IFA effectively decreased the production of K. pneumoniae capsule

The indispensable role of capsule in the pathogenicity of K. pneumoniae renders it an attractive drug target, yet no effective capsule inhibitor has been identified thus far. Herein, we screened potential anti-CPS inhibitors using a uronic acid assay and identified the natural compound IFA as the most promising one (Figs 1A and S1), which dose-dependently reduced the amount of CPS in the hypervirulent K. pneumoniae strain K7 (Fig 1B), but did not exhibit any visible inhibitory effect on bacterial growth and had an MIC exceeding 1024 μg/ml (Fig 1C). Consistently, SDS–PAGE analysis confirmed that the amount of polysaccharide was significantly decreased by IFA treatment (Fig 1D). To profile the changes in capsule architecture, we observed capsule transmission using electron microscopy (TEM). As shown in Fig 1E, wild-type K7 K. pneumoniae was encased in a thick and filamentous capsule, while a dramatic reduction in capsule thickness was observed in cells treated with IFA (Fig 1F), which exhibited greater similarity to the capsule-deficient mutant with deletion of the GT-1 glycosyltransferase (ΔGT-1 K7), with barely visible capsule boundaries around the cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. IFA effectively reduced the production of K. pneumoniae capsule.

(A) Chemical structure of IFA (IFA). (B) Uronic acid assay to confirm the capsule production of K. pneumoniae K7 with the treatment of indicated concentrations of IFA. (C) Growth curves of K. pneumoniae K7 in the presence of the indicated concentrations of IFA. (D) Alcian blue staining of capsule samples extracted from K. pneumoniae K7 with or without IFA treatment. (E) Transmission electron microscopy images of wild-type K. pneumoniae K7 and ΔGT-1 K7 in the presence of DMSO or 32 μg/ml IFA. The black arrows indicate capsule, and its thickness was measured using ImageJ (F). (G) Comparison of cell-attached and unattached capsule content of K. pneumoniae K7 with or without IFA treatment via a uronic acid assay. (H) Capsule production of different K. pneumoniae strains with or without IFA treatment. The data are presented as the means ± SEMs. One-way ANOVA and Tukey’s posttest was carried out to determine the statistical significance of different groups. *P < 0.05, **P < 0.01 and ns, no significance compared with the DMSO control group.

https://doi.org/10.1371/journal.ppat.1012787.g001

To explore whether IFA affects the process of capsule synthesis or retention, we measured cell-attached and unattached CPS, respectively. As with the changes in total CPS (Fig 1B and 1D), both the amounts of capsule bound to the cell and free in the supernatant were dose-dependently suppressed by IFA, indicating that IFA might inhibit capsule synthesis but not the retention process (Fig 1G). To test whether the inhibitory effect of IFA on capsule synthesis was serotype specific, we further performed a uronic acid assay on other strains of different serotypes, including K1, K2, K57 and K61. Promisingly, the amounts of CPS in all the tested strains were dose-dependently reduced by IFA (Fig 1H), suggesting that IFA is a potent broad-spectrum inhibitor of K. pneumoniae capsule. In conclusion, these results confirmed that IFA is an effective and serotype-independent capsule inhibitor that significantly reduced capsule thickness by inhibiting CPS production.

IFA impaired hypercapsule-associated mucoviscosity phenotypes of K. pneumoniae via capsule inhibition

The hypermucoviscosity phenotype (HMV) is a typical feature of hvKP characterized by poor recovery and sedimentation when centrifuged, and this phenotype has been shown to play a distinct role in K. pneumoniae pathogenesis. Although several recent studies have demonstrated discordant changes in capsule production and HMV and also found some new potential regulatory mechanisms [21,22], a genome-wide genetic screening revealed that capsule biosynthesis and HMV were coordinated in most cases [23]. To explore whether IFA also affected the HMV phenotype of K. pneumoniae, we further performed sedimentation assay. Consistent with the changes in capsule production, centrifugation resistance, referred to here as an indicator of excessive mucoidity, was also significantly reduced by increasing dose of IFA (Fig 2A), and this suppressive effect on HMV was also confirmed in different serotypes (Fig 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. IFA abrogated capsule-associated hypermucoviscosity phenotypes of hypervirulent K. pneumoniae.

(A) Hypermucoidy assay of hypervirulent K. pneumoniae K7 treated with DMSO or indicated concentrations of IFA. Bacterial cultures were centrifuged at 1,000 × g in a fixed angle rotor for 5 min to measure the OD₆₀₀ of the supernatants, which was defined relative to the initial OD₆₀₀. (B) Hypermucoidy analysis of different serotypes of K. pneumoniae. (C) Representative colony phenotypes of K. pneumoniae K7 and ΔGT-1 K7 on blood agar plates containing DMSO or 32 μg/ml IFA; the “string test” was performed at the same time. The yellow arrow indicates the string stretched by the tips. (D) The colony morphologies of K. pneumoniae K7 and ΔGT-1 K7 on LB agar plates containing DMSO or 32 μg/ml IFA are shown. (E) Scanning electron microscopy images of wild-type K. pneumoniae K7 and ΔGT-1 K7 in the presence of DMSO or 32 μg/ml IFA. The data are presented as the means ± SEMs. One-way ANOVA and Tukey’s posttest was performed to determine the statistical significance of different groups. **P < 0.01 and ns, no significance compared with the DMSO control group.

https://doi.org/10.1371/journal.ppat.1012787.g002

Building upon the above findings, we next performed a string test to further confirm the inhibitory effect of IFA on HMV. Consistent with previous reports, wild-type K7 formed obviously more large and mucoid colonies on blood agar plates, and these colonies could be easily stretched to an average length of more than 5 mm, which is commonly regarded as a criterion for HMV (Fig 2C). In addition, the wild-type K7 colonies were large, protuberant and more mucoid on the LB agar plates, whereas the colonies in IFA group were smaller, flat, grayish and translucent, similarly to those in the ΔGT-1 K7 group (Fig 2D). Interestingly, the surface of the wild-type K7 bacteria was covered by a mucoid matrix, thus, the cell boundary was not clear according to SEM (Fig 2E). In contrast, the morphology and structure of the bacteria were not impacted by 32 μg/ml IFA, but the mucoid matrix masking the bacterial surface was prominently eliminated, with a clear boundary, as in ΔGT-1 K7.

Given that HMV not only depends on hypercapsule production but also requires the small protein regulator RmpD, we constructed a rmpD-deficient mutant to exclude the possibility that IFA suppresses HMV via acting on rmpD rather than capsule. Consistent with the findings of previous study, ΔrmpD K7 K. pneumoniae completely lost HMV phenotype (S2A–S2C Fig). Interestingly, IFA treatment still reduced uronic acid level of ΔrmpD K7 in a dose-dependent manner, suggesting that IFA suppressed HMV by affecting capsule (S2D Fig). Together, these results elucidated that IFA treatment effectively impaired HMV via reducing CPS production.

IFA reduced CPS biosynthesis through altering bacterial metabolism

To explore how IFA reduced the thickness of K. pneumoniae capsule and capsule-associated HMV, we further analyzed the transcriptome of wild-type K7 K. pneumoniae after different treatments. Accordingly, 295 differentially expressed genes (DEGs) were identified between DMSO- and IFA (32 μg/ml)-treated bacteria, comprising 78 upregulated and 217 downregulated genes in the presence of IFA (Fig 3A). The heat cluster of these DEGs further revealed significant transcriptome alterations caused by IFA (Fig 3B). Gene Ontology (GO) enrichment analysis of these DEGs revealed the top 20 GO terms, including the metabolic and catabolic processes of organic substances and subsequent oxidation–reduction processes (Fig 3C), among which the oxidation–reduction process was the most affected, with 8 upregulated and 23 downregulated genes (adjusted p = 5.73E-05). Cellular metabolism and oxidation–reduction processes are closely related to energy status, and reportedly, high intracellular ATP level is a limiting factor for CPS production and HMV [24]. Hence, we further evaluated the intracellular ATP levels of log-phase K. pneumoniae. As expected, the ATP level of K. pneumoniae was predominantly increased by IFA in a dose-dependent manner (Fig 3D), and in accordance with this result, the ratio of NAD⁺/NADH was also greater in IFA-treated bacteria (Fig 3E). These results further again uncovered the internal integration between CPS biosynthesis and cellular metabolism and energy status.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. IFA reduced the capsular biosynthesis through altering bacterial metabolism.

(A) Volcano plot analysis of differentially expressed genes (DEGs) between DMSO- or 32 μg/ml IFA-treated K. pneumoniae K7. A total of 295 differentially expressed genes (DEGs), including 78 upregulated and 217 downregulated genes, were identified in the IFA-treated group. (B) Heat cluster of DEGs. (C) The top 20 enriched Gene Ontology (GO) terms of all DEGs compared to DMSO control. (D) Intracellular ATP levels in K. pneumoniae K7 cells treated with DMSO or indicated concentrations of IFA. (E) The ratio of NAD⁺/NADH in K. pneumoniae K7 treated with DMSO or indicated concentrations of IFA. (F) The transcriptional levels of capsule synthesis genes and rmp genes. The data are presented as the means ± SEMs. One-way ANOVA and Tukey’s posttest was performed to determine the statistical significance of different groups.*P < 0.05, **P < 0.01 and ns, no significance compared with DMSO control group.

https://doi.org/10.1371/journal.ppat.1012787.g003

Next, we detected the transcriptional levels of core capsule synthesis genes and rmp genes. The qRT-PCR data demonstrated that the transcriptional levels of gnd, wcaJ, wbaP, wzx, wza, wzb, rcsB and rfaH were dose-dependently reduced by IFA, whereas the levels of rmpA, rmpC and rmpD were not affected (Fig 3F), indicating that IFA down-regulated capsule synthesis genes but not HMV-related genes. Taken together, these results suggested that IFA reduced CPS biosynthesis and HMV through suppressing the transcriptional levels of capsule cluster genes, which could be attributed to the alteration in bacterial metabolism and energy status.

IFA sensitized K. pneumoniae to serum killing via suppressing capsule

The complement system present in serum and tissue fluids is the first line of host defense against bacterial invasion. Nevertheless, several gram-negative bacteria have evolved multiple strategies to resist complement-mediated killing through interfering with the activation of the complement cascade or degrading already activated complement molecules [25], which contributes to the survival, multiplication and dissemination of bacteria. Capsule is a critical determinant for K. pneumoniae survival in serum, conferring strong resistance to the binding of C3b and C5b-9 to the cell surface [26]. The remarkable suppression of IFA on capsule synthesis prompted us to further investigate whether IFA sensitized K. pneumoniae to serum killing. Consistently, the viability of the ΔGT-1 K7 strain, which is deficient in capsule production, was significantly reduced in 20% normal human serum (NHS) within 60 min and was eliminated at 120 min, while the serum killing of wild-type K7 was obviously delayed compared to the capsule-deficient mutant and a sharp decrease was observed until 120 min (Fig 4A). As expected, IFA treatment dose-dependently decreased the resistance of wild-type K7 to serum killing. Notably, IFA (64 μg/ml) did not have an obvious effect on the growth of K. pneumoniae either in LB medium or heat-inactivated NHS (56°C, 30 min) (Fig 4B), again supporting its minimal antimicrobial activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. IFA sensitized hypervirulent K. pneumoniae to complement-mediated killing by inhibiting capsule.

(A) Serum-killing assay of K. pneumoniae K7 and ΔGT-1 K7 in the presence of DMSO or indicated concentrations of IFA. Bacteria were incubated with 20% NHS, and the viable count was determined by serial dilution and microbiological plating at the indicated time points. (B) Bacterial viability of K. pneumoniae K7 in heat-inactivated NHS (HNHS) and LB medium supplemented with DMSO or 64 μg/ml IFA. (C) Immunofluorescence microscopy analysis of C3b/C3bi deposition on the surface of 1-hour and 3-hour serum-exposed K. pneumoniae K7 and ΔGT-1 K7 with the treatment of DMSO or 32 μg/ml IFA. C3b/C3bi was stained with APC-conjugated anti-C3b/C3bi antibody. (D) Immunofluorescence microscopy analysis of C5b-9 formation on the surface of 1-hour- and 3-hour serum-exposed K. pneumoniae K7 and ΔGT-1 K7 with the treatment of DMSO or 32 μg/ml IFA. The C5b-9 complex was stained with a mouse anti-C5b-9 antibody and Alexa Fluor 488-conjugated goat anti-mouse IgG. The fluorescence images (top) and transillumination images (bottom) were normalized within each panel. (E) Flow cytometry-based determination of C3b/C3bi binding to K. pneumoniae K7 and ΔGT-1 K7 with the treatment of DMSO or 32 μg/ml IFA was performed after a 3-hour incubation in 20% NHS at 37°C. (F) The data were analyzed using unpaired two-tailed Student’s t-test. (n = 3). (G) Flow cytometry-based determination of C5b-9 formation on K. pneumoniae K7 and ΔGT-1 K7 with the treatment of DMSO or 32 μg/ml IFA was performed after 3 hours of incubation in 20% NHS at 37°C. (H) The data were analyzed using unpaired two-tailed Student’s t-test (n = 3). The means ± SEMs are shown. *P < 0.05, **P < 0.01 and ns, no significance.

https://doi.org/10.1371/journal.ppat.1012787.g004

All three known complement pathways converge at C3 cleavage, with eliciting C3b generation and covalent binding to the target bacterial surface, followed by the assembly of C5 convertases and the appearance of a C5b cleavage product that subsequently forms a C5b-9 membrane attack complex capable of lysing cells [27]. Hence, we examined the deposition of the C3b/C3bi and C5b-9 complexes by immunofluorescence microscopy analysis. Consistent with the serum killing results, serum exposure of ΔGT-1 K7 exhibited considerable binding of C3b/C3bi and C5b-9 within the first hour, while the wild-type K7 strain displayed little to no binding at this time point and only slightly increased over time (Fig 4C and 4D). Promisingly, treatment with 32 μg/ml IFA led to a significant increase in the amount of complement binding to wild-type K7 but not to ΔGT-1 K7 (Figs 4C and 4D and S3). Similarly, these observations were further validated by flow cytometry, which yielded the same results (Fig 4E–4H). Collectively, these results demonstrated that IFA increased the recruitment of complement molecules to the K. pneumoniae surface and thus sensitized bacteria to complement-mediated killing by inhibiting CPS.

IFA impeded capsule-mediated immune evasion of K. pneumoniae

Capsule- and HMV-mediated blockade of adherence and phagocytosis is one of the most important pathogenesis mechanisms of hvKP [22]. To determine whether IFA disarms this in vitro virulence phenotype, we performed an adherence assay using human A549 alveolar epithelial cells and performed phagocytosis assays with mouse J774 macrophages and mouse primary peritoneal macrophages (MPMs), respectively. As shown in Fig 5A, more ΔGT-1 K7 mutant attached to host cells compared with wild-type K7 strain, while IFA treatment significantly increased the adherence of wild-type K7 but not ΔGT-1 K7. In addition, fewer than 10% of the wild-type K7 were phagocytized by J774 cells at 60 min post infection, whereas about 30% ΔGT-1 K7 were intracellular at the same time (Fig 5B), confirming the effective blockade of phagocytosis by capsule. Surprisingly, IFA treatment increased the phagocytosis ratio to a level approaching that of the acapsular mutant, indicating the effective abrogation of capsule and its related HMV. Interestingly, a significant difference between wild-type K7 and ΔGT-1 K7 appeared at 30 min post infection in MPMs cells, and IFA treatment also caused stark increase in wild-type K7 (Fig 5C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. IFA disarmed capsule-mediated immune evasion.

(A) Adhesion assay of K. pneumoniae K7 and ΔGT-1 K7 to A549 cells in the presence of DMSO or 32 μg/ml IFA. Cells were infected with K. pneumoniae K7 or ΔGT-1 K7 at an MOI of 50 for 2 hours and then washed, lysed and plated on LB agar plates to quantify the number of colony-forming units. Data are presented as a percentage of the initial inoculum CFU. (B) Phagocytosis of K. pneumoniae K7 and ΔGT-1 K7 by J774 macrophages in the presence of DMSO or 32 μg/ml IFA. The cells were infected at an MOI of 5 for 2 hours and then washed, followed by a further incubation in medium containing gentamicin (100 μg/ml) to kill the extracellular bacteria. The cells were then rinsed, lysed and plated on LB agar plates after serial dilution. (C) Phagocytosis of K. pneumoniae K7 and ΔGT-1 K7 by mouse primary peritoneal macrophages (MPMs) in the presence of DMSO or 32 μg/ml IFA was also analyzed as described above. The data are presented as the means ± SEMs. Unpaired two-tailed Student’s t-test was performed to determine the statistical significance of two groups. *P < 0.05, **P < 0.01 and ns, no significance.

https://doi.org/10.1371/journal.ppat.1012787.g005

Although the earlier study raised that HMV might be the main factor blocking host cell adherence during hvKP infection [22], our results demonstrated IFA still dose-dependently increased the adherence level of ΔrmpD K7 (S2E Fig). Of note, the deletion of rmpD indeed resulted in a significant increase in the amount of attached bacterial versus wild-type K7 (S2E Fig). Similarly, more ΔrmpD K7 were engulfed by J774 A.1 macrophage, and the phagocytosis ratio was also significantly increased by IFA (S2F Fig), indicating that IFA-mediated enhancement of adherence and phagocytosis might mainly depend on capsule rather than HMV, which requires both capsule and rmpD.

Collectively, these in vitro results confirmed that IFA disarms capsule-mediated prevention of adherence and phagocytosis, which would significantly impede the immune evasion of hvKP.

In vivo IFA therapy rescued G. mellonella from K. pneumoniae infection

Based on the above in vitro virulence-associated analysis, we then performed a G. mellonella killing assay to evaluate the in vivo effect of IFA. G. mellonella were treated with 50 mg/kg IFA in 10 μl of vehicle or an equal volume of vehicle immediately after challenge with 2 × 10⁴ wild-type K7 or ΔGT-1 K7 bacteria. Wild-type K7 infection caused 100% mortality within 32 hours, whereas all the ΔGT-1 K7-infected G. mellonella survived, which confirmed the great importance of capsule for full virulence of K. pneumoniae (Fig 6A). Promisingly, IFA treatment effectively protected G. mellonella from K. pneumoniae killing, with a 37.5% increase in the survival rate (Fig 6A). In addition, the bacterial burden in the hemocoel was determined at 5 hours post infection when all the G. mellonella were alive. As shown in Fig 6B, the number of wild-type K7 bacteria was much greater than the input number, but IFA treatment significantly reduced bacterial dissemination. In contrast, the number of ΔGT-1 K7 bacteria was lower than the input number and was not influenced by IFA, indicating strong host clearance against bacteria deficient in capsule. Melanization is a widely used hallmark reaction of G. mellonella for the assessment of bacterial profile and immune evasion, and is positively correlated with pathogenicity [28]. Consistent with the result of bacterial loading assay, wild-type K7 infection elicited the highest degree of melanization (Fig 6C). In contrast, and as expected, ΔGT-1 K7-induced melanization was much lower compared with wild-type K7, and K7-caused melanization was also significantly attenuated by IFA treatment (Fig 6C). Lactate dehydrogenase (LDH) release, a marker of cell death, is often referred to as another predictor of virulence. We found that, compared with ΔGT-1 K7, wild-type K7 elicited more LDH release, but this effect was significantly diminished by IFA treatment (Fig 6D). Together, these results suggested that IFA treatment effectively rescued G. mellonella from K. pneumoniae killing via inhibiting capsule-conferred pathogenicity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. IFA reduced the virulence of hypervirulent K. pneumoniae in G. mellonella larvae.

(A) Survival of G. mellonella 48 h after K. pneumoniae infection. G. mellonella were treated with 50 mg/kg IFA in 10 μl of 10% DMSO containing vehicle (10% DMSO, 45% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) or an equal volume of vehicle immediately after challenge with 2 × 10⁴ wild-type K. pneumoniae K7 or ΔGT-1 K7 bacteria, and the number of deaths was recorded for survival analysis. Statistical analysis was performed using the log-rank (Mantel–Cox) test (n = 8 larvae each group). (B) Bacterial load in the hemocoels of larvae (presented as cfu/ml). G. mellonella infected and treated as described above were sacrificed at 5 hours postinfection, and the hemocoels were serially diluted and microbiologically plated (n = 6 larvae each group). ND, not detected. The back lines presents the means ± SEMs. (C) Infection-induced melanization of hemocoels from the indicated groups. G. mellonella infected and treated as described above were sacrificed at 5 hours postinfection, and the hemocoels were collected to measure the degree of melanization at OD_490nm (n = 9 larvae each group). The larvae in control group were injected with PBS and treated by equal volume of vehicle. (D) LDH release in the hemocoel from the indicated groups. G. mellonella infected as described above were treated with 50 mg/kg IFA in 10 μl of 1% DMSO containing vehicle (1% DMSO, 54% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) or an equal volume of vehicle and were sacrificed at 5 hours postinfection. Subsequently, the hemocoel was collected to measure LDH release using a cytotoxicity detection kit (n = 9 larvae each group). The results are presented as the OD_490nm values. The data are presented as the means ± SEMs. One-way ANOVA and Tukey’s posttest was used to perform multiple comparisons. *P < 0.05, **P < 0.01 and ns, no significance.

https://doi.org/10.1371/journal.ppat.1012787.g006

IFA conferred significant protection against K. pneumoniae infection in mice

Next, we further used mice lethal K. pneumoniae pneumonia model to better confirm the in vivo therapeutic effect of IFA. Similarly, challenge with 1 × 10⁷ wild-type K7 killed all the mice treated with vehicle within 72 hours, but all the mice infected with ΔGT-1 K7 survived. Promisingly, the survival rate of K7-infected mice was effectively enhanced by IFA treatment (50 mg/kg), with a significant increase of 41.67% (Fig 7A). The bacterial loads were subsequently detected with a 1/4 lethal dose of K. pneumoniae 40 hours post infection when all the mice were alive. Consistent with the 100% survival, almost no bacterial colonization was detected in the lungs of mice infected with ΔGT-1 K7 (Fig 7B), and as expected, IFA treatment reduced the high bacterial burden of wild-type K7 approximately by 22 times. Moreover, we also detected the number of bacteria in the BALF, livers and spleens from different groups. Consistently, considerable bacterial loads were detected in the BALF and distal organs of the mice infected with wild-type K7 (Fig 7C–7E), while ΔGT-1 K7 showed very little to no dissemination from lung to other organs, which indicated effective host clearance due to capsule deficiency. In contrast and as anticipated, IFA treatment significantly reduced the bacterial load in the BALF, livers and spleens of the mice challenged with wild-type K7 (Fig 7C–7E). Pathology analysis of lung sections from the mice infected with wild-type K7 exhibited more severe bronchopneumonia compared to ΔGT-1 K7-infected mice (Fig 7F), accompanied by higher levels of airway inflammation (Fig 7G), neutrophil infiltration (Fig 7H), intralesional bacterial burden and, as a result, a more severe overall histopathology score (Fig 7I and 7J). Collectively, these results confirmed that IFA treatment facilitated effective host clearance of K. pneumoniae and thereby significantly controlled bacterial dissemination through targeting capsule.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. IFA effectively limited bacterial dissemination and protected mice from lethal hypervirulent K. pneumoniae pneumonia.

(A) Survival of mice 72 hours after K. pneumoniae infection. Mice were subcutaneously injected with 50 mg/kg IFA in 50 μl of 10% DMSO containing vehicle (10% DMSO, 45% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) or an equal volume of vehicle immediately after challenge with 1 × 10⁷ wild-type K. pneumoniae K7 or ΔGT-1 K7 bacteria, and the number of deaths was recorded for survival analysis. Statistical analysis was performed using the log-rank (Mantel–Cox) test (n = 12 mice each group). (B) Bacterial burden in the lung tissues of sublethally infected mice. Mice challenged with 2.5 × 10⁶ wild-type K. pneumoniae K7 or ΔGT-1 K7 bacteria were treated as described above and sacrificed at 40 hours post infection, and lung tissues were removed and homogenized in sterilized PBS (10% w/v) to analyze the bacterial burden by microbiological plating (n = 5 mice each group). (C) Bacterial load in the bronchoalveolar lavage fluid (BALF) obtained from sublethally infected mice. Mice that were infected and treated as described above were sacrificed at 32 hours post infection for bronchoalveolar lavage (BAL) analysis, and the BALF was serially diluted and microbiologically plated (n = 6 mice in each group). The bacterial loads in the livers (D) and spleens (E) from the mice mentioned in (B) were also evaluated by microbiologically plating the tissue homogenates in PBS (n = 5 mice each group). ND, not detected. The back lines present the means ± SEMs. (F) H&E-stained lung tissues from mice infected and treated as indicated. The arrows indicate neutrophil infiltration. The degree of airway inflammation (G), neutrophilic infiltration (H), intralesional bacterial burden (I) and total histopathology score (J) were assessed according to standard pathology criteria by blinded scoring. The data are presented as the means ± SEMs. Unpaired two-tailed Student’s t-test was performed to determine the statistical significance of two groups. *P < 0.05, **P < 0.01 and ns, no significance.

https://doi.org/10.1371/journal.ppat.1012787.g007

Consistent with the severe airway inflammation observed via histopathology analysis, sublethal infection with wild-type K7 boosted the secretion of the inflammatory cytokines TNF-α, IL-6 and IL-1β, whereas the levels of these cytokines were significantly reduced by IFA treatment (Fig 8A–8C). Notably, IFA did not have any inhibitory effect on LPS-stimulated inflammation response in vitro (S4 Fig), indicating that IFA was unable to suppress TLR4-MyD88 signaling and that the protective effect of IFA in mice was independent of anti-inflammation activity. Somewhat surprisingly, the production of IL-10 and IFN-γ was much higher in mice infected with wild-type K7 and received IFA treatment (Fig 8D and 8E), indicating that IFA treatment facilitated the development of an inflammation-controlling and infection-limiting response to fight against K. pneumoniae in mice. Of note, the levels of these cytokines in mice infected with ΔGT-1 K7 were all much lower than those in the wild-type K7-infected mice (Fig 8A-8E), which was consistent with effective bacterial clearance resulted from the deficient of capsule. Consistently, the numbers of inflammatory monocytes (CD11b⁺ Ly6C^high) and neutrophils (Cd11b⁺Ly6G⁺) in wild-type K7-infected mice were much greater than those in ΔGT-1 K7-infected mice, but they were also markedly reduced by IFA treatment (Fig 8F and 8G). Conversely, compared with wild-type K7, the numbers of mature macrophages and NK cells were much greater in mice infected with ΔGT-1 K7 (Fig 8H and 8I), which play essential roles in the defense against K. pneumoniae infection [29]. As speculated, IFA treatment effectively increased the number of these populations critical for host innate clearance. The populations of CD4⁺ and CD8⁺ T cells were also much lower in the mice infected with wild-type K7, but the numbers of these anti-infection adaptive immune cells were also significantly increased by IFA (Fig 8J and 8K), which was consistent with the increase in IFN-γ. The lower levels of IL-6, IL-1β and TNF-α and populations of inflammatory monocytes and neutrophils and, in contrast, the greater populations of macrophages, NK cells, and CD4⁺ and CD8⁺ T cells in the lungs of mice treated with IFA explained how IFA provides systemic protection against K. pneumoniae infection in vivo.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. IFA treatment facilitated host immune clearance of hypervirulent K. pneumoniae by targeting capsule.

Mice challenged with 2.5 × 10⁶ wild-type K. pneumoniae K7 or ΔGT-1 K7 bacteria were subcutaneously injected with 50 mg/kg IFA in 50 μl of 10% DMSO containing vehicle (10% DMSO, 45% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) or an equal volume of vehicle, and were then sacrificed at 40 hours postinfection. The concentrations of TNF-α(A), IL-6 (B), IL-1β(C), IL-10 (D) and IFN-γ (E) in lung tissues were measured using ELISA kits following the manufacturer’s instructions (n = 5 mice each group). To determine the number of immune cells in the BALF, mice infected and treated as indicated were sacrificed at 32 h postinfection for BALF collection. The populations of inflammatory monocytes (CD11b⁺Ly6C⁺) (F), neutrophils (CD11b⁺Ly6G⁺) (G), mature macrophages (CD11b⁺F4/80⁺) (H), NK cells (CD3e⁺NK1.1⁺) (I), CD4⁺ T cells (CD3e⁺CD4⁺) (J) and CD8⁺ T cells (CD3e⁺CD8⁺) (K) in the BALF from the indicated groups were quantified by flow cytometry (n = 3 mice each group). The y-axis in F-K means the percentage of gated 30000 cells from individual mice BALF for each marker set. The data are presented as the means ± SEMs. Unpaired two-tailed Student’s t-test was performed to determine the statistical significance of two groups. *P < 0.05, **P < 0.01 and ns, no significance.

https://doi.org/10.1371/journal.ppat.1012787.g008

Finally, we further distinguished the respective contribution of capsule and rmpD-regulated and hypercapsule-associated HMV to IFA-conferred protection against hvKP infection in vivo. Consistent with previous study, rmpD deletion resulted in a significant loss of virulence versus wild-type K7, as 1 × 10⁷ CFU (a 100% lethal dose of wild-type K7) ΔrmpD K7 bacterial failed to cause death in mice (S2G Fig). Surprisingly, 5 × 10⁸ CFU (50 fold of 100% lethal dose of wild-type K7) ΔrmpD K7 mutant still did not result in death within 72 hours (S2G Fig). Therefore, we performed bacterial burden assay by challenging 5 × 10⁷ CFU bacterial (20 fold of the inoculation dose used before). As shown in S2H–S2J Fig, the bacterial burdens of ΔrmpD K7 were much lower compared with wild-type K7 but significantly greater than that of ΔGT-1 K7, which confirmed the distinct contribution of HMV and capsule, as well as the indispensable role of capsule in hvKP pathogenicity. Unfortunately, although IFA treatment reduced the colonization of ΔrmpD K7 in mice lungs by about 6 fold versus vehicle group (P = 0.0618) (S2H Fig), and by 4 fold in livers (P = 0.0716) (S2I Fig), neither of them achieved statistical significance. By contrast, bacterial loading in spleens was significantly reduced by IFA (P = 0.0211) (S2J Fig), indicating that the in vivo protection effect of IFA on hvKP infection is attributed to its inhibition on both capsule and hypercapsule-associated HMV, rather than acting on rmpD.

In conclusion, these results confirmed that IFA treatment promoted host immune clearance of K. pneumoniae by targeting capsule and hypercapsule-associated HMV, thereby systemically protecting mice against K. pneumoniae infection.

Ethics statement

All the animal experiments were approved and strictly conducted in compliance with the relevant guidelines established by the Jilin University Institution Animal Care Committee (approval no. SY202412069).

Reagents and antibodies

IFA (>98%) purchased from Shanghai Yuanye Biotechnology was dissolved in dimethyl sulfoxide (DMSO; Sigma–Aldrich) to make a 40 mg/mL stock solution for in vitro experiments. For in vivo delivery, IFA was dissolved in a vehicle solvent consisting of 10% DMSO, 45% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80 to make a 40 mg/mL stock for injection unless otherwise indicated. Apramycin, spectinomycin, L-arabinose and sucrose used for colony screen were from Shanghai Aladdin Biotechnology. Zwittergent 3–14, citric acid and 3-hydroxydiphenol used for uronic acid measurement were purchased from Sigma–Aldrich. ATP Assay Kit and Enhanced NAD+/NADH Assay Kit with WST-8 were purchased from Beyotime. Lipopolysaccharide from E. coli O26:B6 was purchased from Sigma–Aldrich. Mouse IL-1β, IL-6, TNF-α, IL-10 and IFN-γ kits from Biolegend were used to detect cytokines. A mouse monoclonal allophycocyanin (APC) anti-C3b/iC3b antibody (BioLegend) was used for the detection of C3b/iC3b deposition. The mouse anti-C5b-9 antibody and Alexa Fluor 488-conjugated goat anti-mouse IgG H+L used for the detection of the C5b-9 load were both obtained from Abcam. The following antibodies from BD Biosciences were used for the staining of neutrophils, monocytes, NK cells, mature macrophages, CD4⁺ T cells and CD8⁺ T cells in the bronchoalveolar lavage fluid (BALF): rat anti-mouse CD16/CD32, anti-CD11b-APC, anti-Ly6G-PE-Cy7, anti-Ly6C-FITC, anti-CD3e-Percy-Cy5, anti-NK1.1-APC, anti-F4/80-PE, anti-CD4-APC-Cy7 and anti-CD8-BV421.

Bacterial strains, culture conditions and plasmids

The clinical isolation of carbapenem-sensitive K2-serotype hypervirulent K. pneumoniae strain K7 and other isolates, including WKP5 (K2), WKP22 (K2), WKP11 (K1), WKP13 (K1), WKP25 (K1), WKP35 (K57), KPP29 (K57) and WKP50 (K61) gifted from Gu’s lab, and hvKP ATCC43816 (K2) gifted from Zhang’s lab were all cultured in Luria–Bertani (LB) broth with shaking at 37°C. Notably, the capsule deficient mutant ΔGT-1 K7 from Gu was constructed as previously described and cultured in Luria–Bertani (LB) broth containing 30 μg/ml kanamycin [42]. For comparative assessment of colony mucoidity, isolates were plated on LB agar or blood agar plates (LB agar with 5% defibrinated sheep blood). PCasKP-apr plasmid and pSGKP-spec plasmid gifted from Ji’s lab were used to generate a rmpD-deletion K7 mutant.

Construction of a rmpD-deletion mutant

RmpD-deletion K7 was generated using a CRISPR-based genome-editing approach [43]. In brief, designed single artificial chimeric guide RNA (sgRNA) was ligated into the linearized pSGKP-spec plasmid, and the linear double-stranded DNA (dsDNA) homologous arms of rmpD gene was amplified by fusion PCR as a repair template. Electroporate no less than 200 ng pCasKP-apr plasmid into competent K7 cells to prepare pCasKP-harboring competent K7 cells. Subsequently, 200 ng spacer-induced pSGKP-sgRNA plasmid and 500 ng dsDNA repair template were co-transformed into pCasKP-harboring competent K7 cells by electroporation. Positive transformants were screened using 30 μg/mL apramycin and 300 μg/mL spectinomycin at 30°C and verified by DNA sequencing. Last, plasmids were cured by culturing the single mutant colony at 37°C in the presence of 5% sucrose. Primers used in the in-frame deletion and verification are listed in S1 Table.

Cell lines and mice

Human A549 alveolar epithelial cells and mouse J774 macrophages were cultured in DMEM (Dulbecco’s modified Eagle’s medium, Gibco) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin in a 5% CO₂ incubator. Mouse primary peritoneal macrophages were collected as previously described and cultured in RPMI 1640 (Roswell Park Memorial Institute 1640 Medium) supplemented with 10% FBS [44].

Eight-week-old female C57BL/6 mice were purchased from Changsheng Biotechnology (Shenyang, China) and humanely housed in individually ventilated cages (IVCs) under standard laboratory conditions (25 ± 1°C, 12-h light/dark cycle with free access to food and water).

Extraction and quantification of capsule polysaccharide (CPS)

The extraction and quantification of CPS were performed according to a modified procedure [45]. In brief, overnight K. pneumoniae cultures were transferred to freshly prepared LB broth containing DMSO or the indicated concentrations of IFA at a ratio of 1:50 and further cultivated for 4 hours with shaking at 80 rpm. To determine the total CPS production, 1 ml subcultures were directly mixed with 200 μl of 1% Zwittergent 3–14 detergent in 100 mM citric acid (pH 2.0) and incubated at 50°C for 20 min to extract CPS. To measure cell-attached and unattached CPS, 1 ml aliquots were centrifuged at 13,000 × g for 5 min. The supernatants and pellets resuspended in 1 ml of fresh LB broth were also mixed with 200 μl of 1% Zwittergent 3–14 detergent and incubated under the same conditions. Then, all the aliquots were centrifuged at 13,000 × g for 5 min, and the CPS in the supernatants was precipitated with ethanol (80% final concentration) at 4°C for 20 min and air-dried after discarding the supernatants. Finally, the CPS samples were dissolved in tetraborate/sulfuric acid, developed with 3-hydroxydiphenol, and quantified according to the standard curve of glucuronic acid based on the absorbance of the mixture at 520 nm. In addition, CPS samples extracted from equal amounts of K. pneumoniae strains were separated by 12% SDS-PAGE and followed by alcian blue staining, visualization and quantification by Image J.

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM)

The subcultures of K. pneumoniae described above were washed 3 times with sterilized PBS and fixed in 4% glutaraldehyde overnight at 4°C. The immobilized samples were dehydrated by gradient ethanol (20, 50, 70, 80, 90, 95 and 100) and then dried under flowing nitrogen on cover glasses. Finally, bacterial morphology was observed by ZEISS scanning electron microscopy (SEM) at an acceleration voltage of 2 kV, and images were captured at a magnification of 15,000 ×.

For TEM, the fixed samples were embedded in Lowicryl HM20 Monostep resin and cut into ultrathin sections on an ultramicrotome, followed by staining with uranyl acetate and lead citrate. Images were obtained on an FEI Tecnai Spirit 120 kV transmission electron microscope with a Tietz F4.15 CCD camera.

Susceptibility assay

The minimum inhibitory concentration (MIC) of IFA on K. pneumoniae was determined by the broth microdilution method [46]. The growth curves of K. pneumoniae with or without IFA were determined by spectrophotometry as previously described [47].

Comparative assessment of colony mucoidity

K. pneumoniae strains were serially diluted and plated for fully separated, single colonies on blood agar plates or regular agar plates containing DMSO or 32 μg/ml IFA. After 18 hours of cultivation at 37°C, colony morphology was captured by photography, and the “string test” was performed as previously described [48].

Centrifugation resistance analysis

A semiquantitative sedimentation assay was performed as follows: triplicate aliquots (1 ml) of the subcultures cocultured with DMSO or the indicated concentrations of IFA were centrifuged at 1,000 × g in a fixed angle rotor for 5 min after the original OD₆₀₀ was measured by spectrophotometry, following which the supernatants were removed and transferred into new tubes to measure the OD₆₀₀. The results were defined as the ratio of the supernatant OD₆₀₀ to the original OD₆₀₀.

RNA-seq analysis

RNA sequencing was carried out with 3 replicates to quantify patterns of gene expression in K. pneumoniae treated with DMSO control or 32 μg/ml. Total RNA was extracted using an AMPure XP system (Beckman Coulter, Beverly, USA) according to the manufacturer’s instructions. RNA-seq was performed on an Illumina Novaseq platform (Beijing, China). Differential expression analysis of two groups (two biological replicates per group) was performed using the DESeq R package (1.18.0). Quantitative analysis of gene expression, differential gene expression analysis, and Gene Ontology (GO) analysis were subsequently performed to explore and identify enriched functions and pathways of differentially expressed genes. Notably, GO enrichment analysis of differentially expressed genes was implemented by the GOseq R package. Type-1 errors were reduced by performing false discovery rate (FDR) correction via Benjamini and Hochberg (BH) methods. GO terms and pathways with adjusted p values ≤ 0.05 were significantly enriched in DEGs after multiple testing correction.

Quantitation of intracellular ATP and the ratio of NAD+/NADH

Overnight K. pneumoniae cultures were transferred to freshly prepared LB broth containing DMSO or the indicated concentrations of IFA at a ratio of 1:50 and further cultivated for 4 hours with shaking at 80 rpm. Subsequently, equal amounts of bacterial were collected by centrifugation at 12,000 × g for 5 min. The intracellular ATP levels and the ratio of NAD+/NADH were then analyzed using Enhanced ATP Assay Kit and Enhanced NAD+/NADH Assay Kit with WST-8, respectively.

Serum survival assays

Overnight cultures of K. pneumoniae were subcultured in fresh LB broth supplemented with DMSO or the indicated concentrations of IFA for 4 hours. One milliliter of each culture aliquot was centrifuged, washed once with sterilized PBS, resuspended in 1 ml of PBS and then diluted at a ratio of 1:100. Two hundred microliters of the dilutions were then incubated with 20% (final concentration) prewarmed normal human serum (NHS) in a 37°C incubator with rotation. At each indicated time point, 20 μl aliquots from each assay tube were immediately serially diluted and plated on LB agar for colony enumeration.

Detection of surface-deposited complement components

The subcultures described above were collected by centrifugation and sensitized in 1 ml of GVB²⁺ buffer (0.15 mM CaCl₂, 141 mM NaCl, 0.5 mM MgCl₂, 0.1% gelatin and 5 mM veronal) for 5 min at room temperature. Two hundred microliters of the 1:100 diluted suspensions were incubated with 20% NHS in a 37°C incubator with rotation for the indicated times. Then, the mixtures were centrifuged and washed with PBS. C3b/C3bi was directly labeled with a mouse anti-C3b/C3bi-APC antibody (4 μl per 10⁶ bacterial cells), and C5b-9 membrane attack complex formation was indirectly detected with a mouse anti-C5b-9 antibody as the primary antibody (2 μg/ml) and Alexa Fluor 488 goat anti-mouse IgG H+L as the secondary antibody (1:1000). After 20 minutes of incubation at room temperature, the samples were washed 3 times with PBS and resuspended in 220 μl of PBS. Twenty microliters of each aliquot were transferred to cover glasses for image capture on a laser scanning confocal microscope (Olympus, Japan), and the remaining samples were immediately acquired using a BD FACSCelesta instrument. Approximately 10,000 cell events were collected and analyzed using FlowJo 10 software. Notably, a serum-free control was set to exclude non-specific binding of the secondary antibody. The immunofluorescence intensity was quantified using Image J software. Graph design and statistical analysis were performed using GraphPad 8.4.2 software.

Adhesion and macrophage phagocytosis assay

Adhesion and macrophage phagocytosis assay were performed according to a previously study with some modification [48]. A549 cells were seeded in 24-well plates at a density of 2 × 10⁵ cells per well and cultured overnight in a 5% CO₂ incubator. Overnight cultures of K. pneumoniae were subcultured in fresh LB broth supplemented with DMSO or the indicated concentrations of IFA for 4 hours and then added to the cell wells at an MOI of 50, followed by centrifugation at 1,000 × g for 10 min to synchronize infection. After a 2-hour incubation, the monolayers were washed 3 times with PBS and lysed with 0.1% Triton-X 100. The lysates were serially diluted and plated on LB agar to determine the adherence efficiency by enumerating the colony-forming units. The data are presented as a percentage of the initial inoculum CFU.

J774 cells and mouse primary peritoneal macrophages (MPMs) were seeded in 24-well plates at a density of 5 × 10⁵ cells per well and cultured overnight in a 5% CO₂ incubator. Cells were then infected with K. pneumoniae at an MOI of 5 with different treatments and centrifuged at 1,000 × g for 10 min to synchronize infection. After being incubated for 2 hours, the cells were rinsed 3 times with PBS and further incubated for indicated times in fresh medium containing 100 μg/ml gentamicin to kill extracellular bacteria. The cells were then rinsed, lysed, serially diluted and plated on LB agar to enumerate the colony-forming units. The phagocytosis percentage of each inoculum was defined as the ratio of phagocytized bacteria to the total bacterial count determined before adding gentamicin.

RNA extraction and RT–qPCR

The subcultures described above were collected at log-phase for the detection of transcriptional levels of cps genes and rmp genes. Total RNA was isolated using the TRIzol method. The RT–qPCR primer pairs used in the study are listed in S1 Table. RT–qPCR was carried out on an Applied Bioscience 7500 thermocycler using FastStart Universal SYBR Green Master Mix (Gene Star) according to the manufacturer’s protocol [49].

Galleria mellonella infection

Larvae of G. mellonella were purchased from Yude Biotechnology Ltd (Tianjin, China). Overnight cultures of K. pneumoniae were subcultured at 1:100 in fresh LB broth to an OD₆₀₀ of approximately 0.6. Bacteria were collected, washed 3 times with sterilized PBS and resuspended in PBS. Ten microliters of the suspension containing 2 × 10⁴ bacterial cells or PBS were injected into the foremost right-side proleg of the larvae using a microinjection pump, and 10 μl of vehicle or IFA was delivered via the foremost left-side proleg at a dose of 50 mg/kg. Larvae in the control group were injected with 10 μl of PBS and vehicle. After injection, the larvae were moved to a 37°C incubator, and the number of deaths was scored periodically. To determine the bacterial load in the hemocoel of the larvae, the larvae were placed on ice for 20 min at the fifth hour after infection. The bottom 2 mm of the larvae was aseptically cut off, and the hemocoel was collected into 1.5 ml tubes, followed by serial dilution and microbiological plating. Moreover, melanization and lactate dehydrogenase (LDH) assays were performed as described previously [28]. Of note, in LDH assay, IFA was dissolved in 1% DMSO containing vehicle (1% DMSO, 54% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) rather than 10% DMSO. All experiments were carried out at least in triplicate.

Murine pneumonia model

For the analysis of mouse mortality, female C57BL/6 mice (6–8 weeks old) were challenged with 1 × 10⁷ CFU of K7 or ΔGT-1 K7 bacteria via the intranasal route, followed by subcutaneous injection of a single dose of 50 mg/kg IFA dissolved in 10% DMSO containing vehicle (10% DMSO, 45% stroke-physiological saline solution, 40% PEG400 and 5% Tween-80) or an equal volume of vehicle (50 μl) immediately. IFA and vehicle were administered 3 times daily, and the number of dead mice was counted until 72 hours after inoculation. To assess the histopathology of the lung tissues, mice were infected with 2.5 × 10⁶ CFU of bacteria, treated with IFA or vehicle in the same way, anesthetized with isoflurane and euthanized through rapid cervical dislocation at 40 hours post infection. The mouse lungs were fixed in 10% formalin, stained with hematoxylin and eosin (H&E) and visualized for histopathological examination. Standard pathology criteria were applied to score the histopathology features of the lung lesions [29]. Additionally, the lungs, livers and spleens from infected mice that were treated as described above were homogenized in sterilized PBS to prepare a 10% homogenate (w/v) for the analysis of the bacterial load via microbiological plating. The cytokine levels in the supernatants of lung tissue homogenates were assessed using ELISA. Bronchoalveolar lavage (BAL) was carried out at 32 hours post infection according to a previous study [50], and bacterial loads in BALF were also determined. Total cells in the BALF were collected by centrifugation at 600 × g for 5 min, and erythrocytes were then removed with 1 ml of RBC lysis buffer. To further verify the respective contribution of capsule and HMV in IFA-conferred protection, mice were challenged with 5 × 10⁷ CFU K7, ΔGT-1 K7 and ΔrmpD K7 for bacterial burden analysis.

Staining and analysis of innate and adaptive immune cells by flow cytometry

The cells collected from the BALF described above were resuspended in 1 ml of PBS supplemented with 0.5% BSA, passed through a 70-μm strainer and incubated with a mouse anti-CD16/32 antibody at a dilution of 1:100 to block Fc receptors. After 10 min of incubation at room temperature and washing once with PBS containing 0.5% BSA, half of the cells were transferred to new tubes and further incubated with anti-CD3e-Percy-Cy5, anti-NK1.1-APC, anti-CD4-APC-Cy7 and anti-CD8-BV421 antibodies at 4°C for 30 min, whereas the remaining cells were stained with anti-CD11b+-APC, F4/80-PE, anti-Ly6G-PE-Cy7 and anti-Ly6C-FITC. All the cells were then washed once and resuspended in 300 μl of PBS containing 0.5% BSA. A total of 30,000 cell events were detected on a BD FACSCelesta instrument and analyzed using FlowJo software. The populations of neutrophils (CD11b⁺Ly6G⁺), inflammatory monocytes (CD11b⁺Ly6C⁺), macrophages (CD11b⁺F4/80⁺), NK cells (CD3e⁺NK1.1⁺), CD4⁺T cells (CD3e⁺CD4⁺) and CD8⁺T cells (CD3e⁺CD8⁺) were gated. Unless otherwise indicated, all percentages are of single viable frequency, and the results from 3 mice were analyzed using GraphPad Prism 8.4.2.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 8.4.2 (USA). The numeric data from no less than three independent experiments are presented as the mean ± standard error of the mean (SEM). The statistical significance of differences between two independent groups was determined by unpaired two-tailed Student’s t test. Comparisons of 3 or more data points were analyzed using one-way analysis of variance (ANOVA) and Tukey’s posttest, as indicated. The survival rate was analyzed using the log-rank (Mantel–Cox) test. *P < 0.05, **P < 0.01 and ns, no significance.