. In the context of disease treatment, probiotic bacteria may strengthen immune responses, help maintain stable microflora, outcompete pathogens for critical niches, or antagonize pathogens¹. Regardless of the mechanism, when probiotics are found to be beneficial, they are not curative on their own. Meta-analysis studies examining those that target human pathogens report the best outcomes when probiotics are used alongside antibiotics rather than as the main treatment². Of these, clearance of Clostridioides difficile, a pathogen that causes severe gut infection, is improved by probiotic supplementation, though not all patients were cured. Other studies targeting pathogens with antibiotics found that probiotics alone were significantly less effective and even ineffective². For medical treatment, replacing antibiotic treatments with stand-alone probiotics is considered advantageous^3,4.

Many probiotic studies focus on their ability to reduce infection and correct dysbiosis, or to maintain eubiosis, following disease or antibiotic treatment, but the outcomes vary⁵. There is no medical consensus that that probiotic supplementation during antibiotic treatment completely prevents or reverses dysbiosis⁶. However, it is clear that infection, disease, antibiotic treatment, and the host’s health can exacerbate dysbiosis, thereby further harming the host and increasing symptom severity during disease^7,8. In theory, a probiotic that both prevents infection should also counteract infection-based dysbiosis.

Based on our previous in vitro characterization of Aeromonas dhakensis A603 as a strain with antibiotic properties and preliminary shrimp survival data when larvae are challenged with a Vibrio pathogen⁹, we hypothesized that A603 could act alone as a probiotic therapeutic to prevent infection and disease. We focused on hepatopancreatic necrosis disease (AHPND), a bacterial disease specific to White-legged shrimp (Litopenaeus vannamei). This disease emerged in 2009 and has since spread worldwide, with frequent outbreaks on aquaculture farms costing billions¹⁰. The most prevalent and virulent AHPND agents are Vibrio parahaemolyticus strains that have acquired a common extrachromosomal virulence plasmid. A locus on these plasmids encodes the subunits of a binary toxin belonging to a class of insecticidal toxins; the PirAB toxin distinguishes AHPND strains from other Vibrios commonly found in aquatic hosts^11,12. AHPND also perturbs the endogenous shrimp microbiota, leading to decreased α-diversity and the loss of commensal bacteria, including those considered beneficial^10,12–14. AHPND strains are highly virulent to shrimp larvae and postlarvae, causing damage to and shedding of hepatopancreatic and intestinal epithelial cells, and dispersing pathogens into communal water, thereby infecting other shrimp¹⁵. While antibiotics are a conventional treatment, the prevalence of antibiotic-resistant Vibrio pathogenic strains reduces their efficacy, and their administration further promotes the emergence of antibiotic resistance¹⁶. Environmental management through water treatment and filtration is ineffective and prone to contamination¹⁷. Instead, many farms are seeking alternatives, including probiotics, to prevent initial colonization by AHPND pathogens.

We isolated Aeromonas dhakensis strain A603 from food-grade Litopenaeus vannamei. Our previous characterization of A603 identified it as a potent antibacterial agent in vitro, attributed to its Type VI Secretion System (T6SS) and the phenazine antibiotic AdPhen it produces⁹. In this work, we extend this analysis to characterize the A603 strain colonization ability and its effects in vivo in its original shrimp host. We observe a significant improvement in host survival in the presence of A603 upon infection by AHPND-causing Vibrio parahaemolyticus strain TM-1 and show that both A603’s T6SS and AdPhen production contribute to this positive probiotic phenotype. Alongside these analyses, we present additional insight into compositional and functional changes to the shrimp microbiome during TM-1 infection when shrimp are treated with A603.

Results

A603 improves host survival through its T6SS and AdPhen

Prior studies showed that A603 protected against AHPND-induced shrimp mortality and killed Vibrio species in vitro via the T6SS and its AdPhen antibiotic⁹. To measure the contribution of both antibacterial mechanisms to disease protection, we treated shrimp with A603 and A603 mutant strains defective in T6SS and AdPhen production, then challenged them with the infectious AHPND V. parahaemolyticus strain TM-1, isolated from diseased shrimp in a Thai farm pond. Larval and juvenile shrimp are most susceptible to AHPND^18,17, so we first challenged post-larval shrimp, with and without A603 pre-treatment, to assess survival after TM-1 infection. In infected controls, survival dropped to 53% by day 4 and 40% by day 6 (Fig. 1a). Shrimp pretreated with wild-type A603 showed a survival rate of 93% after 6 days of TM-1 exposure. Shrimp pretreated with A603ΔAdPhen and A603ΔT6SS were more prone to mortality than those in the A603 group. A603ΔT6SSΔAdPhen-pretreated shrimp were the least protected among the pretreated groups. These results suggest that A603 protection relies primarily on both T6SS and AdPhen. Still, we cannot rule out the possibility that A603 employs other, less effective mechanisms that protect TM-1 from infecting and killing shrimp.

Figure 1.

a, Plot showing survival of postlarvae shrimp (PL15) during TM-1 infection and when pretreated with A603 strains (n=14). b, Plot showing survival of postlarvae shrimp (PL15) during TM-1 infection and when A603 strains were added after exposure to TM-1 (n=8). c, Design of treatment and infection of postlarvae to extract and sequence bacteria for metagenomics analyses. d, Plot showing relative abundance of Vibrio in each group. e, Relative abundance of plasmid pVA1. f, Bacterial richness indicated by Chao1 index. g, Overall diversity, including species evenness, is indicated by Shannon Diversity Index. All comparisons shown are one-way ANOVAs against TM-1-infected controls (samples sizes range from n=7 to n=9). h, The similarity of microbial communities to uninfected (mock) shrimp was calculated as Bray-Curtis distance. All mock controls (n=9) are compared with all experimental samples (n=9). Significance is calculated using a one-way ANOVA and displayed above using Compact Letter Display. i, The interindividual similarity of microbiomes within experimental conditions was calculated as Bray-Curtis distance. All individuals within a group (n=9) are compared with all others (n=8). Significance is calculated using a one-way ANOVA and displayed above using Compact Letter Display. When present significance is noted (*P<0.05, **P<0.01, *** P< 0.001, **** P<0.0001, ns P>0.05). Sample sizes in d-i are n=9 except for TM-1 (n=7) and A603DT6SSDAdPhen+TM-1 (n=8).

While A603 provides prophylactic protection against Vibrio-related mortality, its potential as a therapeutic agent for shrimp already infected with TM-1 was also evaluated. A603 was administered 24 hours after TM-1 inoculation, and host survival was monitored for three days. Wild-type A603 prevented shrimp mortality, whereas untreated shrimp exhibited mortality within 72 hours (Fig. 1b). Infected hosts treated with an A603 mutant lacking both antibacterial systems (T6SS and AdPhen production) had a 62.5% survival rate after 72 hours. This reduction likely reflects the inability of A603ΔT6SSΔAdPhen to eliminate or displace TM-1 from infected shrimp, consistent with previous observations.

A603 reduces Vibrio abundance in the shrimp during TM-1 infection

We assessed A603 pretreatment efficacy against TM-1 colonization by analyzing the postlarvae shrimp microbiome, focusing on the total bacterial composition and relative abundance (RA) of each taxon. Shrimp pretreated with A603 strains were infected with TM-1, and after 48 hours—when mortality was about 50%—live shrimp were washed, homogenized, and filtered to enrich bacteria (Fig. 1c). DNA from this filtrate, representing the entire microbiome, was sequenced to determine bacterial RA. To detect finer-scale changes during infection, bacterial reads were mapped and analyzed using the Kraken and Bracken pipelines, thereby increasing classification confidence and re-estimating RA at the genus and species levels¹⁹ .

Vibrio is the most prevalent genus in both mock-treated and TM-1-infected shrimp (Fig. 1d, Supplemental Table S1). To differentiate native endogenous Vibrio species from TM-1 during infection, reads were aligned to the 69 kb TM-1 PirAB toxin-encoding pathogenicity plasmid (pVPA3), which makes up about 1.25% of the TM-1 genome when present as a single copy⁹. We compared shrimp pretreated with mock and A603 to those infected with TM-1. Significant increases in plasmid DNA were observed only in TM-1-infected shrimp (Fig. 1e). At 48 hours post-infection, plasmid-mapped reads comprised roughly 1.0% of all Vibrio reads, suggesting TM-1 became the dominant Vibrio strain. Since TM-1 infection does not notably increase total Vibrio reads (Fig. 1d), these results suggest that infection with an AHPND strain such as TM-1 displaces commensal and endogenous Vibrio strains in healthy shrimp.

In the absence of TM-1 infection, A603 pretreatment had little effect on endogenous Vibrio. In infected shrimp, A603 pretreatment significantly reduced Vibrio levels by about threefold, to levels below those measured in mock-uninfected hosts or in hosts treated with A603 alone (Fig. 1d). Similar reductions were observed with A603ΔAdPhen and A603ΔT6SS. Notably, A603ΔT6SSΔAdPhen treatment modestly lowered Vibrio abundance, despite its demonstrated inability to kill Vibrio strains in our in vitro co-incubation assays⁹. For all A603-treated samples, pVPA3 plasmid reads were undetected (Fig. 1e), indicating TM-1 is absent and other Vibrio persist. Given that PirAB is crucial for AHPND-related mortality¹¹, we surmise that TM-1 elimination from shrimp accounts for the increase in host survival. These findings imply that infection by an AHPND strain, such as TM-1, displaces and outcompetes endogenous and commensal Vibrio. A603 therefore removes endogenous Vibrio only when TM-1 is present. Because TM-1 and other Vibrio can be killed by A603 in vitro, we cannot ascertain whether TM-1 temporarily colonizes shrimp and removes endogenous Vibrio or instead provides an opportunity for A603 to directly target these strains as collateral damage.

A603 preserves microbiota diversity during infection

TM-1 infections significantly reduced microbiota diversity, as indicated by lower Chao1 indices (Fig. 1f). Nonetheless, species richness was preserved in infected shrimp treated with A603 strains, remaining comparable to both mock and A603-only groups. Additionally, the Shannon diversity index was higher in A603-pretreated hosts—regardless of infection status—compared to TM-1 infected and mock controls (Fig. 1g). This increase in richness and diversity in A603-pretreated, TM-1 infected shrimp suggests that Vibrio depletion occurs, likely facilitating the growth of a broader array of other bacterial taxa within the shrimp.

To assess overall changes in the shrimp microbiome, we compared Bray-Curtis distances between experimental and mock-treated replicates and between untreated controls (Fig. 1h). Infection with TM-1 significantly increased Bray-Curtis distances relative to mock controls, indicating shifts in the microbiota, including increased Vibrio abundance and reduced abundance of other taxa. A603 pretreatment does not significantly change microbiota composition in healthy shrimp, based on this measurement. Shrimp pretreated with A603 and infected with TM-1 show only slight, non-significant differences in the microbiome compared to mock controls. These findings suggest that A603 helps preserve the native microbiome composition from notable alterations caused by TM-1 infection.

Qualitatively, the most significant changes in the shrimp microbiome following TM-1 infection reflect intrasample variation, with individuals within experimental conditions showing significantly greater Bray-Curtis distances (Fig. 1i). Notably, this variance is not consistently reversed by A603 pretreatment, and the observed differences between A603 T6SS and Ad-Phen mutants suggest that T6SS plays a larger role.

Infection and and A603 treatment alter key taxa

Diversity indices reveal that RA of many taxa is significantly affected during TM-1 infection. Examining bacterial abundance at the class and order levels within each group reveal that specific key family members are influenced (Fig. 2a). Principal component analysis of RA by order shows that the microbiota in mock- and A603-treated shrimp are quite similar. However, samples infected with TM-1 and those treated with A603 and TM-1 differ substantially from the mock and from each other, highlighting changes in major taxa driven by TM-1 and A603+TM-1 conditions (Fig. 2b).

Figure 2.

a, Proportion of bacterial reads by class from postlarvae shrimp samples merged by group. The top phylogenetic classes across samples are shown above; orders omitted account for <1% of total reads per sample. b, PCA analysis of Mock, A603, TM-1, and A603-TM-1 groups based on bacterial order composition. c, Bacteria taxa by order that increase most during TM-1 infection, independent of A603 treatment. d, Bacteria taxa by order that increase most during infection that are also significantly depleted by A603 treatment. e Significant changes in the order Aeromonadales dependent on A603 and TM-1. f, Bacteria taxa (order) that are depleted during TM-1 infection that A603-treatment restores. When present significance is noted (*P < 0.05, **P <0.01, *** P < 0.001, **** P < 0.0001, ns P > 0.05). . Sample sizes in a-f are n=9 except for TM-1 (n=7) and A603ΔT6SSΔAdPhen+TM-1 (n=8).

A603 treatment failed to reduce infection-induced increases in the bacteria in the orders Rhodobacterales and Hyphomicrobiaceae (Fig. 2c). A603 treatment increased the RA of Aeromonadales (Fig. 2d), whereas for Vibrionales, Alteromonadales, and Psuedomonadales this decreased (Fig. 2e). Members of Bacilli are the only abundant classes that decrease in TM-1-infected shrimp, and A603 treatment restores their levels (Fig. 2f). Other taxa show significant differences between groups, though they are minor components of the microbiota (Supplemental Table S2). When treating TM-1-infected shrimp with A603 strains deficient in T6SS or phenazine production, the shifts in taxonomic composition are less pronounced (Fig. 2a).

During TM-1 infection, A603-induced changes are dependent of T6SS and AdPhen

Next, we analyzed how the RA of genera changes during TM-1 infection using A603 and mutants in its antibacterial mechanisms. Examining RA at the genus level provides higher-resolution insight into which bacteria increase or decrease in shrimp when exposed to different strains (Fig. 3a, Supplemental Table S3). Furthermore, we compared taxa that significantly increased or decreased in A603-pretreated, TM-1-infected shrimp to those in shrimp infected with TM-1 alone, revealing taxa changes that depend solely on A603 pretreatment during TM-1 exposure. This method helped us identify taxa that change significantly in response to A603 strains during TM-1 infection, compared to the mock (marked “F” in Fig. 3a).

Figure 3.

a, Sample diagram Venn diagram key that shows genus-level comparisons of genera between A603-pretreated, TM-1 infected, A603-pretreated+TM-1 infected, and mock postlarvae shrimp reveals distinct taxa altered by A603 during TM-1 infection (designated (F)). b, Bacterial genera found to increase and decrease by RA during TM-1 infection alone. c, Bacterial genera found to increase and decrease when postlarvae shrimp were treated with A603 alone and uninfected. d-g, Bacterial genera found to increase and decrease when postlarvae shrimp were treated with A603 and A603 strains defective in T6SS and AdPhen while infected with TM-1 and when compares to the TM-1 infected. The subset in in purple (designated as “F” in a) are listed. Numbers indicate total bacterial genera in each groups that increase and decrease by RA. Family and Phylum are also presented to show taxonomic context. Sample sizes in d-i are n=9 except for TM-1 (n=7) and A603DT6SSDAdPhen+TM-1 (n=8).

TM-1 infection alone caused the most significant genus-level changes in microbiota composition, as shown in the α-diversity indices (Fig. 1f,g) and RA levels (Supplemental Table S1 and S3). In the mock group, evidence of reads mapping to 1884 distinct genera was detected, with only about 20 exceeding the threshold for dominant representation, defined as more than 1.6% of the microbiome. Compared to the mock group, 192 genera were significantly altered in RA during infection (Fig. 3b, Supplemental Table X). Of these, 103 genera showed a significant decrease—approximately half of which belong to gram-positive Bacillota (47) and Actinomycetota (3), with others from other primarily gram-negative phyla. Conversely, 89 genera increased notably during TM-1 infection, mainly within Pseudomonadota (83), predominantly Alphaproteobacteria (72), especially the Rhodobacter clade (61) within the Roseobacteriales. Increases in Gammaproteobacteria (10) were also observed. No substantial change was noted in the Vibrio genus in the TM-1 infection of shrimp. For uninfected shrimp, the A603 pretreatment alone had little to no effect on bacterial RA, except for a significant increase in a single Actinomycetota (Rothia) (Fig. 3c).

We then compared the bacterial composition in TM-1-infected shrimp with and without A603 pretreatment (Fig. 3b vs 3d). We reasoned that, among the 103 genera measured to decrease during TM-1 infection (Fig. 3b), some may be restored or increase when shrimp were treated (Fig. 3d). Likewise, among the 89 genera that increase during TM-1 infection, some may instead decrease with A603. A603 treatment decreased 6 of the 89 increased genera belonging to Gammaproteobacteria, including Vibrionales, Alteromonadales, Enterobacterales, and Chromatiales (Supplemental table 3). A603 treatment increased 2 of 103: a single Clostridia from Lachnospirales and an Acidobacteriaceae. These findings reveal that, at the genus level, only a subset of microbiota alterations observed during TM-1 infection is reversed by A603 treatment.

Next, we compared the microbiota of TM-1-infected shrimp with those of shrimp pretreated with A603 strains lacking Ad-Phen production (A603 ΔAd-Phen), T6SS (A603 ΔvipA), or both (A603 vipA ΔAd-Phen) in the same manner (Fig. 3b vs Fig. 3efg). Both T6SS and Adphen alone are shown to kill bacteria in vitro⁹, so we hypothesize that they confer partial protection, as indicated by our survival data from challenge experiments (Fig. 1ab). For shrimp pretreated with the A603 ΔAd-Phen mutant, we observe nine genera that are increased and two that decreased significantly compared with infected shrimp (Fig. 3e). All taxa that increased are generally known as commensal and environmental bacteria. The genera that decreased significantly are both in the class Gammaproteobacteria, each represented by a single taxon from Vibrionales and Alteromonadales, respectively. Therefore, we believe that the depletion of Gammaproteobacteria is largely due to T6SS activity.

Pre-treating shrimp with A603 ΔvipA maintains Ad-Phen production but inactivates the T6SS. In this group, a significant rise in a low-RA genus within Bacillota is observed (Fig. 6f). Of the four genera that decrease in RA, all belong to Pseudomonadota and are spread across Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria, including Alteromonadales. One Planctomycota genus increases, while five Pseudomonadota genera decrease. When we compare A603ΔAdphen- and A603-pretreated shrimp (Fig. 3e vs Fig. 3d), and also A603ΔvipA- and A603ΔAdphenΔvipA-pretreated shrimp (Fig. 3f vs Fig. 3g), we measure more similar microbiota compositional changes, indicating that of these two antibiotic mechanisms, T6SS activity alone is more dominant than that of Adphen.

Colonization by A603 is enabled by depletion of the native microbiota

While TM-1 challenge results confirm that A603 offers protection against AHPND mortality, initial data showed minimal colonization of A603 in healthy shrimp and a slight increase in diseased ones (Fig. 1e). A603’s abundance was measured to be less than 0.1% of the microbiota at the time the sample was acquired. Since T6SS facilitates cell-to-cell killing and AdPhen is expected to diffuse and act at higher A603 concentrations, we hypothesized that A603 interacts with AHPND strains such as TM-1 within a common niche in shrimp. We hypothesize that endogenous bacteria might compete for or occupy such niches until A603, AHPND strains, or both are present during inoculation and infection. To test this, we exposed streptomycin-treated shrimp to streptomycin-resistant (SmR) A603 and A603ΔT6SSΔAdPhen strains (Fig. 4a). In this condition, many antibiotic-sensitive shrimp commensal and aquatic bacteria are likely eliminated by streptomycin, making these shrimp more similar to ‘germ-free’ models. After 24 hours, colonization was assessed by homogenizing shrimp tissues and plating on selective media. A603 colonization increased dramatically—almost 5000-fold (Fig. 4b)—in streptomycin-treated shrimp. Over time, colonization levels of both strains decreased by several orders of magnitude by day three. In shrimp cultured in streptomycin-treated water, A603 levels remained consistently higher than in untreated controls (data not shown). On day 3, water was replaced with tank water without antibiotics to reintroduce environmental bacteria. After 24 hours, colonization levels of A603 remained stable, while the A603ΔT6SSΔAdPhen strain was eliminated (Fig. 4b). These findings suggest that endogenous and native bacteria are a substantial barrier to establishing colonization of healthy shrimp and that A603 T6SS and AdPhen assist in maintaining stable colonization once established.

Figure 4.

(a) Design of bacteria-depleted postlarvae shrimp model using antibiotics for A603 colonization study (b) Total CFUs of each A603 strain recovered per postlarvae shrimp (starting at PL14-15) over the course of 5 days (n=3). Inoculation of water with A603 strains after day 1 sample collection and reintroduction of antibiotic-free tank seawater after day 3 sample collection is indicated.

Discussion

This study builds upon our prior work demonstrating the antibiotic effects of A603 T6SS and AdPhen against Vibrio⁹. Our current investigation examines the A603 strain’s role in the host during disease. Both prophylactic and post-exposure treatment with A603 significantly reduce shrimp mortality after TM-1 exposure. Consistent with other studies²⁰, TM-1 infection significantly alters the composition of the shrimp microbiota during disease progression, resulting in the enrichment of pathogenic bacteria and a reduction in other taxa, including potential commensals and beneficial species. A603 pretreatment eliminates TM-1 and reduces the RA of Vibrionales. As observed in single-strain competitions in vitro⁹, both T6SS and AdPhen production are required for maximal protection and reduction of pathogenic taxa within the shrimp bacterial community.

This study highlights significant and dynamic changes in taxonomic abundance during AHPND infection, with some taxa depleted and others enriched, regardless of probiotic treatment. During TM-1 infection, these shifts affect mostly commensals and environmental bacteria. Most bacteria that increase are Pseudomonadota, including Vibrionales and the Roseobacter clade (RCB) Alphaproteobacteria, which are regarded as beneficial, non-pathogenic, commensal, and potential probiotics^21,22. RCB are biomarkers in hatchery tanks with higher shrimp survival post-AHPND exposure. They are dominant members of the juvenile shrimp gut and are key biomarkers for disease ^23,24. RCB abundance is stable in healthy shrimp but also blooms during dysbiosis, possibly due to context-dependent colonization influenced by surrounding microbes^20,24. Likewise, during TM-1 infection, we observe that RCB abundance increases modestly, possibly due to depletion of other taxa. A603 pretreatment doesn’t significantly alter RCB abundance, helping to maintain beneficial bacteria and prevent dysbiosis.

We find that A603 T6SS and AdPhen independently reduce pathogenic bacteria, but the A603 mutant lacking both still affects some Gammaproteobacteria (Fig. 3g). We propose that shrimp niches allow these bacteria to compete with native microbes for colonization, likely requiring T6SS and AdPhen together for optimal protection. We observe a significant decrease in microbiome diversity in TM-1-infected shrimp compared with shrimp pretreated with A603. A603 likely colonizes at low RA, reducing or eliminating pathogenic Gammaproteobacteria during TM-1 infection, with its colonization increasing when TM-1 disrupts the microbiota or when streptomycin is used to eliminate bacteria. Since microbiota is measured here only at 48 hours post-infection, we cannot assess the A603 colonization dynamics before or after this point. Mortality in T6SS and AdPhen mutants rises from days 2 to 6 post-infection, indicating reduced protection. Antibiotic-treated shrimp showed that the A603 mutant is displaced after antibiotic withdrawal and water changes, showing T6SS and AdPhen are crucial for niche occupation. Without AHPND disruption of the microbiota or antibiotic treatment, A603 cannot colonize effectively, suggesting these mechanisms are insufficient alone under normal conditions.

Studies on bacterial communities in shrimp identify taxa linked to health, such as Roseobacter (RCB) and Bacillus spp., which act as antibiotics or probiotics via different mechanisms^25–27. RCB inhibits Vibrio by producing the antibiotic tropodithietic acid (TDA), promising for Vibrio-related diseases, as they are found in marine environments and invertebrates^28–30. TDA acts as a proton antiporter, causing membrane depolarization and stopping bacterial growth. The antibiotic AdPhen, although molecularly distinct from TDA, acts as a protonophore⁹, likely disrupting membrane gradients or the electron transport chain. No resistance to TDA has been observed in Vibrio³¹, and no resistance to AdPhen or T6SS was observed during co-incubations. TDA biosynthetic genes are plasmid-encoded and observed to be easily lost or mutated in biofilms^32,33, whereas T6SS and AdPhen genes are chromosomal, and mutants exhibit no fitness burden or advantage. TDA-expressing strains haven’t been used effectively to treat AHPND in white-legged shrimp but reduce pathogenic Vibrio in copepods used to feed farmed fish³⁰. A603 did not reduce the RA of RBC in this work, and TDA-producing commensals are likely to remain stable in treated shrimp.

This study demonstrates that A603 can serve as a standalone, microbiome-friendly alternative to antibiotics for controlling AHPND spread. Our previous research showed that A603 effectively killed all tested Vibrio strains in vitro, with no signs of spontaneous resistance to its antibacterial mechanisms⁹. Unlike traditional antibiotics, which can significantly disrupt the host microbiota and negatively impact health³⁴, our findings suggest that targeted, bacteria-based antibiotics like A603 could offer a safer, less disruptive option in aquaculture. Widespread antibiotic use in this field leads to environmental drug persistence, resistance gene spread, and increased transfer to human pathogens³⁵. Antibiotics also harm fish and invertebrate health and disturb endogenous bacteria^36,37. Probiotic studies in humans and agriculture that demonstrate that bacteria reduce pathogens successfully show these are an adjunct rather than a standalone solution². Our challenges show A603 may be more effective than other work that shows Bacillus spp. protecting juvenile shrimp against AHPND²⁶. Our results indicate that a bacteria-based antibiotic, active only during infection, could be an effective alternative to purified antibiotics, minimizing or avoiding disruption to the natural, healthy microbiome during diseases like AHPND.

Methods

Strains

Vibrio parahaemolyticus TM-1 and A603 strains used were previously characterized⁹. Spontaneous streptomycin-resistant A603 mutants were selected on streptomycin LB agar (100ug/ml). Strain descriptions are found in Supplemental table S1. All strains were grown in LB at 30°C, with streptomycin (100ug/mL) when appropriate.

Shrimp rearing in tanks and flasks

Litopenaeus vannamei postlarvae shrimp (PL10-11) were obtained from Miami Aquaculture. The shrimp were acclimated to 30°C and then added to a preheated, aerated tank containing 25 ppt artificial seawater (Instant Ocean sea salt). Experiments were performed at 30°C with gentle shaking (80rpm) in 25 ppt sterile artificial seawater in vertical tissue culture flasks with vented caps (Corning 430641) to allow for oxygen exchange and prevent contamination. To prevent water fouling and overcrowding, each flask contained a maximum of 4 animals, and shrimp were fed with ~5mg of sterile food daily.

Shrimp infections and survival

Postlarvae shrimp (PL15) were moved to flasks in groups of 3-4 (total shrimp n=9 for microbiome analysis. We used between n=11-15 for pre-treatment survival challenges and n=8 for post-treatment survival challenges. A603 and its mutant strains were added to the flask water at 10,000 CFUs/mL for microbiome analysis and pre-treatment experiments, and incubated at 30°C for 24h. Vibrio parahaemolyticus TM-1 was then added to flasks at the same concentration. For A603 post-treatment, TM-1 was added first, followed by A603 strains after 24h. For both survival experiments, survivors were quantified every 24h for up to 6 days.

A603 shrimp host colonization

Postlarvae shrimp at 15 days (PL15) were treated with streptomycin (100ug/mL) in sterile seawater. On day one, Streptomycin-resistant (Str^R) A603 or A603ΔT6SSΔAdPhen was added at 10,000 CFUs/mL to water. Shrimp were sacrificed just prior to the addition of A603, and at 24 and 48 hours post-inoculation, washed, and homogenized. Through 72 hours, ~30% of flask water was replaced with sterile streptomycin-supplemented seawater every 24h to prevent fouling. After 72h, before sacrificing hosts, ~30% of the flask water was replaced with the hosts’ original tank water to replenish the native flora. Shrimp in these flasks were sacrificed and processed as above at 24h and 48h after this water change.

Shrimp were homogenized for 1 min in 1 ml of LB using a Tissue-Tearor^™ (BioSpec 985370). The homogenate was filtered through a 5 μm syringe filter, then diluted and plated on LB-Str and TCBS to enumerate Str^R strains and Vibrio species.

Microbiome metagenomic sequencing

Shrimp were pretreated with the indicated A603 strains and infected with TM-1 as above. Individual surviving animals were harvested, washed in distilled water, and homogenized 48 hours after the addition of strain TM-1. The homogenate was filtered through a 5-micron syringe filter to remove large particles and intact shrimp cells and then stored at −20°C. Total gDNA was extracted from thawed samples using the Zymo Quick-DNA Miniprep Kit (Zymo D4069). An Illumina library was constructed using NEBNext Multiplex Oligos 96 Unique Dual Index Primer Pairs (sets 1-2, NEB E6440 and E6442) according to manufacturer instructions. Libraries were pooled and sequenced in equal amounts on an Illumina NovaSeq 6000 S1 flow cell.

Microbiome analysis

Raw Illumina NGS reads were filtered by quality and length mapped to the Standard Refseq database (9/4/2024) using Kraken2^38,19. Identified bacterial-specific reads in Taxa domain 2 were extracted and used in all downstream metagenomics analyses. Species-level read mapping was performed using Bracken with the Standard prebuilt database (build date 12/28/2024)³⁹. Further covariance and diversity analyses were performed using RA (bacterial genus abundance/all bacterial abundance) to eliminate differences in read counts/sample.

Relative genus abundances were visualized using GraphPad Prism 10.4.0. Diversity calculations and Bray-Curtis distances were generated using the Phyloseq package in R 4.40⁴⁰. Genus and species-level enrichment analyses were performed in OmicsBox Bioinformatics (version 4.0.51) https://www.biobam.com/omicsbox ^41,42. Statistical analyses were performed in GraphPad Prism 11.0.1 Windows version, GraphPad Software, Boston, Massachusetts USA, https://www.graphpad.com.

Supplementary Material

Supplement 1

Supplemental Data S1. Table of strains used in this work. Raw data used in Figure 1. Bracken-generated counts for bacteria for all samples.

Supplement 2

Supplemental Data S2. Table of NGS read data for all samples. Source data used in Figure 2.

Supplement 3

Supplemental Data S3. Differential abundance statistic, counts, and fold-change data used in Figure 3.