Isolation and identification of K. pneumoniae from housefly larval intestines was achieved through the traditional isolation and culture method, the culturable aerobic K. pneumoniae KX was isolated from housefly larval intestines under an aerobic environment, and the culturable facultative anaerobic K. pneumoniae KY was isolated from housefly larval intestine under an anaerobic environment. The results showed that K. pneumoniae KY and K. pneumoniae KX clustered into different branches of the K. pneumoniae genus, indicating that they are two different strains of K. pneumoniae ( Figure 1 ).

The phages obtained from sewage were purified by a double-layer agar plate method and uniform plaque was produced on the double-layer plates ( Figures 2A–D ). Host specificity analysis revealed that phage KXP lysed only its susceptible bacteria K. pneumoniae KX and had no apparent impact on the other commensal bacteria isolated from the intestine of housefly larvae, including K. pneumoniae KY, Providencia sneebia DSM 19967, Pseudocitrobacter faecalis, Morganella morganii, Enterobacter hormaechei, Enterococcus casseliflavus, Providencia. stuartii and Enterococcus faecalis ( Table S1 ). It was observed by transmission electron microscopy that the head of phage KXP has a typical icosahedral structure and short tail, which is consistent with the characteristics of Podoviridae ( Figure 2B ). Phage KXP had almost no latent period and tended to become stable after a sharp increase in titer, and the burst size was 3.73 × 10¹³/3.80 × 10³ = 9.82 × 10⁹ particles/infected cell ( Figure 3B ). When the multiplicity of infection (MOI, the ratio of the number of phages to the number of host bacteria) was 10^–2, the phage KXP titer reached the highest level, approximately 12.55 log plaque forming unit (PFU) mL-1, indicating that the optimal multiplicity of infection (OMOI) (OMOI, the highest phage titration of the proportion of phage and its host) of phage KXP was 10^-2 ( Figure 3A ). The phage could withstand a pH 5-11 environment for 1 h ( Figure 3C ). In addition, it withstood a temperature from 4°C to 60°C for 1 h ( Figure 3D ). In a certain range of temperature and pH, the phage KXP shows good stability. Genome sequencing analysis revealed that the genome of KXP had a total length of 42,946 bp, and the average GC content was 54.2%, the number of protein-coding genes is 45, the gene length/genome ratio is 0.885 ( Figure S1A , Table S2 ).

Similarly, host specificity analysis revealed that phage KYP lysed only its susceptible bacteria K. pneumoniae KY and had no apparent impact on the other commensal bacteria isolated from the intestine of housefly larvae, including K. pneumoniae KX, P. sneebia DSM 19967, P. faecalis, M. morganii, E. hormaechei, E. casseliflavus, P. stuartii and E. faecalis ( Table S1 ). It was observed by transmission electron microscopy that the head of phage KYP had a typical icosahedral structure and a short tail, which was in line with the short-tailed phage family (Podoviridae) ( Figure 2D ). The latent period of phage KYP was approximately 20 min, and then the titer increased sharply and reached the highest value, with a burst size of 3.28 × 10¹²/4.08 × 10⁸ = 8.04 × 10³ particles/infected cell ( Figure 4B ). When the MOI was 10^–2, the phage KYP titer reached the highest level, approximately 12.61 log PFU mL-1, indicating that the OMOI of phage KYP was 10^-2 ( Figure 4A ). It withstood a pH 5-11 environment for 1 h ( Figure 4C ). In addition, it withstood temperatures from 4°C to 60°C for 1 h ( Figure 4D ). In summary, the stability of phage KYP is good. Genome sequencing analysis showed that the total length of the KYP genome was 43,195 bp, and the average GC content was 55.99%, the number of protein-coding genes is 48, the gene length/genome ratio is 0.919 ( Figure S1B , Table S3 ).

In order to analyze the effects of K. pneumoniae KX/KY and KX-KY mixture on larval development, feeding experiments were carried out. Based on our results, single feeding of K. pneumoniae KX (10⁹ colony forming unit (CFU)/mL), K. pneumoniae KY (10⁹ CFU/mL), and mixed feeding of K. pneumoniae KX (10⁹ CFU/mL) and K. pneumoniae KY (10⁹ CFU/mL) significantly promoted the growth and development of housefly larvae, but there was no synergistic effect on the growth and development of housefly larvae fed with mixed K. pneumoniae KX and KY. Feeding phage KXP (10⁷ PFU/mL), phage KYP (10⁷ PFU/mL), and mixed feeding of phages KXP (10⁷ PFU/mL) and KYP (10⁷ PFU/mL) significantly inhibited the growth and development of housefly larvae, but mixed feeding of phages KXP and KYP had no synergistic effect (F (24, 308) = 102.7, P<0.0001)( Figure 5 ). These results showed that compared with the control group, addition of K. pneumoniae KX/KY had a positive impact on the growth and development of housefly larvae, and the phage-mediated decrease in K. pneumoniae KX/KY had a negative impact on the growth and development of housefly larvae, thereby compromising larval health.

To study the effect of the addition or lack of bacteria on the composition of the bacterial community, the 16S rRNA gene of intestinal bacteria of housefly larvae in group MLB, MKX, MKY, MKXY, MKXP, MKYP and MKXYP was sequenced. MLB, MKX, MKY, MKXY, MKXP, MKYP and MKXYP were cultured in LB liquid medium, K. pneumoniae KX (10⁹ CFU/mL), K. pneumoniae KY (10⁹ CFU/mL), K. pneumoniae KX (10⁹ CFU/mL) and K. pneumoniae KY (10⁹ CFU/mL), K. pneumoniae phage KXP (10⁷ PFU/mL), K. pneumoniae phage KYP (10⁷ PFU/mL), K. pneumoniae phage KXP (10⁷ PFU/mL) and K. pneumoniae phage KYP (10⁷ PFU/mL) were fed to housefly larvae. A total of 2,235,406 high-quality bacterial sequences with sequence numbers from 100025 to 116160 were produced from all samples. These sequences were normalized and clustered into 3,768 OTUs at a 97% similarity level among all the samples ( Table 1 ). Chao1 indices showed that compared with the control group, the bacterial richness of housefly larvae in the phage-treated group (MKXP, MKYP and MKXYP) and bacteria-treated group (MKX, MKY and MKXY) decreased ( Figure 6A ). Shannon indices suggested that compared with the control group, the bacterial diversity of housefly larvae increased in the phage-treated group (MKXP, MKYP and MKXYP) but decreased in the bacteria-treated groups (MKX, MKY) ( Figure 6B ). PCoA analysis showed that there were significant differences in the intestinal microflora of housefly larvae between the phage treatment groups and the bacterial treatment groups ( Figure 6C ). The phage treatment groups (MKXP, MKYP and MKXYP) and bacterial treatment groups (MKX, MKY and MKXY) clustered together, and neither of them clustered with the control group (MLB). An unweighted pair-group method with arithmetic mean (UPGMA) tree further proved the aggregation of phage-treated group (MKXP, MKYP and MKXYP) and bacteria-treated group (MKX, MKY and MKXY) samples ( Figure 6D ). Therefore, the addition or lack of K. pneumoniae (KX, KY) significantly interfered with the larval intestinal flora.

The Venn diagram showed the common and unique OTUs of all samples, of which 248 OTUs (6.58%) were common to all samples ( Figure 6E ). There were 1319 (35.01%), 60 (1.59%), 56 (1.49%), 103 (2.73%), 82 (2.18%), 333 (8.84%) and 176 (4.67%) unique OTUs in the MLB, MKX, MKY, MKXY, MKXP, MKYP and MKXYP samples, respectively ( Figure 6E ). The Venn diagram showed the differences in the intestinal microflora of housefly larvae in different samples.

To further study the composition of the bacterial community in different groups, the composition and structure of the intestinal bacterial community were analyzed at different levels. Top 5 phyla and top 12 genera with relatively high abundance were selected for analysis. The results showed that the housefly larvae of different treatment groups had only slight differences at the phylum level, but had different bacterial community structures at the genus level. ( Figures 7A, B ). Proteobacteria accounted for the highest proportion of all intestinal bacteria in housefly larval samples ( Figure 7A ). In addition, in the bacterial treatment groups (MKX, MKY and MKXY) and phage treatment groups (MKXP, MKYP and MKXYP), the proportion of Proteobacteria increased, while Firmicutes and Bacteroidota decreased ( Figure 7A ). Among the 12 most abundant genera in the intestines of housefly larvae, the genus Klebsiella was significantly increased in abundance in the bacterial treatment group (MKX, MKY and MKXY), but the genus Providencia, Serratia, Morganella, and Vagococcus showed a significantly decreasing trend ( Figure 7B ). In the phage treatment group (MKXP, MKYP and MKXYP), the genus Klebsiella and Pseudomonas increased significantly in abundance, the genus Proteus increased slightly, and the genus Vagococcus decreased significantly ( Figure 7B ). The relative abundances of the genus Serratia and Enterobacter were significantly increased in the phage treatment group (MKYP, MKXYP) ( Figures 7B, C ). Therefore, the composition of the intestinal bacterial community of housefly larvae was greatly altered by the bacterial treatment group. Although phages have host specificity, they induce cascading effects on microbiota species that are not directly targeted. Similarly, the composition of the intestinal bacterial community of housefly larvae is greatly affected. Bacterial genera with great changes in abundance and proportion were investigated. In the group of larvae fed with K. pneumoniae KX/KY or with mixed KX and KY, the proportions of the genus Klebsiella and Bordetella were significantly increased while the abundance of the genus Providencia, Serratia, Morganella and Vagococcus decreased significantly. Significantly, the abundance of the genus Paenalcaligenes and Enterobacter showed the opposite trend in the two bacterial treatment groups (MKX, MKY). In the group of larvae fed with phage KXP/KYP or mixed KXP and KYP, the proportions of the genus Klebsiella, Pseudomonas, Proteus and Enterobacter were significantly increased, while the abundance of the genus Providence decreased slightly, and that of the genus Vagococcus decreased significantly. The proportion of the genus Serratia was significantly increased, while the abundance of the genus Paenalcaligenes was significantly decreased in the phage treatment group (MKYP, MKXYP). Similarly, the abundance of the genus Paenalcaligenes, Ignatzschineria, Serratia and Bordetella showed a reverse trend in the two phage treatment groups (MKXP, MKYP). Therefore, we speculate that in addition to the increase or phage-mediated decrease in K. pneumoniae (KX, KY), the change in the proportion of dominant bacteria, such as the genus Klebsiella, Providence, Serratia, Morganella, Bordetella, Pseudomonas, Vagococcus and Enterobacter in the intestinal tract of housefly larvae is another important factor affecting the health of houseflies. To determine the interaction between culturable bacteria and K. pneumoniae KX/KY, antagonism assays were carried out in aerobic cultures and facultative anaerobic cultures. The antagonism assay showed that the growth of K. pneumoniae KY, P. aeruginosa Y12 and P. stuartii Ps was inhibited by high abundances of K. pneumoniae KX, and the growth of K. pneumoniae KX, P. aeruginosa Y12 and P. stuartii Ps was also inhibited by high concentrations of K. pneumoniae KY ( Figures S2 , 3 ).

To analyse the bacterial correlations within the bacterial community, we constructed a network based on the Spearman correlation. Bacterial correlations among the larval intestinal flora were significantly changed after the larvae were treated with the two K. pneumoniae strains and K. pneumoniae-specific phages ( Figure 8 ). Compared to the control group, oral administration of K. pneumoniae and phages significantly reduced bacterial correlations, especially in Firmicutes (except MKXYP), Actinobacteriota (except MKXY) and Bacteroidota (except MKY) ( Figure 8 ). The total number of nodes, total links, average degree and average cluttering coefficient (except MKXY) of the interaction network in the gut microflora of the larvae decreased in the bacterial treatment group (MKX, MKY and MKXY) and phage treatment group (MKXP, MKYP and MKXYP) ( Table S1 ). We speculated that there was less contact between harmful bacteria in the bacterial treatment group and less contact between beneficial bacteria in the phage treatment group. However, compared with the bacterial treatment groups (MKX, MKY and MKXY), the phage-treated groups (MKXP, MKYP and MKXYP) had more total links and total nodes indicating that phage treatment increased the association between harmful bacteria ( Table S4 ). In all samples, a high degree of connectivity between Proteobacteria was observed ( Figure 8 ). However, compared with the bacterial treatment group (47.62% in MKX, 65.00% in MKY and 42.31% in MKXY), the interaction between Proteobacteria increased in the phage-treated groups (59.26% in MKXP, 60.71% in MKYP and 67.44% in MKXYP). We speculate that there is an increased association between some harmful bacteria in Proteobacteria in phage-treated groups (MKXP, MKYP and MKXYP). Based on the results, alteration of the loads of specific bacteria in the intestine could greatly affect the bacterial interactions among the microbial communities.

The dynamic analysis of K. pneumoniae KX/KY in housefly larvae under different oxygen concentrations revealed that phage KXP showed a higher titer in the environment under a high oxygen content while phage KYP revealed a higher titer in the environment under a low oxygen content. As phage KXP/KYP could rapidly amplify only in the presence of their host bacteria K. pneumoniae KX/KY, the titers of phages KXP/KYP revealed the result that a higher bacterial load of aerobic K. pneumoniae KX in the environment under a high oxygen content, while a higher bacterial load of anaerobic K. pneumoniae KY in the environment under a low oxygen content ( Figures 9A, B ). The results indicated that aerobic K. pneumoniae KX plays a major role in environments with a high oxygen content, while facultative anaerobic K. pneumoniae KY plays a major role in environments with a low oxygen content.

The midgut and hindgut maintain different oxygen concentration in housefly larvae. To analyze the dynamic alterations of K. pneumoniae KX/KY in the gut of housefly larvae with different oxygen concentrations, the bacterial load of K. pneumoniae KX/KY in the midgut and hindgut of housefly larvae was assayed. Based on the results, higher KXP titers was found in the midgut while higher KYP titers was found in hindgut indicating a higher bacterial load of aerobic K. pneumoniae KX in the high oxygen midgut while a higher bacterial load of anaerobic K. pneumoniae KY in the low oxygen hindgut. Our results revealed that aerobic K. pneumoniae KX plays a major role in the aerobic midgut, while facultative anaerobic K. pneumoniae KY plays a major role in the anaerobic hindgut ( Figures 9C, D ).

To further analyse the in vivo interactions between K. pneumoniae KX and KY in housefly larvae, the bacterial load of K. pneumoniae KX/KY in housefly larvae feeding K. pneumoniae KX/KY alone or mixed K. pneumoniae KX-KY was assayed through phage titer analysis. Phage titer analysis revealed that the phage KXP titer increased in the MKX and MKXY groups and decreased in the MKY group ( Figures 9E, G ). The phage KYP titer increased in groups MKY and MKXY, and the KYP titer decreased in group MKX ( Figures 9F, H ). The above experiments showed that the bacterial load of K. pneumoniae KX increased and that of KY decreased after the larvae were fed KX, while K. pneumoniae KY increased and KX decreased after the larvae were fed KY; both KX and KY increased after the larvae were fed mixed K. pneumoniae KX-KY. Therefore, we assume that K. pneumoniae KX and KY could maintain a dynamic balance in the housefly larvae and competitively facilitate larval development.

A housefly (Musca domestica) colony was reared and maintained in the Laboratory of Vector and Insect Diseases of Shandong First Medical University since 2005 (36).

Using traditional isolation and culture methods and anaerobic techniques (anaerobic airbag and anaerobic tank), K. pneumoniae KX was isolated from the intestine of 3-day-old housefly larvae under aerobic condition, and K. pneumoniae KY was isolated from the intestine of 3-day-old housefly larvae under facultative anaerobic condition (7). The isolated strains were sent to Qingdao sequencing Department of Beijing Ruibo Xingke Biotechnology Co., Ltd. for sequencing. The 16s rDNA was amplified with the primers 27F (AGAGTTTGATCCTGGCTCAG) and 1492R (TACGGCTACCTTGTTACGACTT). The sequence was blast in the GenBank database to search for the most similar sequences. The retrieved homologous sequences were analyzed with MEGA11.0 software to construct phylogenetic tree by ML bayesian method.

In this experiment, sewage from Taian treatment plant in Shandong Province was collected for phage isolation. K. pneumoniae KX isolated from the intestinal tract of housefly larvae under aerobic condition and K. pneumoniae KY isolated from facultative anaerobic state were used as host bacteria to screen phages. Phage KXP were purified by a double-layer agar plate method as reported in our previous research (36). Except that the culture state is facultative anaerobic, the process of phage KYP separation is the same as that of phage KXP. The high concentration phage stocks were stored at 4°C. The purified phage preparation was imaged under transmission electron microscope (H-7700, Hitachi, Japan).

In order to confirm the host specificity (47, 48) of phage KXP, the spot titer technique was used. Take 10 μL 10⁸ PFU/mL phage KXP and drop it on the double-layer agar plate overlays of each bacteria (without phage) including P. sneebia DSM 19967, P. faecalis, M. morganii,E. hormaechei, E. casseliflavus, P. stuartii, E. faecalis, K. pneumoniae KX, K. pneumoniae KY were isolated from the intestine of housefly larvae and plates were incubated overnight at 37°C aerobically depending on the bacterial culture conditions. Under the condition of co-culture with K. pneumoniae KX, the OMOI of phage KXP was determined (49). Under aerobic condition, K. pneumoniae KX was cultured in LB liquid medium (yeast extract 5.0 g liter^-1, tryptone 10.0 g liter^-1, NaCl 10.0 g liter^-1) to the logarithmic growth phase (OD 600 = 0.6). Phage KXP and host K. pneumoniae KX (10⁸ CFU/mL) were co-cultured at different MOIs ((10, 1, 10^-2, 10^-4, 10^-6)). After incubated at 37°C for 3.5h, the phage KXP titer of supernatant was determined. The group with the highest titer is the OMOI of the phage KXP. The one-step growth curve experiment was carried out using the method reported in our previous research (37). One hundred microliter of phage KXP was mixed with 900 µL of SM buffer in different pH values (4–12), then incubated in a water bathe at 37°C. Samples were taken out after 3.5 h, and the potency was measured. One milliliter of phage KXP was incubated in a water bath at 4°C, 25°C, 37°C, 50°C, 60°C and 70°C, respectively. Samples were taken out after 1h to measure the titers. The titer of phage was determined by double-layer agar plate method. All analyses were performed as previously described, with some modifications, and were performed in triplicate. Except that the culture state is facultative anaerobic, the biological determination method of phage KYP was the same as that of phage KXP.

Phage chromosomal DNA was isolated using the λ phage genomic DNA purification kit (ABigen) following the manufacturer’s instructions. Whole-genome sequencing was performed with an Illumina HiSeq 4000 platform. The METAVIRALSPADES pipeline (50) was used to identify the phage in the sample (mainly including sequence assembly, phage sequence identification, and phage integrity identification). The whole gene sequence was predicted by GeneMarkS software, the tRNA gene in the whole genome was predicted by tRNAscan-SE, the rRNA gene was predicted by Barrnap, and the prediction of other non-coding RNA was mainly obtained by comparing with Rfam database. The functional annotation of protein coding genes was compared with Swiss-Prot, NR, pfam, GO, KEGG and other databases by blastp. Annotated genome map was made using the criclize package in R. Phage sequences were deposited in the NCBI (KXP: ON755175, KYP: ON755176).

To study the effects of K. pneumoniae KX/KY on the growth and development of housefly larvae. K. pneumoniae KX/KY was inoculated in freshly prepared LB liquid medium and placed in a constant temperature shaker. The concentration of K. pneumoniae KX/KY was about 109 CFU/mL after shaking at 110 rpm at 37°C for 24 h, which could be used as a high concentration of K. pneumoniae. To study the effects of K. pneumoniae KX-KY mixture on the growth and development of housefly larvae, K. pneumoniae KX and KY were mixed at 1:1 at the same concentration of 10⁹ CFU/mL. Effects of K. pneumoniae phage KXP on the growth and development of housefiy larvae. The feeding experiment of phage KXP was designed as follows: polyethylene glycol (PEG) precipitation was used to enrich the phage particles, and the high concentration of phage KXP was about 10¹¹ PFU/mL. Phage KXP was diluted into 10⁷ PFU/mL with LB liquid medium. The feeding experiment on the effects of K. pneumoniae phage KYP on the growth and development of housefly larvae was the same as above. Housefly larvae from different groups that were fed LB liquid medium, K. pneumoniae KX (10⁹ CFU/mL), K. pneumoniae KY (10⁹ CFU/mL), K. pneumoniae KX (10⁹ CFU/mL) and K. pneumoniae KY (10⁹ CFU/mL), K. pneumoniae phage KXP (10⁷ PFU/mL), K. pneumoniae phage KYP (10⁷ PFU/mL), K. pneumoniae phage KXP (10⁷ PFU/mL) and K. pneumoniae phage KYP (10⁷ PFU/mL) were named MLB, MKX, MKY, MKXY, MKXP, MKYP and MKXYP respectively. The solutions of different treated groups were mixed with sterilized wheat bran at 1:1 ratio, respectively, and placed in a 5 mL perforated tube to ensure ventilation. The same amount of wheat bran (1.5 g~1.6 g) was placed in each tube, and then 10 normal-breeding, good-growing, uniform-sized 1-day-old larvae were added to each centrifuge tube. Each group is repeated three times and a piece of gauze is placed between the tube and the lid to prevent the larvae from escaping. They were placed in an artificial climate chest at 25 ± 1°C, a 50-55% relative humidity (RH), and a photoperiod of LD 12:12 h. Each group of experiments were carried out in three perforated test tubes, and four samples were taken from each tube every day as repetition. The body length and weight of housefly larvae were recorded regularly every day.

The surface debris of housefly larvae was removed and placed in a 1.5 mL centrifuge tube containing 75% alcohol, soaked and disinfected 10min, and then rinsed with sterile deionized water to remove bacteria adhering to the surface. After strict body surface disinfection, the housefly larva samples were stored in -80°C, and the samples of housefly larval on the day with the largest difference were selected for 16S rRNA high-throughput sequencing. Five larvae taken from different treatment groups and control groups were taken as sampling units, and there were 3 repeats in each group (38).

In order to determine the interaction between culturable bacteria and K. pneumoniae (KX and KY) and between K. pneumoniae KX and K. pneumoniae KY of housefly larvae, plate confrontation assay were carried out in nutrient agar (NA) medium plates (peptone 10.0 g liter^-1, agar 20 g liter^-1, NaCl 5.0 g liter^-1, beef extract 3.0 g liter^-1) under aerobic and facultative anaerobic culture conditions. K. pneumoniae KX, K. pneumoniae KY, P. aeruginosa Y12 and Providencia stuartii Ps were inoculated in LB liquid medium (yeast extract 5.0 g liter^-1, tryptone 10.0 g liter^-1, NaCl 10.0 g liter^-1) and cultured at 37°C overnight (OD600 >1.0). K. pneumoniae KX cultures was inoculated on half of the nutrient agar plate using the spread plate method with a sterile cotton swab, and the opposite side of the agar plate was used as a negative control. Two 6mm-diameter sterile filter papers were placed symmetrically on both sides of the agar medium. 10 µL of the isolated cultivable gut bacteria, including K. pneumoniae KY, P. aeruginosa Y12 and P. stuartii Ps were added to the filter papers. Similarly, K. pneumoniae KY cultures was inoculated on half of the nutrient agar plate using the spread plate method with a sterile cotton swab, and the opposite side of the agar plate was used as a negative control. Two 6mm-diameter sterile filter papers were placed symmetrically on both sides of the agar medium. 10 µL of the isolated cultivable gut bacteria, including K. pneumoniae KX, P. aeruginosa Y12 and P. stuartii Ps were added to the filter papers. All plates were cultured at 37°C and cultured in aerobic and anaerobic conditions for 48 h. Finally, the growth diameter of bacteria was recorded. The interaction between them was evaluated by measuring the colony size of K. pneumoniae KX, K. pneumoniae KY, P. aeruginosa Y12 and P. stuartii Ps. The experiments were conducted with six independent biological replications.

DNA extraction of the intestinal bacteria and determining changes in gut composition using Illumina MiSeq Sequencing were carried out using the same method as described in our previous research (36). MENAP is used to analyze the OTU data of each sample, and the edge and node data are obtained. The network was performed using Gephi. The parameters used in drawing mainly include: node size is degree (degree: the degree of connection, the more connections, the higher the degree), the node color is phylum classification, the edge color is positive and negative correlation, and the node layout is Fruchterman Reingold.

Based on the highly specificity of phages, phage KXP/KYP could rapidly amplify only in the presence of their host bacteria. Therefore, bacterial load of K. pneumoniae KX/KY in housefly could be analyzed by measuring the titers of their specific phages in vitro. In order to analyze the load of K. pneumoniae KX and K. pneumoniae KY in housefly larvae grown under different oxygen concentrations, phage titer analysis was carried out. In the normal oxygen content culture group, LB liquid medium was mixed with sterilized wheat bran at a 1:1 ratio while the ration of LB liquid medium was increased (1.5:1) to create the low oxygen environment. The same amount of wheat bran (1.5 g~1.6 g) was placed in each tube, and then 10 uniform-sized 1-day-old larvae with normal breeding behavior and good growth were added to each centrifuge tube. Each group was repeated three times, and a piece of gauze was placed between the tube and the lid to prevent the larvae from escaping. Larvae were placed in an artificial climate chest at 25 ± 1°C, 50-55% relative humidity (RH), and a photoperiod of LD 12:12 h. After the larvae were 3 days old, the larvae cultured in normal oxygen and low oxygen environment were collected and ground thoroughly to obtain a uniform grinding solution as the stock solution. The larvae cultured in the normal oxygen group were taken as an example. The larvae were removed from the 5 mL centrifuge tube and placed in a 1.5 mL centrifuge tube. The larvae were killed by being placed in the freezer at -20°C for a few minutes. Then, the larvae were removed, soaked in alcohol for 2 min, and rinsed with sterile water three times to thoroughly disinfect their body surface. PBS (1000 μL) was added to a 1.5 mL centrifuge tube, ground thoroughly with a grinding rod after high-pressure sterilization, and shaken with a shaker for 1 min to obtain a uniform grinding solution as the stock solution. Then, PBS buffer was used to dilute the uniform grinding solution to obtain a 10^-4 grinding solution. The treatment of larvae in the low oxygen content culture group was the same as above. Each group had 3 parallels. Then, 3 groups were established. Control group: 0.1 mL KXP (10⁷ PFU/mL) and 9.9 mL LB were mixed, and 0.1 mL KYP (10⁷ PFU/mL) and 9.9 mL LB were mixed. Normal oxygen content culture group: after mixing 1 mL grinding solution (10^–4) and 20 mL LB, the bacteria were shaken on a shaking table for 12 h to obtain bacterial solution X1. Then, 1 mL bacterial solution X1, 0.1 mL KXP (10⁷ PFU/mL) and 8.9 mL LB were mixed, and 1 mL bacterial solution X1, 0.1 mL KYP (10⁷ PFU/mL) and 8.9 mL LB were mixed. Low oxygen content culture group: The experimental process was the same as above. Each group was analyzed in triplicate. The mixed solutions of different treatment groups were placed on a shaking table for 2 h and then filtered to measure the titers of KXP and KYP in different groups to determine the load of K. pneumoniae KX and KY in the housefly larvae under the two culture conditions.

As the midgut and hindgut contain different oxygen conditions, phage titers analysis was incorporated to analyze the dynamic bacterial load of K. pneumoniae KX and K. pneumoniae KY in midgut and hindgut.

The normally reared housefly larvae were cultured to 3-day-old larvae, and the midgut and hindgut were taken with a scalpe in clean bench for further analysis. The specific operations are as follows. First, the 3-day-old housefly larvae were taken out and placed in a 1.5 mL centrifuge tube. The larvae were killed by placing them in the refrigerator at -20°C for a few minute. Then, the larvae were removed, soaked in alcohol for 2 min, and rinsed with sterile water three times to thoroughly disinfect their body surface. Then the housefly larvae were dissected and the midgut and hindgut were taken in clean bench.

Take 6 portions of midgut and put it in 1000 μL of PBS in a 1.5 mL centrifuge tube, ground thoroughly with a sterile grinding rod, and shaken with a shaker for 1 min to obtain a uniform grinding solution as the stock solution. Then, PBS buffer was used to dilute the uniform grinding solution to obtain a 10^-4 grinding solution. The treatment of hindgut group of housefly larvae was the same as above. Each group had 3 parallels. Then, 3 groups were established. Control group: 0.1 mL KXP (10⁵ PFU/mL) and 9.9 mL LB were mixed, and 0.1 mL KYP (10⁵ PFU/mL) and 9.9 mL LB were mixed. Midgut group: after mixing 1 mL of 10^-4 grinding solution and 20 mL LB, the bacteria were shaken on a shaking table for 12 h to obtain bacterial solution M1. Then, 1 mL bacterial solution M1, 0.1 mL KXP (10⁵ PFU/mL) and 8.9 mL LB were mixed, and 1 mL bacterial solution M1, 0.1 mL KYP (10⁵ PFU/mL) and 8.9 mL LB were mixed. Hindgut group: The experimental process was the same as above. Each group was analyzed in triplicate. The mixed solutions of different treated groups were placed on a shaking table for 4 h and then filtered to measure the titers of KXP and KYP in different groups to determine the concentrations of K. pneumoniae KX and KY in the midgut and hindgut of housefly larvae.

The 1-day-old larvae in the control group were fed LB medium (group MLB), while others were fed K. pneumoniae KX (group MKX, 10⁹ CFU/mL), K. pneumoniae KY (group MKY, 10⁹ CFU/mL) and K. pneumoniae KX-KY (group MKXY, 10⁹ CFU/mL) until they grew to 3-day-old larvae. Phages KXP and KYP were used to measure the concentrations of K. pneumoniae KX and K. pneumoniae KY in housefly larvae in the four groups. In the control group, LB liquid medium was mixed with sterilized wheat bran at a 1:1 ratio and placed in a 5 mL perforated tube to ensure ventilation. In the MKX group, K. pneumoniae KX (10⁹ CFU/mL) bacterial solution was mixed with sterilized wheat bran at a 1:1 ratio and placed in a 5 mL perforated tube to ensure ventilation. In the MKY group, K. pneumoniae KY (10⁹ CFU/mL) bacterial solution was mixed with sterilized wheat bran at a 1:1 ratio and placed in a 5 mL perforated tube to ensure ventilation. In the MKXY group, both K. pneumoniae KX (109 CFU/mL) and K. pneumoniae KY (109 CFU/mL) were mixed with sterilized wheat bran at a 1:1 ratio and placed in a 5 mL perforated tube to ensure ventilation. Each treatment was repeated three times, and a piece of gauze was placed between the tube and the lid to prevent the larvae from escaping. Larvae were placed in an artificial climate chest at 25 ± 1 °C, 50-55% relative humidity (RH), and a photoperiod of LD 12:12 h. After three days, the 3-day-old larvae were removed from the 5 mL centrifuge tube and placed in a 1.5 mL centrifuge tube. The larvae were killed by placing them in the refrigerator at -20°C for a few minutes. Then, the larvae were removed, soaked in alcohol for 2 min, and rinsed with sterile water three times to thoroughly disinfect their body surface. PBS (1000 μL) was added to a 1.5 mL centrifuge tube, ground thoroughly with a grinding rod after high-pressure sterilization, and shaken for 1 min to obtain a uniform grinding solution as the stock solution. The stock solution was 10^-4 diluted with PBS buffer. Then, 1 mL of the 10^-4 diluted solution was transferred to a sterile conical flask mixed with 20 mL of LB, and the conical flasks were cultured for 12 h to obtain bacterial solution C1. Then, 1 mL of bacterial solution C1, 0.1 mL of KXP (10⁵ PFU/mL) and 8.9 mL of LB were mixed, and 1 mL of bacterial solution C1, 0.1 mL of KYP (10⁵ PFU/mL) and 8.9 mL of LB were mixed. After shaking for 1 and 4 h, the cultures were filtered to measure the titers of KXP and KYP to determine the concentrations of K. pneumoniae KX/KY in housefly larvae in different groups. The experimental processes for the MKX, MKY and MKXY groups were the same as above. Each group was analysed in triplicate.