Klebsiella pneumoniae Outer Membrane Porins OmpK35 and OmpK36 Play Roles in both Antimicrobial Resistance and Virulence

Abstract

OmpK35 and OmpK36 are the major outer membrane porins of Klebsiella pneumoniae. In this study, a virulent clinical isolate was selected to study the role of these two porins in antimicrobial resistance and virulence. The single deletion of ompK36 (ΔompK36) resulted in MIC shifts of cefazolin, cephalothin, and cefoxitin from susceptible to resistant, while the single deletion of ompK35 (ΔompK35) had no significant effect. A double deletion of ompK35 and ompK36 (ΔompK35/36) further increased these MICs to high-level resistance and led to 8- and 16-fold increases in the MICs of meropenem and cefepime, respectively. In contrast to the routine testing medium, which is of high osmolarity, susceptibility tests using low-osmolarity medium showed that the ΔompK35 mutation resulted in a significant (≥4-fold) increase in the MICs of cefazolin and ceftazidime, whereas a ΔompK36 deletion conferred a significantly (4-fold) lower increase in the MIC of cefazolin. In the virulence assays, a significant (P < 0.05) defect in the growth rate was found only in the ΔompK35/36 mutant, indicating the effect on metabolic fitness. A significant (P < 0.05) increase in susceptibility to neutrophil phagocytosis was observed in both ΔompK36 and ΔompK35/36 mutants. In a mouse peritonitis model, the ΔompK35 mutant showed no change in virulence, and the ΔompK36 mutant exhibited significantly (P < 0.01) lower virulence, whereas the ΔompK35/36 mutant presented the highest 50% lethal dose of these strains. In conclusion, porin deficiency in K. pneumoniae could increase antimicrobial resistance but decrease virulence at the same time.

Klebsiella pneumoniae is a Gram-negative opportunistic pathogen and a common cause of nosocomial infections. Over the past 2 decades, an invasive form of community-acquired K. pneumoniae causing pyogenic liver abscesses has emerged in Asia, and complications, including endophthalmitis or meningitis, are common in these patients (17, 29). Invasive K. pneumoniae has also been reported to be an emerging infectious disease in non-Asian countries (16, 28, 30). Cephalosporins are one of the most common drugs for the treatment of K. pneumoniae infection. However, increased resistance due to production of extended-spectrum β-lactamase (ESBL) or AmpC-type β-lactamase by K. pneumoniae has complicated treatment with this class of antibiotics. Although carbapenems are still effective drugs for these strains, resistance to this class could result from the additional loss of outer membrane porins (11, 42) or the production of carbapenemases (46).

The cell envelope of Gram-negative bacteria consists of three principal layers: the outer membrane, the peptidoglycan cell wall, and the inner membrane. Outer membrane porins exist as trimers and act as water-filled protein channels that allow the transport of small hydrophilic molecules such as iron, nutrients, and antibiotics, including β-lactams, across the outer membrane. Porins also serve as receptors for phages and bacteriocins and, in conjunction with peptidoglycan and lipopolysaccharide (LPS), they have a significant structural role in maintaining the integrity of the cells (1, 3). Previous studies have also shown that porins may act as virulence factors during bacterial infection (2, 6, 9, 14, 18).

Most studies examining the role of porins have been performed in Escherichia coli (especially the laboratory strain K-12), and two major porins, OmpC and OmpF, have been extensively characterized. The functional pore of OmpF is slighter larger than that of OmpC; therefore, it is easier for molecules to pass through the OmpF pore (39, 40). In K. pneumoniae, two major porins, OmpK35 and OmpK36, are homologous to OmpF and OmpC, respectively. Clinically, most of the ESBL-producing K. pneumoniae strains express only OmpK36, whereas the majority of K. pneumoniae that do not produce ESBLs synthesize both OmpK35 and OmpK36 (21). The absence of OmpK35 may be one of the factors contributing to the antibiotic resistance of ESBL-producing K. pneumoniae, and this hypothesis has been supported by Palasubramaniam et al. (43). Reports also revealed that OmpK36 may play an important role in the resistance or reduced susceptibility to carbapenems in K. pneumoniae that produce ESBL or AmpC-type β-lactamases (36, 50, 52). However, previous studies have shown that the loss of the expression of one porin, either OmpK35 or OmpK36, did not significantly alter the MICs of K. pneumoniae strains without other mechanisms of antimicrobial resistance, such as ESBLs (22, 27), while these results were different from others’ studies examining OmpK35 (31) and OmpK36 (9).

To our knowledge, no in-depth study has been performed to further evaluate the importance of these two major porins. Our previous study has showed that inactivation of the ompK36 gene by insertion-duplication mutagenesis could reduce virulence in a mouse peritonitis model for a virulent K. pneumoniae with a 50% lethal dose (LD₅₀) at 5 × 10³ CFU (9). However, polar effect and fusion protein caused by the integration of the insertion vector into gene may also occur and influence the virulence of bacteria as previously described (7). In present study, we used in-frame deletion mutagenesis to generate the ΔompK35, ΔompK36, and double-deletion mutants (53). A highly virulent K. pneumoniae with an LD₅₀ < 10 CFU in a mouse peritonitis model was selected to evaluate the role of porins contributing to the antimicrobial resistance and virulence.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in the present study are listed in Table 1. K. pneumoniae NVT2001, serotype 2, was isolated from a patient in Taiwan with liver abscesses. Unless otherwise noted, E. coli and K. pneumoniae and its derivatives were cultured at 37°C in Mueller-Hinton broth (MHB) or Luria-Bertani (LB) broth with appropriate antibiotics. The growth of cells was monitored by determining the optical density at 600 nm (OD₆₀₀) using an Eppendorf BioPhotometer (Vaudaux-Eppendorf AG, Basel, Switzerland).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source or reference
Strains
K. pneumoniae
NVT2001	Clinical isolates	This study
NVT2001S	Isolate of NVT2001; Smr	This study
ΔompK35 mutant	ompK35 deletion strain of NVT2001S; Smr	This study
ΔompK36 mutant	ompK36 deletion strain of NVT2001S; Smr	This study
ΔompK35/36 mutant	ompK35 and ompK36 deletion strain of NVT2001S; Smr	This study
ΔompK35/36::35 mutant	ompK35-complemented strain of the ΔompK35/36 mutant; Smr	This study
ΔompK35/36::36 mutant	ompK36-complemented strain of the ΔompK35/36 mutant; Smr	This study
ΔompK35/36::35/36 mutant	ompK35- and ompK36-complemented strain of the ΔompK35/36 mutant; Smr	This study
E. coli
S17-1 λpir	hsdR recA pro RP4-2 (Tc::Mu; Km::Tn7) (λpir)	49
Plasmids
pKAS46	Suicide vector, rpsL; Apr Kmr	49
pKO35	2,202-bp fragment containing an 816-bp deletion in ompK35 locus cloned into pKAS46	This study
pKO36	2,161-bp fragment containing an 867-bp deletion in ompK36 locus cloned into pKAS46	This study
pCOM35	3,018-bp fragment containing the entire ompK35 locus cloned into pKAS46	This study
pCOM36	3,028-bp fragment containing the entire ompK36 locus cloned into pKAS46	This study
pCTXM14	Plasmid containing blaCTX-M-14 from clinical isolate	33
pCMY2	Plasmid containing blaCMY-2 from clinical isolate	54

^aAp^r, resistance to ampicillin; Km^r, resistance to kanamycin; Sm^r, resistance to streptomycin.

DNA sequencing and sequence analysis.

Based on the nucleotide sequence of the complete genome of K. pneumoniae 342 (15) (GenBank accession number CP000964), primers were designed and used for PCR amplification of the corresponding K. pneumoniae NVT2001 sequence by using Phusion Flash High-Fidelity PCR master mix (Finnzymes Oy, Espoo, Finland). The lengths of ompK35 and ompK36, including their amplified flanking regions, were both ∼3.4 kb using the primer sets 35F-35R and 36F-36R (see Table S1 in the supplemental material). All sequence analyses and protein homology searches were conducted using the NCBI database (http://www.ncbi.nlm.nih.gov/).The free energy of formation was calculated through the Vienna RNA websuite (19).

Construction of ompK35 and ompK36 deletion mutants.

K. pneumoniae NVT2001S mutants disrupted specifically in the ompK35 and/or ompK36 genes were constructed by in-frame deletion mutagenesis (53) (Fig. 1). Two DNA fragments (∼1.1 kb in size) that flanked the regions to be deleted were PCR amplified using specific primer sets (see Table S1 in the supplemental material). For the ompK35 deletion construction, the primer sets 5XbaF-5LapR and 5LapF-5SacR were used. Primer 5LapR contains 23 bp of DNA that is complementary to the sequence of primer 5LapF. The two gel-purified PCR products containing complementary ends were mixed and amplified using primers 5XbaF and 5SacR to create an 816-bp deletion in ompK35 by overlap PCR (23, 24). The resulting 2.2-kb PCR fragment was digested with XbaI and SacI and then cloned into pKAS46 that was similarly digested, resulting in plasmid pKO35. A ompK36 deletion mutant was constructed in an analogous manner, except that primer sets 6XbaF-6LapR and 6LapF-6SacR were used. Primer 6LapR contains 26 bp of DNA that is complementary to the sequence of primer 6LapF. The two gel-purified PCR products containing complementary ends were mixed and amplified using primers 6XbaF and 6SacR to create an 867-bp deletion in ompK36 by overlap PCR. The resulting 2.2-kb PCR fragment was digested with XbaI and SacI and then cloned into pKAS46 that was similarly digested, resulting in plasmid pKO36.

FIG. 1.

Creation of the porin deletion and complemented strains. (A) Physical map of the ompK35 and ompK36 genes. The positions of the putative promoters (arrows) and terminators (stem-loop structures) identified are highlighted. DNA fragments KO (dotted lines) are the region to be deleted from porin genes. DNA fragments 1 and 2 (dark lines) are the flanking regions of fragment KO. (B and C) Amplify the DNA fragments from K. pneumoniae NVT2001S using PCR. The generated DNA fragments were ligated into the suicide vector pKAS46, resulting in plasmid pKO35, pKO36, pCOM35, and pCOM36. The porin deletion (B) and complemented (C) constructs were transformed into K. pneumoniae to generate the porin deletion and complemented strains, respectively. (D) A single recombination event through DNA fragment 1 or 2 gave rise to one of two possible single-crossover strains. (E) The second recombination event can result in either restoration of the origin gene or a mutant double-crossover strain.

For homologous recombination, each of the porin deletion constructs in pKAS46 was transformed into E. coli S17-1 λpir by electroporation and then mobilized into the streptomycin-resistant strain K. pneumoniae NVT2001S via conjugation. Plasmid pKAS46 is a suicide vector containing the E. coli rpsL gene, which encodes wild-type ribosomal protein S12. When this gene is expressed on the integrated pKAS46, it confers a streptomycin-sensitive phenotype upon streptomycin-resistant strains, which allows for positive selection with streptomycin to detect loss of the vector (49). Single-crossover strains were selected from brilliant green containing inositol-nitrate-deoxycholate (BIND) plates supplemented with kanamycin (50 μg/ml), while the growth of non-K. pneumoniae contaminants was effectively suppressed on BIND plates (41). The kanamycin-resistant transconjugant was selected, and the insertions of pKO35 or pKO36 were verified by PCR using the primer pairs that flanked the target gene. After incubation overnight in 4 ml of MHB in the absence of kanamycin at 37°C, the fully grown cultures were diluted (1:1,000) in fresh MHB supplemented with 500 μg of streptomycin/ml, incubated at 37°C until the late-exponential growth phase, and then spread onto MHB plates supplemented with streptomycin (500 μg/ml). After double crossover occurred, streptomycin-resistant and kanamycin-sensitive colonies were selected, and the deletions of ompK35 or ompK36 were confirmed by PCR. The resulting mutants were designated K. pneumoniae ΔompK35 and K. pneumoniae ΔompK36; K. pneumoniae ΔompK35 was further used to generate the double-deletion (ΔompK35 ΔompK36) strain K. pneumoniae ΔompK35/36 (Table 1).

Complementation of mutants.

The allelic exchange method was used to restore the entire ompK35 and/or ompK36 gene in K. pneumoniae ΔompK35/36 (Fig. 1). Briefly, DNA fragments of the entire ompK35 and ompK36 genes with their flanking regions were amplified from K. pneumoniae NVT2001S using PCR with the primer sets 5XbaF-5SacR and 6XbaF-6SacR, respectively. Each of the two 3-kb PCR fragments generated were digested with XbaI and SacI and then cloned into pKAS46 that was similarly digested. The resulting plasmids, pCOM35 and pCOM36, were then transformed into E. coli S17-1 λpir by electroporation and mobilized into the porin double-deletion strain K. pneumoniae ΔompK35/36 via conjugation. After single crossover occurred, the kanamycin-resistant transconjugant was selected, and the insertions of pCOM35 or pCOM36 were verified by PCR. After double crossover occurred, streptomycin-resistant and kanamycin-sensitive colonies were selected, and the restorations of ompK35 or ompK36 were also confirmed by PCR. The resulting strains were designated K. pneumoniae ΔompK35/36::35 (ompK35-complemented strain) and ΔompK35/36::36 (ompK36-complemented strain); K. pneumoniae ΔompK35/36::35 was further used to generate the double-complemented strain K. pneumoniae ΔompK35/36::35/36 (Table 1). In addition, the experiments achieved in the present study for these complemented strains and for NVT2001S, ΔompK35, and ΔompK36 strains were performed independently to eliminate the possibility of contamination from each other.

Analysis of OMPs.

Bacterial cells were grown in high-osmolarity MHB or low-osmolarity nutrient broth (NB) to the logarithmic phase and were lysed by sonication. Outer membrane proteins (OMPs) were extracted with sodium lauroyl sarcosinate (Sigma, St. Louis, MO) and recovered by ultracentrifugation, as described previously (21). The OMP profiles were determined by SDS-PAGE using 12% SDS gels, followed by Coomassie blue staining (Gibco-BRL, Grand Island, NY). The intensity of bands was analyzed by densitometry using ImageJ software (http://rsb.info.nih.gov/ij/).

Antimicrobial susceptibility test and growth rate determination.

The MICs of 17 antimicrobial agents were determined using a broth microdilution test according to the recommendations of the Clinical and Laboratory Standards Institute (10). The following antimicrobial agents were used: ampicillin, piperacillin-tazobactam, ticarcillin-clavulanic acid (CLA), cefazolin, cephalothin, cefoxitin, ceftriaxone, cefpodoxime, cefotaxime, cefotaxime-CLA, ceftazidime, ceftazidime-CLA, cefepime, imipenem, meropenem, ciprofloxacin, and gentamicin.

Growth rates were determined by incubating bacteria in LB broth at 37°C with rotation at 200 rpm. Cells from overnight LB cultures were transferred to fresh LB broth to give an initial OD₆₀₀ of 0.01, and the growth was quantified every 30 min for 6 h based on the OD₆₀₀. The maximum generation time was determined from the logarithms of the values measured in the exponential phase.

Serum bactericidal assay.

Normal human serum, pooled from healthy volunteers, was divided into equal volumes and stored at −70°C before use. The serum bactericidal activity was measured by using the method described by Podschun et al. (44), which was modified by using 10-fold reaction volume and 2-fold bacterial concentration. Briefly, bacteria were grown in brain heart infusion (BHI) broth until an OD₆₀₀ of 0.35 (∼10⁸ cells/ml) was reached. The cultures were washed and then diluted 25-fold using phosphate-buffered saline (PBS). Portions (250 μl) of the cell suspension and 750 μl of pooled human serum were placed into 1.5-ml Eppendorf tubes, mixed, followed by incubation at 37°C. To determine the number of viable bacteria, an aliquot of each bacterial suspension was removed immediately and after every hour of incubation. The number of viable bacteria was determined by dilution and plating on Mueller-Hinton agar for colony counts. The results were expressed as a percentage of the inoculum, and the responses in terms of viable counts were graded from 1 to 6 as Table S2 in the supplemental material (25, 44). Each isolate was classified as highly susceptible (grade 1 or 2), intermediately susceptible (grade 3 or 4), or resistant (grade 5 or 6).

Neutrophil phagocytosis assay.

Phagocytosis was measured by using a standard assay (20). A FACscan (Becton Dickinson Immunocytometry Systems, San Jose, CA) was used to measure the phagocytic rate. The labeling of bacteria with fluorescein isothiocyanate was performed as described by Heinzelmann et al. (20), and the isolation of neutrophils from three healthy volunteers was performed as previously described (8). A mixture of labeled bacteria, the neutrophil suspension, the pooled normal human serum, and PBS was incubated in a shaking water bath at 37°C. The percentage of neutrophils that had phagocytosed bacteria was counted at 5, 15, and 30 min. An unincubated tube served as the 0-min time point. The experimental procedures and fluorescence-activated cell sorting settings have been described previously (32).

Mouse lethality assay.

Pathogen-free, 6- to 8-week-old, male BALB/c mice were obtained from the National Laboratory Animal Center (Taipei, Taiwan) and maintained in the pathogen-free vivarium of the Laboratory Animal Center of NHRI (Miaoli, Taiwan). All animal care procedures and protocols were approved by the Institutional Animal Care and Use Committee of the National Health Research Institute (NHRI-IACUC-096004-A). The tested bacteria were cultured overnight at 37°C in BHI broth and then diluted (1:100) in fresh BHI broth. The culture was incubated until the mid-exponential growth phase, and the cells were then washed once, resuspended in PBS, and adjusted to the desired concentrations according to the OD₆₀₀. The actual concentrations were verified by plating the cells to determine viable counts. Six mice for each group were injected intraperitoneally with 0.1 ml of the cell suspension, and the mice were monitored daily for 14 days to measure the severity of illness and survival.

Statistical analysis.

A two-tailed t test or a log-rank test (for survival analysis) was used for statistical analysis. A P value of <0.05 was considered statistically significant.

Nucleotide sequence accession numbers.

The nucleotide sequences of the K. pneumoniae NVT2001 ompK35 and ompK36 genes, including their flanking regions reported here, have been deposited in GenBank under accession numbers GU461278 and GU461279, respectively.

RESULTS

Sequence analysis of ompK35 and ompK36.

The protein sequence identities of K. pneumoniae NVT2001 OmpK35 and OmpK36 are 56.6 and 80.2% similar to OmpF and OmpC in E. coli K-12 (U00096), respectively. Putative ribosome-binding sites and promoters could be found upstream of the start codons of ompK35 and ompK36, while putative terminators were located 31 and 17 bp downstream of the stop codons of these two genes. (see Fig. S1 in the supplemental material). Similar osmoregulatory elements that have been defined for E. coli ompC or ompF, including OmpR-binding sites, integration host factor, and the micF gene, could also be found in the upstream region of K. pneumoniae ompK35 or ompK36 promoter (see Fig. S1 in the supplemental material). These results indicate that K. pneumoniae NVT2001 ompK35 and ompK36 can be transcribed independently and are likely regulated by osmolarity.

OMP profiles and phenotype comparison.

The allelic exchange strategy was used to delete and repair the porin genes from K. pneumoniae NVT2001S (Table 1 and Fig. 1). The presence or absence of ompK35 and ompK36 in K. pneumoniae NVT2001S and its derived strains was confirmed by PCR (data not shown) and by SDS-PAGE analysis of OMPs (Fig. 2A). In Fig. 2A, densitometric analysis revealed that the expression of the two major porins, OmpK35 and OmpK36, under MHB (lane 1) was at a 1:9 ratio, while that under NB (lane 2) was at a 4:9 ratio. This result indicates that OmpK35 is more abundant in low-osmolarity NB than cultured in high-osmolarity MHB. When we compared the concentrated regions of colonies on blood agar, the ΔompK36 and ΔompK35/36 strains both displayed glistening colony morphology that was distinguishable from that of the NVT2001S and ΔompK35 strains (Fig. 2B). This phenotype could be restored by complementing ompK36 (Fig. 2B).

FIG. 2.

OMP profiles (A) and phenotype comparison (B) of evaluated strains. (A) OMP profiles from strains grown in high-osmolarity MHB (lane 1) and low-osmolarity NB (lanes 2 to 8). Lanes 1 and 2, NVT2001S; lane 3, ΔompK35 strain; lane 4, ΔompK36 strain; lane 5, ΔompK35/36 strain; lane 6, ΔompK35/36::35 strain; lane 7, ΔompK35/36::36 strain; lane 8, ΔompK35/36::35/36 strain. (B) The K. pneumoniae stains on blood agar incubated at 37°C overnight. When the concentrated regions of colonies are compared on blood agar, the phenotype of these OmpK36-loss strains formed more glistening colonies than that of strains without OmpK36 loss.

Antimicrobial susceptibility testing.

The effect of porin loss on antimicrobial resistance was determined by using the broth microdilution test (Table 2). The ΔompK35 strain had comparable MICs to those of the parental NVT2001S strain, which was susceptible to all antibiotics tested, except ampicillin (this strain has intrinsic resistance due to an endogenous copy of bla_SHV in K. pneumoniae). The ΔompK36 strain showed an increase in the MICs of cefazolin, cephalothin, and cefoxitin from susceptible to intermediate resistance. The ΔompK35/36 strain became highly resistant to these cephalosporins, and had an 8- and 16-fold increase in the MICs of meropenem and cefepime, respectively. The increased MICs could be restored to the original levels following complementation (data not shown).

TABLE 2.

MICs of antimicrobial agents against evaluated strains using high-osmolarity MHB for the susceptibility test

Antimicrobial agenta	MIC (μg/ml)b
	No plasmid				CTX-M-14c				CMY-2
	WT	Δ35	Δ36	Δ35/36	WT	Δ35	Δ36	Δ35/36	WT	Δ35	Δ36	Δ35/36
Ampicillin	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32
Piperacillin-TZB	≤4	≤4	≤4	≤4	≤4	≤4	8	8	8	8	32	64
Ticarcillin-CLA2	8	8	8	32	32	32	64	≥512	64	64	≥512	≥512
Cefazolin	1	2	16	32	≥1,024	≥1,024	≥1,024	≥1,024	≥1,024	≥1,024	≥1,024	≥1,024
Cephalothin	≤8	≤8	16	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32	≥32
Cefoxitin	4	4	16	64	4	4	16	64	128	128	≥512	≥512
Ceftriaxone	≤1	≤1	≤1	≤1	32	32	128	≥256	16	16	128	128
Cefpodoxime	≤0.25	≤0.25	≤0.25	0.5	32	32	≥64	≥64	≥64	≥64	≥64	≥64
Cefotaxime	≤0.25	≤0.25	≤0.25	≤0.25	16	16	≥128	≥128	8	8	64	64
Cefotaxime-CLA4	≤0.12	≤0.12	≤0.12	0.25	≤0.12	≤0.12	0.25	0.5	8	8	64	64
Ceftazidime	≤0.25	≤0.25	≤0.25	0.5	1	2	4	4	32	32	64	128
Ceftazidime-CLA4	≤0.12	0.25	0.25	0.5	0.25	0.5	0.5	1	32	32	64	128
Cefepime	0.06	0.06	0.125	1	4	4	16	≥512	0.5	0.5	2	4
Imipenem	0.25	0.25	0.25	0.5	0.25	0.25	0.5	1	0.5	0.5	2	4
Meropenem	0.03	0.03	0.06	0.25	0.06	0.06	0.125	1	0.06	0.06	0.5	2
Ciprofloxacin	≤1	≤1	≤1	≤1	≤1	≤1	≤1	≤1	≤1	≤1	≤1	≤1
Gentamicin	≤4	≤4	≤4	≤4	≤4	≤4	≤4	≤4	≤4	≤4	≤4	≤4

^aTZB, tazobactam with a fixed concentration of 4 μg/ml; CLA₂, clavulanic acid with a fixed concentration of 2 μg/ml; CLA₄, clavulanic acid with a fixed concentration of 4 μg/ml. Strains: WT, K. pneumoniae NVT2001S; Δ35, K. pneumoniae ΔompK35; Δ36, K. pneumoniae ΔompK36; Δ35/36; K. pneumoniae ΔompK35/36.

^bBoldface numbers indicate a significant difference in the MICs of the porin deletion strains and the parental NVT2001S strain of at least 4-fold.

^cA plasmid from a clinical isolate containing bla_CTX-M-14 or bla_CMY-2 was conjugated into NVT2001S and its porin deletion strains.

To evaluate the effect of porin loss combined with the presence of ESBL or AmpC-type β-lactamase on antimicrobial resistance, two plasmids from clinical isolates containing bla_CTX-M-14 (ESBL) or bla_CMY-2 (AmpC) were conjugated into the NVT2001S, ΔompK35, ΔompK36, and ΔompK35/36 strains. The antimicrobial susceptibility profiles for these isolates are presented in Table 2. The NVT2001S strain containing the conjugated plasmid (CTX-M-14 or CMY-2) showed significant (≥4-fold) increases in the MICs of ticarcillin-CLA and the cephalosporins tested (except for the MIC of cefoxitin for the CTX-M-14 strain). When the additional loss of OmpK35 was involved, no significant increase in the MICs was observed. In contrast, the combination of OmpK36 loss resulted in a 2- to 8-fold or more increase in the MICs of penicillin/β-lactamase inhibitor, cephalosporins, and carbapenems. The highest MICs were seen in the ΔompK35/36 strains producing CTX-M-14 or CMY-2. With the loss of OmpK35/36, the CTX-M-14 strain showed significant (4- and 16-fold) decreases in susceptibility to imipenem and meropenem, while CMY-2 strain showed significantly (8- and 32-fold) increased MICs to imipenem and meropenem.

To evaluate the effect of incubation medium on the role of porins in antimicrobial resistance, the susceptibility tests on NVT2001S and its derived mutants were performed in high-osmolarity MHB and low-osmolarity NB (Table 3). Unlike when the strains were cultured in MHB, susceptibility tests using low-osmolarity NB showed that ΔompK35 resulted in a significant (≥4-fold) increased MICs of cefazolin and ceftazidime, whereas ΔompK36 conferred a significantly (4-fold) lower increase in the MIC of cefazolin.

TABLE 3.

Differences in antimicrobial susceptibility profile due to low (NB)- or high (MHB)-osmolarity medium against evaluated strains

Antimicrobial agent	MIC (μg/ml)a
	NVT2001S		ΔompK35 mutant		ΔompK36 mutant		ΔompK35/36 mutant
	MHB	NB	MHB	NB	MHB	NB	MHB	NB
Cefazolin	1	1	2	4	16	4	32	32
Cefoxitin	4	4	4	8	16	8	64	64
Ceftazidime	≤0.25	≤0.25	≤0.25	1	≤0.25	≤0.25	0.5	1

^aBoldface numbers indicate a significant difference in the MICs of the porin deletion strains and the parental NVT2001S strain of at least 4-fold.

Growth rate and serum resistance analyses.

Similar growth rates were observed for the NVT2001S, ΔompK35, and ΔompK36 strains grown in rich medium, LB broth (Fig. 3A1), and the generation time was approximately 27 min for each strain. In contrast, the ΔompK35/36 strain grew significantly (P < 0.05) slower, with a generation time of approximately 30 min. The faster growth rate could be restored by complementing ompK35, ompK36, or both (Fig. 3A2).

FIG. 3.

Effect of OmpK35 and OmpK36 on bacterial growth (A1 and A2), susceptibility to normal human serum (B1 and B2), and phagocytosis (C1 and C2). (A1, A2, B1, and B2) The data illustrated are representative of three independent experiments, all of which showed the similar results. (A1 and A2) Generation time (derived from the three growth curve data between 1 and 3 h) for OmpK35/36-loss strain was about 30 ± 0.5 min, while that for other strains was all about 27 ± 0.5 min. (B1 and B2) Generation times (derived from the three growth curve data between 2 and 6 h) for OmpK35/36-loss strain was about 76 ± 4 min, while that for other strains was all about 57 ± 3 min. (C1 and C2) Each point on the curves represents the means of three independent experiments ± the standard deviation. In time points of 15 and 30 min, the phagocytic resistances of both OmpK36-loss and OmpK35/36-loss strains were significantly (P < 0.05) lower than those of OmpK35-loss and OmpK35/36-non-loss strains.

In the serum complement killing assay, K. pneumoniae NVT2001S and its porin deletion strains all showed grade 6 serum resistance (Fig. 3B1), indicating that loss of OmpK35 and/or OmpK36 did not significantly influence the ability of the strains to resist the bactericidal effects of serum. Although no difference in serum killing was seen, the ΔompK35/36 strain demonstrated a significantly (P < 0.05) slower generation time, approximately 76 min in 75% serum, while the generation times of the NVT2001S, ΔompK35, and ΔompK36 strains were all approximately 56 min (Fig. 3B1). In addition, a slight decrease in growth rate was noted in the NVT2001S and ΔompK35 strains within the first hour in all three independent experiments (Fig. 3B1). These results have been further validated by testing the serum resistance of the complemented strains (Fig. 3B2).

Phagocytosis assay and virulence in mice.

To evaluate the effect of porin loss in the neutrophil phagocytosis assay, neutrophils from three healthy volunteers were isolated and tested with K. pneumoniae NVT2001S and its porin deletion strains. At time points of 15 and 30 min, the ΔompK36 and ΔompK35/36 strains showed significantly (P < 0.05) lower phagocytic resistance than strains NVT2001S and ΔompK35 (Fig. 3C1). No significant difference in phagocytosis was observed between the ΔompK36 and ΔompK35/36 strains or between the NVT2001S and ΔompK35 strains (Fig. 3C1). The decreased resistance to phagocytosis in ΔompK36 and ΔompK35/36 strains was restored in the ΔompK35/36::36 and ΔompK35/36::35/36 complemented strains (Fig. 3C2).

To further assess the role of the porins in virulence, a mouse peritonitis model was used. Mice were inoculated with 300 CFU of each strain (Fig. 4); the LD₅₀ of K. pneumoniae NVT2001S was less than 10 CFU. No significant differences in the survival of mice inoculated with the ΔompK35 and NVT2001S strains were observed (Fig. 4A). When the experiment was repeated at a lower dose (10 CFU), the mice inoculated with the ΔompK35 strain again showed a survival similar to those inoculated with the NVT2001S strain (data not shown). However, unlike the mice inoculated with the ΔompK35 strain, the mice inoculated with the ΔompK36 strain demonstrated a significantly (P < 0.01) longer survival time and slightly decreased death rate than the NVT2001S-inoculated mice. Furthermore, the mice inoculated with the ΔompK35/36 strain showed a >30-fold increase in the LD₅₀ compared to the NVT2001S-inoculated mice (Fig. 4A). The ΔompK35/36::36 and ΔompK35/36::35/36 complemented strains showed virulence comparable to that of the parental NVT2001S strain, while the complemented ΔompK35/36::35 strain demonstrated significantly (P < 0.01) lower virulence than the NVT2001S strain (Fig. 4B). These results suggest that OmpK36 has a more important role than OmpK35 in the pathogenicity of K. pneumoniae.

FIG. 4.

Effect of OmpK35 and/or OmpK36 loss on mice lethality. The survivals after intraperitoneal injection with 3 × 10² CFU of K. pneumoniae were counted over 14 days. The data points represent the percentage of mice surviving in each experimental group over time (n = 12 mice per strain). In experiment A or B, each of the strains was inoculated at different groups of 6 mice at the same day, while the data illustrated are pooled from two independent experiments. (A) Survival curves for K. pneumoniae NVT2001S and its porin deletion strains. (B) Survival curves for K. pneumoniae ΔompK35/36 and its porin-complemented strains. In both experiments, the survival curves for both OmpK36-loss and OmpK35/36-loss strains were significantly (log-rank test, P < 0.01) different from that for the OmpK35/36-non-loss strain, while no significantly different was found between the curves for the OmpK35-loss strain and OmpK35/36-non-loss strain.

DISCUSSION

OmpK35 and OmpK36 provide a channel that allows a wide range of antibiotics to penetrate the K. pneumoniae cell wall; this was initially shown by the cloning and ectopic expression of ompK35 or ompK36 in a strain lacking both OmpK35 and OmpK36 (5, 12, 34, 35). These studies have shown that OmpK35 allows for more efficient penetration of cephalosporin than OmpK36 does and that OmpK35 is not normally expressed in high-osmolarity medium. In the present study, the allelic exchange method was used to directly delete and then restore copies of the ompK35 and/or ompK36 genes from the K. pneumoniae chromosome. In comparison with the parental strain, the strain lacking OmpK36 alone became resistant to cefazolin, cephalothin, and cefoxitin, as measured by antimicrobial susceptibility testing. The additional loss of OmpK35 further increased the MICs, demonstrating that the double-deletion strain is highly resistant to these agents. In addition, the MICs of meropenem and cefepime were significantly increased 8- and 16-fold, respectively, in the double-deletion strain. Significant (≥4-fold) increases in the MICs were observed for the ompK35 single-deletion strain only in low-osmolarity NB. These results suggest that OmpK35 play a relative minor role in antibiotic resistance when cultured in high-osmolarity MHB. The susceptibility data also revealed that meropenem was more active than imipenem in vitro against K. pneumoniae NVT2001S. On the contrary, the three ΔompK35/36 mutants with or without an additional plasmid mediating β-lactamase showed significantly (8- to 32-fold) decreased susceptibility to meropenem and to a lesser (2- to 8-fold) extent to imipenem compared to the corresponding strain without OmpK35/36 loss. This result supports previous observations comparing 65 ESBL-producing clinical strains (13) and suggests that the loss of porins affects resistance to meropenem more than to imipenem. One intriguing finding is that OmpK35/36 loss, with or without an additional plasmid bearing β-lactamase, could significantly (8- to 128-fold or more) decrease susceptibility to cefepime, a late generation cephalosporin for both ESBL and AmpC-type β-lactamase-producing bacteria. To our knowledge, this observation has not been previously reported.

In the present study, the loss of OmpK35 conferred more antibiotic resistance when the strains were cultured in low-osmolarity medium compared to high-osmolarity medium, while the inverse result was seen after the loss of OmpK36. This result is in agreement with the theory that OmpF is preferentially expressed at low osmolarity in E. coli, whereas OmpC is most abundant at high osmolarity (45). Because the susceptibility results showed that the disruption of ompK36 resulted in greater drug resistance compared to the ompK35 mutant in high-osmolarity MHB, a medium more similar to human body fluid, the emergence of clinical antibiotic resistance due to OmpK36 loss is predicted.

In K. pneumoniae, most pathogenesis studies have been devoted to the roles of the capsule and LPS, while the porins have garnered less attention. Our previous study has shown that the inactivation of ompK36 by insertion results in a 100-fold increase in the LD₅₀ in a mouse peritonitis model for a virulent strain of serotype K1 K. pneumoniae NVT1001, (9). In the present study, nonpolar in-frame deletion in ompK36 could also significantly (P < 0.01) decreased the virulence of the highly virulent strain of serotype K2 K. pneumoniae NVT2001S. The increased susceptibility to neutrophil phagocytosis was observed in both of these ompK36-deficient strains. Moreover, this work further compares the colony morphology on blood agar, demonstrating that the loss of OmpK36 can modify the surface structure of K. pneumoniae. These results indicate that the modification may alter the binding of neutrophils, leading to increased susceptibility to phagocytosis. A previous study has shown that K. pneumoniae OmpK36 can directly activate the classical complement pathway by binding to its first component, C1q, in vivo (2). However, our ompK36 deletion strains showed the same serum resistance as the parental strain, while the similar result could also be obtained for the study of the serum-susceptible strain K. pneumoniae NVT1001 (9). Whether this result is due to the C1q-binding sites were covered by the LPS core (3) requires further study.

Porins are important for bacterial survival because of their role in the exchange of substances, including nutrients and toxic metabolites. Previous studies have shown that few ESBL-producing K. pneumoniae clinical isolates lack both OmpK35 and OmpK36 (21), although strains lacking both porins exhibit higher antibiotic resistance than strains that express either only one or both porins (13, 22). In the present study, the deletion of both ompK35 and ompK36 in K. pneumoniae NVT2001S showed a >30-fold increase in the LD₅₀ in a mouse peritonitis model, while its growth rate was significantly (P < 0.05) reduced when cultured in LB broth and in 75% normal human serum. These results suggest that the loss of these two major porins could significantly affect the metabolic fitness of K. pneumoniae, and the slower growth rate would likely cause it to be eliminated in vivo by the immune system much more easily. In comparison to the parental strain, the ΔompK35 strain had comparable MICs and virulence; however, the ΔompK36 strain demonstrated increased MICs and a reduced virulence, while the ΔompK35/36 strain presented the highest MICs and the weakest virulence of these strains. These results suggest that porin deficiency in K. pneumoniae could increase antimicrobial resistance but decrease virulence at the same time.

The production of E. coli OmpF and OmpC is regulated by many environmental factors, such as osmolarity and temperature, with high osmolarity or temperature favoring the expression of OmpC over OmpF (45). Regulation in response to osmolarity is mediated by the OmpR-binding sites, integration host factor, and the micF gene (47, 48, 55), while the response to temperature is mediated by the micF gene (4). Previous studies have proposed that under low-osmolarity lake water at ambient temperature, OmpF is the major porin expressed; in contrast, under high-osmolarity animal intestine at 37°C, OmpC is the predominant porin expressed (38, 45). In the present study, the putative promoters and regulatory elements of ompK35 or ompK36 were similar to that of E. coli ompF or ompC (26, 47, 48, 37, 51, 55). Our results of the OMP profiling and antimicrobial susceptibility testing also showed that the osmoregulation of OmpK35 or OmpK36 followed to that found for OmpF or OmpC. Unlike OmpK36-loss strains, the deletion of ompK35 alone did not significantly affect the virulence of K. pneumoniae NVT2001S, and this may be because ompK35 expression was already repressed in vivo. However, the antimicrobial resistance and virulence of the ΔompK36 strain were significantly affected by the further deletion of ompK35. These results suggest that OmpK35 and OmpK36 both play dual roles in K. pneumoniae infection.