Carbapenem-Resistant Klebsiella pneumoniae Exhibit Variability in Capsular Polysaccharide and Capsule Associated Virulence Traits

Abstract

Background. Novel therapies are urgently needed to treat carbapenem-resistant Klebsiella pneumoniae (CR-Kp)-mediated infection, which constitute a major health threat in the United States. In order to assess if it is feasible to develop anticapsular antibodies as a potential novel therapy, it is crucial to first systematically characterize capsular polysaccharide (CPS) and virulence traits in these strains.

Methods. Forty CR-Kp were genotyped by pulsed field gel electrophoresis, multilocus sequence typing (MLST), and molecular capsule typing (C-patterns and wzi sequencing). Their biofilm formation, serum resistance, macrophage-mediated killing, and virulence in Galleria mellonella were compared. MAb (1C9) was generated by co-immunization with 2 CPSs, and cross-reactivity was investigated.

Results. MLST assigned 80% of CR-Kp isolates to the ST258-clone. Molecular capsule typing identified new C-patterns, including C200/wzi-154, which was widely represented and associated with bla_KPC-3-bearing strains. Heterogeneity was detected in biofilm formation and macrophage-mediated killing. Differences in serum resistance correlated with virulence in G. mellonella. ST258 strains carrying bla_KPC-3 were less virulent than those with bla_KPC-2. MAb 1C9 cross-reacted with 58% of CR-Kp CPSs.

Conclusions. CR-Kp ST258 strains exhibit variability of virulence-associated traits. Differences were associated with the type of KPC gene and CPS. Identification of cross-reacting anti-CPS mAbs encourages their development as adjunctive therapy.

In the past decades carbapenem-resistant Klebsiella pneumoniae (CR-Kp) strains have emerged in the United States since 2001 and worldwide [1]. Currently, the most common carbapenemase in the United States is K. pneumoniae carbapenemase (KPC), an Ambler molecular class A enzyme that facilitates hydrolysis of a broad variety of β-lactams. Recent CDC surveillance data estimate that the prevalence of CR-Kp in the United States among healthcare-associated infections increased from 1.6% in 2001 to 10.4% in 2011 [2]. The majority of clinical CR-Kp isolates in the United States are of MLST–defined clonal background ST258 that carry KPCs (bla_KPC-2 or bla_KPC-3) [3]. CR-Kp infections have high mortality rates (40%–50%) and result in increased treatment and hospitalization costs [4, 5]. With no novel antimicrobials for emerging CR-Kp in sight, efforts to explore alternative treatment options and prevention of global dissemination are warranted [6].

One of the main virulence factors of K. pneumoniae is its capsular polysaccharide (CPS) [7]. CPS is expressed in vivo, promotes biofilm formation, and exerts an anti-opsonic effect, all of which evade the host immune response. Strategies targeting the CPS have been successful both in vaccine development as well as passive immunotherapy for other encapsulated pathogens. For K. pneumoniae protective efficacy of anticapsular antibodies has been demonstrated in animal models, further supporting efforts to develop antibodies as adjunctive therapy [8]. CPS genes in K. pneumoniae strains are chromosomally encoded and clustered in the cps genomic locus [9, 10]. Over 77 capsular (K) serotypes have been described. However, strains of ST258 have not been extensively characterized for their K-serotype or molecular methods of cps cluster analysis such as C-pattern [10] and wzi sequencing [11]. In this study we characterized 40 CR-Kp strains from the Bronx with respect to their CPS, biofilm formation, resistance to serum and macrophage killing, as well as virulence in a Galleria mellonella and mouse model. This study is the first to our knowledge to document significant CPS-associated variability including novel C-patterns and wzi alleles among CR-Kp strains of the ST258 clone. Despite variability, cross-reactive antibodies could be generated. In addition, significant variability was documented with respect to virulence-associated traits. The implications of these findings for efforts of developing anti-capsular antibodies are discussed.

MATERIAL AND METHODS

K. pneumoniae Strains

CR-Kp strains were collected from inpatients at Montefiore Medical Center (MMC) in Bronx, New York, that presented with CR-Kp bacteremia between December 2010 and November 2012. Retrospective chart review of patient data was performed with IRB approval. For comparison, 8 carbapenem-susceptible K. pneumoniae strains (CS-Kp) were collected during the same time period. K. pneumoniae was cultured in Luria-Bertani (LB) broth or agar plates at 30°C or 37°C. Hypermucoviscosity phenotype was determined with the string test as described elsewhere [12].

Determination of Genetic Relatedness

Pulsed-field gel electrophoresis (PFGE) typing of K. pneumoniae isolates was performed according to the PulseNet protocol (http://www.cdc.gov/pulsenet/protocols.htm) analyzing XbaI restriction enzyme patterns with a CHEF-DR II system (Bio-Rad, USA). MLST was carried out following the guidelines of the Institut Pasteur K. pneumoniae MLST Database (www.pasteur.fr/mlst) [13]. Novel wzi alleles were incorporated into the K. pneumoniae sequence typing database at bigsdb.web.pasteur.fr.

CPS Typing and Glycosyl Composition Analyses

HincII restriction enzyme pattern of the polymerase chain reaction (PCR)-amplified cps cluster, C-typing [10], and typing by wzi sequencing, which is strongly associated with K-type [11] was performed as described elsewhere [10, 11]. K-serotyping was performed at Statens Serum Institute (Copenhagen, Denmark). CPS was purified as described elsewhere [14, 15] with minor modifications (Supplementary methods). Carbohydrate composition and linkage analysis was performed at the Complex Carbohydrate Research Center (Athens, GA) as described elsewhere [16, 17].

Biofilm Formation (BF) Assays

BF assays were performed at 37°C as described elsewhere [18, 19] (Supplementary Methods). Data obtained were used to classify the strains as high (OD > 0.6), median (OD ≤ 0.6 and >0.4), or low producers (OD ≤ 0.4).

Serum Resistance Assays

In vitro virulence assays were performed as published [12, 20] and described (Supplementary Methods). K. pneumoniae strains were categorized into 3 different groups: no serum resistance, meaning unable to grow (survival ratio ≤1); moderate serum resistance, meaning those strains with moderate growth (survival ratio >1 and ≤5); or high serum resistance, which included K. pneumoniae strains that exhibited high rate of replication (survival ratio >5).

Macrophage-mediated Killing

In vitro killing of CR-Kp strains was investigated in the J774.16 macrophage cell line as published [21] and described (Supplementary methods). Intracellular killing was based on the decrease of viable bacteria 30 minutes after initial coincubation relative to time 0.

G. mellonella and Murine Infection Models

Virulence of CR-Kp strains was assessed in G. mellonella by injecting 20 larvae with 10⁴ CFU of K. pneumoniae in 10 µL phosphate-buffered saline (PBS). Control animals were injected with PBS only. Larvae were kept at 37°C in the dark on sterile Petri plates, and survival was assessed for 3 weeks. To compare in vivo replication dynamics of CR-Kp strains, 5 µL of hemolymph were pooled from 20 larvae at different time points. Colony-forming units (CFU) were calculated from 50 µL of hemolymph [22]. Intratracheal and intravenous infection in mice were performed as described elsewhere [23] (Supplementary methods) and approved by the Animal Care and Use Committee.

MAb Generation and Agglutination Assays

Monoclonal antibody (MAb) 1C9 to CPS C200 and C186 was generated by immunization with CPS in complete Freund adjuvant (CFA) followed by booster of CPS in incomplete Freund adjuvant (IFA) of BALB/c mice. Fusion and cloning was performed as described elsewhere [24]. Agglutination was carried out as previously described on glass slides [8].

Statistical Analysis

Data are presented in mean ± standard deviation or median, range. Differences between patient data were analyzed by a Fisher exact test. Survival data were analyzed with a log–rank test. Statistical tests were performed with GraphPad Prism 6 for Mac.

RESULTS

Patient Characteristics

Forty CR-Kp strains were derived from blood of 38 septicemic patients, who were hospitalized at MMC. Patients were elderly with average age of 64 years, and 71% resided in nursing homes. Mortality was high (52%; Table 1). Characteristics of survivors and nonsurvivors were comparable except for percent of patients identified as white and with diabetes. Days until effective treatment was started were variable in both groups and ranged from −33 to 18 days. Four CR-Kp infected patients never received effective treatment prior to death. Antibiotic susceptibility of the strains was performed by the microbiology laboratory, which identified CR in 40 strains by standard CLSI laboratory practice [25], 87% were also resistant to ciprofloxacin, 53% to amikacin, 36% to gentamicin, and 16% to polymyxin B.

Table 1.

Characteristics of Patients Infected With CR-Klebsiella pneumoniae

Characteristic	Total (n = 38)	Survivors (n = 18)	Nonsurvivors (n = 20)
Age in years, mean ± SD	64 ± 16	61.5 ± 16.6	67 ± 15
Gender, male, n (%)	20 (52.6)	10 (58.5)	10 (50)
Race, black, n (%)*	13 (34)	9 (50)	4 (20)
Race, white, n (%)*	8 (21)	1 (5.5)	8 (40)
Nursing home residency (%)	27 (71)	14 (77.8)	13 (65)
Hospital day blood culture positive for CR-KP, median (range)	8 (0–80)	14 (0–80)	3 (0–46)
Abx treatment in prior 6 mo, n (%)	36 (94.7)	16 (88.9)	20 (100)
Day after CR-KP diagnosis patient received effective treatment, Median (range)	0 (−33,18)	1 (−4, 8)	−0.5 (−33, 18)
Day patient died or was discharged after positive culture, median (range)	17.5 (1–112)	25 (2–112)	14 (1–72)
Hospital length of stay, mean ± SD	39 ± 31	42 ± 36	36 ± 28
ICU admission, n (%)	15 (39.5)	6 (33.3)	9 (45)
ICU length of stay, mean ± SD	20 ± 17	20 ± 6	21 ± 21
Diabetes mellitus**	23 (60.5)	15 (83.3)	8 (40)
CAD	14 (36.8)	7 (37.9)	7 (35)
COPD	18 (47.4)	7 (38.9)	11 (55)
Liver disease	4 (10.5)	1 (5.3)	3 (15)
Chronic kidney disease	4 (10.5)	2 (11.1)	2 (10)
Solid organ transplant	1 (2.6)	0 (0)	1 (5)
Surgery (%)	10 (26.3)	3 (16.7)	7 (35)
Central venous catheter	34 (84)	14 (77.7)	18 (90)
Mechanical ventilation	27 (71)	10 (55.5)	17 (85)
Immunocompromised	4 (10.5)	2 (11.1)	2 (10)
Dialysis	4 (10.5)	2 (11.1)	2 (10)
Ciprofloxacin	38 (97.4)	17 (94.4)	20 (100)
Amikacin	20 (52.6)	10 (55.5)	10 (50)
Gentamycin	17 (44.7)	8 (44.4)	9 (45)
Polymyxin B	8 (21)	5 (27.7)	3 (15)

Abbreviations: CAD, community-acquired pneumonia; COPD,chronic obstructive pulmonary disease; CR-Kp, carbapenem-resistant Klebsiella pneumoniae; ICU, intensive care unit; SD, standard deviation.

*P-value < .05; **P-value < .01.

Strain Typing

MLST and PFGE were performed on the 40 CR-Kp strains and 8 concomitantly collected CS-Kp strains. MLST identified coinfection with distinct isolates in 4 patients. Two patients were infected with 2 distinct CR-Kp isolates, and 2 patients with a CR- and CS-Kp isolate. Most (32/40, 80%) of CR-Kp strains belonged to ST258. Other sequence types were identified including ST37 (2/40, 5%), ST502 (1/40, 2.5%), and ST14 (1/40, 2.5%; Supplementary Table 1) and were unrelated. Also, 3 new types, ST1403 (2/40), ST1404 (1/40), and ST1406 (1/40) were identified of which the latter differs from ST258 only in the phoE allele. High (76%) similarity among PFGE patterns confirmed relatedness of CR-Kp strains assigned to the ST258 cluster (Figure 1). The other sequence types clustered into different PFGE profiles. Of the 8 CS-Kp strains, 3 belonged to ST258 (33%), and ST23, ST111, ST15, ST502, and the new ST1405 each included 1 isolate. Sequence analysis identified bla_KPC-2 in 20/40 strains and the bla_KPC-3 variant in 17/40 isolates. No significant correlation between type of bla_KPC gene and patient outcome was found.

Figure 1.

PFGE analysis of the Klebsiella pneumoniae strains. Clustering of the K. pneumoniae isolates on basis to their PFGE profiles after digestion with XbaI. Percent similarities are shown above the dendogram. Strain number, sequence types, and profile names are shown on the right of the PFGE gel. Abbreviation: PFGE, pulsed-field gel electrophoresis.

Capsule Typing (C-typing)

Molecular typing of the cps gene cluster was performed by restriction enzyme digestion of the polymerase chain reaction (PCR)-amplified capsule gene region [10] and by wzi sequencing [11]. C-typing identified 17 different C-patterns (Figure 2, Supplementary Table 1). The largest cluster included 11 CR-Kp strains that exhibited a novel C-pattern, C200, which is related to C14a, the C-type of reference strain 138 with a K14 serotype [10]. Five CR-Kp strains with C-patterns C193, C194, C195, C196, and C197 had a highly similar C-pattern to C200 (75%). Strains with these C-patterns and C200 contain in common the same and newly identified wzi154 allele. In addition, 10 unique novel C-patterns were identified in this study and now named C186 (wzi50), C184 (wzi154), C189 (wzi50), C102-like (wzi29), C190 (wzi29), C191 (wzi29), C192 (wzi29), C201 and C202 (both wzi150). Five other previously described C-types were identified in 9 other CR-Kp strains. They include C102-like, which could be further differentiated by wzi sequencing (wzi29, wzi153, wzi150), C23a (wzi83), C51a (wzi50), C15a (wzi50), and C16a (wzi16), respectively. Even among CS-Kp strains 2 novel C-types C189 (wzi50), C198 (wzi151) were identified in addition to the already described C200/wzi154, C102-like/wzi29, C15a/wzi50, C51a/wzi29, and C1a/wzi1. Classical K-serotyping was performed in 4 CR-Kp strains. One C200 strain was identified as K34, whereas the other C200 and a C186 were untypable, and a C102 strain was typed as K48. As expected, a hypermucoviscous capsule (string test positive) was documented in a CS-Kp strain (no. 20, Supplementary Table 1) with a C1a/wzi1 molecular serotype that belongs to the hypervirulent clone ST23 of serotype K1 [26]. Interestingly, 3 CR-Kp strains, 2 with a C200 C-pattern (nos. 2 and 40, Supplementary Table 1) and one with a C15a C-pattern (no. 9) also expressed a hypermucoviscous capsule phenotype. Furthermore, this study found a significant association of C200 or related C-patterns and the presence of bla_KPC-3 gene (11/16 vs 4/20 Fisher test, P-value .0021). Taken together substantial variability in CPS was documented in CR-Kp strains, including ST258 strains. In addition, the lack of typability of some isolates with molecular CPS variants highlights the limitations of traditional K-typing.

Figure 2.

C-patterns of the Klebsiella pneumoniae strains. Clustering of the C-types was achieved using the UPGMA method based on Dice similarity coefficient. Percent similarities are shown above the dendrogram; the scale bar in base pairs indicates the size of DNA fragments. Abbreviation: UPGMA, unweighted group method with arithmetic mean.

Agglutination of Klebsiella by Anti-CPS IgM

Mab (1C9), an immunoglobulin M (IgM), was produced by mouse immunized with 2 purified CPSs (C200, the one typed as K34, and C186) for which monosaccharide composition analysis demonstrated distinct sugar content (Table 2). Enzyme-linked immunosorbent assay (ELISA) confirmed that 1C9 binds with high affinity to both C200 and C186 and not to bovine serum albumin (data not shown). To assess the cross-reactivity of mAb- 1C9 its ability to agglutinate CR-Kp strains expressing different CPSs was investigated. MAb 1C9 agglutinated 7 of the 14 distinct C-types and 26 of the 40 tested CR-Kp strains. Specifically, agglutination was observed in all (16/16) strains expressing C200 and related C-patterns, and in all 3 CR-Kp strains with C186 C-pattern. In addition the mAb-1C9 agglutinated all C23a strains, C189, C192, and C184 and C102-like CR-Kp strains (Figure 3). Seven C-types were not agglutinated. Agglutination was consistent for all isolates of a specific C-pattern except the C102a group, of which 1 isolate agglutinated and 5 did not.

Table 2.

CPS Sugar Composition of Strains 34 (C200) and 38 (C186)

Sugar Composition	No. 34 C200, %	No. 38 C186, %
Galactose	37	4
Rhamnose	38
Glucose		5
Galacturonic acid	11
Glucoronic acid		2
Mannose		68

Abbreviation: CPS, capsular polysaccharide.

Figure 3.

Effect of the 1C9 mAb IgM on the agglutination of clinical isolates. A, 10 mg of mAb were added to cells and incubated 1 hour. Positive agglutination is shown of strain no. 11, part of the cluster C200, and negative effect is shown for strain no. 52, C201. C-patterns having a positive or negative agglutination are shown. B, Relative presence of each of the C-types found. Each C-type is presented in one color. C, Relative abundance of the agglutinated C-types. Purple means strains agglutinated after the mAb incubation, whereas magenta means some of the isolates agglutinated. Nonagglutinated strains have their own color. Abbreviations: IgM, immunoglobulin M; mAb, monoclonal antibody.

Phenotypic Traits Analyses

Planktonic Growth and Biofilm Formation

Growth of K. pneumoniae strains in suspension varied only slightly (median doubling time of 23.7 minutes ranging from 16.7 to 33.9 minutes, Supplementary Table 1); however, biofilm formation varied considerably (Figure 4A and Supplementary Table 1). The highest biofilm producer was CS-Kp#20 (ST23), which exhibited the C1a/wzi1-K1 molecular serotype and hypermucoviscous phenotype, followed by CS-Kp no. 32 (ST15) and CR-Kp no. 34 (ST258). Biofilm production did not correlate with doubling time of bacteria in suspension.

Figure 4.

Phenotypic characterization of CR-Kp ST258 cluster strains. A, Levels of biofilm production onto polystyrene plates measured by CV quantity. Line represents the average biofilm production, A540 ≅ 0.5. Cyan denotes low BF, red high producers, and gray average BF. B, Macrophage-dependent killing of ST258 cluster strains. Viable phagocytized CR-Kp cells were measured after incubation with J766.16 macrophages (10:1 ratio KP:MO). Cyan denotes strains with macrophage killing above the media (53.59%) and red below the media. C, Growth of CR-Kp strains in the presence of NHS. Levels of viable CFU of the 3 different group-types of Kp strains after incubation for 2 hours with NHS were assessed. Levels of viable cells relative to the number of CFU at time 0′ are shown. D, Virulence of Kp strains on G. mellonella. Survival rates of 20 G. mellonella individuals infected with 10⁴ CR-Kp ST258 CFU were assessed. Median days are displayed as a line within the box; maximum and minimum of the boxes display the 25th and 75th percentiles. Cyan represents avirulence, grey average virulence and red is high virulence. E, Kp replication rates within the G. mellonella during 8 hours postinfection. F, Correlation of G. mellonella survival and human serum resistance. Abbreviations: CFU, colony-forming unit; CR-Kp, carbapenem-resistant Klebsiella pneumonia.

Killing Experiments

CPS shields K. pneumoniae bacteria from phagocytosis and killing by phagocytic cells. Marked differences among K. pneumoniae strains with respect to resistance to killing by the J744.2 macrophage cell line (Figure 4B and Supplementary Table 1) were found. Although killing of individual strains was highly variable and ranged from 26% to 86%, no consistent correlation with a C-pattern, wzi allele, or clonal background was established.

Serum Resistance Assays

Resistance to human serum was compared, as it constitutes an important virulence trait that allows K. pneumoniae to persist in vivo. Again, considerable variability in serum resistance was documented for CR-Kp strains (Supplementary Table 1). Among CR-Kp strains, 37.5% were highly resistant, 52.5% moderate, and 10% not resistant to serum. Growth dynamics in human serum confirmed impaired growth of CR-Kp strains classified as highly serum resistant (Figure 4C). Furthermore, significant association of bla_KPC-2-bearing strains (non-C200 and related) and high serum resistance was established (unpaired t-test, P-value .05).

Virulence in In Vivo Infection Models

Virulence of CR-Kp strains was compared in the moth G. mellonella infection model and for selected strains also tested in a murine tracheal and blood stream infection (BSI) model. Infection with 10⁴ CFU from CR-Kp strains resulted in profound variability of G. mellonella waxworm survival ranging from avirulence to death within days (Figure 4D, Supplementary Table 1). Specifically, 9/40 of the strains were avirulent at that dose in the waxworm model (median survival 13–17 days, phosphate-buffered saline [PBS] control 14 days), whereas 12/40 CR-Kp strains induced rapid death (median survival 1–2 days). Interestingly, strains exhibiting high virulence in G. mellonella and high resistance to human serum more frequently exhibit wzi50, wzi150 alleles compared to strains exhibiting low resistance and less virulence (9/14 vs 2/34, Fisher test P-value <.0001). In addition, the bla_KPC-2-bearing K. pneumoniae strains were significantly more virulent than bla_KPC-3 bearing strains (χ², P-value .0186). Consequently, G. mellonella survival also correlated with the strain's ability to grow in serum (Figure 3F; r = −0.7, P-value < .001 by Pearson) and in vivo replication dynamics of 6 strains with low, medium, or high virulence (Figure 4E). Of note is that neither macrophage-mediated killing nor biofilm predicted G. mellonella virulence or serum resistance. Virulence of strain CR-Kp no. 34 (C200), CR-Kp no. 38 (C186), CR-Kp no. 39 (C200), CR-Kp no. 26 (C200), CR-Kp no. 40 (C200), and CS-Kp no. 20 (C1a) was also tested in 2 murine infection models. Only CS-Kp no. 20 could kill mice (1-day median survival), consistent with it belonging to virulent clone CC23-K1, whereas none of the tested ST258 strains, including the hypermucoviscous CR-Kp no. 40, were virulent in mice even when high inocula were used (10⁸ CFU).

DISCUSSION

This study is the most comprehensive characterization of CPS variability in CR-Kp strains and the first systematic comparison to our knowledge of virulence-associated traits in these strains. Most importantly, despite extensive CPS variability, cross-reactive antibodies could be generated with immunization of 2 distinct CPS. Despite enormous progress in developing antibodies for oncology, rheumatology, and transplant only 2 anti-infective monoclonal antibodies (mAbs) are licensed to date. One major challenge in generating anti-infective Abs is the diversity of pathogen populations, which has to be investigated prior to developing adjunctive therapies with Abs.

The majority of CR-Kp mediated BSIs in this study were nosocomial infections in older patients with comorbidities. Their high mortality rate could not be attributed to ineffective treatment and was similar to other reports [27]. MLST assigned 80% of the CR-Kp-strains to ST258 [3] and also identified ST37, ST14, and ST502, which cause K. pneumoniae infections worldwide [28, 29]. In addition, MLST and CPS-typing demonstrate coinfection with different K. pneumoniae strains in some patients, which is relevant as it may promote the selection of nontargeted K. pneumoniae strains, and horizontal gene transfer (HGT) resulting in the possible emergence of novel CR-Kp clones. Despite this restricted clonal background, high variability was evident by PFGE as well as CPS-typing and consistent with estimates that only 65%–75% of predicted protein-coding genes were shared among ST258 isolates [30–32]. Whole genome sequencing has recently been applied [33] in a CR-Kp outbreak. Eventually, comparison of different CR-Kp strains will better delineate genomic microevolution in the ST258 clone.

Pathogen variability constitutes a challenge to vaccination strategy if targeted antigens are affected. K. pneumoniae strains are encapsulated, and over 77 distinct K-serotypes have been described [34]. K-serotyping is expensive, inconsistent, and only available in a few reference laboratories, which explains the paucity of K-serotyping data on CR-Kp strains. Our data validate earlier studies that novel molecular CPS-typing methods, PCR-RFLP of the cps region and wzi-sequencing, are more discriminatory than K-serotyping [10, 11]. Among 40 CR-Kp strains, 11 novel C-patterns were identified. The largest cluster among the ST258 strains exhibited a novel C-pattern that was defined as C200 and 5 related ones, C200-like (16/33), with a banding pattern similar to the previously described C14a pattern. Results from wzi sequencing [11] correlated with C-types. In this study novel alleles were found that never had formal K-typing assignments. Glycosyl-composition analysis of the C200 CPS (strain no. 34), exhibited monosaccharide components similar to the only 2 CPSs from CR-Kp strains described elsewhere [35]. Those 2 CPSs showed cross-reactivity with anti-K34 poly-serum, which also bound to the tested CPS-C200 of K. pneumoniae strain no. 34, whereas a different CPS-C200 (K. pneumoniae strain no. 26) was untypable, highlighting CPS heterogeneity among ST258 strains (and further underlining the limitations of K-serotyping). Variation of K-types among strains from a single ST has been attributed to HGT of cps cluster region [36]. However, further deep sequencing in CR-Kp strains of the cps cluster will be required to establish the precise molecular mechanisms leading to this extensive cps heterogeneity. Of note is the strong association of C200 and related C-patterns with the presence of bla_KPC-3 in these CR-Kp strains, suggesting clonal dissemination of strains with these characteristics.

CPS heterogeneity in the internationally emerging ST258 lineage raises the concern that passive immunotherapy may be challenging. Our agglutination data, however, is encouraging. We demonstrate significant cross-reactivity of IgM antibody 1C9, which was generated by co-immunization with CPS-C200 and CPS-C186. MAb 1C9 agglutinated 58% of the CR-Kp strains, including C200-like and C186 CR-Kp strains and 5 other unrelated capsule types. Studies that reported cross-reactive Abs in serum of volunteers vaccinated with CPS derived from one K. pneumoniae strain [37, 38] or with 23-valent Pneumovax [39], further support the notion that hybridomas producing cross-reactive Abs can be detected. Future immunization strategies could be optimized and include conjugated CPS as well as booster regimens with different CPS to increase the pool of hybridomas that produce cross-reactive Abs. In addition, more extensive knowledge about the precise sugar components in individual CPS exhibiting distinct C-patterns may help to predict the chance of cross-reactivity based on the degree of carbohydrate content similarity.

Given the detection of novel C-patterns and our intention to develop anti-CPS Abs as adjunctive therapy, we investigated if differences in CPS would also show differences in virulence traits, especially those potentially affected by anti-CPS mAb-binding. Such assays also constitute useful tests to screen for efficacy of anti-CPS Abs.

Biofilm has been studied in hypervirulent K. pneumoniae strains with a hypermucoviscous K1 or K2 CPS [40, 41]. In this study only 1 of 4 hypermucoviscous strains was identified as ST23-K1. Biofilm formation was variable among all tested CR-Kp strains but no correlation with C-type, serum resistance, Galleria lethality or resistance to macrophage killing was documented. This lack of correlation is not unexpected as increased capsule production in K1 K. pneumoniae strains did not affect virulence in mice [42], and disruption of hypermucoviscous-associated genes did not alter resistance to human serum [43].

Macrophages carry out important defense functions by phagocytizing and killing K. pneumoniae, which is mediated by antimicrobial peptides, reactive nitrogen intermediates, and ROIs [44–46]. An earlier study with 10 ST258 CR-Kp strains reported that all were killed by macrophages [47]. In contrast, we found variable killing rates ranging from 25% to 86% but no relationship to the C-pattern or other virulence traits of the CR-Kp strains. It is possible that other factors differ among ST258 strains such as lipopolysaccharide and outer membrane proteins [48].

Galleria has been used to assess virulence of infectious microorganisms including K. pneumoniae [49]. Interestingly, resistance to human serum and survival in the Galleria host correlated in our CR-Kp strains. However, unlike K1 and K2 strains [49], for ST258 strains, Galleria mortality did not predict high virulence in mice. All 5 tested ST258 strains, which exhibited different degrees of virulence in Galleria, were avirulent in the murine infection models. This study lacked the power to identify a clear correlation of CPS-type, serum resistance and virulence, but the data suggest that strains typed with the wzi alleles 50 and 150 are the most virulent in Galleria and the most resistant to serum, whereas bla_KPC-3-bearing strains, which exhibit C200-like and wzi154 allele CPS are consistently less virulent in Galleria and resistant to serum. This could be related to the CPS of these strains or to other inherent factors of the bla_KPC-3-harboring isolates. Isogenic comparison and bla_KPC-3 plasmid cure will have to be performed to determine the contribution of this plasmid to virulence.

Based on our data using opsonophagocytic killing assays (OPKA) to determine if mAbs promote phagocyte CR-Kp killing in addition to CPS agglutination assays represents a reasonable screening strategy for efficacy of CPS-specific Abs. For pneumovax vaccine-elicited sera that promote Streptococcus pneumoniae killing are good surrogate markers for vaccine response and efficacy [50] and OPKAs are used to assess functional responses to pneumococcal conjugate vaccines in clinical studies. In addition, virulence of CR-Kp in Galleria correlates with serum resistance both of which can be easily quantified to compare antibody efficacy.

One major concern is that CR-Kp strains are avirulent in mice. The lack of a good animal model is likely the reason for paucity on research examining the pathogenicity of CR-Kp, despite its high clinical relevance. Recent data suggest that virulence can be achieved by induction of neutropenia in mice [47]. Going forward, efforts have to be undertaken to further optimize this murine model so that Ab-efficacy can be studied. We conclude from our data that cocktails of several cross-reactive mAbs will have to be generated to facilitate broad coverage of all CR-Kp strains. Although this constitutes a major challenge, antibody technology and FDA regulations have been rapidly evolving, and consequently several antibody cocktails against other infectious agents are already in advanced clinical testing.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.