The type strain of K. pneumoniae ATCC 4352 was purchased from American Type Culture Collection. K. pneumoniae ATCC 700603 was provided by Juncai Luo (Tiandiren Biotech, Changsha, China). K. pneumoniae XDR clinical isolate strains of KPLUO and KPWANG were provided by Cha Chen (Traditional Chinese Medicine Hospital of Guangdong Province, Guangzhou, China). K. pneumoniae PDR clinical strain LH2020 was collected from the Third Xiangya Hospital of Central South University (Changsha, China). The K. pneumoniae clinical isolates of KPLUO, KPWANG and LH2020 were both obtained from the sputum of hospitalized patients with lung infections from Jan to Apr in 2020. K. pneumoniae strains were cultured in Luria-Bertani (LB) broth (Solarbio, Shanghai, China). All bacteria were grown at 37°C with shaking at 180–200 rpm. SPR741, E, CLA, and other drugs (including PE: Polymyxin E; PMBN: Polymyxin B nonapeptide) were purchased from MedChem Express (NJ, USA).

The minimal inhibitory concentrations (MICs) of antibiotics were determined using broth microdilution assay (Song et al., 2020), as recommended by the Clinical and Laboratory Standards Institute (CLSI, 2021). Briefly, bacteria were cultured to mid-log phase in LB medium at 37°C, 180 rpm, and then diluted to approximately 1×10⁶ CFU/mL. The tested antibiotics were serially diluted 2-fold in Mueller-Hinton (MH) broth II (cation regulation) and mixed with an equal volume of bacterial suspension in a 96-well plate (Corning Costar, USA). The MIC was defined as the lowest concentration without visible bacterial growth after incubation at 37°C for 16–18 h. To determine the minimum bactericidal concentration (MBC), 10 μL of bacterial culture was spotted on sheep blood agar plates (AutoBio, Zhengzhou, Henan, China). MBC is considered the minimum drug concentration without bacterial colony growth on the plate, after incubation at 37°C overnight.

An overnight culture of K. pneumoniae was adjusted to a McFarland standard equivalent of 0.5, and then spread onto a MH agar plate using a sterile cotton swab. The blank disks loaded with antibiotics were manually placed on the surface of agar and incubated at 37°C for 16-18 h. Then, the outcomes were determined according to diameter of inhibition zones (CLSI, 2021).

A microdilution checkerboard assay was used to assess the interactions of double or triple antibiotic combinations against K. pneumoniae isolates. For the double antibiotic combination, bacterial suspension at mid-log phase was diluted to 1×10⁶ CFU/mL with MH broth, 50 μL of 2-fold serially diluted drug added horizontally to the 96-well plate, and another 50 μL serially diluted drug with bacterial suspension was added vertically to the well. After incubation at 37°C for 16–18 h, the results were determined by measuring the optical density at 630 nm (OD₆₃₀) using a microplate spectrophotometer (Bio-Rad, USA), and the fractional inhibitory concentration index (FICI) was calculated using the formula: FICI = MIC_A (in combination)/MIC_A (singly) + MIC_B (in combination)/MIC_B (singly). FICI was judged as follows: FICI ≤ 0.5, synergism; 0.5<FICI ≤ 1, additive; 1<FICI ≤ 4, indifference; and >4, antagonism (Dupieux et al., 2017). For the triple antibiotic combination (SPR741+E+CLA), antibiotic E was added to the MH medium at the expected concentration, and the other two diluted drugs (SPR741 and CLA) were added separately to the 96-well plate, as described previously for the double antibiotic combination. OD₆₃₀ was detected after incubation at 37°C for 16–18 h.

A single colony of K. pneumoniae was inoculated in 10 mL LB medium and cultured overnight at 37°C, with shaking at 180 rpm. The bacterial cells were diluted to 1×10⁶ CFU/mL, and mono- and combinational therapy of SPR741, CLA, and E were added to 5 mL of suspension; 0.1% dimethyl sulfoxide (DMSO) was added as a control treatment. Samples were collected at 0, 2, 4, 6, 8, and 24 h, and viable counts were determined by means of plate counting (Liu et al., 2020).

The biofilm eradication assay was performed as described previously (Chatzimoschou et al., 2015), with minor modifications. Specifically, K. pneumoniae cultured overnight was diluted 1:100 with MH broth containing 2% glucose, following which 100 µL bacterial suspension was added to the 96-well microplates and incubated at 37°C for 48 h, to form mature biofilms. The unattached planktonic cells were removed by washing with 1× PBS (pH=7.4) twice, following which 100 µL of MH broth containing antimicrobials was added at the indicated concentrations. After incubation at 37°C for 24 h, the unattached cells were gently washed with 1× PBS, and the biofilm biomass was determined by measuring the OD₆₃₀.

The 48-h pre-formed biofilms were established as previously described. After treatment with the indicated concentrations of antibiotics alone or in combination for 24 h, the biofilms were washed with 1×PBS and dispersed by vigorous pipetting in 100 µL saline, to ensure separation from the wall of each well. The bacterial mixtures were subsequently transferred to a new 96-well microplate, continuously diluted 10-fold in saline, following which the viable bacterial cells were counted by means of plate counting (Tan et al., 2019).

CLSM was used to visually evaluate the effect of SPR741 combined with macrolides in eradicating pre-formed biofilms. The overnight culture of K. pneumoniae was diluted 1:100 with MH broth containing 2% glucose, following which 2 mL of K. pneumoniae cultures were subsequently added to a 6-well microplate (Corning Costar, USA), and the sterile glass sheet was placed into the wells. After incubation at 37°C for 48 h, the bacterial suspension was subjected to mono-, dual- and triple-therapy with SPR741, CLE, and E; 1% DMSO treatment served as a control. After incubation for 24 h, a mixture of SYTO9 and Propidium Iodide (PI) (Thermo Fisher Scientific, Shanghai, China) was used to dye the biofilm for 15 min in the dark. The biofilms were visualized using CLSM (Zeiss LSM800, Jena, Germany), with excitation and emission wavelengths of 485 nm/530 nm and 485/630 nm for SYTO9 and PI, respectively. (Liu et al., 2015).

Carbonyl cyanide 3-chlorophenylhydrazone (CCCP)-induced persister cells were cultured as previously described (Narayanaswamy et al., 2018), and overnight cultures of K. pneumoniae were incubated with 200 μg/mL CCCP at 37°C, with shaking at 180 rpm for 3 h, to induce persister cell formation. The bacteria were then centrifuged, washed twice with saline, and re-suspended in MH medium to obtain a final density of 1×10⁸ CFU/mL. Drugs were added at the indicated concentrations and the cells were incubated with them for 6 h, following which the surviving persister cells were determined by means of plate counting method.

All animal procedures were performed in accordance with the Ethics Committee of the Third Xiangya Hospital, Central South University (no: 2019sydw0233). Specific pathogen-free ICR female mice (Hunan, China), aged 8 weeks and weighing 24–26 g, were used. Mice (n = 6 -7 mice/group) were rendered transiently neutropenic by intraperitoneal injection of cyclophosphamide at a dose of 150 mg/kg and 100 mg/kg, at 4 d and 1 d before bacterial infection. To induce deep thigh infection, overnight culture of K. pneumoniae LH2020 was adjusted to a concentration of approximately 1×10⁶ CFU/mL, and the final inoculum concentrations were confirmed by means of serial dilution and plating techniques. Fifty microliters of the bacterial suspension was then injected intramuscularly into the right thigh muscle of the mice. Anti-microbial therapy was initiated 1 h after the infection. Treatments with 30 mg/kg SPR741, 40 mg/kg CLA, and 30 mg/kg E or their combinations were administered by means of subcutaneous injection into the neck (Wang et al., 2019). Saline with 1% DMSO was used as the negative control. Mice were euthanized after 24 h, and the thigh muscles were excised and homogenized with saline. The number of viable cells was counted using the plate counting method.

ICR mice weighing 24–26 g were randomly divided into two groups (n = 5 mice/group): the test group was administered a subcutaneous injection of 30 mg/kg SPR741 combined with 40 mg/kg CLA and 30 mg/kg E. Normal saline with 1% DMSO was included as a vehicle control. After 24 h of administration, plasma and serum samples were collected for whole blood cell analysis and alanine aminotransferase (ALT), blood urea nitrogen (BUN), and creatine kinase (CK) levels were measured in them. The mice were euthanized, and the liver, kidney, spleen, heart, and lungs were excised and preserved in 4% paraformaldehyde (Servicebio, Wuhan, China), sectioned, and stained with hematoxylin-eosin for histopathological analysis, and the results were confirmed by a pathologist.

All experiments were performed at least three independent experiments. GraphPad Prism software 8.0 was used for statistical analysis, and all data were expressed as means ± S.D. Unless otherwise noted, P-values were calculated using Student’s t-test for two-group comparisons, while multiple comparisons was analysed using one-way ANOVA followed by Tukey tests or Dunn’s multiple comparison test. P<0.05 was considered statistically significant.*: p<0.05; **: p<0.01; ***: p<0.001; ****: p<0.0001.

The antibiotic resistance pattern of K. pneumoniae LH2020, KPLUO and KPWANG clinical isolates were confirmed by K-B test and indicated LH2020 was PDR K. pneumoniae, KPWANG and KPLUO were XDR K. pneumoniae ( Table 1 ). We attempted to identify polypeptide antibiotics [including PE, PB, PMBN, and SPR741] that had a synergistic combination with macrolide antibiotics against PDR K. pneumoniae LH2020. The results are shown in Table 2 and Figure S1 , only SPR741 showed a synergistic combination with macrolide antibiotics. Although SPR741 and macrolide antibiotics lack of intrinsic antibacterial activity against K. pneumoniae, SPR741 could sensitized macrolide antibiotics to K. pneumoniae, and showed effective synergistic anti-microbial effects with E or CLA ( Figure 1A ), while other polypeptide antibiotics such as PE, PB, and PMBN only showed additive or indifferent effects ( Figure S1 ). Therefore, we chose SPR741 as a candidate antibiotic combination. Interestingly, E combined with CLA did not exhibit synergistic activity ( Figure 1B ), whereas the triple antibiotic combination of SPR741 with CLA and E showed the most potent synergistic effect ( Figure 1C ). To exclude the effect of the solvent DMSO, we added DMSO in checkerboard assay and found that DMSO did not affect the OD₆₃₀ value (P>0.05), indicating that the presence of DMSO did not affect the results of the combination ( Figure S2 ). To better compare the synergetic antibacterial difference between the dual- and triple-combinations, we set the MIC value >128 µg/mL for SPR741 in the checkerboard assay to equal 128 µg/mL to calculate the FIC values. As shown in Table S1 , when 8, 16 and 32 µg/mL of E were added, the FIC values for the triple antibiotic combinations were 0.187, 0.125 and 0.125, respectively, which was lower than that of the dual-combinations, with FIC values of 0.375 (SPR741+CLA) and 0.266 (SPR741+E), indicating that the triple-drug combination may have better antibacterial efficacy than the dual-combinations.

As efflux pump inhibitors have been reported to reduce antibiotic resistance (Lowrence et al., 2019), a checkerboard dilution assay was also conducted to study the triple anti-microbial combinational effects of SPR741+ macrolide (E or CLA) + efflux pump inhibitors (PAβN/STE) against PDR K. pneumoniae LH2020, to identify the most active combination in vitro. We found that although the double drug combination of SPR741 with E/CLA had a synergistic effect, the combined anti-microbial effect was not enhanced when PAβN or STE was added ( Figure 1D ), besides, synergistic effect of SPR741+C or SPR741+E diminished after the addition of PAβN, suggesting that there may be an antagonistic effect. Hence, in the following study, we selected the anti-microbial agents SPR741, E, and CLA, to investigate their double or triple synergistic anti-bacterial effect.

SPR741 also showed a synergistic anti-bacterial effect when combined with E or CLA against XDR K. pneumoniae KPLUO, KPWANG, and type strain ATCC 700603 ( Figure 2A ). Although the combination of E and CLA showed no interaction ( Figure 2B ), the anti-bacterial activity was significantly improved by the triple antibiotic combination of SPR741, E, and CLA ( Figure 2C ). Compared with any pair of double antibiotic combinations, the triple antibiotic combination exhibited the most significant anti-microbial effect. As seen in Table S2 , for the clinical isolate KPLUO, the FICI values for SPR741 in combination with CLA or E were 0.187 and 0.156, respectively, and when 8 µg/mL E was added, a lower FIC value of 0.0468 was observed. Similarly, the FIC values for the triple-antibiotic combination groups were lower than that of the dual-antibiotic combination for K. pneumoniae KPWANG and ATCC 700603, demonstrating that E enhanced the antibacterial activity of the SPR741 + CLA, which indicates the synergistic antimicrobial activity of the triple-antibiotic combination. Besides, we have selected four multi-drug resistant clinical strains of K. pneumoniae to determine whether the synergistic antibacterial effects were strain-dependent ( Figure S3 ), interestingly, synergistic antimicrobial activity was still observed with triple combination therapy, it demonstrates triple combinations may achieve better antimicrobial efficacy than double-drug combinations.

To verify the combined effects of SPR741 with E and CLA, we performed a time-killing assay. The results showed that the triple antibiotic combination group showed the most effective bactericidal activity against XDR and PDR K. pneumoniae compared with mono- or double- antimicrobials treated groups. For the type strain ATCC 700603, the triple combination of SPR741 (7 µg/mL), CLA (5 µg/mL) and E (7 µg/mL) reduced the bacterial count by 2.88 Log10 CFU/mL at 6 h, compared to the control group ( Figure 3A ). For XDR strains KPLUO and KPWANG, the single or double combination of SPR741 (2 µg/mL), CLA (2 µg/mL) and E (4 µg/mL) showed no bactericidal activity, but the triple drug group reduced the number of bacteria from over 8 to 2.34 Log10 CFU/mL (KPLUO), and over 9 to 6.91 Log10 CFU/mL (KPWANG), respectively, after treatment for 6 h ( Figure 3B ). Similarly, for the PDR strain LH2020, the triple combination groups among SPR741 (8 µg/mL), CLA (8 µg/mL) and E (16 µg/mL) could also reduce the number of bacteria from over 8.81 to 7.18 Log10 CFU/mL at the time point of 6h ( Figure 3C ).

Based on the significant synergistic effect of the triple combination on the planktonic bacteria of K. pneumoniae, we proceeded to ask whether this combination was effective against highly resistant phenotype of persister cells. We therefore investigated the synergistic bactericidal activity against CCCP-induced persisters by means of colony counting. The persister cell-killing assay showed that, in contrast to the control or single used group, the double or triple combination based on SPR741, CLA and E showed the greatest bactericidal effect against persister cells of type strain ATCC 700603 ( Figure 4A ), XDR ( Figure 4B ), and PDR ( Figure 4C ) strains. For example, in case of the ATCC 700603 ( Figure 4A ), whether it was the control or single drug group, the number of persister cells was between 7.6 to 8.6 Log10 CFU/mL, while the combination of 8 µg/mL SPR741 and 16 µg/mL CLA could significantly reduce the number of persisters to 5.5 Log10 CFU/mL. In case of the XDR strain of K. pneumoniae KPWANG and KPLUO, as compared to the control, the combination of 16 µg/mL SPR741 and 32 µg/mL CLA could reduce the number of persisters by 1.1 and 1.08 Log10 CFU/mL, respectively ( Figure 4B ), while the addition of 32 µg/mL E reduced the persisters by 1.41 and 0.58 Log10 CFU/mL. As for the PDR strain K. pneumoniae LH2020, the combination of 16 µg/mL SPR741 and 32 µg/mL CLA could reduce the number of persisters by 0.74 Log10 CFU/mL, when 32 µg/mL of E was added, the bactericidal potency against persister cells was further enhanced, as confirmed by reduced the bacterial count by 1.14 Log10 CFU/mL ( Figure 4C ). Although the triple combination did not significantly enhance the persisters-killing activity against the type strain ATCC 700603 and the XDR strain KPLUO, the double combination of SPR741 and CLA was effective enough in killing persister cells.

K. pneumoniae ATCC 4352 and KPLUO were chosen for biofilm eradication and viable bacteria counting assays due to their strong biofilm-forming ability, while K. pneumoniae KPWANG, LH2020 and ATCC 700603 had a weaker biofilm forming ability. When 8 µg/mL SPR741 was combined with 16 µg/mL CLA and 16 μg/mL E, the combination demonstrated the most effective biofilm-eradicating effect against the 48-h pre-formed biofilms of K. pneumoniae ( Figure 5A ). The results of live bacterial counting ( Figure 5B ) showed that, in contrast to the control, the number of live bacteria in the biofilms was significantly reduced after treatment with triple combinations of SPR741, CLA, and E. For example, the combination of 8 μg/mL SPR741 with 16 μg/mL E showed a moderate effect against ATCC 4352 biofilms among the single and dual combinations, resulting in a 3.33 Log10 CFU/mL reduction in bacterial numbers, when 16 µg/mL CLA was added, presented the best killing effect, reducing the number of bacteria by 4.21 Log10 CFU/mL. For the KPLUO strain, the combination of 8 µg/mL SPR741 with 16 µg/mL CLA and 16 µg/mL E presented the best bactericidal activity, reducing the number of bacteria in the biofilm by 3.0 Log10 CFU/mL. Similarly, the CLSM images also showed that SPR741 combined with CLA and E exerted significant biofilm eradication activities, as compared to the control or single and dual combinations ( Figure 5C ), with the green fluorescence signal intensity of viable bacteria markedly decreased and the density of biofilm significantly weakened.

Before in vivo efficacy testing, the in vivo toxicity was evaluated by subcutaneous administration of 30 mg/kg SPR741, 30 mg/kg E, and 40 mg/kg CLA into ICR mice. In the routine biochemical examination, there was no statistical difference among the various parameters including ALT, BUN, and CK (P>0.05) ( Figure 6A ). The blood cell analysis results showed that there was no statistical difference among the various parameters, including white blood cell count, red blood cell count, hemoglobin quantification, platelet count, neutrophil classification (N%), and lymphocyte classification (L%) ( Figure 6B ). Furthermore, histopathological analysis showed that compared with the control group, the triple antibiotic combination group did not cause significant injury to the tissues, and no granulocyte infiltration, hemorrhage phenomenon, or morphological changes were observed ( Figure 6C ). These results suggested that no obvious toxicity was found after treatment with SPR, CLA, and E, indicating that these triple antibiotics are relatively safe to be used together.

A neutropenic murine thigh infection model was used to evaluate therapeutic efficacy in vivo. Results showed that the triple combination of 30mg/kg SPR741, 30 mg/kg E, and 40 mg/kg CLA had the most significant anti-microbial effect, as demonstrated by the reduced bacterial load of 4.52 Log10 CFU/thigh compared to vehicle group, which was below the detection limit ( Figure 6D ). Even lowering the administration doses of CLA and E in the triple-drug combination to 30 mg/kg and 20 mg/kg, respectively, it is still demonstrated excellent in vivo efficacy, resulting in a reduction in bacterial load of 1.83 Log10 CFU/thigh. It was better than the double-drug combination of 30 mg/kg SPR741 + 40 mg/kg CLA, which reduced the bacterial count by 1.58 Log10 CFU/thigh. Moreover, the anti-bacterial efficacy of the triple combination of 30 mg/kg SPR741 + 30mg/kg E + 40mg/kg CLA was more effective than those of the 30 mg/kg SPR741 + 30 mg/kg E + 30 mg/kg E and 30mg/kg SPR741 + 40mg/kg CLA + 40mg/kg CLA combinations. This indicates significant synergistic anti-microbial efficacy of SPR741+E+CLA, rather than additional effects of SPR741 and macrolide antibiotics in vivo. It could be seen that only when SPR741 was used in combination with CLA and E (triple combination) could it demonstrate remarkable bactericidal activity in vivo.