In this study, crucian carp (average weight 35 ± 1 g) specimens obtained from a fish farm were used. All experiments were performed rigorously under the Regulations for the Administration of Affairs Concerning Experimental Animals that were issued by the State Council of the People’s Republic of China (1988.11.1), as well as in accordance with the guidelines of the Animal Welfare and Research Ethics Committee at Jilin Agriculture University (JLAU08201409).

Aeromonas hydrophila 152 was used for phage isolation, and 39 additional Aeromonas strains (including 21 A. hydrophila and 18 A. veronii) were used for host range analysis (Table 1). All the bacterial strains were cultured in LB broth at 37°C.

Aeromonas hydrophila 152 was adopted for phage propagation in the current study. PZL-Ah152 was isolated from sewage systems in Changchun, China. Phage isolation, plaque assays, and spot tests were conducted according to previously reported methods (Gu et al., 2011; Hyman, 2019). A measure of 5 μL of phage (1.0 × 10⁷ PFU/ml) was spotted on double-layered plates containing different A. hydrophila strains for the detection of the host range. After overnight incubation, the zone of lysis of the susceptible host was observed.

One-step growth curve determination was performed according to a previously described method with slight modifications (Kropinski, 2017). Briefly, A. hydrophila 152 was incubated in an LB medium until the logarithmic phase (OD600 nm = 0.5), and the culture was infected with bacteriophage PZL-Ah152 with an MOI of 0.1 and kept at 37°C for 5 min for adsorption. The cultures were centrifuged to remove the unabsorbed bacteriophage, and the precipitate was resuspended in 10 ml of LB. A total of 0.1 ml of culture was transferred to 9.9 ml of LB. A 10-fold serial dilution of the mixture was carried out two times and incubated at 37°C. Then, 0.1 ml of the sample was collected every 5 min for PFU determination by plaque determination on a double-layer LB plate. The average burst size was quantified as the difference between the final and the initial phage titers divided by the initial phage titer.

For pH stability tests, the effect of varying pH on a 100 μL PZL-Ah152 (1.0 × 10⁷ PFU/ml) was determined; for this purpose, the bacteriophage was cultured in LB broth adjusted to pH 2–13 for 1 h, and aliquots were taken to measure phage titers at different pH values. For thermal stability tests, 2 ml of bacteriophage (1.0 × 10⁷ PFU/ml) was incubated at 30°C, 40°C, 50°C, 60°C, 70°C, and 80°C. Then, 100 μL aliquot samples were collected at intervals of 10 min. Moreover, we incubated aliquot samples of phage PZL-Ah152 suspensions at 4°C for 1 year. All tests were performed in triplicate.

Bacteriophage was condensed and modified as previously studied (Gong et al., 2016). Bacteriophage PZL-Ah152 was amplified in 800 ml LB and centrifuged at 8,000 × g for 15 min at 4°C to remove bacterial debris. Phage particles were precipitated by the addition of 10% polyethylene glycol 8000 and 1 M NaCl and subsequently disbanded in 5 ml PBS. CsCl (1.45, 1.50, 1.70 g/ml) was added to the supernatant and centrifuged at 120,000 × g for 3 h at 4°C. The phage band was harvested and dialyze.

Phage morphology was observed by negative phosphotungstic acid staining. A 20 μL aliquot of the concentrated PZL-Ah152 suspension was applied to copper grids stained negatively with 1% phosphotungstic acid (pH 7) for 30 s. Electron microscopy of JEOL (JEM-1400, Japan) was operated at 80 kV. Afterward, 10 μL of phage particles purified by CsCl density-gradient centrifugation were mixed with a gel electrophoresis sample buffer, boiled for 10 min, and subjected to SDS–polyacrylamide gel electrophoresis (8–16% gradient). Protein bands were visualized by using Instant Blue staining (Expedeon Protein Solutions Ltd., Cambridge, United Kingdom).

Phages (≥ 10¹¹ PFU/ml) were filtered through a 0.22 μm filter. The genomic DNA of the phage was extracted using a Universal Phage Genomic DNA Extraction Kit (Knogen, Guangzhou, China). The extracted phage DNA was sequenced at Sangon Biotech (Shanghai, China) by using an Illumina HiSeq 2500 sequencing system. Open reading frames (ORFs) were predicted with BLAST and GeneMarkS.

Aeromonas hydrophila 152 mixed with PZL-Ah152 at MOIs of 1, 0.1, and 0.01 was cultured in tubes filled with LB broth (1 × 10⁸ CFU/ml). After incubation at 37°C for 1, 2, 3, 6, 9, and 12 h, the cell survival rate was detected. A bacterial culture without the phage was used as the control, and a colony count was carried out.

For the safety assay, PZL-Ah152 was intraperitoneally administered to wild-type crucian carps (1.0 × 10¹⁰ PFU/ml, 200 μL/fish) for 12 consecutive days, while the control group was exposed to the same volume of PBS. Tissue sampling (n = 3) was performed on days 1, 4, and 12 postexposure. Liver, spleen, kidney, gut, and gill tissues were surgically removed to conduct hematoxylin and eosin (H&E) staining analysis and qRT-PCR. The mRNA expression levels of TGF-β, IFN-γ, TNF-α, IL-1β, and IL-10 were quantified by qPCR. The primers used for the immunity-related genes and β-actin, which was considered a housekeeping gene, are listed in Table 2.

To determine the minimal lethal dose (MLD), six crucian carps in each experimental group were injected intraperitoneally (i.p.) with A. hydrophila 152 (at concentrations of 1 × 10⁶, 1 × 10⁷,1 × 10⁸, 1 × 10⁹, and 1 × 10¹⁰CFU/ml, 100 μL/fish). No bacterial inoculation was performed in the control group, and 2 × MLD was used as the infective inoculum.

Crucian carps infected with 2 × MLD (2 × 10⁹ CFU/ml, 100 μL/fish) of A. hydrophila 152 were treated with PZL-Ah152 at different concentrations (1 × 10⁸, 1 × 10⁹, and 1 × 10¹⁰ PFU/ml, 100 μL/fish) (n = 20 in the respective groups) after 1 h. The crucian carps were infected with 2 × MLD (2 × 10⁹ CFU/ml, 100 μL/fish) of A. hydrophila 152, and 12 h and 24 h later, the crucian carps were treated with PZL-Ah152 (1 × 10¹⁰ PFU/ml, 200 μL/fish, n = 20 in each group). The mortality rates of the fish were recorded every 12 h for 7 days to determine the mortality (Jia et al., 2020).

The bacterial loads in the crucian carp gut were determined. The crucian carp intestinal structure was harvested, weighed, and suspended in filter-sterilized PBS. The collected intestinal structure slurry was diluted. Then, 100 μL of each dilution was used to determine the bacterial loads.

A lipopolysaccharide (LPS) ELISA kit (Jiangsu Jingmei Biological Technology Co., Ltd., China) was used to measure the level of LPS in the crucian carp intestinal contents 24 h after phage treatment. The operation method complied with the product instructions.

Histopathological analysis of the liver, spleen, kidney, gut, and gill was performed. Briefly, the crucian carp tissues were removed and placed in 4% formalin, stained with H&E, and analyzed by microscopy.

Totally, 24 crucian carps were randomly and equally divided into eight groups (Bg, Bge, Pg, Pge, BPg, BPge, Ng, and Nge). The fish in the Bg group were challenged i.p. with 2 × 10⁸ CFU of A. hydrophila 152 (per fish). The fish in the BPg group were challenged with A. hydrophila 152 (2 × 10⁸ CFU/fish) and injected i.p. with 2 × 10⁹ PFU of PZL-Ah152 after 1 h. The fish in the Pg group were injected i.p. with 2 × 10⁹ PFU of PZL-Ah152. The Ng group was not treated. The intestinal contents of each fish were collected at 24 h post-infection. The procedures for Bge, Pge, BPge, and Nge were the same as those for the previous four groups, except that the intestinal epithelial mucus was collected (Supplementary Table 1).

For each intestinal content and intestinal epithelial mucus sample, total genomic DNA was extracted by using a QIAamp DNA Stool Mini Kit (Qiagen, West Sussex, United Kingdom). Sequencing libraries were constructed by PCR amplification of the V3 + V4 regions of the 16S rRNA gene using the primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′) (Drengenes et al., 2021). The amplicons were purified using a Qiagen Gel Extraction Kit (Qiagen, Germany). Sequencing libraries were generated using the TruSeq^@ DNA PCR-Free Sample Preparation Kit (Illumina, United States), and then, index codes were added. Library quality was assessed using a Qubit@ 2.0 fluorometer (Thermo Scientific) and an Agilent Bioanalyzer 2100 system.

Trimmomatic and PEAR were used to screen the FASTQ data and to obtain high-quality sequences. Raw tag quality filtering was performed under specific filtering conditions to obtain high-quality clean tags using QIIME (V1.7.0 3) (Caporaso et al., 2010; Bokulich et al., 2013). Sequence analyses were performed using UPARSE (UPARSE v7.0.1001 6) (Edgar et al., 2011). Sequence similarities (minimum identity of 97%) were assigned to the same OTUs (Febvre et al., 2019). Other subsequent biological analyses were conducted based on the results of the OTU cluster analysis.

All statistical analyses were performed using the SPSS (version 25). Standard deviations were calculated in all experiments. Redundancy analysis (RDA) was used to determine the correlation between intestinal cytokines and the composition of the intestinal microbiome. R software (version 2.15.3) was used for data analysis.

The phage PZL-Ah152 was isolated from sewage systems in Changchun by plaque purification. PZL-Ah152 formed plaques after incubation for 12 h (3–4 mm diameter, Figure 1A). TEM images showed that PZL-Ah152 had a capsid diameter of 83 ± 1 nm and a tail length of 8 ± 1 nm. The morphological characteristics of PZL-Ah152 showed that it belonged to the Podoviridae family (Figure 1B). The structural protein of the most prominent protein band from phage capsids corresponded to the major capsid protein. The size indicated in the gel (38 kDa) (Figure 1C) was approximately the same as the ORF-01 protein (39.9 kDa) (Supplementary Table 2), as a potential major capsid protein.

We determined the host range of PZL-Ah152 against a group of 40 Aeromonas spp. strains. Apart from A. hydrophila 152, PZL-Ah152 could lyse nine other Aeromonas strains, including eight strains of A. hydrophila (Ah-TPS, Ah-119, Ah-69, Ah-150, Ah-103, Ah-142, Ah-107, and Ah-87) and one strain of Aeromonas veronii AV-6 (Table 1). The results of the one-step growth curve showed that phage PZL-Ah152 expressed a short latent period of less than 20 min. The rise period was about 20 min. In addition, the burst size of phage PZL-Ah152 was approximately 91 PFUs per infected cell on strain A. hydrophila 152 (Figure 2A). PZL-Ah152 was stable at pH values in the range of 4-11 and acted more effectively at pH 6-8 (Figure 2B). The phage retained full activity at 30°C, 40°C, and 50°C but retained half of the activity at 60°C for the full time of exposure, and after 50 min of exposure to 70°C, the activity to kill bacteria disappeared. It was completely inactivated at 80°C after 10 min (Figure 2C). Furthermore, the titer of PZL-Ah152 remained almost constant after 1 year of storage at 4°C (Figure 2D).

To examine PZL-Ah152 at the genetic level, we obtained and analyzed the complete genome sequence of PZL-Ah152 and deposited it in GenBank (MW671054). The complete genome of PZL-Ah152 was 40,975 bp. The composition of the phage was 23.50% A, 24.76% T, 26.85% C, and 24.90% G (Figure 3A). The phage genome contained 54 putative ORFs (Supplementary Table 2). Among all 54 ORFs in PZL-Ah152, 52 ORFs (96.3%) had putative functions, and the sizes of the proteins encoded by the 52 ORFs ranged from 4.55 kDa (ORF15) to 149.96 kDa (ORF48).

BLASTp analysis showed that 23 proteins (42.6%) were annotated predictive functions, with the remaining 31 (53.7%) putative gene products as unknown functions. The 23 proteins according to their functions were divided into four categories: metabolism and replication of nucleic acids (ORFs 08, 18, 19, 22, 30, 31, and 34), host lysis (ORFs 17 and 46), structure/morphogenesis (ORFs 01, 02, 03, 27, 47, 48, 49, 50, 51, 53, and 54), and DNA packaging/maturation (ORF 44). The analysis did not identify any toxic genes in the PZL-Ah152 genome (Figure 3B).

The PZL-Ah152 genome was similar to the Aeromonas phage T7-Ah (95.76%), Klebsiella phage vB_KpnP_Sibilus (79.03%), Dickeya phage vB_DsoP_JA10 (78.85%), Erwinia phage pEp_SNUABM_12 (74.81%), Vibrio phage JSF30 (71.82%), Vibrio phage VP4 (71.41%), and Vibrio phage VP3 (71.35%) (Figures 4A–C), but the phage PZL-Ah152 genome also had its own unique sequences compared to those phages, such as the sequences from 6,529-6,645 bp, 7,132-7,290 bp, 9,361-9,570 bp, 15,917-16,279 bp, 17,798-18,361 bp, 18,405-18,563 bp, 22,670-22,861 bp, and 28,018-29,739 bp. This finding suggested that PZL-Ah152 is a new phage strain.

To evaluate the activity of PZL-Ah152 against A. hydrophila 152, we performed a time killing assay. The results showed that when MOI = 0.1, the presence of phages after 2 h of co-incubation reduced A. hydrophila 152 counts by 4.2 log units. When MOI = 1 or MOI = 0.01, A. hydrophila 152 counts decreased by 3.95 and 3.67 log units after 2 h, respectively (Figure 5).

The safety of using PZL-Ah152 as a potential therapeutic was examined by exposing crucian carps to PZL-Ah152 for 12 consecutive days. In the safety test, the survival rate of crucian carps reached 100%. In addition, the tissues collected from immunized crucian carps treated with phage showed no pathological changes, as shown in Figure 6A. The liver, spleen, kidney, gut, and gill tissues were surgically removed. The mRNA expression levels of TGF-β, IL-1β, TNF-α, IFN-γ, and IL-10 were quantified with qPCR. Compared with the control group, the transcription level of IL-10 mRNA in the liver and spleen began to increase on the first day, while the expression level of TNF-α decreased (Figure 6B). In the intestinal structure, the transcription levels of the five cytokines began to increase on the first day, and the mRNA expression level decreased to a non-significant level on the 12^th day.

The crucian carps were injected intraperitoneally with 2 × MLD (2 × 10⁸ CFU/fish) of A. hydrophila 152, and all of them died within 3 days. We administered intraperitoneal injections of different doses of the phage PZL-Ah152 1 h after challenging with A. hydrophila 152. Compared with PBS, bacteriophage (2 × 10⁹ PFU/fish) treatment significantly improved the survival rate (Figure 7A). After 1 h of A. hydrophila 152 (2 × 10⁸ CFU/fish) infection, the bacterial load reached > 10⁷ CFU/ml in the gut (Figure 7B). At the same time, crucian carps were administered with the phage PZL-Ah152 (1 × 10¹⁰ PFU/ml, 200 μL/fish). The bacterial load decreased by 3.8 log units in the gut tissues after 24 h (Figure 7B). The test produced a 100% survival rate within the experimental period (Figure 7A). In contrast, the bacterial loads in the gut were approximately 8.2 log units 12 h after the crucian carps were infected with A. hydrophila 152 (Figure 7D). Treatment with bacteriophage PZL-Ah152 (1 × 10¹⁰ PFU/ml, 200μL/fish) reduced the gut A. hydrophila by 2.6 log units after 18 h (Figure 7D) and produced 80% survival over 2 days (Figure 7C). However, 24 h after the crucian carps were infected with A. hydrophila 152, the use of bacteriophage PZL-Ah152 reduced the A. hydrophila 2 log unit in the crucian carp gut (Figure 7F) and resulted in a 65% survival rate within 3 days (Figure 7E). The results indicated that crucian carps should be treated with phage as soon as possible after bacterial infection.

By determining the LPS content in crucian carp intestines, we found that when the phage was applied, the LPS content increased along with the decrease in the A. hydrophila 152 level. Although the LPS content in the A. hydrophila 152-challenged group and the phage supplementation group increased, the increase was less significant than that in the control group (Figure 8).

Intestinal morphology is an important indicator of intestinal health. We evaluated the ameliorative effect of the phage PZL-Ah152 on intestinal pathological injury in crucian carps (Figure 9). Infection of crucian carps with A. hydrophila 152 caused damage to intestinal mucosa integrity and resulted in the loss of intestinal crypts, accompanied by inflammatory cell infiltration in the mucosa. The glands were not intact and showed obvious bleeding spots (Figure 9A). In contrast, after 24 h of treatment with the phage PZL-Ah152, the inflammation was gradually alleviated, and the intestinal villus injury progressively decreased, leading to intestinal crypts being observable. Accompanied by goblet cell production, the gut structure slowly returned to normal (Figure 9B).

The gut is the most important digestive organ of crucian carps. We analyzed the changes in the gut microbiota in crucian carps treated with phage therapy by selecting 1,538,144 effective tags. The plateauing of rarefaction curves in all the samples suggested that the sequencing depth was sufficient (Figure 10A). Moreover, the observed species of Ng in the intestinal contents was the smallest, but it was not significantly different from other treatments. The PCoA showed that there was a high similarity in the microbial community structures of the intestinal contents (Figure 10B). Upon comparing groups with and without phage treatment, we found that phage treatment had stronger effects on the intestinal microbial community of crucian carps. Compared with the Ng group, the Proteobacteria content increased significantly in each group (P < 0.05), especially in the Pge group, which showed an increase of 21.5%. However, we discovered that the Fusobacteriota content decreased in either the intestinal contents (37.4%) or intestinal mucus (14.5%) in only the phage-treated groups (Figure 10C). In the intestinal contents and the intestinal epithelial mucus, the Cetobacterium abundance decreased by 10% and 5.3%, respectively, after challenged with A. hydrophila 152. In the Pg and Pge groups, the Cetobacterium abundance decreased by 37.3% and 14.3%, respectively, and in the BPg and BPge groups, the Cetobacterium abundance decreased by 22% and 12.3%, respectively. The Vibrio abundance increased by 5.8% and 6.4% in the intestinal contents and intestinal epithelial mucus, respectively, after A. hydrophila 152 was challenged; in the Pg and Pge groups, the Vibrio abundance increased by 9.4% and 1%, respectively, and in the BPg and BPge groups, the Vibrio abundance increased by 24.4% and 15.6%, respectively (Figure 10D). In addition, the crucian carps attacked by A. hydrophila 152 exhibited greater bacterial changes at the genus level than normal crucian carps. Brevinema, Acidovorax, and Erythrobacter in the intestinal epithelial mucus increased, as did the abundance of Bacteroides, in the intestinal contents (Supplementary Figure 1B). We also found that when the phage PZL-Ah152 was used to treat A. hydrophila 152 infection, the abundance of Lactobacillus in the intestinal epithelial mucus increased (Supplementary Figure 1B). The results indicated that Aeromonas in the Bge group was increased compared to that in the Nge group, while they were decreased in the Pge group. Thus, the addition of the bacteriophage resulted in a change in Aeromonas abundance in the epithelial mucus. However, this change in Aeromonas abundance was not observed in the microbial community of the intestinal contents (Figure 10D). Our study found that the intestinal flora abundance of crucian carps was also regulated by intestinal cytokines (Supplementary Figure 1A). The first and second axes accounted for 54.28% and 21.49% of the total variations, respectively. TNF-α was positively correlated with IL-1β, while TGF-β and IFN-γ were positively correlated with IL-10. Moreover, Actinobacteriota was closely related to TGF-β and IFN-γ transcription, Firmicutes was closely related to IL-10 expression, and Proteobacteria and Bacteroidota were closely related to TNF-α and IL-1β, respectively.