Nucleus-forming vibriophage cocktail reduces shrimp mortality in the presence of pathogenic bacteria

The global aquaculture industry has suffered significant losses due to the outbreak of Acute Hepatopancreatic Necrosis Disease (AHPND) caused by Vibrio parahaemolyticus. Since the use of antibiotics as control agents has not been shown to be effective, an alternative anti-infective regimen, such as phage therapy, has been proposed. Here, we employed high-throughput screening for potential phages from 98 seawater samples and obtained 14 phages exhibiting diverse host specificity patterns against pathogenic VPAHPND strains. Among others, two Chimallinviridae phages, designated Eric and Ariel, exhibited the widest host spectrum against vibrios. In vitro and in vivo studies revealed that a cocktail derived from these two nucleus-forming vibriophages prolonged the bacterial regrowth of various pathogenic VPAHPND strains and reduced shrimp mortality from VPAHPND infection. This research highlights the use of high-throughput phage screening that leads to the formulation of a nucleus-forming phage cocktail applicable for bacterial infection treatment in aquaculture.

Global food security has deteriorated rapidly in recent years due to a number of factors, of which climate change stands out as a major factor.Specifically, adverse effects of climate abnormalities were observed across all agricultural sector value chains 1,2 .The yield of many agricultural goods is severely impacted by the increased possibility of major disease outbreaks brought on by unpredictable weather patterns, and aquaculture products are not spared from this inevitable fate.In many coastal food exporting countries, shrimp was one of the most crucial fishery products 3,4 , contributing to economic prosperity and uplifting millions of people's lives.In recent years, however, the shrimp industry has been confronted with an increasing number of disease epidemics, especially acute hepatopancreatic necrosis disease (AHPND) [5][6][7] , resulting in a precipitous decline in shrimp production.
AHPND is typically characterized by the unusual and acute mortality of shrimp within the first 35 days in cultivation ponds and the mortality rate could easily reach 100% in larvae stage 8 .The disease is caused by the infection of a gram-negative rod-shaped bacteria Vibrio parahaemolyticus 9 .The presence of V. parahaemolyticus is dependent on many environmental conditions, such as pH, temperature, and salinity 9,10 .It is detected year-round in areas, in which the water temperature is above 15 °C, and increases in quantity are detected with increased temperatures 11 .Rising ocean temperatures due to global warming significantly propel the increased range and frequency of infection with V. parahaemolyticus 12 .In 2009, a new strain called V. parahaemolyticus AHPND (VP AHPND ) emerged in southern China and spread into major shrimp production hubs in east and southeast Asian countries 13 , causing a long-lasting economic impact to the region.To prevent large-scale losses, an abundance of antibiotics is used due to their low cost and accessibility 14 .However, the misuse and overuse of antibiotics are promoting the spread of antibiotic resistance in bacteria 15 .The extensive misuse of antibiotics in aquaculture has also caused prolific increases in resistant strains and multi-antibiotic resistance in shrimp pathogens 16 .In www.nature.com/scientificreports/addition, the resistance could be transferred to other bacterial species, including human pathogens, leading to uncontrollable resistome overspill and eventually severe consequences for human health 17 .This issue urges the need for effective alternative therapeutics that could mitigate the adverse effect on antibiotic resistance.
One of the most promising antibiotic alternatives is bacteriophage therapy, the method in which evolutionary races between two organisms are exploited for therapeutic purposes 18,19 .Phages, viruses infecting their specific bacterial host, typically enter either a lytic or lysogenic (temperate) life cycle 20 .Since one of the key events during the lysogenic cycle is characterized by the insertion of a phage genome and maintaining its genetic material within the host without destroying the host cell DNA 20,21 , lytic phages are highly preferred for use as therapeutic agents.In addition, integration of the phage genome could potentially contribute to more virulent bacterial strains if genes encoding virulence factors are carried within the genome 22 .The use of lytic phages has many advantages over the use of antibiotics, including auto-dosing, biofilm penetration, and high specificity, which disallows cross-infection, thereby minimizing its impact on bacterial community interference 20,23 .With these advantages, the use of phages and mixtures of phages, known as phage cocktails, for biocontrol agents in aquaculture recently came under the spotlight.In particular, phages targeting various pathogenic vibrios, including V. harveyi, V. coralliilyticus, V. campbellii, V. anguillarum, V. alginolyticus, V. splendidus, V. cyclitrophicus, and V. parahaemolyticus, have been extensively isolated and studied 24 , some of which demonstrated successful outcomes.Notably, a previous study showed that the use of phage cocktails in shrimp infected with pathogenic Vibrio sp.resulted in a 91.4% survival rate, a comparable outcome to that of antibiotic treatments at 91.6% 25 .
Here, we first developed a high-throughput phage screening method in order to swiftly screen for the vibriophages capable of infecting a collection of pathogenic VP AHPND strains that were collected from local farms where the bacterial spread has been reported.The method was then used to screen through 98 seawater samples against the bacteria panel and led to the discovery of two virulent nucleus-forming vibriophages that were specific to a wide range of VP AHPND strains.A cocktail derived from these vibriophages displayed high efficiency in both in vitro and in vivo experiments by suppressing the regrowth of bacteria and rescuing shrimp from infection within 24 h.Moreover, the vibriophages were able to colonize the digestive tract of shrimp, offering alternative aspects for prophylaxis against the bacteria in the future.

Phage isolation by the high-throughput screening (HiTS) method leads to the discovery of two phages that exhibit a broad host range
Here, we employed the high-throughput screening (HiTS) method for phages 26 , with slight modifications (Fig. 1).A collection of V. parahaemolyticus strains (28 strains) from Songkla Aquatic Animal Health Research and Development Center (SAAHRDC) that were previously isolated from infected shrimp in local farms were used as parental hosts for extensive screening for phages in 98 different seawater samples.The enrichment cultures obtained from the method were then tested against corresponding Vibrio isolates to determine the presence of phages.The appearance of plaques on the bacterial cell lawns was recorded and counted as a hit.Out of 98 seawater samples, 72 samples revealed potential hits where putative plaques were observed (Table S1).The appearances of plaques from the identified hits included clear, turbid, and bull's eye morphology.Since the clear plaque morphology was assumed to contain lytic phages, only 12 enrichment cultures that produced the clear plaque morphology were further processed for the purification step (Table S1; red).After the phage purification using their parental bacterial hosts, we eventually obtained 14 phages from the screen as designated: phiKT1015, phiKT1016, phiKT1017, phiKT1018, phiKT1019, phiKT1020, phiKT1021, phiKT1022, phiKT1023, phiKT1024, phiKT1025, phiKT1026, phiKT1027, and phiKT1028 (Table 1).
Since various VP AHPND strains and their multidrug-resistant variants have spread widely in aquaculture, it is important to have effective phages that can kill a wide range of vibrios.To determine the host spectrum of these Figure 1.High throughput phage screening (HiTS) method for phages, with slight modifications, showing an integrated workflow for searching phages from 98 seawater samples and 28 bacterial hosts.Screening in a checkerboard manner can combine all samples with all bacteria of interest.The enrichments were examined whether phages are present by the spot test against corresponding bacterial hosts.Plaque morphologies are then recorded and represented by following colors; red for clear plaque, orange for turbid plaque, yellow for bull's eye plaque, and white for no plaque, for further purification step.(See also Table S1).

The phage candidates are effective and safe for use in biocontrol
To characterize whether the candidates are appropriate for use in biocontrol, phage biology and conventional phage kinetics of phages phiKT1019 and phiKT1028 were first conducted.Both phages were capable of lysing various V. parahaemolyticus strains (Table 1) and produced 0.5 mm-clear circular plaques on V. parahaemolyticus KT1018B cell lawns (Fig. 2a and c).The negative straining of phage structures under the Transmission Electron Microscope (TEM) revealed that these phages belonged to the order Caudovirales and the family Myoviridae.Both phages had a large hexagonal capsid (a diameter of 104.57± 7.31 nm in phiKT1019 and 129.25 ± 2.7 nm in phiKT1028) and a contractile tail (a length of 226.63 ± 12.50 nm in phiKT1019 and 232.35 ± 2.30 nm in phiKT1028) (Fig. 2b and d; n = 3).According to the adsorption assay, both phages had somewhat different adsorption rates on the bacterial host cells.PhiKT1019 exhibited a higher adsorption rate than phiKT1028, reaching approximately 74% of absorbed phages within 25 min.About 42% of phiKT1028, on the other hand, attached to the host within 10 min and remained at a similar level until 30 min (Fig. 2e).However, both phages required a similar latent period of approximately 40 min for maturation to reproduce around 25-35 particles per cell in phiKT1019 and phiKT1028, respectively (Fig. 2f).Under environmental factors, both phages appeared to be stable over a relatively wide range of temperatures and pHs.The phages remained active at temperatures ranging from 4 to 50 °C, but their infectivity drastically decreased when the temperatures were over 50 °C (Fig. 2g).Phage phiKT1028, however, was more tolerant of the extreme pHs than phiKT1019.PhiKT1028 remained Table 1.Host range determination of 14 isolated phages.To determine the host spectrum of the isolated phages, various strains of V. parahaemolyticus, including non-AHPND and pathogenic AHPND strains, were used as the hosts.Grey and white boxes indicate appearance of plaque and no plaque, respectively.The total number of susceptible hosts to each phage was recorded as shown in the last row of the table.
To gain more insights into the phage candidates at the genetic level, we sequenced and analyzed the whole genomes of the phages phiKT1019 and phiKT1028.The whole genome sequencing of both phages resulted in more than 7 million reads of FastQ sequence after host genome subtraction.After genome assembly of the genomes with coverage over 700X, we obtained the whole genomes of the phages phiKT1019 and phiKT1028 (Fig. 3).Both phages had large double-stranded DNA genomes: with draft genome size of 238,840 bp for phiKT1019 and 238,838 bp for phiKT1028, encoding 218 open reading frames (ORFs) with 48.81% CG content in phage phiKT1019 and 209 ORFs with 44.83% CG content in phage phiKT1028 (Fig. 3a and b; Accession numbers MZ622713 and OP123707).Due to the size of the genome larger than 200 kbp, the dsDNA phages phiKT1019 and phiKT1028 are classified as giant or jumbophages.The annotated ORFs were assigned to the basic phagefunctional modules as follows: (I) DNA replication, transcription, and protein translation; (II) DNA metabolism and modification; (III) Phage regulation; (IV) Phage structural and assembly; (V) Lysis-related proteins; and (VI) Hypothetical proteins (Fig. 3).We did not find any tRNA genes in these phages.Moreover, the genomes of both phages did not harbor any unwanted genes.Neither the genes related to putative toxins and antimicrobial resistance nor the lysogeny-related genes were found in these phage candidates.Since genome analysis suggested

The phages phiKT1019 (Eric) and phiKT1028 (Ariel) are nucleus-forming vibriophages
To investigate the genetic relationship of the phages phiKT1019 and phiKT1028 with other phages that are available in the NCBI database, we first conducted a nucleotide blast search using our phage genomes as queries against the database.The Blastn search revealed that the genome of phage phiKT1019 was highly similar to that of vibriophage Aphrodite1 with 76.20% identity and 78% coverage, while the genome of phage phiKT1028 was highly similar to that of vibriophage pTD1 with 75.73% identity and 71% coverage.We then selected the closely and distantly related phages to both phages phiKT1019 and phiKT1028 as available in the database and constructed a phylogenetic tree to investigate the relationship among the selected phages.The tree showed that both phages phiKT1019 and phiKT1028 were grouped together into the clade of Myoviridae, further supporting the classification of viruses, based on the phage structure as observed previously by TEM (Figs. 2b, d, and 4a, asterisks).Interestingly, both phages were also genetically clustered together with the recently proposed Chimalliviridae family 27 , with a close genetic relationship with vibriophages BONAISHI, Aphrodite1, and pTD1 (Fig. 4a).The recently proposed Chimalliviridae family is a novel viral family whose phages share conserved core genes necessary for nucleus-based replication.To examine the similarity of our phages with other closelyrelated phages in the Chimalliviridae family, we used the VIRIDIC tool to evaluate the intergenomic similarity among the phages 28 .We found that phiKT1019 is most similar to vibriophage Aphrodite1 and phiKT1028 is most similar to vibriophage pTD1, with a similarity score of 59% and 54.8%, respectively.Based on the cutoff for identifying novel species (95%) and novel genus (75%) 28 , the intergenomic similarity values among the phages indicated that phages phiKT1019 and phiKT1028 are novel vibriophages in the Chimalliviridae family (Fig. 4b).
The fact that phiKT1019 and phiKT1028 belong to the Chimalliviridae family, which is possibly capable of forming a phage nucleus, is particularly interesting.It has been previously demonstrated that, during the replication inside the bacterial host, the nucleus-forming phages encode the chimallin (ChmA) protein with other accessory proteins to assemble the phage nucleus.The phage nucleus not only serves crucial roles in facilitating phage replication and maturation processes but also in protecting the phage genome from bacterial defense mechanisms [29][30][31][32][33][34][35] , highlighting their superior antibacterial properties.Thus, we further sought to investigate the genome organization and pinpoint the genes encoding for the defining structures of the nucleus-forming phages 29 .We performed genome comparisons of our phages with the closely-related vibriophages and the wellstudied nucleus-forming phage, Pseudomonas phage 201Phi2-1 30 .The result demonstrated that despite the low intergenomic similarity score (42.8%) between phages phiKT1019 and phiKT1028, their genome organizations were quite similar and the genes were mostly organized in the same direction (Fig. 4c).However, most regions in their genomes were differently organized compared to those of their closely related phages: vibriophages Aphrodite1 and pTD1 (Fig. 4c).ChmA, which is a major component of the phage nucleus 34 , was present in all of the analyzed phage genomes, including ORF105, ORF71, ORF65, ORF203, ORF180, and ORF65 for Pseudomonas phage 201Phi2-1, vibriophages BONAISHI, Aphrodite1, phiKT1019, phiKT1028, and pTD1 (Fig. 4c; pink), supporting the classification of phiKT1019 and phiKT1028 in the Chimalliviridae family by the phylogenetic tree.However, the phuZ gene, which encodes phage tubulin filaments that orchestrate the process of the phage lytic cycle [36][37][38] , appeared not much conserved in the vibriophages.Unlike Pseudomonas phage 201Phi2-1 (ORF59), vibriophages BONAISHI (ORF59), Aphrodite1 (ORF56), and phiKT1019 (ORF192), the genomes of vibriophages pTD1 and phiKT1028 did not contain the phuZ gene (Fig. 4c; yellow), suggesting variations in temporally subcellular regulation for the replication of vibriophages.
To assure that our phages phiKT1019 and phiKT1028 are nucleus-forming vibriophages, we further confirmed experimentally whether the phage nucleus assembles during their infection.We first constructed a fusion of superfolder GFP (sfGFP) to the ChmA proteins (ORF203 for phiKT1019 and ORF180 for phiKT1028) for labeling the wildtype protein and employed fluorescence microscopy to observe the formation of the phage nucleus during infection.In uninfected wildtype cells (Figure S2a; upper panels), the nucleoid of V. parahaemolyticus (KT1018B used as a bacterial host; for more genomic data, see Figure S3) formed an irregularly shaped nucleoid that distributed throughout the cells.Moreover, the ChmA proteins of both vibriophages expressed in uninfected cells diffused throughout the cells and did not assemble any specific structures similar to sfGFP control (Fig. 4d  www.nature.com/scientificreports/and S2b, green).During infection by the vibriophages phiKT1019 and phiKT1028 at 60 min post-infection (mpi), a bright DAPI-staining DNA blob appeared close to the midcell in phiKT1019-infected cells and at the cell center in phiKT1028-infected cells (Figure S2a; middle and bottom panels).The blob of the phage genome is enclosed inside sfGFP-tagged ChmA homologs, which assemble into a ring structure as seen in the Z-cross section (Fig. 4e; green and blue or grey), similar to those previously identified phage nucleus structures 27,30,32,35,39 .
No extranuclear DAPI staining due to the host genomic DNA was observed in the cells (Fig. 4e; blue or grey), suggesting that these vibriophages encode some nucleases to destroy the host chromosome during infections.Taken together, our jumbophages phiKT1019 and phiKT1028, as we redesignated them to "Eric" and "Ariel", respectively, are new members in the Chimalliviridae family and are vibriophages that assemble the nucleus-like structure during their lytic life cycle.

The nucleus-forming vibriophage cocktail prolongs the regrowth of VP AHPND and reduces shrimp mortality in the presence of VP AHPND
With the aim to formulate a cocktail of phages "Eric" and "Ariel" to tackle the pathogenic VP AHPND that currently spreads and causes mass mortality in aquaculture (Table S2), we selected the strains of VP AHPND from the collection of SAAHRDC that appeared to be sensitive to both of our vibriophages (Table 1) and investigated the killing profile of the cocktail against these sensitive isolates (12 isolates).However, due to the difficulty of estimating the target pathogen quantity, as described by Chen et al., (2019), that results in an inaccuracy of the actual optimal MOI, we formulated a phage cocktail containing both phages Eric and Ariel; each at a multiplicity of infection (MOI input ) of 5, resulting in a total MOI input of 10 relative to the starting bacterial cell number 25 .The in vitro killing profile revealed that the phage cocktail could prolong the regrowth of all tested VP AHPND isolates; however, some bacterial isolates recovered and resumed to grow, resulting in a slight increase in cell density at 20 h of incubation (Fig. 5a).Among the 12 tested isolates, the cocktail efficiently suppressed the bacterial regrowth of 3 VP AHPND isolates throughout 20 h, including strains 1021A, 1021B, and 1022A, with the most prevalent outcome on VP AHPND strain 1021B (Fig. 5a), indicating the potency of the cocktail against these isolates.The use of the combination of phages (Eric + Ariel) against the strain 1021B was proven to be more effective than the use of single phage alone.At the same MOI input , the cocktail initially suppressed the bacterial growth as early as 6 h of treatment, followed by the decrease in bacterial cell density to the baseline at 15 h.Moreover, the regrowth of bacteria was not observed during the treatment of the cocktail until 20 h of incubation (Fig. 5b).Together with the high pathogenicity of VP AHPND strain 1021B, which causes 95% mortality in Pacific white shrimp (Penaeus vannamei) (Table S2), this strain was thus selected as a representative of AHPND strains to further validate the performance of the cocktail against VP AHPND infection in vivo.
We first focused on two main aspects of the cocktail, including the prophylaxis efficiency against VP AHPND and the phage stability in the real environmental tanks.According to the prophylactic activity of the cocktail against VP AHPND , we evaluated the survival rate of shrimp over a window of 7 days in the following conditions: media control, challenge with VP AHPND in the presence of the phage cocktail, and challenge with VP AHPND .With the concern of phage-mediated toxicity, we also included a condition of the phage cocktail alone.The result revealed that, with a challenge with VP AHPND , almost 20% of shrimp died upon the challenge at day 1, followed by a sharp increase in mortality leading to only 40% surviving shrimp at day 7 (Fig. 5c; red line).The presence of the phage cocktail can prolong and reduce shrimp mortality.The phage cocktail retained the survival rate of infected shrimp at a similar level compared to the media control at day 1 (Fig. 5c; yellow and blue lines).However, since the presence of the vibriophages sharply declined at day 5, in the cultured conditions (Fig. 5d), the cocktail failed to rescue the infected shrimp at the following period, resulting in a 60% survival rate at day 7 (Fig. 5c; yellow line).The survival rate of shrimp in the presence of the cocktail was 20% higher than that in the absence of the cocktail, suggesting the activity of the cocktail in preventing the animals from infection (Fig. 5c; yellow and red lines).The shrimp survival rates in the conditions of the media control and the phage cocktail alone were at a comparable level (Fig. 5c; blue and green lines), suggesting undetected side effects from the phage cocktail.
Since we observed the low stability of our vibriophages in shrimp-rearing water, we questioned whether the phages would be able to colonize the open digestive tract of shrimp.To evaluate the colonization of phages in the shrimp digestive system, we dissected the stomach and hepatopancreas of shrimp in the experiment and determined the phage titer for at least 7 days in comparison to the phage titer in shrimp rearing water (Fig. 5e).At day 0, the phage titer in shrimp-rearing water was approximately 10 5 -10 6 pfu/ml, which was similar to the concentration of the phage cocktail initially added to the water.The number of phages, again, dropped substantially to around 10 3 pfu/ml at day 3 and was almost completely diminished (less than 10 pfu/ml) at day 5 and day 7 (Fig. 5e; blue), suggesting the vibriophages' low tolerance in the rearing water.Surprisingly, the number of phages in the digestive tract of shrimp increased significantly from an undetectable level at day 0 to approximately 10 3 pfu/ml at day 1 (Fig. 5e; purple).Moreover, the phage numbers in the digestive tract remained relatively constant throughout the 7-day experimental period (Fig. 5e; purple).Taken together, this phage cocktail, which is derived from the nucleus-forming vibriophages Eric and Ariel, is effective in preventing shrimp from VP AHPND infection and is capable of colonizing the digestive tract of shrimp.

Discussion
The spread of pathogenic VP AHPND contributes to global losses in the shrimp farming industry 5,7 .Unfortunately, the use of antibiotics, a conventional approach, has limited success due to multidrug-resistant (MDR) variants 40 .Bacteriophage therapy, the use of prokaryotic viruses targeting bacteria, is now considered a future alternative strategy to combat pathogenic infection and has been proven by numerous successful cases worldwide 23,41 .Here, we present an integrated workflow starting from high-throughput phage screening, followed by characterization of phage kinetics and genomic data, and the application of phage as a biocontrol.With our high-throughput www.nature.com/scientificreports/phage screening adapted from the previous published HiTS method 26 , we can rapidly identify many potential natural resources that contain phages targeting the bacterial isolates of interest.Since the screening of a variety of bacterial isolates and various natural samples was conducted in a checkerboard manner, this screening generally increases the possibility of obtaining diverse phages in a single screen, thus avoiding the laborious work of traditional phage screening.However, with this screening, it is worth mentioning here that the elementary criteria, including clear plaque morphology and host spectrum, are still needed for the selection of phage candidates appropriate for use.In this screening of 98 resources against 28 pathogenic VP AHPND , we obtained 12 potential resources that contain at least 14 different phages.Among these, phages phiKT1019 (Eric) and phiKT1028 (Ariel) that displayed the broadest host spectrum were selected as candidates due to our primary selection criteria described above.Selection of phages for biocontrol and therapy is a crucial step, and selection of inappropriate phages could result in ineffective treatment 42 .Lytic or virulent phages are highly recommended for the use in bacterial infection treatment and prophylaxis to avoid horizontal gene transfer that would make the bacteria much more threatening, as in the case of temperate phages 43 .Here, we have carefully determined that phages, Eric and Ariel, are safe and appropriate for use in biocontrol.Their genomes did not contain any toxin genes and antibiotic-resistance genes.In addition, the absence of lysogenic-related genes as of the genomic data and the lysogeny assay in phageresistant isolates experimentally support that the phages are indeed lytic phages.However, using these phages might have some drawbacks due to the relatively low adsorption rate to the bacterial host and the production of low progeny after the lytic cycle compared to other phages 44 .This is unsurprising as it has been known that www.nature.com/scientificreports/giant phages that harbor large genomes would require more replication period and highly organized replication that would result in the low production of phage progeny 45 .Phages Eric and Ariel are nucleus-forming vibriophages that belong to the Chimalliviridae family due to the presence of the chmA gene in the genome.The chmA gene, as one of the defining structures for nuclear-based replication, is conserved and present across diverse nucleus-forming jumbophages, including Erwinia, Ralstonia, Escherichia, Salmonella, Pseudomonas, and Vibrio phages 27,29 .During infection by the phages in this family, they encode the ChmA protein along with other accessory proteins to assemble a proteinaceous shell to enclose the phage genome and partition proteins according to their functions, allowing enzymes involved in DNA replication and RNA transcription inside while excluding other enzymes involved in protein translation and nucleotide synthesis on the outside 30,34 .This structure is referred to as the phage nucleus because of its biological functions that resemble those of the eukaryotic cell nucleus 32,46 .In addition to serving as the nucleus during phage replication, in an evolutionary arms race between them and their bacteria host, the phage nucleus also protects the phage genome from host defense mechanisms that target DNA, such as Clustered Regularly Interspaced Short Palindromic Repeats (CRISPRs) and Cas enzymes, and restriction-modification enzymes, thus overcoming the host defense 33,35 .Therefore, the ability of vibriophages Eric and Ariel to assemble the phage nucleus to protect their genome from the DNA-targeting defenses of the bacteria is greatly advantageous from an application standpoint, as they would be more tolerant in long-term application than phages in other families.
PhuZ-based filaments, previously thought to be another defining structure for nuclear-based replication as they spatially and temporally organize the phage nucleus position and capsid maturation, are not obligatory conserved in the Chimalliviridae family 27,29 .Previous studies have revealed that disruption of the dynamic stability of PhuZ filaments leads to the mispositioning of the phage nucleus and a failure to deliver procapsids to the nucleus for DNA encapsidation, resulting in decreased phage progeny [36][37][38]47 . Ourstudy demonstrated that phage Ariel has a slightly higher number of phage progenies than phage Eric, which assembles the PhuZ-based filaments.This suggests that nucleus-forming phages that lack the phuZ gene in their genome might exploit alternative subcellular organizations to achieve optimal phage progeny production.Other organizations that might complement the absence of PhuZ filaments require further investigation.Our recent research has implicated PhuZ as a factor contributing to virogenesis incompatibility for viral speciation, as homologs of PhuZ from distinct phages can co-assemble but produce non-functional static hybrid filaments 48 .Therefore, infection of a single bacterial cell with 2 nucleus-forming phages encoding PhuZ homologs that would occur in the phage cocktail might hamper the effectiveness of the phage combination.Fortunately, since one of the phages in our combination lacks a PhuZ homolog, the viral incompatibility mediated by the non-functional hybrid filaments would not occur, rendering the combination of our vibriophages compatible and effective for use.Since the combination of phages, which possess different cellular organizations, enhances the overall efficiency of the cocktail 49 , insights into host-phage interaction at the cellular level during co-infection of these phages will be necessary to warrant the compatibility of phages in the future.
According to the appropriate properties of the candidates, phages Eric and Ariel, described above, the phage cocktail was formulated and was shown to produce desirable outcomes in both in vitro and in vivo studies.The cocktail completely suppressed the regrowth of three highly pathogenic strains of VP AHPND , previously isolated from infected shrimp in local farms.In addition, the use of a cocktail containing polyvalent phages was more effective than the use of monovalent phages in reducing the bacterial density, ensuring the advantage of the phage cocktail.The maximum efficiency of the phage cocktail (VP AHPND + cocktail) appears to be maintained within the first 24 h, as evidenced by shrimp survival rates, which are comparable to those observed in the absence of VP AHPND but are higher than those observed in the presence of VP AHPND alone (Fig. 5c).However, the activity of the cocktail in the condition (VP AHPND + cocktail) was reduced thereafter as the cumulative mortality reached 40% at day 7.The cumulative mortality is still lower than that of the conditions in the presence of VP AHPND alone at day 7, indicating the prophylactic effect of the cocktail.This is partly due to the low phage stability in the cultured tanks, which is likely influenced by unknown environmental factors (Fig. 5d and 5e).Further development for maximizing phage stability in environments will be needed to enhance the phage cocktail efficiency.Even though the phages are not tolerant of the environment, they are capable of colonizing the open digestive tract of shrimp and being retained for at least a week (Fig. 5e).It has been reported in moribund shrimps that phages are able to penetrate into the hepatopancreas through hemolymph 50 , and the phage particles in the digestive tract are likely maintained within the mucosal surface of the shrimp stomach by the interaction between Immunoglobulin (Ig)-like domains on the phage capsid and mucin peptidoglycans 51,52 .In this study, the organs in shrimp where phages colonize are still ambiguous, as phages were titered from the mixture of the ground digestive tract.This will further form the basis for an examination of the process of colonization by phages and specific organs where phages interact.Moreover, since shrimp immersion in seawater containing a phage cocktail appears to allow the phages to colonize in the gut, but at low quantities (~ 10 3 pfu/ml), feeding shrimps a diet supplemented with the cocktail may contribute to more phage particles adhering within the tract and would be a further application worth pursuing.This research on a phage cocktail derived from nucleus-forming vibriophages still has room for advancement and would soon provide a potential preventive strategy against AHPND infection.

Bacterial strains and growth
To make an overnight culture, V. parahaemolyticus (Table 1) were grown on tryptic soy agar plates supplemented with 1.5% NaCl (TSA-salt) at 30 °C and then inoculated into tryptic soy broth supplemented with 1.5% NaCl (TSB-salt) at 30 °C, for at least 16 h, vigorously shaking at 250 rpm.To make a day culture, the overnight culture was inoculated 1:100 into TSB-salt until the early exponential growth (the optical density at 600 nm = 0.2 or 0.4) by shaking at 250 rpm, 30 °C.The day culture was used in following experiments: phage adsorption, one-step

Vibriophage screening
According to the phage isolation method of Olsen et al. 26 , we employed the high-throughput screening (HiTS) method with modifications.To easily handle a large number of seawater samples, we adjust smaller final volumes per well and use chloroform to eliminate bacteria instead of filtration.On top of that, we allowed the phage to propagate for 2 days, which was longer than the HiTS method.Our method is able to apply for isolating various phages from other environmental samples by adjusting the host medium, buffers and incubation conditions.To this, our developed method was roughly explained into 4 steps as follows; (I) Sample preparation, (II) Phage propagation, (III) Bacterial elimination and (IV) Spot test, as illustrated in Fig. 1.

Phage purification and propagation, plaque morphology, and phage structure
In order to purify and propagate phages, a double-layer agar (DLA) technique was used as described by Thammatinna et al. 53 .Briefly, for phage purification, the mixture of V. parahaemolyticus overnight culture (100 µl) and phage lysate (100 µl) was added into 5 ml semi-solid TSB-salt agar (0.35% agar) and then immediately poured on TSB-salt agar followed by incubation at 30 °C for 24 h.A single plaque was picked from the double-layer agar plate and suspended in 100 µl of SM buffer (100 mM NaCl, 8 mM MgSO 4 , 50 mM Tris-HCl pH 7.4, 0.002 w/v gelatin).To ensure a single phage, this step was repeated at least three times.
For phage propagation, phage was amplified by a double-layer agar technique plate as described above.After incubation, 5 ml of SM buffer was added to the web lysis plate followed by incubation at room temperature for 5 h allowing phage to release into the buffer.Then, the phage lysate was collected by centrifugation at 9000 rpm for 15 min and the supernatant was filtrated through a 0.45 µm filter.Phage lysate was maintained in SM buffer and subsequently stored at 4 °C until use.
To investigate phage morphology, a transmission electron microscope (TEM) was performed.The clear plaque on the lawn of the parental host was picked and resuspended in SM buffer.After centrifugation at 9000 rpm for 2 min, the supernatant was dropped onto copper grids and negatively stained with 2% w/v uranyl acetate (pH4.5).The phage morphology was observed by a Hitachi HT7700 at 80-100 keV.

Host range determination
The host spectrum of the isolated phage against 31 strains of V. parahaemolyticus (Table 1) was determined by using the spot test.Briefly, 5 µl of phage lysate was dropped onto the soft-agar plate surface, which was prepared by mixing 100 µl of bacterial overnight culture with 5 ml of the 0.35% semi-solid agar and overloading onto the TSB-salt plate.The formation of the translucent spot on the bacterial lawn was recorded after 16 h of incubation at 30 °C.

Phage adsorption and one-step growth curve
The evaluation of phage adsorption and the one-step growth curve was performed as described in a previous study by Thammatinna et al. 53 .V. parahaemolyticus was cultured in TSB-salt as mentioned previously.For the adsorption assay, the bacterial culture at an optical density at 600 nm (O.D. 600 ) of 0.4 indicating ~ 4 × 10 8 cfu/ml was infected with phage at a multiplicity of infection (MOI input ) of 0.01 or 4 × 10 6 pfu/ml in a static condition at 30 °C.The 100 µl of samples were immediately taken to 900 µl of SM buffer at the following time points: 0, 1, 2, 5, 7.5, 10, 15, 20, 25, and 30 min, and then they were centrifuged at 12,000 rpm at 4 °C for 1 min.The nonadsorbed free phage in the supernatant at each time point was evaluated using the double-layer agar method.
A one-step growth curve analysis was performed to determine the latent period and burst size of the isolated phages.Mid-log phase growth V. parahaemolyticus cultures were infected with phage as described in the above condition.Phage particles were allowed to attach to their host for 15 min at 30 °C followed by removing the non-adsorbed free phage by centrifugation at 10,000 rpm for 5 min.The infected cells were suspended in 10 ml of TSB-salt media and incubated in a static condition at 30 °C for 2 h.At the desired time points, the bacterial cells were then removed by centrifugation at 9000 RPM for 2 min, and the supernatant was collected to evaluate phage number using the double-layer agar method.The calculation of the burst size is provided below.

Phage tolerance: pH and temperature
The stability of the isolated phages was examined at different pHs and temperatures.To evaluate the pH stability, 50 µl of each phage lysate containing around 1 × 10 6 pfu/ml was incubated with 450 µl of SM Buffer with the following pH values; 2, 4, 6, 7.4, 8, 10, and 12 at 30 °C for 1 h.For thermal stability of the phage, 50 µl of 1 × 10 6 pfu/ ml phage lysate was kept at different temperatures for 1 h as follows: 4, 35, 30, 37, 40, 50, 60, and 70 °C.The survival phages were titered by using a spot test, as described previously.

Lysogeny test
To determine whether phage is temperate phage, a phage-resistant strain containing prophage in its genomes was investigated by induction with mitomycin C. Firstly, a phage-resistant strain was isolated from the parental host lawn containing a high titer of phages.Secondly, to confirm the resistant strains, high-titer phage lysate was dropped onto the cross-streaks of the isolated strains compared with the wild-type strain.Lastly, the wild-type strain was grown in semi-solid TSB-salt agar supplement with 0.05% mitomycin C and then resistant-strain was stabbed using a sterile toothpick.The DLA plate was incubated at 30 °C for at least 16 h.Positive clear zone indicated the presence of prophage in V. parahaemolyticus genome.The phage lysate was used as positive zone.
The triplicate experiment was conducted.
To further confirm the presence/absence of prophage in the phage-resistant genomes, conserved genes including capsid and recA genes of each phage were detected in the genome by PCR method.Briefly, genomic DNA of the phage-resistant strains was extracted by using the colony boiling method.The colony of resistant bacteria was picked, transferred to 100 µl deionized water, boiled at 100˚C for 10 min, and placed on ice for 15 min.The oligonucleotide primers of both conserved genes of each phage were designed including capsid gene of phiKT1019; forward: 5′-GCT GCA GGA GGC AGC CAA AAA ATG AAA AGA AAT ATC GTA GCT C-3′ and reverse: 5′-GCC TGC AGG TCG ACT CTA GAC TAT GGC AAC TTA CCT GCATC-3′, recA gene of phiKT1019; forward: 5′-GCT GCA GGA GGC AGC CAA AAA ATG TCC ACC GCT TTT AAT TTC -3′ and reverse: 5′-GCC TGC AGG TCG ACT CTA GAT TAC TTT TTC TTA GGA CGACC-3′, capsid gene of phiKT1028; forward: 5′-GCT GCA GGA GGC AGC CAA AAA ATG AAA AGA AAT ATC GTA GCT C-3′ and reverse: 5′-GCC TGC AGG TCG ACT CTA GAT TAT GGT AAC TGA CCA GTTTC-3′, and recA gene of phiKT1028; 5′-GCT GCA GGA GGC AGC CAA AAA ATG TCA ATT CCG AAT TTCG-3′ and reverse: 5′-GCC TGC AGG TCG ACT CTA GAT TAT TTT TTT GCA GGA CGTC-3′.PCR mixture contained 1X PCR buffer (0.25 mM dNTP, 1.5 mM MgCl 2 ), 1 µM of each forward and reverse primer, 1 U Taq DNA polymerase (Thermo scientific), and 5 µl of DNA template in a total volume of 25 µl.The reaction was carried out in Bio-Rad T100 Thermal cycler as follows: 94 °C for 1 min; 35 cycles of 30 s at 94 °C, 45 s at 55 °C, and 1 min at 72 °C; 10 min at 72 °C.The expected size of the capsid and recA genes were 2195 bp and 1434 bp for phiKT1019 and 2180 bp and 1422 bp for phiKT1028.The PCR products were investigated by agarose gel electrophoresis in 1x TBE buffer (0.04 M Tris-acetate, 0.001 M EDTA [pH 8.0]) and visualized by UV transilluminator.

Phage DNA extraction, phage genome sequencing and genome analysis
Phage particles were precipitated with a 20% phage precipitant solution (30% w/v PEG-8000, 3.3 M NaCl, and sterile distilled water) at 4 °C overnight and centrifuged at 10,000 rpm for 30 min.To degrade bacterial DNA and RNA, the pellet was resuspended in 1 × DNaseI buffer and incubated with 5U DNaseI and 25 µg RNaseA at 37 °C for 2 h.The nuclease activity was inhibited by adding 25 mM EDTA, and after that, phage capsids were digested with 0.5% SDS and 25 µg proteinase K at 60 °C for 2 h.Lastly, phage genomic DNA was extracted using the conventional phenol-chloroform method.
Whole-genome libraries were carried out on the Illumina HiSeq platform and qualified by FastQC 54 .The contaminated host DNA fragments were filtered out by using Bowtie2 55 and Samtools 56 against V. parahaemolyticus genomes.Lastly, the filtered reads of each phage were assembled as a single contig by SPADES 57 .
The formation of the phage genome (in linear or circular form) was determined by PhageTerm 58 .Open reading frames (ORFs) were predicted by GeneMark 59 and assigned by the algorithms of BLASTp 60 and PSI-BLAST 60 (cutoff e-value < 10 −3 ) against the following databases: non-redundant (nr) protein databases of NCBI, InterPro 75.0 61 , NCBI conserved domain 62 and ACLAME 63 .The predicted genes were also automatically confirmed by the RAST server annotation service 64 and PHASTER 65 .The presence of antimicrobial resistance genes and putative toxins was determined using BLAST against RESFINDER and VirulenceFinder 66 .The tRNAScanSE 67 was used to examine the putative transfer RNA (tRNA).The genome map was produced by the DNA plotter tool of Artemis software 68 .
A phylogenetic relationship was constructed by the VICTOR Virus Classification web tool 69 , Tree Building Online Resource, Molecular Evolutionary Genetics Analysis (MEGA) version 10.0 70 , and Figtree program against complete phage genomes that were selected from the GenBank database.The intergenomic similarity and comparative genome of selected phage genomes were conducted by the VIRIDIC web tool 28 and EasyFig 2.1 71 , respectively.

Bacterial whole genome sequencing, genome assembly and annotation
The DNA of V.parahaemolyticus KT1018B (or J42) was extracted using a bacterial DNA extraction kit (Vivantis).Its quality was evaluated by Nanodrop and agarose gel electrophoresis with A 260 /A 280 > 1.90 and no degraded pattern.The paired-end 2 × 150-bp sequencing library was prepared, and the sequencing was performed using the DNBSeq platform (BGI Tech Solutions (Hongkong) Co., Ltd.).The sequencing output was evaluated for quality with Q20 > 95%.The raw reads were removed from low quality sequences and remaining adapters by SOAPnuke software 72 .
Cleaned reads were assembled and annotated using NCBI Read Assembly and Annotation Pipeline Tool v0.5.3 (RAPT, https:// www.ncbi.nlm.nih.ov/ rapt).In brief, the pipeline consists of three major parts: genome assembly using SKESA 73 , taxonomic assignment inferred from the average nucleotide identity (ANI) matrix and annotation using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) 74 , and production of an annotated genome of quality comparable to RefSeq.Associated gene ontology (GO), cluster of orthologous groups (COG), and KEGG pathway terms were then identified using EggNOG-mapper v2.1.9 75.Assembled contigs and associated features were visualized as a circos plot using the Proksee (https:// proks ee.ca) web application.

Figure 2 .Figure 3 .
Figure 2. Characterization of morphological and biological properties of the selected phage candidates; phiKT1019 and phiKT1028.Plaque morphology of phage phiKT1019 (a) and phage phiKT1028 (c).Scale bars, 2 mm.Morphology of phages: phiKT1019 (b) and phiKT1028 (d) observed by TEM and negative staining.Scale bars, 100 nm.Biological studies of phages phiKT1019 (blue) and phiKT1028 (red); phage adsorption assay (e), one-step growth curve (f), and stability of phages at various temperatures (g) and pHs (h).The experiments were performed in at least 3 independent biological replicates and the data are represented as mean ± standard deviation.

Figure 4 .
Figure 4.The phages phiKT1019 (Eric) and phiKT1028 (Ariel) are classified into the Chimalliviridae family and are nucleus-forming vibriophages.(a) Phylogenetic analysis of the relationship between genome of various selected phages from the database.Bootstrap values (%) of the tree as constructed by VICTOR and MEGA are shown at each main clade.The phage families, including Chimalliviridae, Siphoviridae, Demerecviridae, and Myoviridae, are included in the tree.Asterisks represent the branch of vibriophages Eric and Ariel.Scale bar represents the branch length of taxa.(b) VIRIDIC heatmap showing the intergenomic similarities of selected phages' genomes; vibriophages BONAISHI, Aphrodite1, pTD1, Eric, Ariel, and Pseudomonas phage 201 Phi2-1.The intergenomic similarities between each pair of phage genomes are shown as percentage on the right half of the figure.The closely-relationship of the phages is represented by the dark blue color (scaled by color annotation of intergenomic similarity).(c) Genome comparison of 6 selected phages.The dark blue arrows represent the annotated ORFs with their directions.The ORFs encoding for PhuZ and ChmA are color-coded in yellow and pink, respectively.The similarity of phage genome regions is indicated by gradient shades of grey.(d) Single-cell level assay revealing the assembly of the phage nucleus during infection with vibriophages Eric and Ariel.Fluorescence images of uninfected V. parahaemolyticus KT1018B expressing sfGFP-tagged ChmA (upper panel, ORF203, lower panel, ORF180) and V. parahaemolyticus KT1018B expressing sfGFP-tagged ChmA infected with corresponding vibriophages at MOI input 1 (upper panel, ORF203 infected with Eric, lower panel, ORF180 infected with Ariel).DNA was stained with DAPI (blue or grey) and cell membrane was stained with FM4-64 (red).Dashed lines indicate cell borders.Scale bars, 1 micron.(See also Figure S2).

Figure 5 .
Figure 5.The efficiency of the phage in vitro and in vivo.(a) Heatmap of growth inhibition of 12 V. parahaemolyticus strains in the presence of the phage cocktail at MOI input 10 (left columns) and MOI input 0 (right column; control) throughout 20 h.The O.D. values range from 0 (white) to 1 (yellow).(b) Bacterial growth of V. parahaemolyticus AHPND strain KT1021B in the absence of phages, in the presence of individual phage (Eric or Ariel at MOI input 10), and in the presence of the cocktail (MOI input 10) during 20 h of incubation.(c)The prophylactic performance of the phage cocktail against the infection of V. parahaemolyticus AHPND strain KT1021B throughout 7 days and (d) the stability of phages in the environmental tanks.The experimental groups were set up as following; TSB-salt media alone (blue), the phage cocktail alone (green), challenge with VP AHPND in the presence of the phage cocktail (yellow), and challenge with VP AHPND (red).(e) Colonization of phages in the digestive tract of shrimp.Phage numbers were evaluated in the stomach and hepatopancreas of shrimp (purple) and shrimp rearing water (blue) throughout the 7-day treatment, showing that the phages were retained within the shrimp tract, while becoming not infective in the shrimp rearing water.The experiment was carried out triplicate and the data represent mean ± standard deviation.