Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Population Genomic Analysis of 1,777 Extended-Spectrum Beta-Lactamase-Producing Klebsiella pneumoniae Isolates, Houston, Texas: Unexpected Abundance of Clonal Group 307

Population Genomic Analysis of 1,777 Extended-Spectrum Beta-Lactamase-Producing Klebsiella... RESEARCH ARTICLE crossm Population Genomic Analysis of 1,777 Extended-Spectrum Beta-Lactamase- Producing Klebsiella pneumoniae Isolates, Houston, Texas: Unexpected Abundance of Clonal Group 307 a,b a,b a a S. Wesley Long, Randall J. Olsen, Todd N. Eagar, Stephen B. Beres, a c,d c,d c,d Picheng Zhao, James J. Davis, Thomas Brettin, Fangfang Xia, a,b James M. Musser Center for Molecular and Translational Human Infectious Diseases Research, Department of Pathology and Genomic Medicine, Houston Methodist Research Institute and Houston Methodist Hospital, Houston, Texas, USA ; Department of Pathology and Laboratory Medicine, Weill Cornell Medical College, New York, New York, b c USA ; Computing, Environment and Life Sciences, Argonne National Laboratory, Argonne, Illinois, USA ; Computation Institute, University of Chicago, Chicago, Illinois, USA ABSTRACT Klebsiella pneumoniae is a major human pathogen responsible for high Received 1 April 2017 Accepted 18 April morbidity and mortality rates. The emergence and spread of strains resistant to mul- 2017 Published 16 May 2017 tiple antimicrobial agents and documented large nosocomial outbreaks are espe- Citation Long SW, Olsen RJ, Eagar TN, Beres SB, cially concerning. To develop new therapeutic strategies for K. pneumoniae,itisim- Zhao P, Davis JJ, Brettin T, Xia F, Musser JM. 2017. Population genomic analysis of 1,777 perative to understand the population genomic structure of strains causing human extended-spectrum beta-lactamase-producing infections. To address this knowledge gap, we sequenced the genomes of 1,777 Klebsiella pneumoniae isolates, Houston, Texas: unexpected abundance of clonal group 307. extended-spectrum beta-lactamase-producing K. pneumoniae strains cultured from mBio 8:e00489-17. https://doi.org/10.1128/ patients in the 2,000-bed Houston Methodist Hospital system between September mBio.00489-17. 2011 and May 2015, representing a comprehensive, population-based strain sample. Editor Steven J. Projan, MedImmune Strains of largely uncharacterized clonal group 307 (CG307) caused more infections Copyright © 2017 Long et al. This is an open- access article distributed under the terms of than those of well-studied epidemic CG258. Strains varied markedly in gene content the Creative Commons Attribution 4.0 and had an extensive array of small and very large plasmids, often containing anti- International license. microbial resistance genes. Some patients with multiple strains cultured over time Address correspondence to James M. Musser, were infected with genetically distinct clones. We identified 15 strains expressing the jmmusser@houstonmethodist.org. This article is a direct contribution from a New Delhi metallo-beta-lactamase 1 (NDM-1) enzyme that confers broad resistance Fellow of the American Academy of to nearly all beta-lactam antibiotics. Transcriptome sequencing analysis of 10 phylo- Microbiology. External solicited reviewers: genetically diverse strains showed that the global transcriptome of each strain was Michael Dunne, bioMérieux, Inc.; Robert Bonomo, Louis Stokes Veterans Affairs Medical unique and highly variable. Experimental mouse infection provided new information Center. about immunological parameters of host-pathogen interaction. We exploited the large data set to develop whole-genome sequence-based classifiers that accurately predict clinical antimicrobial resistance for 12 of the 16 antibiotics tested. We con- clude that analysis of large, comprehensive, population-based strain samples can as- sist understanding of the molecular diversity of these organisms and contribute to enhanced translational research. IMPORTANCE Klebsiella pneumoniae causes human infections that are increasingly difficult to treat because many strains are resistant to multiple antibiotics. Clonal group 258 (CG258) organisms have caused outbreaks in health care settings world- wide. Using a comprehensive population-based sample of extended-spectrum beta- lactamase (ESBL)-producing K. pneumoniae strains, we show that a relatively uncom- mon clonal type, CG307, caused the plurality of ESBL-producing K. pneumoniae infections in our patients. We discovered that CG307 strains have been abundant in Houston for many years. As assessed by experimental mouse infection, CG307 strains May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 1 ® Long et al. were as virulent as pandemic CG258 strains. Our results may portend the emergence of an especially successful clonal group of antibiotic-resistant K. pneumoniae. lebsiella pneumoniae is a major human pathogen responsible for high morbidity Kand mortality rates (1–3). Numerous studies have investigated the genetic epide- miology of K. pneumoniae infections in hospitalized patients in the United States and elsewhere (3–8). These studies have often used low-resolution molecular typing such as multilocus sequence typing (MLST) or have studied relatively small collec- tions of organisms. Studies of hospitalized patients in the United States commonly have reported a predominance of clonal group 258 (CG258) strains (9–12). CG258 was first identified in 2009 as a single-locus variant of sequence type 11 (ST11) from Hungary (13) and has since been reported worldwide. Data suggest that CG258 diverged from its most recent common ancestor in 1995, after acquisition of the K. pneumoniae carbapenemase (KPC) gene (14). K. pneumoniae strains producing KPC carbapenemase were first identified in North Carolina in 1996 (15) and now cause multidrug-resistant hospital-acquired infections globally (16). Evidence suggests that there is circulation of K. pneumoniae clones within hospitals, with transmissible resistance plasmids shared between Klebsiella spp. and other pathogens (5, 17–20). This sharing of resistance elements is believed to have contributed to the emergence of multidrug resistance in K. pneumoniae and large, hospital-based outbreaks. The increasing reports of these nosocomial outbreaks are of great clinical concern (21, 22). To develop new therapeutics and control strategies for K. pneumoniae, it is imper- ative to understand the population genomic structure of strains causing human infections. To begin filling this knowledge gap, we sequenced the genomes of 1,777 extended-spectrum beta-lactamase (ESBL)-producing K. pneumoniae strains (Table 1; see Table S1A to C in the supplemental material) cultured from patients in the 2,000-bed Houston Methodist Hospital system between September 2011 and May 2015. These strains were all positive for ESBL production, which accounted for an average of 40% of the K. pneumoniae strains in our laboratory during the study period. The Houston Methodist Hospital system serves patients in the greater Houston metropolitan area (population of approximately 6,000,000), now the most ethnically diverse population center in the United States (23). In addition to our large and diverse local population, patients routinely travel for treatment to Houston Methodist Hospital from across the United States and internationally. The genome sequence data identified unexpectedly high diversity. Strains varied markedly in gene content and had an extensively diverse assortment of large plasmids. These transmissible plasmids often contain antimicrobial resistance genes and represent important elements contributing to the dissemination of resistance markers. RESULTS Whole-genome sequencing of 1,777 ESBL-producing K. pneumoniae strains identified extensive genomic diversity within and between major clades. We sequenced the genome of 1,777 ESBL-producing K. pneumoniae strains (Table 1 and TABLE 1 Summary of clonal groups and specimen sources of 1,777 K. pneumoniae isolates No. of isolates (% of total) Source(s) CG258 CG307 Other CG Total Blood 34 (1.9) 54 (3.0) 59 (3.3) 147 (8.3) Respiratory 130 (7.3) 111 (6.2) 143 (8.0) 384 (21.6) Urine 220 (12.4) 374 (21.0) 333 (18.7) 927 (52.2) Other 90 (5.1) 105 (5.9) 124 (7.0) 319 (18.0) All 474 (26.7) 644 (36.2) 659 (37.1) 1,777 (100) Other specimen sources include tissue, wounds, and body fluids. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 2 ® ESBL K. pneumoniae CG307 Diverse CG258 ST Clade 0.025 FIG 1 Estimates of genetic relationships among 1,777 ESBL-producing K. pneumoniae strains recovered from infected patients in a multiple-hospital system in the Houston metropolitan area. Polymorphisms were called against the genome of CG258 K. pneumoniae reference strain NJST258_2 (NCBI GenBank accession no. CP006918.1). Colors represent the major clonal groups, including CG258 (red) and CG307 (blue), and the remaining heterogeneous STs (gray). Phylogenetic relationships were defined by the neighbor-joining method in FastTreeMP with double precision. The core genome was defined as the chromosomal sequence with phage sequence regions excluded. To simplify presentation, outliers have been cropped out of frame at 3 and 9 o’clock. The outlier strains, which were identified as K. pneumoniae by matrix-assisted laser desorption ionization mass spectrometry and MLST, may be allied with the genomes of some strains referred to as K. quasipneumoniae and K. variicola. Fig. 1; see Table S1A to C and Fig. S1) cultured between September 2011 and May 2015 from infected patients in the 2,000-bed Houston Methodist Hospital system. The sample represents virtually all of the ESBL-producing K. pneumoniae strains recovered during this time period. Over half of the strains were cultured from urine (927 strains, 52.2%). The remainder were cultured from respiratory (384 strains, 21.6%), blood (147 strains, 8.3%), and other tissue, wound, and miscellaneous specimens (319 strains, 18.0%). Much of the literature describing nosocomial outbreaks or hospital strain samples in the United States and elsewhere has reported a predominance of CG258 strains, regardless of the population studied (17, 24–29). In striking contrast, on the basis of genome-wide phylogenetic analysis, our strains were broadly classified into three major genetic clades (Fig. 1; Table 1). Approximately one-fourth (474 strains, 26.7%) were assigned to CG258. These organisms represent progeny of a globally disseminated epidemic clone reported to be the major cause of K. pneumoniae infections in the United States (1, 13, 14, 30–32). The average minimum pairwise genetic distance (a measure of genomic diversity in a population of related strains) between the CG258 strains in our sample was 346 single nucleotide polymorphisms (SNPs). More than one-third of the strains (635, 35.7%) were assigned to CG307. These organisms were cultured from patients at all of the hospitals in our system; that is, no single hospital was responsible for the abundance of CG307 infections (Fig. 2). Similarly, CG258 strains May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 3 ® Long et al. CG Hospital A CG258 ybt 17 clb 3 Mortality Hospital B CG307 ybt 10 Hospital C Other ST Other ybt Hospital D Hospital E FIG 2 Spatial relationships of ESBL-producing K. pneumoniae strains from the multiple hospitals in the Houston Methodist system with clonal groups and associated patient deaths. This circular cladogram displays the first isolate from each patient, color coded by the five hospitals of origin (teal, yellow, red, purple, and gray). The circles surrounding the cladogram, from innermost to outermost, indicate the clonal group of the strain, yersiniabactin locus, colibactin locus, and patient death. In the innermost circle, the CG258 strains are red and the CG307 strains are blue. Other clonal types are indicated with a gray bar. The yersiniabactin circle is color coded to indicate which yersiniabactin locus is present, if any, with the two most common loci, ybt17 and ybt10, in yellow and teal, respectively; all other ybt loci are gray. The colibactin circle is purple to indicate the presence of clb3, the only colibactin synthesis locus detected in our collection. In the outermost circle, in-hospital death is indicated by a black line. For patients with multiple strains, only the first isolate from the patient is represented on the tree. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 4 ® ESBL K. pneumoniae FIG 3 ESBL-producing K. pneumoniae strains recovered during the course of this study. The first strain recovered from each patient is graphed (circles, green linear regression). The number of strains classified as CG307 (triangles, blue linear regression) and CG258 (inverted triangles, red linear regression) are also shown. This figure demonstrates that CG258 and CG307 strains were abundant throughout the study period. caused infections in patients at all of our hospitals (Fig. 2). CG307 and CG258 strains were common in our population throughout the study period (Fig. 3). Although strains of CG307 have been identified sporadically in diverse locations globally in recent years (10, 33–40), they have not been reported to be abundant causes of infections in a large unbiased population over a multiyear period. The average minimum pairwise genetic distance between the CG307 strains was 948 SNPs. Com- pared to CG258 organisms that have been extensively studied, very little is known about strains of CG307. The remaining strains (668 strains, 37.6%) in our sample were a genetically heterogeneous array of multilocus STs reported to cause infections worldwide. The average minimum pairwise genetic distance among these 668 strains is 6,580 SNPs, reflecting the far greater diversity present in these strains than in CG258 and CG307 organisms. There were 99 unique previously defined STs present in the strains sequenced (1,430 strains, 80.4%) (see Table S2). An additional 51 strains (2.9%) had novel combinations of known ST alleles, whereas the remaining 296 strains (16.7%) have novel ST alleles resulting in previously undefined ST types. After CG258 and CG307, the three more common STs were ST16 (92 strains, 5.2%), ST15 (49 strains, 2.8%), and ST280 (32 strains, 1.8%). Closure and annotation of five new K. pneumoniae reference genomes. Five strains were sequenced to closure using PacBio long-read sequencing (see Table S3). This allowed the annotation of new reference strains to aid our studies of K. pneu- moniae population genomics and pathogenesis. The five strains were chosen to represent regions of the phylogenetic tree for which existing reference genomes deposited in publically available databases were lacking. In addition, genomes con- taining the bla and OXA-48 genes (see below) were chosen to allow more in-depth NDM-1 analysis of these important strains. KPN11 is representative of the CG307 strains in our patient population. KPN528 was chosen for PacBio sequencing and closure because it was the first K. pneumoniae strain in our sample identified with the bla metallo-beta-lactamase. KPN1481 (ST906) NDM-1 was chosen because it was one of the last strains in our sample containing the bla NDM-1 metallo-beta-lactamase gene, and a closed reference genome has not been reported for this clonal type. Similarly, KPN1482 (ST37) is a clonal type for which a closed reference genome is not available. BK13043, a representative CG258 strain collected in New York City in 2004, was chosen to provide a historical CG258 reference genome. Each new closed genome contained two to five plasmids (see Table S3). The plasmids matched previously identified reference plasmids in GenBank with sequence coverage ranging from 21 to 100%. The bla containing plasmids present in NDM-1 KPN528 and KPN1481 differed. KPN528 had a 292-kb plasmid that had 99% coverage of the pPKPN1 plasmid of reference strain PittNDM01, whereas strain KPN1481 had a May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 5 ® Long et al. 342-kb plasmid that had 74% coverage of the pNDM-MAR plasmid (41, 42). Strain KPN528 contained a second plasmid (76 kb) with 99% coverage of plasmid 3 present in strain PittNDM01. The PittNDM01 K. pneumoniae strain is an ST14 organism isolated from a patient in Pittsburgh, PA, in 2013 (41). Of note, KPN528 also is an ST14 strain and was isolated in December 2012. Taken together, these results suggest that these Pittsburgh and Houston strains are lineally descended from a common ancestor organism. Strains of CG307 are unexpectedly abundant. KPC-producing K. pneumoniae were initially described in the Houston region in 2009 (43). That initial report was followed by the identification of multiple clonal types in 2010, including CG258 and CG307 (10). The CG307 clade of K. pneumoniae was first identified in a Dutch teaching hospital in 2008 (44). Since then, CG307 strains have been identified periodically in Europe, Africa, the Middle East, Asia, Central and South America, and New York City (20, 33–40, 45–47), and a hospital-based outbreak was recently reported in South Korea (48). However, they have not been documented to be of great abundance in any of these areas. Thus, in our genomic data set based on comprehensive, population-based sampling of ESBL-producing K. pneumoniae strains, we demonstrate the existence of a numerically highly successful but poorly characterized genetic group (CG307) with disease prevalence exceeding that of the well-known epidemic CG258 organisms. We identified no statistically significant difference between the observed in-hospital mor- tality rate of all patients infected with CG307 strains and that of patients infected with the CG258 clonal type (chi-square test) (Fig. 2). Similarly, the mortality rate did not differ statistically significantly between patients with bloodstream infections caused by CG307 and CG258 strains (chi-square test). Furthermore, no significant differences between CG307 and CG258 strains were identified in any clinical feature (see Ta- ble S1A), including the hospital of origin, organ transplant status, infection type (urinary tract versus invasive infection), and number of episodes of care (single versus multiple strains recovered longitudinally from the same patient). Comparison of CG258 and CG307 genomes: abundant variation in gene content. The CG258 reference genome (NJST258_2; NCBI GenBank accession no. CP006918.1) has 5,188 annotated genes. Comparison of these genes to our assembled CG307 reference genome (KPN11) by bidirectional BLAST analysis found that 429 of the annotated NJST258_2 genes do not have a significant homolog present in the KPN11 CG307 genome (Fig. 4). Most of these genes (253 out of 429) are annotated as hypothetical proteins. Many of the others are associated with phage, secretion systems, or metabolism. Of the 5,237 annotated genes in the KPN11 CG307 reference assembly, there are 414 annotated genes that lack a homolog in NJST258_2, most of which (281 of 414) are annotated as hypothetical proteins. Most of the other genes that are missing are associated with phage, transduction elements, or metabolism. Some of the genes present in or absent from either clonal type are related to the capsule gene locus. Genetic variation in this locus is well established (17, 49, 50). A summary of the annotated genes present in or absent from NJST258_2 compared to KPN11 CG307 is presented in Tables S4 and S5. Variations in yersiniabactin locus, antimicrobial resis- tance gene content, and capsule locus are discussed in more detail below. Presence of yersiniabactin and colibactin loci in ESBL-producing K. pneu- moniae. Recent studies have emphasized the important contribution of siderophore gene acquisition to severe Klebsiella invasive disease (51, 52). Holt et al. reported that the yersiniabactin locus (ybt) was present in 40.0% of the CG258 and 32.2% of the overall K. pneumoniae genomes analyzed (52). In contrast, we found that only 274 of our 1,777 strains (15.4%) had the ybt locus. Only 13 of the 644 CG307 strains (2.0%) and 85 of the 474 CG258 strains (17.9%) contained the ybt locus. Of the 474 CG258 strains, 76 (16.0%) had both the colibactin synthesis locus (clb) and the ybt locus. In contrast, none of the CG307 strains carried a clb locus. The presence or absence of the yersiniabactin and colibactin loci is shown in Fig. 2 and presented in Table S1A. In contrast to results from other studies using smaller strain collections (51, 52), the presence of the yersiniabactin and colibactin loci was not significantly associated with May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 6 2.6 5.2 2.5 5.1 2.4 5.0 2.3 4.9 2.2 4.8 2.1 4.7 2.0 1.9 4.5 1.8 4.4 1.7 4.3 1.6 4.2 1.5 4.1 1.4 4.0 ESBL K. pneumoniae ST258 ST307 Percent Protein Sequence Identity ST258 ST307 FIG 4 Gene content differences between the reference genomes of CG258 and CG307. CG258 is the globally abundant strain of K. pneumoniae, and CG307 is the locally abundant strain in our study. Bidirectional BLAST was performed by using the PATRIC resource to illustrate the differences in gene content between these two reference genomes. The color indicates the percent identity of the BLAST hit for each gene, with darker shading indicating a bidirectional hit and lighter shading indicating a unidirectional hit (top). A progressive Mauve alignment of the CG258 and CG307 reference genomes is shown. Local colinear blocks are organized by color (bottom). NJST258_2 is the CG258 reference genome, and KPN11 is the CG307 reference assembly. any clinical feature, including invasive disease, chronic infection, or death (no statisti- cally significant difference, chi-square test). Extensive plasmid diversity within ESBL-producing K. pneumoniae. K. pneu- moniae strains commonly have large plasmids that often contain antibiotic resistance May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 7 1.3 3.9 1.2 3.8 1.1 3.7 1.0 3.6 0.9 3.5 0.8 3.4 0.7 3.3 0.6 3.2 0.5 3.1 0.4 3.0 0.3 2.9 0.2 2.8 0.1 2.7 0.0 ® Long et al. genes and multiple replicons. The genome data identified extensive plasmid diversity among the 1,777 strains. The average number of plasmid replicons per isolate was 4.5 (standard deviation, 2.5; range, 0 to 16). Six replicons were each present in more than 25% of the 1,777 strains, i.e., FIBk (1,327 strains), FIIk (1,237 strains), FII (693 strains), R (667 strains), ColRNAI (660 strains), and FIBpQil (494 strains). The distribution of the replicons across the phylogenetic tree is shown in Fig. S1 and Table S1C. The FIBk and FIIk replicons are present in strains of diverse clonal back- grounds, with some clustering in CG307 organisms. The FII, ColRNAI, and FIBpQiL replicons appear to cluster predominantly with CC258 strains, whereas the R replicons are associated with the CG307 organisms. The most common combination of replicons in a single isolate was FIBk, FII, ColRNAI, and FIBpQil (248 strains), followed closely by FII and FIBpQil (244 strains) and FIBk and FIBpQiL (228 strains). The only replicon of the top six to be found in the absence of any of the other five was FIBpQiL (177 strains). Diversity of ESBL genes. Our analysis identified an extensive array of ESBL genes in the 1,777 genomes, including those encoding carbapenemases. Genes in the SHV- LEN-OKP family of ESBLs were most common, being present in 1,442 of the 1,777 strains. The second most common ESBL gene was bla , which was present in 1,161 CTX-M of 1,777 strains (see Table S1B). Among the eight bla alleles identified, bla , CTX-M CTX-M-15 was the most abundant. Other ESBL-encoding genes identified included a member of the TEM family (797 strains) and the KPC gene (581 strains). OXA genes characterized as encoding ESBLs (OXA-23, OXA-24, OXA-48, and OXA-83) were far less common, being present in only 23 strains. bla -positive strains were cultured from six NDM-1 patients, two of whom had strains that also had the OXA-48 gene. Some of the antimicrobial resistance genes detected in these strains, such as TEM30, TEM33, and SHV26, are known to be inhibitor resistant. Overall, KPC alleles were present in 64.6% of the CG258 strains (306 strains) compared to 34.3% of the CG307 strains (221 strains). Among all of our strains with a bla carbapenemase gene (581 strains), bla was the dominant allele (572 strains, KPC KPC-2 98.4%), including 293 CG258 strains (96.4%) and 197 CG307 strains (93.4%). Our results differ from those of other studies reporting that both bla and bla are found in KPC-2 KPC-3 CG258 K. pneumoniae (10, 13, 17, 20, 45, 53) and that CG307 was strongly associated with bla (39). CG307 strains expressing bla have been recently reported in KPC-3 KPC-2 Korea and Colombia (34–36). In our sample, no CG258 strains and only three CG307 strains had the bla allele. The difference between bla and bla is one KPC-3 KPC-2 KPC-3 nonsynonymous SNP (C814T) that results in an H272Y amino acid replacement. Whereas both bla and bla are very efficient at hydrolyzing penicillins, cepha- KPC-2 KPC-3 losporins, and carbapenems, bla has increased enzymatic activity against and KPC-3 confers higher resistance to cephamycins and ceftazidime (15). Four newer KPC variants (bla , bla , bla , and bla ) likely derived from KPC-7 KPC-8 KPC-9 KPC-10 bla have a second nonsynonymous SNP that confers greater catalytic activity KPC-3 against ceftazidime than bla (54). That is, the use of ceftazidime worldwide may be KPC-2 selecting for KPC variants with increasing levels of resistance. Of these newer bla KPC variants, there are four CG258 strains in our collection with bla . Although the KPC-8 reason for the predominance and persistence of bla in our hospital system is KPC-2 unclear, it may be related in part to the relatively conservative use of ceftazidime and ceftazidime-avibactam compared to other beta-lactam antibiotics. The potential for a shift from a CG258-centered K. pneumoniae epidemic to a polyclonal epidemic was first reported in Italy in 2014, and concern has been raised in Korea (33–36). Analysis of bla -carrying patient strains suggests horizontal transfer of a NDM-1 bla -containing plasmid from a classic Indian bla -positive clone into the NDM-1 NDM-1 Houston CG307 population. The identification of the gene encoding New Delhi metallo-beta-lactamase 1 (bla ) in our patient strains is concerning because that NDM-1 enzyme confers very broad resistance to nearly all beta-lactam antibiotics and can be readily transferred between strains by plasmids. Since its discovery in 2009, the bla gene has spread globally among members of the family Enterobacteriaceae; NDM-1 however, the number of reported human infections remains relatively low. As of May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 8 ® ESBL K. pneumoniae TABLE 2 bla -containing strains recovered from six patients (A to F) NDM-1 Strain Patient ST Date collected Source ST geographic association(s) 528 A ST14 12/2012 Sputum India 537 A ST14 12/2012 Trachea India 553 A SLV of ST14 12/2012 Wound India 603 A SLV of ST14 1/2013 Urine India 658 A ST14 2/2013 Blood India 755 B SLV of ST14 4/2013 Urine India 764 B ST14 4/2013 Urine India 1010 B ST14 10/2013 Urine India 1273 B SLV of ST14 1/2014 Urine India 1145 C ST14 12/2013 Urine India 1402 C SLV of ST14 4/2014 Bile India 1426 D ST37 5/2014 Urine India/United Kingdom 1458 D ST37 5/2014 Urine India/United Kingdom 1481 E ST906 6/2014 Urine Israel 1844 F ST307 10/2014 Urine Houston SLV, single-locus variant. 6 January 2017, only 175 cases have been reported in the United States, including 2 in Texas (https://www.cdc.gov/hai/organisms/cre/TrackingCRE.html). To identify strains with bla in our strain sample, assembled contigs were mapped to a custom NDM-1 database of all published bla -bearing plasmids. We identified 15 strains from six NDM-1 patients with the bla gene (Table 2; see Fig. S2). All contigs containing the NDM-1 bla gene are 100% identical to published plasmid sequences, indicating that no NDM-1 novel bla sequences have been introduced into or evolved de novo in Houston NDM-1 patients. Contigs from three strains were identical. Consistent with the idea that strains with the bla gene were imported to Houston, the majority of the bla - NDM-1 NDM-1 containing strains belong to ST14 and ST37, two STs endemic in India, Pakistan, and elsewhere in southern Asia (55). Unexpectedly, the bla gene was detected in one NDM-1 CG307 strain. It is possible that this strain acquired the bla plasmid while circu- NDM-1 lating locally in Houston. Consistent with this idea, the bla -containing plasmid in NDM-1 our CG307 strain (Table 2, patient F) is closely similar to a bla -containing plasmid NDM-1 in two Houston patients (Table 2, patients A and B). Extensive diversity in the capsule locus. K. pneumoniae capsule is an important virulence factor, and the capsule synthesis locus has considerable diversity in gene content between clonal groups (17, 49–51). Diverse associations among different wzi genotypes, bla genotypes, and STs have been reported, and some clonal groupings KPC have been identified (45, 53). Recently, Holt et al. (44) proposed a new nomenclature for capsule locus genotyping (KL type) that encompasses the diversity of the known locus and described a software tool (Kaptive) for capsule locus genotyping. Using Kaptive, we identified 72 unique KL capsule locus types and potential novel variants (see Table S1A). The two most common KL capsule locus types among CG258 strains were KL106 (n  259) and KL107 (n  117). KL107 has the wzi154 gene allele. Although previous studies reported that the wzi154 allele in CG258 strains was associated with the bla gene, this was not the case in our strain sample. We found that almost all KPC-3 of our CG258 strains with a KPC gene contain bla . The CG307 strains we studied KPC-2 had less diversity in the capsule locus. We found that 524 of 526 strains had the KL102 capsule locus, including all 119 ST307 strains with the bla gene. KPC-2 In general, a given clonal group had a predominant KL capsule locus type, although some exceptions were observed (see Table S1A). CG258 is notable for having two capsule gene loci common in the group. Other clonal types with notable capsule locus diversity include ST15 (49 strains, four KL capsule locus types) and ST37 (18 strains, eight KL capsule locus types). K. pneumoniae clinical strains recovered over time from the same patient. Many Houston Methodist patients (n  357) had two or more K. pneumoniae strains cultured from different specimens, in some cases more than 3 years apart (Fig. 5). To investigate whether these different episodes in the same patient represent infections caused by May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 9 ® Long et al. FIG 5 Genetic variation in ESBL-producing K. pneumoniae clinical strains in the same patient over time. (A) Clonal group of each strain recovered from 196 patients with multiple strains. (B) Clonal group of each isolate recovered from 10 patients (numbered 1 to 10) with multiple strains spanning the longest period. Each circle represents one isolate. The vertically overlapping circles shown for patient 5 indicate that two different strains were recovered from different specimens on the same day or consecutive days. the same or a different clone, we used the whole-genome sequence data to determine the relatedness of the organisms recovered. For this analysis, a second episode was defined as a strain isolated more than 35 days after the prior infection event. The 35-day-criterion was selected by using the Infectious Diseases Society of America definition of a late cure of cystitis (cystitis was the most common infection type in our collection) (56). Of the 196 patients with multiple strains recovered in different epi- sodes, we found that 42 (21.4%) were infected with different clones during the study period (Fig. 5A). Similarly, 10 patients had two genomically distinct strains recovered on the same day (Fig. 5B). Consistent with the hypothesis that CG307 strains are as virulent and able to cause recurrent or chronic infections as CG258 strains, the frequency of CG307 and CG258 strains recovered from patients with multiple episodes did not differ significantly (Fisher exact test, no statistically significant difference). Of note, no simple pattern of change in capsule locus or antimicrobial resistance genes was identified among the multiple-episode patients. Extensive transcriptome diversity among and within clades. Whole-genome phylogenetic analysis revealed that the ESBL-producing K. pneumoniae strains causing infections in our health care system could be broadly classified into three major genetic clades. To begin investigating the effect of the extensive genomic diversity on global gene expression, 10 rationally selected strains were analyzed by genome-wide tran- scriptome sequencing (RNA-seq) analysis. The 10 strains were selected to broadly represent the overall population structure, including the year of collection, phage content, antimicrobial resistance gene content, and numerical dominance of CG (see Table S6). Given the extensive genomic variation present, we hypothesized that each strain would have a unique transcriptome. The 10 strains had nearly superimposable growth curves, suggesting that observed differences in global transcriptome were attributable to genetic differences among strains rather than growth characteristics (Fig. 6A). Transcript libraries were prepared by standard methods and sequenced to very deep coverage (mean, 22.2 [range, 14.8 to 33.6] million reads per replicate) to minimize the false discovery of significantly altered gene expression. Transcripts were counted relative to CG258 reference strain NJST258_2. Consistent with our hypothesis, principal- component analysis showed differences in the transcriptomes of these strains, both within and between the CGs (Fig. 6B). On average, each strain differentially expressed 427 genes compared to the reference strain (Fig. 6C). In the aggregate, the maximum number of differentially expressed genes was 617, indicative of the substantial diversity in these strains (Fig. 6D). Also consistent with gene content differences identified among the 10 strains, approximately half of the reads for each strain did not map to the NJST258_2 reference chromosome (mean, 46.3% reads unmapped per duplicate strain pair; range, 42.2 to 53.5%). BLAST analysis revealed that most unmapped reads derived from plasmid-borne genes, insertion sequences, and other mobile elements. Similar May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 10 ® ESBL K. pneumoniae FIG 6 RNA-seq analysis of 10 genomically diverse ESBL-producing K. pneumoniae strains. Strains of CG258 (n  4, red), CG307 (n  3, blue), and other STs (n  3, gray) were analyzed. (A) Growth curves of strains. RNA-seq analysis of duplicate cultures was done (the mean absorbance of each strain pair is shown); the growth curves of all of the strains are similar. Cells were harvested at the mid-exponential phase of growth (OD of 0.6; horizontal dashed green line). (B) RNA-seq transcriptome data for chromosomally located genes of each strain were compared to those of CG258 reference strain BK30684. Plotted are expression relationships among the strains based on principal component 1 (PC1) and principal component 2 (PC2), which account for the two largest unrelated variances in the data. The strains are distributed across both axes, showing transcriptome diversity within and between CGs. (C) Number of genes with significantly altered expression relative to reference strain BK30684. Plotted are the numbers of genes with significantly increased (positive y axis) and decreased (negative y axis) expression for each strain. (D) Box plot summarizing data from the 10 strains combined or within CGs. results were generated when reads were mapped to CG307 strain KPN11 (see Fig. S3A) or CG14 strain KPN528 (see Fig. S3B). Also, the RNA-Seq data confirmed expression of the genes encoding NDM-1, OXA-48, KPC, and yersiniabactin by strains carrying these genes (Table 2; see Fig. S3C to F and Table S6). That is, among these 10 strains, mobile gene content contributed to nearly half of all differentially expressed gene transcripts. After accounting for differences in gene content, we identified no simple patterns of transcriptome differences, either within or between the CGs. Mouse model of ESBL-producing K. pneumoniae pneumonia. A very unexpected finding from the whole-genome sequence data was that CG307 organisms were very abundant among the ESBL-producing K. pneumoniae strains recovered from patients in our Houston Methodist Hospitals. To test the hypothesis that CG307 strains are as virulent as the globally disseminated and numerically common CG258 clone, refer- ence strains KPN1411 (CG307) and KPN113 (CG258) were compared using a mouse model of pneumonia. Mice were inoculated by intranasal instillation and necropsied at 24 h, and the numbers of CFU recovered from the infection site (lung) and a site of dissemination (spleen) were determined. CFU were recovered from the lungs of all of the mice. There was no significant difference in the number of CFU cultured from the lungs of mice infected with CG307 strain KPN1411 and CG258 strain KPN113 5 4 (mean, 1.22  10 CFU for KPN1411 compared to 8.98  10 CFU for KPN113; data not shown). Similarly, there was no significant difference in dissemination to the May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 11 ® Long et al. FIG 7 Host immune response to ESBL-producing K. pneumoniae in a mouse model of pneumonia. Flow cytometry analysis of leukocytes recovered from infected lungs, including neutrophils (A), monocytes (B), T cells (C), B cells (D), alveolar macrophages (E), CD4 T cells (F),  T cells (G), IL-17-producing CD4 T cells (H), and IL-17-producing  T cells (I). (J) IL-17-producing  T cells were also recovered from the spleen. spleen (four of five mice for KPN1411 compared to three of five mice for KPN113; data not shown). To further investigate host-pathogen interactions, the host immune response was examined in infected mice. Mice were inoculated by intranasal instillation, monitored for 7 days, and sacrificed, and flow cytometry analysis of immune cells recovered from their lungs and spleens was conducted (Fig. 7). Compared with uninfected control mice, animals infected with ESBL-producing K. pneumoniae had significantly more total CD45 leukocytes in their lungs, including neutrophils (9.04-fold; Fig. 7A), monocytes (4.06-fold; Fig. 7B), T cells (2.63-fold; Fig. 7C), and  T cells (25.21-fold; Fig. 7D). There was no significant difference in the numbers of B lymphocytes or alveolar macrophages (Fig. 7E and F). K. pneumoniae-infected lungs also had significantly more  and CD4 T cells expressing IL-17 (9.50- and 12.07-fold, respectively, Fig. 7G and H). However, compared to the CG258 strain, the CG307 strain elicited a significantly smaller fraction of IL-17-producing  T cells (Fig. 7I). IL-17-expressing CD4 T cells were also increased in the spleens of K. pneumoniae-infected animals, likely representing a reservoir of proliferating reactivated memory cells (4.78-fold; Fig. 7J). Taken together, the mouse pneumonia data demonstrate that the CG307 reference strain was as virulent as the epidemic CG258 reference strain in this disease model. Prediction of resistance to antimicrobial agents based on whole-genome se- quence data. The large number of genome sequences generated for this study of drug-resistant organisms gave us the opportunity to use machine learning-based classifiers to predict resistance to antimicrobial agents. To test the ability to use whole-genome sequence data to predict antimicrobial resistance phenotypes, we used the AdaBoost algorithm (57, 58) to generate classifiers for 16 agents commonly used to treat human K. pneumoniae infections. Results of the classifier prediction strongly correlated with the clinical phenotype, as measured by the area under the curve (AUC) and the F1 score (Fig. 8). Importantly, the prediction data generated the three highest scores for the carbapenem antibiotics. DISCUSSION In the aggregate, our data demonstrate that ESBL-producing K. pneumoniae strains causing infections in our large hospital system over several years have substantial genomic diversity. Analysis of large, comprehensive, population-based strain samples May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 12 ® ESBL K. pneumoniae FIG 8 Receiver operating characteristic curves of 16 AdaBoost-based classifiers predicting antimicrobial resistance. Antibiotic agent abbreviations: AMK, amikacin; ATM, aztreonam; FEP, cefepime; FOX, cefoxitin; CIP, ciprofloxacin; ETP, ertapenem; FOF, fosfomycin; GEN, gentamicin; IPM, imipenem; LVX, levofloxacin; MEM, meropenem; TZP, piperacillin-tazobactam; TET, tetracycline; TGC, tigecycline; TOB, tobramycin; SXT, trimethoprim-sulfamethoxazole. AUC is area under the concentration-time curve. The F1 score, the harmonic mean of precision and recall, is commonly used to compare classification methods. Data are the results of a 10-fold cross validation. can assist our understanding of the molecular pathogenesis of these organisms and thereby contribute to enhanced translational research. We identified the heretofore unrecognized numerical success of CG307 strains in the Houston region (Fig. 1). CG307 strains caused more infections than the pandemic CG258 strains throughout the study period (Fig. 3). We confirmed that CG307 strains continue to be an abundant cause of infections in our hospital system by sequencing an additional 96 ESBL-producing K. pneumoniae strains recovered in 2017 (S. W. Long, unpublished data). Although CG307 strains have been previously identified in our area, the extent to which this clonal type has successfully coexisted with CG258 to achieve a state of codominance in our region was a very unexpected discovery. The reason(s) the CG307 clone has become abundant in Houston but not other parts of North America is not understood. CG307 strains have not been reported previously to be a prominent cause of infections in a large, unbiased population over a multiyear period. The relatively recent success of CG307 progeny in Italy and Korea has caused alarm (33, 34, 36, 47). By all measures, including the in-hospital mortality rate, the CG307 strains causing infections in Houston appear to be as virulent as CG258 strains. Over the 45-month course of our study, the incidence of ESBL-producing K. pneumoniae steadily increased from approximately 18 to 28 strains recovered per month (Fig. 3). There was no preferential increase in CG258 or CG307 strains compared to other ST types during the study period. This increasing incidence mirrors a trend reported elsewhere and stresses the detrimental impact of multidrug-resistant K. pneumoniae on human health. The use of whole-genome sequence data to infer antimicrobial resistance pheno- types is an area of intense interest (59, 60). Two very different approaches have been pursued. The most common strategy attempts to predict phenotypes by identifying the presence or absence of genes and gene mutations conferring antimicrobial resistance (61). Although modestly successful, this strategy has two major disadvantages. First, it requires the curation of a comprehensive database that is updated as new discoveries are made. Second, this strategy is likely to fail to accurately predict resistance if the molecular mechanism is multifactorial or unknown. Since this study provides a very large, well-annotated data set, we sought an alternative approach using machine learning-based classifiers to predict antimicrobial resistance (Fig. 8). We have recently used this method to build accurate predictors of antimicrobial resistance phenotypes for several important human pathogens, including Acinetobacter baumannii, Mycobac- terium tuberculosis, Staphylococcus aureus, and Streptococcus pneumoniae (62). Machine May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 13 ® Long et al. learning approaches are advantageous because they require no a priori knowledge of the underlying molecular mechanism of resistance and may potentially identify new antimicrobial resistance-related genomic features. The results of this study suggest that for many antibiotics, it may be possible to build diagnostically valid classifiers for the prediction of antimicrobial resistance directly from the genome sequence data. Paired with new rapid sequencing technologies, these classifiers could provide clini- cally actionable data (63). We also attempted to build classifiers for the prediction of other clinical phenotypes. No classifiers could be generated to accurately predict the site of infection, length of stay, chronic or repeated infection, or death. The lack of predictive classifiers for these clinical phenotypes may reflect the extensive genomic and transcriptomic diversity present in the Houston strains or the complex comorbidi- ties present in patients with K. pneumoniae infections. Our data from the mouse model of pneumonia demonstrated that the host immune response to K. pneumoniae is multifaceted. We found a significant infiltration of neutrophils, monocytes, and T cells in infected lungs. We also detected IL-17 produc- tion by two different T-cell subsets, CD4 and  T cells. IL-17 is well known to play a key role in innate immunity to extracellular pathogens, particularly in patients with chronic disease or immune compromise. Chen et al. recently reported that IL-17 secretion by CD4 T cells provided protection against eight strains of K. pneumoniae belonging to three different capsular serotypes (STs not reported) recovered from patients in Southeast Asia, an area where very few CG307 strains have been reported (64). Herein, we are the first to suggest that IL-17 production by  T cells may also contribute to immunity to K. pneumoniae. Compared to the globally pandemic CG258 strain, the locally prominent CG307 strain elicited a significantly diminished IL-17 response by  T cells. In addition, the lack of a yersiniabactin locus in the great majority of CG307 strains may, in part, be related to this diminished IL-17 production. Nearly all K. pneumoniae produce their own siderophore, enterobactin, that binds iron in the body (52). Enterobactin is susceptible to inactivation by lipocalin 2 (Lnc2), which also stimulates an inflammatory response (65). K. pneumoniae may acquire the yersini- abactin locus to produce a siderophore that does not bind Lnc2, providing a survival benefit in exchange for the energy required to express yersiniabactin (52, 66). By limiting IL-17 production in the immune response, CG307 K. pneumoniae may be enhancing the effectiveness of the enterobactin siderophore system, negating the need for yersiniabactin loci in CG307 organisms for survival and success in the host. These discoveries provide one potential explanation for the numerical success of CG307 strains among patients in Houston. That is, the CG307 clone circulating in Houston, through an undetermined genomic or transcriptomic mechanism, may elicit a different innate immune response than CG258 strains. Additional studies are under way to investigate this idea. MATERIALS AND METHODS Strain sample. Strains were cultured from patient specimens submitted to the Diagnostic Microbi- ology Laboratory at Houston Methodist Hospital between September 2011 and May 2015. The Diag- nostic Microbiology Laboratory serves as the central diagnostic laboratory for the 2,000-bed Houston Methodist Hospital system. Strains were collected by the research team after routine microbiology diagnostics were completed. All of the strains collected were ESBL-producing K. pneumoniae, as deter- mined by automated methods (Phoenix ESBL test; BD). A summary of the isolate collection dates and sources is provided in Table S1A. Strains were collected and patient data were examined under human subjects protocol IRB1010-0199 as approved by the Institutional Review Board, Houston Methodist Research Institute. DNA extraction and whole-genome sequencing. Genomic and plasmid DNA was extracted in accordance with the Gram-negative protocol of the DNeasy Blood and Tissue kit (Qiagen). DNA was quantitated with the Qubit (Invitrogen) instrument, and whole-genome sequencing libraries were prepared with NexteraXT reagents (Illumina). Libraries were sequenced with Illumina MiSeq and NextSeq 500 instruments. Bioinformatic analysis. Initial bioinformatic analysis was performed on the high-performance computer cluster in the Houston Methodist Research Institute, a Beowulf cluster with 336 CPU cores and 1.8 TB of RAM. FASTQ files were processed with Trimmomatic (v 0.35) to remove adapter sequence 67, 68). SMALT (v 0.7.6) was used contamination and Musket (v 1.1) to correct short-read sequence errors ( to align reads, and FreeBayes (v 0.9.20) was used to identify variants (69). The in-house-developed scripts May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 14 ® ESBL K. pneumoniae prephix and phrecon were used to generate multi-FASTA files of SNPs for phylogenetic tree generation (https://github.com/codinghedgehog/). Phylogenetic trees were generated with FastTreeMP (v 2.1.8) using double precision and annotated with CLC Genomics workbench 9.5.2 (Qiagen) (70). Short-read sequence typing for bacterial pathogens (SRST2) (71) was used to identify the MLST assignment of each strain with the K. pneumoniae database of 2,470 STs available on 4 November 2016 curated at the Institut Pasteur (http://bigsdb.web.pasteur.fr/klebsiella/). SRST2 was also used to identify antibiotic resistance gene content and plasmid replicon content by using the SRST2 gene database files ARGannot.r1.fasta and PlasmidFinder.fasta. Reads were assembled into contigs with SPAdes v.3.6.2 (72). Kaptive was used to identify the capsule locus genotype and capsule locus gene content from assembled contigs (50). Kleborate was used to identify yersiniabactin, colibactin, and other siderophore locus content from assembled contigs (52). MEGA7 and MEGA-CC were used to generate distance matrices for pairwise distance analysis (73, 74). Gene content analysis was performed with the Proteome Comparison tool in PATRIC (75). Pacific Biosciences genome sequence data. Strains were sequenced to closure by the Pacific Biosciences SMRT long-read strategy in the laboratory of C. Mason at Weill Cornell Medical College. This technology permitted closed assembly of the core chromosome and plasmid DNA sequences from the target strains. Polished contigs were generated with HGAP. The plasmid sequences were compared to the NT database at NCBI with BLAST. Genome-wide RNA transcript analysis. RNA-Seq analysis was conducted as previously described (76). Briefly, 10 strains were selected to be broadly representative of our population, on the basis of the year of collection, phage type, CRE type, bla type, and numerical dominance of CGs (see Table S6). NDM-1 Two of these 10 strains were previously sequenced to closure (17), and 1 was sequenced to closure in this study with PacBio as described above. Duplicate cultures of each strain were inoculated into prewarmed LB medium and grown at 37°C to the mid-exponential phase (optical density at 600 nm [OD ] of 0.60 to 0.65). Total RNA was extracted with an RNeasy kit (Qiagen) according to the manufacturer’s instructions, and RNA-seq libraries were prepared with an Epicenter ScriptSeq kit (Illumina). The libraries were sequenced by using a 75-bp single-end protocol with a high-output v 3 reagent kit (Illumina). To generate a sufficient number of sequence reads for transcript analysis (average of 12 million reads per library), two runs were performed with one MiSeq instrument at the default software settings for the Generate FASTQ on instrument workflow. The two FASTQ files generated for each library were concatenated, and the number and quality of sequence reads were determined with FastQC (Babraham Bioscience Technologies, Cambridge, United Kingdom). The RNA transcripts were quantified from the concatenated FASTQ files with the default software settings relative to the chromosome of strain BK30684 with CLC Genomics Workbench v 9.5.2 (Qiagen). Differences of 1.5-fold by Baggerly’s test and a P value of 0.05 after applying Bonferroni’s correction for multiple comparisons were considered significant. Strain BK30684 was selected as the chromosome reference because the genome has been sequenced to closure, it is genetically representative of the globally disseminated CG258 clone, and it has been used in previous pathogenomic studies (17). To confirm that the many differences in gene expression observed between strains were due to differences in gene content, RNA-Seq reads were also mapped to CG307 reference strain KPN11 and CG14 reference strain KPN528 (see Table S3). To confirm expression of the KPC gene, reads were mapped to reference plasmid pNFST258C2 (GenBank accession no. CP006919.1). To confirm expression of the bla and OXA48 genes, reads were mapped to reference plasmids pPKPN1 NDM-1 (GenBank accession no. CP006799.1) and pKP112 (GenBank accession no. LN864819.1). Mouse model of pneumonia. A mouse model of pneumonia was used to compare the virulence of reference CG307 and CG258 strains KPN1411 and KPN113, respectively. C57BL/6 mice were inoculated by intranasal instillation of 1  10 CFU in 30 l (15 l in each naris). At 24 h postinoculation, mice were sacrificed. The lungs and spleens were removed, homogenized (Omni), serially diluted in sterile phosphate- buffered saline, plated on tryptic soy agar supplemented with 5% sheep blood, and incubated overnight at 37 C. CFU were counted, graphed as the mean  the standard error of the mean, and statistically compared by using the Mann-Whitney test (GraphPad). Similarly, mice were infected and sacrificed after 7 days and their lungs were harvested for flow cytometry analysis by standard methods (77). Preparation of genomic data for antimicrobial resistance prediction. Antimicrobial susceptibility testing for each strain was performed as part of the routine clinical workup. Machine learning-based classifiers for antimicrobial resistance phenotype prediction were created as previously described (62), by using the genomic data and clinical antimicrobial susceptibility data generated in this study. Briefly, the 1,777 genomes were assembled by using the PATRIC Full SPAdes assembly strategy and annotated by using the PATRIC annotation service (75). For each antimicrobial agent, the genomes were evenly divided into resistant and susceptible sets, matrices were built with 15-bp k-mers, and 10 iterative rounds of the AdaBoost algorithm were performed. Tenfold cross validation was applied to the data set for each classifier (62). The final set of classifiers generated for each antimicrobial agent is accessible in the PATRIC and RAST annotation services. Accession number(s). The whole genome sequence FASTQ files, RNAseq FASTQ files, and genomes of the strains sequenced by PacBio (see Table S3) have been deposited in the NCBI database under BioProject no. PRJNA376414. SUPPLEMENTAL MATERIAL Supplemental material for this article may be found at https://doi.org/10.1128/mBio .00489-17. FIG S1, PDF file, 0.4 MB. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 15 ® Long et al. FIG S2, PDF file, 0.04 MB. FIG S3, TIF file, 1.3 MB. TABLE S1, XLSX file, 0.3 MB. TABLE S2, PDF file, 0.1 MB. TABLE S3, PDF file, 0.04 MB. TABLE S4, PDF file, 0.1 MB. TABLE S5, PDF file, 0.1 MB. TABLE S6, PDF file, 0.1 MB. ACKNOWLEDGMENTS This work was supported by funds from the Fondren Foundation and the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Service (contract HHSN272201400027C). We thank Frank DeLeo and Barry Kreiswirth for providing reference strains BK30660 and BK30684. We thank Concepcion Cantu, Sarah Linson, Waleed Nasser, and Matthew Ojeda Saavedra for technical assistance. We thank Frank DeLeo, Timothy Palzkill, Katherine Perez, and Kathryn Stockbauer for suggestions to improve the manuscript. REFERENCES 1. Munoz-Price LS, Poirel L, Bonomo RA, Schwaber MJ, Daikos GL, Cormican CA, Doi Y. 2011. Colistin-resistant, Klebsiella pneumoniae carbapenemase M, Cornaglia G, Garau J, Gniadkowski M, Hayden MK, Kumarasamy K, (KPC)-producing Klebsiella pneumoniae belonging to the international Livermore DM, Maya JJ, Nordmann P, Patel JB, Paterson DL, Pitout J, epidemic clone ST258. Clin Infect Dis 53:373–376. https://doi.org/10 Villegas MV, Wang H, Woodford N, Quinn JP. 2013. Clinical epidemiology .1093/cid/cir401. of the global expansion of Klebsiella pneumoniae carbapenemases. 10. Castanheira M, Farrell SE, Wanger A, Rolston KV, Jones RN, Mendes RE. Lancet Infect Dis 13:785–796. https://doi.org/10.1016/S1473-3099(13) 2013. Rapid expansion of KPC-2-producing Klebsiella pneumoniae iso- 70190-7. lates in two Texas hospitals due to clonal spread of ST258 and ST307 2. Endimiani A, Choudhary Y, Bonomo RA. 2009. In vitro activity of NXL104 lineages. Microb Drug Resist 19:295–297. https://doi.org/10.1089/mdr in combination with beta-lactams against Klebsiella pneumoniae isolates .2012.0238. producing KPC carbapenemases. Antimicrob Agents Chemother 53: 11. Chen L, Chavda KD, Mediavilla JR, Zhao Y, Fraimow HS, Jenkins SG, Levi 3599–3601. https://doi.org/10.1128/AAC.00641-09. MH, Hong T, Rojtman AD, Ginocchio CC, Bonomo RA, Kreiswirth BN. 2012. 3. Snitkin ES, Zelazny AM, Thomas PJ, Stock F, NISC Comparative Sequenc- Multiplex real-time PCR for detection of an epidemic KPC-producing Kleb- ing Program Group, Henderson DK, Palmore TN, Segre JA. 2012. Tracking siella pneumoniae ST258 clone. Antimicrob Agents Chemother 56: a hospital outbreak of carbapenem-resistant Klebsiella pneumoniae with 3444–3447. https://doi.org/10.1128/AAC.00316-12. whole-genome sequencing. Sci Transl Med 4:148ra116. https://doi.org/ 12. Fowler RC, Scharn CR, Ali MA, Cavalieri SJ, Goering RV, Hanson ND. 2014. 10.1126/scitranslmed.3004129. Whole genome mapping of the first reported case of KPC-2–positive 4. Centers for Disease Control and Prevention (CDC). 2013. Notes from the Klebsiella pneumoniae ST258 in Nebraska. Diagn Microbiol Infect Dis field: hospital outbreak of carbapenem-resistant Klebsiella pneumoniae 79:384–386. https://doi.org/10.1016/j.diagmicrobio.2014.03.023. producing New Delhi metallo-beta-lactamase—Denver, Colorado, 2012. 13. Kitchel B, Rasheed JK, Patel JB, Srinivasan A, Navon-Venezia S, Carmeli Y, MMWR Morb Mortal Wkly Rep 62:108. https://www.cdc.gov/mmwr/ Brolund A, Giske CG. 2009. Molecular epidemiology of KPC-producing preview/mmwrhtml/mm6206a5.htm. Klebsiella pneumoniae isolates in the United States: clonal expansion of 5. Mathers AJ, Stoesser N, Sheppard AE, Pankhurst L, Giess A, Yeh AJ, multilocus sequence type 258. Antimicrob Agents Chemother 53: Didelot X, Turner SD, Sebra R, Kasarskis A, Peto T, Crook D, Sifri CD. 2015. 3365–3370. https://doi.org/10.1128/AAC.00126-09. Klebsiella pneumoniae carbapenemase (KPC)-producing K. pneumoniae 14. Bowers JR, Kitchel B, Driebe EM, MacCannell DR, Roe C, Lemmer D, de at a single institution: insights into endemicity from whole-genome se- Man T, Rasheed JK, Engelthaler DM, Keim P, Limbago BM. 2015. Genomic quencing. Antimicrob Agents Chemother 59:1656–1663. https://doi.org/10 analysis of the emergence and rapid global dissemination of the clonal .1128/AAC.04292-14. group 258 Klebsiella pneumoniae pandemic. PLoS One 10:e0133727. 6. Stoesser N, Giess A, Batty EM, Sheppard AE, Walker AS, Wilson DJ, https://doi.org/10.1371/journal.pone.0133727. Didelot X, Bashir A, Sebra R, Kasarskis A, Sthapit B, Shakya M, Kelly D, 15. Yigit H, Queenan AM, Anderson GJ, Domenech-Sanchez A, Biddle JW, Pollard AJ, Peto TEA, Crook DW, Donnelly P, Thorson S, Amatya P, Joshi Steward CD, Alberti S, Bush K, Tenover FC. 2001. Novel carbapenem- S. 2014. Genome sequencing of an extended series of NDM-producing hydrolyzing beta-lactamase, KPC-1, from a carbapenem-resistant strain Klebsiella pneumoniae isolates from neonatal infections in a Nepali hos- of Klebsiella pneumoniae. Antimicrob Agents Chemother 45:1151–1161. pital characterizes the extent of community- versus hospital-associated https://doi.org/10.1128/AAC.45.4.1151-1161.2001. transmission in an endemic setting. Antimicrob Agents Chemother 58: 16. Nordmann P, Cuzon G, Naas T. 2009. The real threat of Klebsiella pneu- 7347–7357. https://doi.org/10.1128/AAC.03900-14. moniae carbapenemase-producing bacteria. Lancet Infect Dis 9:228–236. 7. Wright MS, Perez F, Brinkac L, Jacobs MR, Kaye K, Cober E, van Duin D, https://doi.org/10.1016/S1473-3099(09)70054-4. Marshall SH, Hujer AM, Rudin SD, Hujer KM, Bonomo RA, Adams MD. 17. Deleo FR, Chen L, Porcella SF, Martens CA, Kobayashi SD, Porter AR, Chavda 2014. Population structure of KPC-producing Klebsiella pneumoniae iso- KD, Jacobs MR, Mathema B, Olsen RJ, Bonomo RA, Musser JM, Kreiswirth BN. lates from midwestern U.S. hospitals. Antimicrob Agents Chemother 2014. Molecular dissection of the evolution of carbapenem-resistant multi- 58:4961–4965. https://doi.org/10.1128/AAC.00125-14. locus sequence type 258 Klebsiella pneumoniae. Proc Natl Acad SciUSA 8. Chung The H, Karkey A, Pham Thanh D, Boinett CJ, Cain AK, Ellington M, 111:4988–4993. https://doi.org/10.1073/pnas.1321364111. Baker KS, Dongol S, Thompson C, Harris SR, Jombart T, Le Thi Phuong T, 18. Rodrigues C, Machado E, Ramos H, Peixe L, Novais Â. 2014. Expansion of Tran Do Hoang N, Ha Thanh T, Shretha S, Joshi S, Basnyat B, Thwaites G, ESBL-producing Klebsiella pneumoniae in hospitalized patients: a suc- Thomson NR, Rabaa MA, Baker S. 2015. A high-resolution genomic cessful story of international clones (ST15, ST147, ST336) and epidemic analysis of multidrug-resistant hospital outbreaks of Klebsiella pneu- plasmids (IncR, IncFIIK). Int J Med Microbiol 304:1100–1108. https://doi moniae. EMBO Mol Med 7:227–239. https://doi.org/10.15252/emmm .201404767. .org/10.1016/j.ijmm.2014.08.003. 9. Bogdanovich T, Adams-Haduch JM, Tian G-B, Nguyen MH, Kwak EJ, Muto 19. Munoz-Price LS, Quinn JP. 2009. The spread of Klebsiella pneumoniae May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 16 ® ESBL K. pneumoniae carbapenemases: a tale of strains, plasmids, and transposons. Clin Infect tertiary care hospitals reveals high frequency of non-CG258 clones of Dis 49:1739–1741. https://doi.org/10.1086/648078. carbapenem-resistant Klebsiella pneumoniae with distinct clinical char- 20. Richter SN, Frasson I, Franchin E, Bergo C, Lavezzo E, Barzon L, Cavallaro acteristics. Antimicrob Agents Chemother 60:332–342. https://doi.org/ A, Palù G. 2012. KPC-mediated resistance in Klebsiella pneumoniae in two 10.1128/AAC.01775-15. hospitals in Padua, Italy, June 2009–December 2011: massive spreading 36. Ahn C, Yoon SS, Yong TS, Jeong SH, Lee K. 2016. The resistance mech- of a KPC-3-encoding plasmid and involvement of non-intensive care anism and clonal distribution of tigecycline-nonsusceptible Klebsiella units. Gut Pathog 4:7. https://doi.org/10.1186/1757-4749-4-7. pneumoniae isolates in Korea. Yonsei Med J 57:641–646. https://doi.org/ 21. Pitout JD, Laupland KB. 2008. Extended-spectrum beta-lactamase- 10.3349/ymj.2016.57.3.641. producing Enterobacteriaceae: an emerging public-health concern. 37. Ruiz-Garbajosa P, Hernández-García M, Beatobe L, Tato M, Méndez MI, Lancet Infect Dis 8:159–166. https://doi.org/10.1016/S1473-3099(08) Grandal M, Aranzábal L, Alonso S, Lópaz MÁ, Astray J, Cantón R. 2016. A 70041-0. single-day point-prevalence study of faecal carriers in long-term care 22. President’s Council of Advisors on Science and Technology. 2014. Report hospitals in Madrid (Spain) depicts a complex clonal and polyclonal to the president on combating antibiotic resistance. Centers for Disease dissemination of carbapenemase-producing Enterobacteriaceae. J Anti- Control and Prevention, Atlanta, GA. https://www.cdc.gov/drugresistance/ microb Chemother 71:348–352. https://doi.org/10.1093/jac/dkv355. pdf/report-to-the-president-on-combating-antibiotic-resistance.pdf. 38. Mansour W, Grami R, Ben Haj Khalifa A, Dahmen S, Châtre P, Haenni M, 23. Emerson MO, Bratter J, Howell J, Jeanty PW, Cline M. 2012. Houston Aouni M, Madec JY. 2015. Dissemination of multidrug-resistant blaCTX- region grows more racially/ethnically diverse, with small declines in M-15/IncFIIk plasmids in Klebsiella pneumoniae isolates from hospital- segregation. A joint report analyzing census data from 1990, 2000, and and community-acquired human infections in Tunisia. Diagn Microbiol 2010. Rice University, Houston, TX. http://kinder.rice.edu/uploadedFiles/ Infect Dis 83:298–304. https://doi.org/10.1016/j.diagmicrobio.2015.07 Urban_Research_Center/Media/Houston%20Region%20Grows%20More .023. %20Ethnically%20Diverse%202-13.pdf. 39. Gona F, Barbera F, Pasquariello AC, Grossi P, Gridelli B, Mezzatesta ML, 24. Gomez SA, Pasteran FG, Faccone D, Tijet N, Rapoport M, Lucero C, Caio C, Stefani S, Conaldi PG. 2014. In vivo multiclonal transfer of Lastovetska O, Albornoz E, Galas M, Melano RG, Corso A, Petroni A. 2011. blaKPC-3 from Klebsiella pneumoniae to Escherichia coli in surgery Clonal dissemination of Klebsiella pneumoniae ST258 harbouring KPC-2 patients. Clin Microbiol Infect 20:O633–O635. https://doi.org/10 in Argentina. Clin Microbiol Infect 17:1520–1524. https://doi.org/10 .1111/1469-0691.12577. .1111/j.1469-0691.2011.03600.x. 40. Habeeb MA, Haque A, Nematzadeh S, Iversen A, Giske CG. 2013. High 25. Di Carlo P, Gulotta G, Casuccio A, Pantuso G, Raineri M, Farulla CA, prevalence of 16S rRNA methylase RmtB among CTX-M extended- Bonventre S, Guadagnino G, Ingrassia D, Cocorullo G, Mammina C, spectrum beta-lactamase-producing Klebsiella pneumoniae from Islam- Giarratano A. 2013. KPC-3 Klebsiella pneumoniae ST258 clone infection in abad, Pakistan. Int J Antimicrob Agents 41:524–526. https://doi.org/10 postoperative abdominal surgery patients in an intensive care setting: .1016/j.ijantimicag.2013.02.017. analysis of a case series of 30 patients. BMC Anesthesiol 13. https://doi 41. Doi Y, Hazen TH, Boitano M, Tsai Y-C, Clark TA, Korlach J, Rasko DA. 2014. .org/10.1186/1471-2253-13-13. Whole-genome assembly of Klebsiella pneumoniae coproducing NDM-1 26. Valentín-Martín A, Valverde-De Francisco A, Bosque-Vall M, Cantón- and OXA-232 carbapenemases using single-molecule, real-time se- Moreno R. 2013. First report of colistin-resistant KPC-2 producing ST258- quencing. Antimicrob Agents Chemother 58:5947–5953. https://doi.org/ Klebsiella pneumoniae in Spain. Enferm Infecc Microbiol Clin 31:489–491. 10.1128/AAC.03180-14. https://doi.org/10.1016/j.eimc.2012.12.002. 42. Villa L, Poirel L, Nordmann P, Carta C, Carattoli A. 2012. Complete 27. Rodríguez-Zulueta P, Silva-Sanchez J, Barrios H, Reyes-Mar J, Velez-Perez sequencing of an IncH plasmid carrying the blaNDM-1, blaCTX-M-15 and F, Arroyo-Escalante S, Ochoa-Carrera L, Delgado-Sapien G, Morales- qnrB1 genes. J Antimicrob Chemother 67:1645–1650. https://doi.org/10 Espinoza MdR, Tamayo-Legorreta E, Hernandez-Castro R, Garza-Ramos .1093/jac/dks114. U. 2013. First outbreak of KPC-3-producing Klebsiella pneumoniae 43. Hirsch EB, Chang KT, Lasco TM, Caeiro JP, Tam VH. 2011. Emergence of (ST258) clinical isolates in a Mexican medical center. Antimicrob Agents KPC-producing Klebsiella pneumoniae in Texas. Diagn Microbiol Infect Chemother 57:4086–4088. https://doi.org/10.1128/AAC.02530-12. Dis 69:234–235. https://doi.org/10.1016/j.diagmicrobio.2010.09.003. 28. Marsh JW, Krauland MG, Nelson JS, Schlackman JL, Brooks AM, Pasculle 44. Paltansing S, Kraakman MEM, Ras JMC, Wessels E, Bernards AT. 2013. AW, Shutt KA, Doi Y, Querry AM, Muto CA, Harrison LH. 2015. Genomic Characterization of fluoroquinolone and cephalosporin resistance mech- epidemiology of an endoscope-associated outbreak of Klebsiella pneu- anisms in Enterobacteriaceae isolated in a Dutch teaching hospital moniae carbapenemase (KPC)-producing K. pneumoniae. PLoS One 10: reveals the presence of an Escherichia coli ST131 clone with a specific e0144310. https://doi.org/10.1371/journal.pone.0144310. mutation in parE. J Antimicrob Chemother 68:40–45. https://doi.org/10 29. Tedja R, Nowacki A, Fraser T, Fatica C, Griffiths L, Gordon S, Isada C, van .1093/jac/dks365. Duin D. 2014. The impact of multidrug resistance on outcomes in 45. Gomez-Simmonds A, Greenman M, Sullivan SB, Tanner JP, Sowash MG, ventilator-associated pneumonia. Am J Infect Control 42:542–545. Whittier S, Uhlemann A-C. 2015. Population structure of Klebsiella pneu- https://doi.org/10.1016/j.ajic.2013.12.009. moniae causing bloodstream infections at a New York City tertiary care 30. Chen L, Mathema B, Chavda KD, DeLeo FR, Bonomo RA, Kreiswirth BN. hospital: diversification of multidrug-resistant isolates. J Clin Microbiol 2014. Carbapenemase-producing Klebsiella pneumoniae: molecular and 53:2060–2067. https://doi.org/10.1128/JCM.03455-14. genetic decoding. Trends Microbiol 22:686 – 696. https://doi.org/10 46. Dropa M, Lincopan N, Balsalobre LC, Oliveira DE, Moura RA, Fernandes .1016/j.tim.2014.09.003. MR, da Silva QM, Matté GR, Sato MIZ, Matté MH. 2016. Genetic back- 31. Lee CR, Lee JH, Park KS, Kim YB, Jeong BC, Lee SH. 2016. Global dissemi- ground of novel sequence types of CTX-M-8- and CTX-M-15-producing nation of carbapenemase-producing Klebsiella pneumoniae: epidemiology, Escherichia coli and Klebsiella pneumoniae from public wastewater treat- genetic context, treatment options, and detection methods. Front Mi- ment plants in Sao Paulo, Brazil. Environ Sci Pollut Res Int 23:4953–4958. crobiol 7:895. https://doi.org/10.3389/fmicb.2016.00895. https://doi.org/10.1007/s11356-016-6079-5. 32. Pitout JDD, Nordmann P, Poirel L. 2015. Carbapenemase-producing 47. Geraci DM, Bonura C, Giuffrè M, Saporito L, Graziano G, Aleo A, Fasciana Klebsiella pneumoniae, a key pathogen set for global nosocomial dom- T, Di Bernardo F, Stampone T, Palma DM, Mammina C. 2015. Is the inance. Antimicrob Agents Chemother 59:5873–5884. https://doi.org/10 monoclonal spread of the ST258, KPC-3-producing clone being replaced .1128/AAC.01019-15 in southern Italy by the dissemination of multiple clones of carbapenem- 33. Bonura C, Giuffrè M, Aleo A, Fasciana T, Di Bernardo F, Stampone T, nonsusceptible, KPC-3-producing Klebsiella pneumoniae? Clin Microbiol Giammanco A, Palma DM, Mammina C. 2015. An update of the evolving Infect 21:e15–e17. https://doi.org/10.1016/j.cmi.2014.08.022. epidemic of blaKPC carrying Klebsiella pneumoniae in Sicily, Italy, 2014: 48. Kim JO, Song SA, Yoon EJ, Shin JH, Lee H, Jeong SH, Lee K. 2017. emergence of multiple non-ST258 clones. PLoS One 10:e0132936. Outbreak of KPC-2-producing Enterobacteriaceae caused by clonal dis- https://doi.org/10.1371/journal.pone.0132936. semination of Klebsiella pneumoniae ST307 carrying an IncX3-type plas- 34. Park DJ, Yu JK, Park KG, Park YJ. 2015. Genotypes of ciprofloxacin- mid harboring a truncated Tn4401a. Diagn Microbiol Infect Dis 87: resistant Klebsiella pneumoniae in Korea and their characteristics accord- 343–348. https://doi.org/10.1016/j.diagmicrobio.2016.12.012. ing to the genetic lineages. Microb Drug Resist 21:622–630. https://doi .org/10.1089/mdr.2015.0001. 49. Wyres KL, Gorrie C, Edwards DJ, Wertheim HFL, Hsu LY, Van Kinh N, 35. Ocampo AM, Chen L, Cienfuegos AV, Roncancio G, Chavda KD, Kre- Zadoks R, Baker S, Holt KE. 2015. Extensive capsule locus variation and iswirth BN, Jiménez JN. 2015. A two-year surveillance in five Colombian large-scale genomic recombination within the Klebsiella pneumoniae May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 17 ® Long et al. clonal group 258. Genome Biol Evol 7:1267–1279. https://doi.org/10 Sengupta A, Mohades Z, Dery S, Deng W, Earl E, Demeter DV, Mills K, .1093/gbe/evv062. Mihai G, Ruzic L, Ketz N, Reineberg A, Reddan MC, Goddings A, Gonzalez- 50. Wyres KL, Wick RR, Gorrie C, Jenney A, Follador R, Thomson NR, Holt KE. Castillo J, Gorgolewski KJ, Froehlich C, Dekel G, Margulies DS, Craddock 2016. Identification of Klebsiella capsule synthesis loci from whole ge- RC, Fulcher BD, Glatard T, Das S, Madjar C, Sengupta A, Mohades Z, Dery nome data. Microb Genom 2:e000102. https://doi.org/10.1099/mgen.0 S, Deng W, Earl E, Demeter DV, Mills K, Mihai G, Ruzik L, et al. 2016. 2015 .000102. Brainhack proceedings. Gigascience 5(Suppl 1):46. https://doi.org/10 51. Holt KE, Wertheim H, Zadoks RN, Baker S, Whitehouse CA, Dance D, .1186/s13742-016-0147-0. Jenney A, Connor TR, Hsu LY, Severin J, Brisse S, Cao H, Wilksch J, Gorrie 64. Chen K, Eddens T, Trevejo-Nunez G, Way EE, Elsegeiny W, Ricks DM, Garg C, Schultz MB, Edwards DJ, Nguyen KV, Nguyen TV, Dao TT, Mensink M, AV, Erb CJ, Bo M, Wang T, Chen W, Lee JS, Gaffen SL, Kolls JK. 2016. IL-17 Minh VL, Nhu NTK, Schultsz C, Kuntaman K, Newton PN, Moore CE, receptor signaling in the lung epithelium is required for mucosal chemo- Strugnell RA, Thomson NR. 2015. Genomic analysis of diversity, popula- kine gradients and pulmonary host defense against K. pneumoniae. Cell tion structure, virulence, and antimicrobial resistance in Klebsiella pneu- https://doi.org/10.1016/j.chom.2016.10.003. Host Microbe 20:596–605. moniae, an urgent threat to public health. Proc Natl Acad SciUSA 65. Bachman MA, Miller VL, Weiser JN. 2009. Mucosal lipocalin 2 has pro- 112:E3574–E3581. https://doi.org/10.1073/pnas.1501049112. inflammatory and iron-sequestering effects in response to bacterial 52. Lam MMC, Wick RR, Wyres KL, Gorrie C, Judd LM, Brisse S, Jenney A, Holt enterobactin. PLoS Pathog 5:e1000622. https://doi.org/10.1371/journal KE. 2017. Frequent emergence of pathogenic lineages of Klebsiella pneu- .ppat.1000622. moniae via mobilisation of yersiniabactin and colibactin. bioRxiv https:// 66. Bachman MA, Oyler JE, Burns SH, Caza M, Lepine F, Dozois CM, Weiser doi.org/10.1101/098178. JN. 2011. Klebsiella pneumoniae yersiniabactin promotes respiratory tract 53. Diago-Navarro E, Chen L, Passet V, Burack S, Ulacia-Hernando A, Kodi- infection through evasion of lipocalin 2. Infect Immun 79:3309–3316. yanplakkal RP, Levi MH, Brisse S, Kreiswirth BN, Fries BC. 2014. https://doi.org/10.1128/IAI.05114-11. Carbapenem-resistant Klebsiella pneumoniae exhibit variability in capsu- 67. Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for lar polysaccharide and capsule associated virulence traits. J Infect Dis Illumina sequence data. Bioinformatics 30:2114–2120. https://doi.org/10 210:803–813. https://doi.org/10.1093/infdis/jiu157. .1093/bioinformatics/btu170. 54. Mehta SC, Rice K, Palzkill T. 2015. Natural variants of the KPC-2 carbap- 68. Liu Y, Schroder J, Schmidt B. 2013. Musket: a multistage k-mer spectrum- enemase have evolved increased catalytic efficiency for ceftazidime based error corrector for Illumina sequence data. Bioinformatics 29: hydrolysis at the cost of enzyme stability. PLoS Pathog 11:e1004949. 308–315. https://doi.org/10.1093/bioinformatics/bts690. https://doi.org/10.1371/journal.ppat.1004949. 69. Garrison E, Marth G. 2012. Haplotype-based variant detection from 55. Bushnell G, Mitrani-Gold F, Mundy LM. 2013. Emergence of New Delhi short-read sequencing. arXiv arXiv:1207.3907v2 [q-bio.GN]. http://arXiv metallo-beta-lactamase type 1-producing enterobacteriaceae and non- .org/abs/1207.3907v2. enterobacteriaceae: global case detection and bacterial surveillance. Int 70. Price MN, Dehal PS, Arkin AP. 2010. FastTree 2—approximately J Infect Dis 17:e325–e333. https://doi.org/10.1016/j.ijid.2012.11.025. maximum-likelihood trees for large alignments. PLoS One 5:e9490. 56. Gupta K, Hooton TM, Naber KG, Wullt B, Colgan R, Miller LG, Moran GJ, https://doi.org/10.1371/journal.pone.0009490. Nicolle LE, Raz R, Schaeffer AJ, Soper DE. 2011. International clinical 71. Inouye M, Dashnow H, Raven LA, Schultz MB, Pope BJ, Tomita T, Zobel practice guidelines for the treatment of acute uncomplicated cystitis J, Holt KE. 2014. SRST2: rapid genomic surveillance for public health and and pyelonephritis in women: a 2010 update by the Infectious Diseases hospital microbiology labs. Genome Med 6:90. https://doi.org/10.1186/ Society of America and the European Society for Microbiology and s13073-014-0090-6. Infectious Diseases. Clin Infect Dis 52:e103– e120. https://doi.org/10 72. Nurk S, Bankevich A, Antipov D, Gurevich A, Korobeynikov A, Lapidus A, .1093/cid/ciq257. Prjibelsky A, Pyshkin A, Sirotkin A, Sirotkin Y, Stepanauskas R, McLean J, 57. Freund Y, Schapire RE. 1997. A decision-theoretic generalization of on-line Lasken R, Clingenpeel SR, Woyke T, Tesler G, Alekseyev MA, Pevzner PA. learning and an application to boosting. J Comput Syst Sci 55:119–139. 2013. Assembling genomes and mini-metagenomes from highly chi- https://doi.org/10.1006/jcss.1997.1504. meric reads, p 158 –170. In Deng M, Jiang R, Sun F, Zhang X (ed), 58. Hastie T, Rosset S, Zhu J, Zou H. 2009. Multi-class AdaBoost. Stat Inter- Research in computational molecular biology: 17th Annual Interna- face 2:349–360. https://doi.org/10.4310/SII.2009.v2.n3.a8. tional Conference, RECOMB 2013, Beijing, China, April 7–10, 2013. 59. Ellington MJ, Ekelund O, Aarestrup FM, Canton R, Doumith M, Giske C, Springer, Berlin, Germany. http://link.springer.com/chapter/10.1007/ Grundman H, Hasman H, Holden MTG, Hopkins KL, Iredell J, Kahlmeter 978-3-642-37195-0_13. G, Köser CU, MacGowan A, Mevius D, Mulvey M, Naas T, Peto T, Rolain 73. Kumar S, Stecher G, Tamura K. 2016. MEGA7: molecular evolutionary J-M, Samuelsen Ø, Woodford N. 2017. The role of whole genome se- genetics analysis version 7.0 for bigger datasets. Mol. Biol. Evol. quencing in antimicrobial susceptibility testing of bacteria: report from 33:1870 –1874. https://doi.org/10.1093/molbev/msw054. the EUCAST subcommittee. Clin Microbiol Infect 23:2–22. https://doi 74. Kumar S, Stecher G, Peterson D, Tamura K. 2012. MEGA-CC: computing .org/10.1016/j.cmi.2016.11.012. core of molecular evolutionary genetics analysis program for automated 60. Metcalf BJ, Chochua S, Gertz RE, Jr., Li Z, Walker H, Tran T, Hawkins PA, and iterative data analysis. Bioinformatics 28:2685–2686. https://doi.org/ Glennen A, Lynfield R, Li Y, McGee L, Beall B, Active Bacterial Core 10.1093/bioinformatics/bts507. Surveillance Team. 2016. Using whole genome sequencing to identify 75. Wattam AR, Abraham D, Dalay O, Disz TL, Driscoll T, Gabbard JL, Gillespie resistance determinants and predict antimicrobial resistance pheno- JJ, Gough R, Hix D, Kenyon R, Machi D, Mao C, Nordberg EK, Olson R, types for year 2015 invasive pneumococcal disease isolates recovered in Overbeek R, Pusch GD, Shukla M, Schulman J, Stevens RL, Sullivan DE, the United States. Clin Microbiol Infect 22:1002.e1–1002.e8. https://doi Vonstein V, Warren A, Will R, Wilson MJC, Yoo HS, Zhang C, Zhang Y, .org/10.1016/j.cmi.2016.08.001. Sobral BW. 2014. PATRIC, the bacterial bioinformatics database and 61. Moran RA, Anantham S, Holt KE, Hall RM. 2017. Prediction of antibiotic analysis resource. Nucleic Acids Res 42:D581–D591. https://doi.org/10 resistance from antibiotic resistance genes detected in antibiotic- .1093/nar/gkt1099. resistant commensal Escherichia coli using PCR or WGS. J Antimicrob 76. Olsen RJ, Fittipaldi N, Kachroo P, Sanson MA, Long SW, Como-Sabetti KJ, Chemother 72:700–704. https://doi.org/10.1093/jac/dkw511. Valson C, Cantu C, Lynfield R, Van Beneden C, Beres SB, Musser JM. 2014. 62. Davis JJ, Boisvert S, Brettin T, Kenyon RW, Mao C, Olson R, Overbeek R, Clinical laboratory response to a mock outbreak of invasive bacterial Santerre J, Shukla M, Wattam AR, Will R, Xia F, Stevens R. 2016. Antimi- infections: a preparedness study. J Clin Microbiol 52:4210–4216. https:// crobial resistance prediction in PATRIC and RAST. Sci Rep 6:27930. https://doi.org/10.1038/srep27930. doi.org/10.1128/JCM.02164-14. 63. Craddock RC, Bellec P, Margules DS, Nichols BN, Pfannmöller JP, Badh- 77. Eagar TN, Tang Q, Wolfe M, He Y, Pear WS, Bluestone JA. 2004. Notch 1 war A, Kennedy D, Poline J, Toro R, Cipollini B, Rokem A, Clark D, signaling regulates peripheral T cell activation. Immunity 20:407–415. Gorgolewski KJ, Craddock RC, Craddock RC, Clark DJ, Das S, Madjar C, https://doi.org/10.1016/S1074-7613(04)00081-0. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 18 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png mBio Pubmed Central

Population Genomic Analysis of 1,777 Extended-Spectrum Beta-Lactamase-Producing Klebsiella pneumoniae Isolates, Houston, Texas: Unexpected Abundance of Clonal Group 307

Loading next page...
 
/lp/pubmed-central/population-genomic-analysis-of-1-777-extended-spectrum-beta-lactamase-LIPOF6V6rm

References (156)

Publisher
Pubmed Central
Copyright
Copyright © 2017 Long et al.
ISSN
2150-7511
eISSN
2150-7511
DOI
10.1128/mBio.00489-17
Publisher site
See Article on Publisher Site

Abstract

RESEARCH ARTICLE crossm Population Genomic Analysis of 1,777 Extended-Spectrum Beta-Lactamase- Producing Klebsiella pneumoniae Isolates, Houston, Texas: Unexpected Abundance of Clonal Group 307 a,b a,b a a S. Wesley Long, Randall J. Olsen, Todd N. Eagar, Stephen B. Beres, a c,d c,d c,d Picheng Zhao, James J. Davis, Thomas Brettin, Fangfang Xia, a,b James M. Musser Center for Molecular and Translational Human Infectious Diseases Research, Department of Pathology and Genomic Medicine, Houston Methodist Research Institute and Houston Methodist Hospital, Houston, Texas, USA ; Department of Pathology and Laboratory Medicine, Weill Cornell Medical College, New York, New York, b c USA ; Computing, Environment and Life Sciences, Argonne National Laboratory, Argonne, Illinois, USA ; Computation Institute, University of Chicago, Chicago, Illinois, USA ABSTRACT Klebsiella pneumoniae is a major human pathogen responsible for high Received 1 April 2017 Accepted 18 April morbidity and mortality rates. The emergence and spread of strains resistant to mul- 2017 Published 16 May 2017 tiple antimicrobial agents and documented large nosocomial outbreaks are espe- Citation Long SW, Olsen RJ, Eagar TN, Beres SB, cially concerning. To develop new therapeutic strategies for K. pneumoniae,itisim- Zhao P, Davis JJ, Brettin T, Xia F, Musser JM. 2017. Population genomic analysis of 1,777 perative to understand the population genomic structure of strains causing human extended-spectrum beta-lactamase-producing infections. To address this knowledge gap, we sequenced the genomes of 1,777 Klebsiella pneumoniae isolates, Houston, Texas: unexpected abundance of clonal group 307. extended-spectrum beta-lactamase-producing K. pneumoniae strains cultured from mBio 8:e00489-17. https://doi.org/10.1128/ patients in the 2,000-bed Houston Methodist Hospital system between September mBio.00489-17. 2011 and May 2015, representing a comprehensive, population-based strain sample. Editor Steven J. Projan, MedImmune Strains of largely uncharacterized clonal group 307 (CG307) caused more infections Copyright © 2017 Long et al. This is an open- access article distributed under the terms of than those of well-studied epidemic CG258. Strains varied markedly in gene content the Creative Commons Attribution 4.0 and had an extensive array of small and very large plasmids, often containing anti- International license. microbial resistance genes. Some patients with multiple strains cultured over time Address correspondence to James M. Musser, were infected with genetically distinct clones. We identified 15 strains expressing the jmmusser@houstonmethodist.org. This article is a direct contribution from a New Delhi metallo-beta-lactamase 1 (NDM-1) enzyme that confers broad resistance Fellow of the American Academy of to nearly all beta-lactam antibiotics. Transcriptome sequencing analysis of 10 phylo- Microbiology. External solicited reviewers: genetically diverse strains showed that the global transcriptome of each strain was Michael Dunne, bioMérieux, Inc.; Robert Bonomo, Louis Stokes Veterans Affairs Medical unique and highly variable. Experimental mouse infection provided new information Center. about immunological parameters of host-pathogen interaction. We exploited the large data set to develop whole-genome sequence-based classifiers that accurately predict clinical antimicrobial resistance for 12 of the 16 antibiotics tested. We con- clude that analysis of large, comprehensive, population-based strain samples can as- sist understanding of the molecular diversity of these organisms and contribute to enhanced translational research. IMPORTANCE Klebsiella pneumoniae causes human infections that are increasingly difficult to treat because many strains are resistant to multiple antibiotics. Clonal group 258 (CG258) organisms have caused outbreaks in health care settings world- wide. Using a comprehensive population-based sample of extended-spectrum beta- lactamase (ESBL)-producing K. pneumoniae strains, we show that a relatively uncom- mon clonal type, CG307, caused the plurality of ESBL-producing K. pneumoniae infections in our patients. We discovered that CG307 strains have been abundant in Houston for many years. As assessed by experimental mouse infection, CG307 strains May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 1 ® Long et al. were as virulent as pandemic CG258 strains. Our results may portend the emergence of an especially successful clonal group of antibiotic-resistant K. pneumoniae. lebsiella pneumoniae is a major human pathogen responsible for high morbidity Kand mortality rates (1–3). Numerous studies have investigated the genetic epide- miology of K. pneumoniae infections in hospitalized patients in the United States and elsewhere (3–8). These studies have often used low-resolution molecular typing such as multilocus sequence typing (MLST) or have studied relatively small collec- tions of organisms. Studies of hospitalized patients in the United States commonly have reported a predominance of clonal group 258 (CG258) strains (9–12). CG258 was first identified in 2009 as a single-locus variant of sequence type 11 (ST11) from Hungary (13) and has since been reported worldwide. Data suggest that CG258 diverged from its most recent common ancestor in 1995, after acquisition of the K. pneumoniae carbapenemase (KPC) gene (14). K. pneumoniae strains producing KPC carbapenemase were first identified in North Carolina in 1996 (15) and now cause multidrug-resistant hospital-acquired infections globally (16). Evidence suggests that there is circulation of K. pneumoniae clones within hospitals, with transmissible resistance plasmids shared between Klebsiella spp. and other pathogens (5, 17–20). This sharing of resistance elements is believed to have contributed to the emergence of multidrug resistance in K. pneumoniae and large, hospital-based outbreaks. The increasing reports of these nosocomial outbreaks are of great clinical concern (21, 22). To develop new therapeutics and control strategies for K. pneumoniae, it is imper- ative to understand the population genomic structure of strains causing human infections. To begin filling this knowledge gap, we sequenced the genomes of 1,777 extended-spectrum beta-lactamase (ESBL)-producing K. pneumoniae strains (Table 1; see Table S1A to C in the supplemental material) cultured from patients in the 2,000-bed Houston Methodist Hospital system between September 2011 and May 2015. These strains were all positive for ESBL production, which accounted for an average of 40% of the K. pneumoniae strains in our laboratory during the study period. The Houston Methodist Hospital system serves patients in the greater Houston metropolitan area (population of approximately 6,000,000), now the most ethnically diverse population center in the United States (23). In addition to our large and diverse local population, patients routinely travel for treatment to Houston Methodist Hospital from across the United States and internationally. The genome sequence data identified unexpectedly high diversity. Strains varied markedly in gene content and had an extensively diverse assortment of large plasmids. These transmissible plasmids often contain antimicrobial resistance genes and represent important elements contributing to the dissemination of resistance markers. RESULTS Whole-genome sequencing of 1,777 ESBL-producing K. pneumoniae strains identified extensive genomic diversity within and between major clades. We sequenced the genome of 1,777 ESBL-producing K. pneumoniae strains (Table 1 and TABLE 1 Summary of clonal groups and specimen sources of 1,777 K. pneumoniae isolates No. of isolates (% of total) Source(s) CG258 CG307 Other CG Total Blood 34 (1.9) 54 (3.0) 59 (3.3) 147 (8.3) Respiratory 130 (7.3) 111 (6.2) 143 (8.0) 384 (21.6) Urine 220 (12.4) 374 (21.0) 333 (18.7) 927 (52.2) Other 90 (5.1) 105 (5.9) 124 (7.0) 319 (18.0) All 474 (26.7) 644 (36.2) 659 (37.1) 1,777 (100) Other specimen sources include tissue, wounds, and body fluids. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 2 ® ESBL K. pneumoniae CG307 Diverse CG258 ST Clade 0.025 FIG 1 Estimates of genetic relationships among 1,777 ESBL-producing K. pneumoniae strains recovered from infected patients in a multiple-hospital system in the Houston metropolitan area. Polymorphisms were called against the genome of CG258 K. pneumoniae reference strain NJST258_2 (NCBI GenBank accession no. CP006918.1). Colors represent the major clonal groups, including CG258 (red) and CG307 (blue), and the remaining heterogeneous STs (gray). Phylogenetic relationships were defined by the neighbor-joining method in FastTreeMP with double precision. The core genome was defined as the chromosomal sequence with phage sequence regions excluded. To simplify presentation, outliers have been cropped out of frame at 3 and 9 o’clock. The outlier strains, which were identified as K. pneumoniae by matrix-assisted laser desorption ionization mass spectrometry and MLST, may be allied with the genomes of some strains referred to as K. quasipneumoniae and K. variicola. Fig. 1; see Table S1A to C and Fig. S1) cultured between September 2011 and May 2015 from infected patients in the 2,000-bed Houston Methodist Hospital system. The sample represents virtually all of the ESBL-producing K. pneumoniae strains recovered during this time period. Over half of the strains were cultured from urine (927 strains, 52.2%). The remainder were cultured from respiratory (384 strains, 21.6%), blood (147 strains, 8.3%), and other tissue, wound, and miscellaneous specimens (319 strains, 18.0%). Much of the literature describing nosocomial outbreaks or hospital strain samples in the United States and elsewhere has reported a predominance of CG258 strains, regardless of the population studied (17, 24–29). In striking contrast, on the basis of genome-wide phylogenetic analysis, our strains were broadly classified into three major genetic clades (Fig. 1; Table 1). Approximately one-fourth (474 strains, 26.7%) were assigned to CG258. These organisms represent progeny of a globally disseminated epidemic clone reported to be the major cause of K. pneumoniae infections in the United States (1, 13, 14, 30–32). The average minimum pairwise genetic distance (a measure of genomic diversity in a population of related strains) between the CG258 strains in our sample was 346 single nucleotide polymorphisms (SNPs). More than one-third of the strains (635, 35.7%) were assigned to CG307. These organisms were cultured from patients at all of the hospitals in our system; that is, no single hospital was responsible for the abundance of CG307 infections (Fig. 2). Similarly, CG258 strains May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 3 ® Long et al. CG Hospital A CG258 ybt 17 clb 3 Mortality Hospital B CG307 ybt 10 Hospital C Other ST Other ybt Hospital D Hospital E FIG 2 Spatial relationships of ESBL-producing K. pneumoniae strains from the multiple hospitals in the Houston Methodist system with clonal groups and associated patient deaths. This circular cladogram displays the first isolate from each patient, color coded by the five hospitals of origin (teal, yellow, red, purple, and gray). The circles surrounding the cladogram, from innermost to outermost, indicate the clonal group of the strain, yersiniabactin locus, colibactin locus, and patient death. In the innermost circle, the CG258 strains are red and the CG307 strains are blue. Other clonal types are indicated with a gray bar. The yersiniabactin circle is color coded to indicate which yersiniabactin locus is present, if any, with the two most common loci, ybt17 and ybt10, in yellow and teal, respectively; all other ybt loci are gray. The colibactin circle is purple to indicate the presence of clb3, the only colibactin synthesis locus detected in our collection. In the outermost circle, in-hospital death is indicated by a black line. For patients with multiple strains, only the first isolate from the patient is represented on the tree. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 4 ® ESBL K. pneumoniae FIG 3 ESBL-producing K. pneumoniae strains recovered during the course of this study. The first strain recovered from each patient is graphed (circles, green linear regression). The number of strains classified as CG307 (triangles, blue linear regression) and CG258 (inverted triangles, red linear regression) are also shown. This figure demonstrates that CG258 and CG307 strains were abundant throughout the study period. caused infections in patients at all of our hospitals (Fig. 2). CG307 and CG258 strains were common in our population throughout the study period (Fig. 3). Although strains of CG307 have been identified sporadically in diverse locations globally in recent years (10, 33–40), they have not been reported to be abundant causes of infections in a large unbiased population over a multiyear period. The average minimum pairwise genetic distance between the CG307 strains was 948 SNPs. Com- pared to CG258 organisms that have been extensively studied, very little is known about strains of CG307. The remaining strains (668 strains, 37.6%) in our sample were a genetically heterogeneous array of multilocus STs reported to cause infections worldwide. The average minimum pairwise genetic distance among these 668 strains is 6,580 SNPs, reflecting the far greater diversity present in these strains than in CG258 and CG307 organisms. There were 99 unique previously defined STs present in the strains sequenced (1,430 strains, 80.4%) (see Table S2). An additional 51 strains (2.9%) had novel combinations of known ST alleles, whereas the remaining 296 strains (16.7%) have novel ST alleles resulting in previously undefined ST types. After CG258 and CG307, the three more common STs were ST16 (92 strains, 5.2%), ST15 (49 strains, 2.8%), and ST280 (32 strains, 1.8%). Closure and annotation of five new K. pneumoniae reference genomes. Five strains were sequenced to closure using PacBio long-read sequencing (see Table S3). This allowed the annotation of new reference strains to aid our studies of K. pneu- moniae population genomics and pathogenesis. The five strains were chosen to represent regions of the phylogenetic tree for which existing reference genomes deposited in publically available databases were lacking. In addition, genomes con- taining the bla and OXA-48 genes (see below) were chosen to allow more in-depth NDM-1 analysis of these important strains. KPN11 is representative of the CG307 strains in our patient population. KPN528 was chosen for PacBio sequencing and closure because it was the first K. pneumoniae strain in our sample identified with the bla metallo-beta-lactamase. KPN1481 (ST906) NDM-1 was chosen because it was one of the last strains in our sample containing the bla NDM-1 metallo-beta-lactamase gene, and a closed reference genome has not been reported for this clonal type. Similarly, KPN1482 (ST37) is a clonal type for which a closed reference genome is not available. BK13043, a representative CG258 strain collected in New York City in 2004, was chosen to provide a historical CG258 reference genome. Each new closed genome contained two to five plasmids (see Table S3). The plasmids matched previously identified reference plasmids in GenBank with sequence coverage ranging from 21 to 100%. The bla containing plasmids present in NDM-1 KPN528 and KPN1481 differed. KPN528 had a 292-kb plasmid that had 99% coverage of the pPKPN1 plasmid of reference strain PittNDM01, whereas strain KPN1481 had a May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 5 ® Long et al. 342-kb plasmid that had 74% coverage of the pNDM-MAR plasmid (41, 42). Strain KPN528 contained a second plasmid (76 kb) with 99% coverage of plasmid 3 present in strain PittNDM01. The PittNDM01 K. pneumoniae strain is an ST14 organism isolated from a patient in Pittsburgh, PA, in 2013 (41). Of note, KPN528 also is an ST14 strain and was isolated in December 2012. Taken together, these results suggest that these Pittsburgh and Houston strains are lineally descended from a common ancestor organism. Strains of CG307 are unexpectedly abundant. KPC-producing K. pneumoniae were initially described in the Houston region in 2009 (43). That initial report was followed by the identification of multiple clonal types in 2010, including CG258 and CG307 (10). The CG307 clade of K. pneumoniae was first identified in a Dutch teaching hospital in 2008 (44). Since then, CG307 strains have been identified periodically in Europe, Africa, the Middle East, Asia, Central and South America, and New York City (20, 33–40, 45–47), and a hospital-based outbreak was recently reported in South Korea (48). However, they have not been documented to be of great abundance in any of these areas. Thus, in our genomic data set based on comprehensive, population-based sampling of ESBL-producing K. pneumoniae strains, we demonstrate the existence of a numerically highly successful but poorly characterized genetic group (CG307) with disease prevalence exceeding that of the well-known epidemic CG258 organisms. We identified no statistically significant difference between the observed in-hospital mor- tality rate of all patients infected with CG307 strains and that of patients infected with the CG258 clonal type (chi-square test) (Fig. 2). Similarly, the mortality rate did not differ statistically significantly between patients with bloodstream infections caused by CG307 and CG258 strains (chi-square test). Furthermore, no significant differences between CG307 and CG258 strains were identified in any clinical feature (see Ta- ble S1A), including the hospital of origin, organ transplant status, infection type (urinary tract versus invasive infection), and number of episodes of care (single versus multiple strains recovered longitudinally from the same patient). Comparison of CG258 and CG307 genomes: abundant variation in gene content. The CG258 reference genome (NJST258_2; NCBI GenBank accession no. CP006918.1) has 5,188 annotated genes. Comparison of these genes to our assembled CG307 reference genome (KPN11) by bidirectional BLAST analysis found that 429 of the annotated NJST258_2 genes do not have a significant homolog present in the KPN11 CG307 genome (Fig. 4). Most of these genes (253 out of 429) are annotated as hypothetical proteins. Many of the others are associated with phage, secretion systems, or metabolism. Of the 5,237 annotated genes in the KPN11 CG307 reference assembly, there are 414 annotated genes that lack a homolog in NJST258_2, most of which (281 of 414) are annotated as hypothetical proteins. Most of the other genes that are missing are associated with phage, transduction elements, or metabolism. Some of the genes present in or absent from either clonal type are related to the capsule gene locus. Genetic variation in this locus is well established (17, 49, 50). A summary of the annotated genes present in or absent from NJST258_2 compared to KPN11 CG307 is presented in Tables S4 and S5. Variations in yersiniabactin locus, antimicrobial resis- tance gene content, and capsule locus are discussed in more detail below. Presence of yersiniabactin and colibactin loci in ESBL-producing K. pneu- moniae. Recent studies have emphasized the important contribution of siderophore gene acquisition to severe Klebsiella invasive disease (51, 52). Holt et al. reported that the yersiniabactin locus (ybt) was present in 40.0% of the CG258 and 32.2% of the overall K. pneumoniae genomes analyzed (52). In contrast, we found that only 274 of our 1,777 strains (15.4%) had the ybt locus. Only 13 of the 644 CG307 strains (2.0%) and 85 of the 474 CG258 strains (17.9%) contained the ybt locus. Of the 474 CG258 strains, 76 (16.0%) had both the colibactin synthesis locus (clb) and the ybt locus. In contrast, none of the CG307 strains carried a clb locus. The presence or absence of the yersiniabactin and colibactin loci is shown in Fig. 2 and presented in Table S1A. In contrast to results from other studies using smaller strain collections (51, 52), the presence of the yersiniabactin and colibactin loci was not significantly associated with May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 6 2.6 5.2 2.5 5.1 2.4 5.0 2.3 4.9 2.2 4.8 2.1 4.7 2.0 1.9 4.5 1.8 4.4 1.7 4.3 1.6 4.2 1.5 4.1 1.4 4.0 ESBL K. pneumoniae ST258 ST307 Percent Protein Sequence Identity ST258 ST307 FIG 4 Gene content differences between the reference genomes of CG258 and CG307. CG258 is the globally abundant strain of K. pneumoniae, and CG307 is the locally abundant strain in our study. Bidirectional BLAST was performed by using the PATRIC resource to illustrate the differences in gene content between these two reference genomes. The color indicates the percent identity of the BLAST hit for each gene, with darker shading indicating a bidirectional hit and lighter shading indicating a unidirectional hit (top). A progressive Mauve alignment of the CG258 and CG307 reference genomes is shown. Local colinear blocks are organized by color (bottom). NJST258_2 is the CG258 reference genome, and KPN11 is the CG307 reference assembly. any clinical feature, including invasive disease, chronic infection, or death (no statisti- cally significant difference, chi-square test). Extensive plasmid diversity within ESBL-producing K. pneumoniae. K. pneu- moniae strains commonly have large plasmids that often contain antibiotic resistance May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 7 1.3 3.9 1.2 3.8 1.1 3.7 1.0 3.6 0.9 3.5 0.8 3.4 0.7 3.3 0.6 3.2 0.5 3.1 0.4 3.0 0.3 2.9 0.2 2.8 0.1 2.7 0.0 ® Long et al. genes and multiple replicons. The genome data identified extensive plasmid diversity among the 1,777 strains. The average number of plasmid replicons per isolate was 4.5 (standard deviation, 2.5; range, 0 to 16). Six replicons were each present in more than 25% of the 1,777 strains, i.e., FIBk (1,327 strains), FIIk (1,237 strains), FII (693 strains), R (667 strains), ColRNAI (660 strains), and FIBpQil (494 strains). The distribution of the replicons across the phylogenetic tree is shown in Fig. S1 and Table S1C. The FIBk and FIIk replicons are present in strains of diverse clonal back- grounds, with some clustering in CG307 organisms. The FII, ColRNAI, and FIBpQiL replicons appear to cluster predominantly with CC258 strains, whereas the R replicons are associated with the CG307 organisms. The most common combination of replicons in a single isolate was FIBk, FII, ColRNAI, and FIBpQil (248 strains), followed closely by FII and FIBpQil (244 strains) and FIBk and FIBpQiL (228 strains). The only replicon of the top six to be found in the absence of any of the other five was FIBpQiL (177 strains). Diversity of ESBL genes. Our analysis identified an extensive array of ESBL genes in the 1,777 genomes, including those encoding carbapenemases. Genes in the SHV- LEN-OKP family of ESBLs were most common, being present in 1,442 of the 1,777 strains. The second most common ESBL gene was bla , which was present in 1,161 CTX-M of 1,777 strains (see Table S1B). Among the eight bla alleles identified, bla , CTX-M CTX-M-15 was the most abundant. Other ESBL-encoding genes identified included a member of the TEM family (797 strains) and the KPC gene (581 strains). OXA genes characterized as encoding ESBLs (OXA-23, OXA-24, OXA-48, and OXA-83) were far less common, being present in only 23 strains. bla -positive strains were cultured from six NDM-1 patients, two of whom had strains that also had the OXA-48 gene. Some of the antimicrobial resistance genes detected in these strains, such as TEM30, TEM33, and SHV26, are known to be inhibitor resistant. Overall, KPC alleles were present in 64.6% of the CG258 strains (306 strains) compared to 34.3% of the CG307 strains (221 strains). Among all of our strains with a bla carbapenemase gene (581 strains), bla was the dominant allele (572 strains, KPC KPC-2 98.4%), including 293 CG258 strains (96.4%) and 197 CG307 strains (93.4%). Our results differ from those of other studies reporting that both bla and bla are found in KPC-2 KPC-3 CG258 K. pneumoniae (10, 13, 17, 20, 45, 53) and that CG307 was strongly associated with bla (39). CG307 strains expressing bla have been recently reported in KPC-3 KPC-2 Korea and Colombia (34–36). In our sample, no CG258 strains and only three CG307 strains had the bla allele. The difference between bla and bla is one KPC-3 KPC-2 KPC-3 nonsynonymous SNP (C814T) that results in an H272Y amino acid replacement. Whereas both bla and bla are very efficient at hydrolyzing penicillins, cepha- KPC-2 KPC-3 losporins, and carbapenems, bla has increased enzymatic activity against and KPC-3 confers higher resistance to cephamycins and ceftazidime (15). Four newer KPC variants (bla , bla , bla , and bla ) likely derived from KPC-7 KPC-8 KPC-9 KPC-10 bla have a second nonsynonymous SNP that confers greater catalytic activity KPC-3 against ceftazidime than bla (54). That is, the use of ceftazidime worldwide may be KPC-2 selecting for KPC variants with increasing levels of resistance. Of these newer bla KPC variants, there are four CG258 strains in our collection with bla . Although the KPC-8 reason for the predominance and persistence of bla in our hospital system is KPC-2 unclear, it may be related in part to the relatively conservative use of ceftazidime and ceftazidime-avibactam compared to other beta-lactam antibiotics. The potential for a shift from a CG258-centered K. pneumoniae epidemic to a polyclonal epidemic was first reported in Italy in 2014, and concern has been raised in Korea (33–36). Analysis of bla -carrying patient strains suggests horizontal transfer of a NDM-1 bla -containing plasmid from a classic Indian bla -positive clone into the NDM-1 NDM-1 Houston CG307 population. The identification of the gene encoding New Delhi metallo-beta-lactamase 1 (bla ) in our patient strains is concerning because that NDM-1 enzyme confers very broad resistance to nearly all beta-lactam antibiotics and can be readily transferred between strains by plasmids. Since its discovery in 2009, the bla gene has spread globally among members of the family Enterobacteriaceae; NDM-1 however, the number of reported human infections remains relatively low. As of May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 8 ® ESBL K. pneumoniae TABLE 2 bla -containing strains recovered from six patients (A to F) NDM-1 Strain Patient ST Date collected Source ST geographic association(s) 528 A ST14 12/2012 Sputum India 537 A ST14 12/2012 Trachea India 553 A SLV of ST14 12/2012 Wound India 603 A SLV of ST14 1/2013 Urine India 658 A ST14 2/2013 Blood India 755 B SLV of ST14 4/2013 Urine India 764 B ST14 4/2013 Urine India 1010 B ST14 10/2013 Urine India 1273 B SLV of ST14 1/2014 Urine India 1145 C ST14 12/2013 Urine India 1402 C SLV of ST14 4/2014 Bile India 1426 D ST37 5/2014 Urine India/United Kingdom 1458 D ST37 5/2014 Urine India/United Kingdom 1481 E ST906 6/2014 Urine Israel 1844 F ST307 10/2014 Urine Houston SLV, single-locus variant. 6 January 2017, only 175 cases have been reported in the United States, including 2 in Texas (https://www.cdc.gov/hai/organisms/cre/TrackingCRE.html). To identify strains with bla in our strain sample, assembled contigs were mapped to a custom NDM-1 database of all published bla -bearing plasmids. We identified 15 strains from six NDM-1 patients with the bla gene (Table 2; see Fig. S2). All contigs containing the NDM-1 bla gene are 100% identical to published plasmid sequences, indicating that no NDM-1 novel bla sequences have been introduced into or evolved de novo in Houston NDM-1 patients. Contigs from three strains were identical. Consistent with the idea that strains with the bla gene were imported to Houston, the majority of the bla - NDM-1 NDM-1 containing strains belong to ST14 and ST37, two STs endemic in India, Pakistan, and elsewhere in southern Asia (55). Unexpectedly, the bla gene was detected in one NDM-1 CG307 strain. It is possible that this strain acquired the bla plasmid while circu- NDM-1 lating locally in Houston. Consistent with this idea, the bla -containing plasmid in NDM-1 our CG307 strain (Table 2, patient F) is closely similar to a bla -containing plasmid NDM-1 in two Houston patients (Table 2, patients A and B). Extensive diversity in the capsule locus. K. pneumoniae capsule is an important virulence factor, and the capsule synthesis locus has considerable diversity in gene content between clonal groups (17, 49–51). Diverse associations among different wzi genotypes, bla genotypes, and STs have been reported, and some clonal groupings KPC have been identified (45, 53). Recently, Holt et al. (44) proposed a new nomenclature for capsule locus genotyping (KL type) that encompasses the diversity of the known locus and described a software tool (Kaptive) for capsule locus genotyping. Using Kaptive, we identified 72 unique KL capsule locus types and potential novel variants (see Table S1A). The two most common KL capsule locus types among CG258 strains were KL106 (n  259) and KL107 (n  117). KL107 has the wzi154 gene allele. Although previous studies reported that the wzi154 allele in CG258 strains was associated with the bla gene, this was not the case in our strain sample. We found that almost all KPC-3 of our CG258 strains with a KPC gene contain bla . The CG307 strains we studied KPC-2 had less diversity in the capsule locus. We found that 524 of 526 strains had the KL102 capsule locus, including all 119 ST307 strains with the bla gene. KPC-2 In general, a given clonal group had a predominant KL capsule locus type, although some exceptions were observed (see Table S1A). CG258 is notable for having two capsule gene loci common in the group. Other clonal types with notable capsule locus diversity include ST15 (49 strains, four KL capsule locus types) and ST37 (18 strains, eight KL capsule locus types). K. pneumoniae clinical strains recovered over time from the same patient. Many Houston Methodist patients (n  357) had two or more K. pneumoniae strains cultured from different specimens, in some cases more than 3 years apart (Fig. 5). To investigate whether these different episodes in the same patient represent infections caused by May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 9 ® Long et al. FIG 5 Genetic variation in ESBL-producing K. pneumoniae clinical strains in the same patient over time. (A) Clonal group of each strain recovered from 196 patients with multiple strains. (B) Clonal group of each isolate recovered from 10 patients (numbered 1 to 10) with multiple strains spanning the longest period. Each circle represents one isolate. The vertically overlapping circles shown for patient 5 indicate that two different strains were recovered from different specimens on the same day or consecutive days. the same or a different clone, we used the whole-genome sequence data to determine the relatedness of the organisms recovered. For this analysis, a second episode was defined as a strain isolated more than 35 days after the prior infection event. The 35-day-criterion was selected by using the Infectious Diseases Society of America definition of a late cure of cystitis (cystitis was the most common infection type in our collection) (56). Of the 196 patients with multiple strains recovered in different epi- sodes, we found that 42 (21.4%) were infected with different clones during the study period (Fig. 5A). Similarly, 10 patients had two genomically distinct strains recovered on the same day (Fig. 5B). Consistent with the hypothesis that CG307 strains are as virulent and able to cause recurrent or chronic infections as CG258 strains, the frequency of CG307 and CG258 strains recovered from patients with multiple episodes did not differ significantly (Fisher exact test, no statistically significant difference). Of note, no simple pattern of change in capsule locus or antimicrobial resistance genes was identified among the multiple-episode patients. Extensive transcriptome diversity among and within clades. Whole-genome phylogenetic analysis revealed that the ESBL-producing K. pneumoniae strains causing infections in our health care system could be broadly classified into three major genetic clades. To begin investigating the effect of the extensive genomic diversity on global gene expression, 10 rationally selected strains were analyzed by genome-wide tran- scriptome sequencing (RNA-seq) analysis. The 10 strains were selected to broadly represent the overall population structure, including the year of collection, phage content, antimicrobial resistance gene content, and numerical dominance of CG (see Table S6). Given the extensive genomic variation present, we hypothesized that each strain would have a unique transcriptome. The 10 strains had nearly superimposable growth curves, suggesting that observed differences in global transcriptome were attributable to genetic differences among strains rather than growth characteristics (Fig. 6A). Transcript libraries were prepared by standard methods and sequenced to very deep coverage (mean, 22.2 [range, 14.8 to 33.6] million reads per replicate) to minimize the false discovery of significantly altered gene expression. Transcripts were counted relative to CG258 reference strain NJST258_2. Consistent with our hypothesis, principal- component analysis showed differences in the transcriptomes of these strains, both within and between the CGs (Fig. 6B). On average, each strain differentially expressed 427 genes compared to the reference strain (Fig. 6C). In the aggregate, the maximum number of differentially expressed genes was 617, indicative of the substantial diversity in these strains (Fig. 6D). Also consistent with gene content differences identified among the 10 strains, approximately half of the reads for each strain did not map to the NJST258_2 reference chromosome (mean, 46.3% reads unmapped per duplicate strain pair; range, 42.2 to 53.5%). BLAST analysis revealed that most unmapped reads derived from plasmid-borne genes, insertion sequences, and other mobile elements. Similar May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 10 ® ESBL K. pneumoniae FIG 6 RNA-seq analysis of 10 genomically diverse ESBL-producing K. pneumoniae strains. Strains of CG258 (n  4, red), CG307 (n  3, blue), and other STs (n  3, gray) were analyzed. (A) Growth curves of strains. RNA-seq analysis of duplicate cultures was done (the mean absorbance of each strain pair is shown); the growth curves of all of the strains are similar. Cells were harvested at the mid-exponential phase of growth (OD of 0.6; horizontal dashed green line). (B) RNA-seq transcriptome data for chromosomally located genes of each strain were compared to those of CG258 reference strain BK30684. Plotted are expression relationships among the strains based on principal component 1 (PC1) and principal component 2 (PC2), which account for the two largest unrelated variances in the data. The strains are distributed across both axes, showing transcriptome diversity within and between CGs. (C) Number of genes with significantly altered expression relative to reference strain BK30684. Plotted are the numbers of genes with significantly increased (positive y axis) and decreased (negative y axis) expression for each strain. (D) Box plot summarizing data from the 10 strains combined or within CGs. results were generated when reads were mapped to CG307 strain KPN11 (see Fig. S3A) or CG14 strain KPN528 (see Fig. S3B). Also, the RNA-Seq data confirmed expression of the genes encoding NDM-1, OXA-48, KPC, and yersiniabactin by strains carrying these genes (Table 2; see Fig. S3C to F and Table S6). That is, among these 10 strains, mobile gene content contributed to nearly half of all differentially expressed gene transcripts. After accounting for differences in gene content, we identified no simple patterns of transcriptome differences, either within or between the CGs. Mouse model of ESBL-producing K. pneumoniae pneumonia. A very unexpected finding from the whole-genome sequence data was that CG307 organisms were very abundant among the ESBL-producing K. pneumoniae strains recovered from patients in our Houston Methodist Hospitals. To test the hypothesis that CG307 strains are as virulent as the globally disseminated and numerically common CG258 clone, refer- ence strains KPN1411 (CG307) and KPN113 (CG258) were compared using a mouse model of pneumonia. Mice were inoculated by intranasal instillation and necropsied at 24 h, and the numbers of CFU recovered from the infection site (lung) and a site of dissemination (spleen) were determined. CFU were recovered from the lungs of all of the mice. There was no significant difference in the number of CFU cultured from the lungs of mice infected with CG307 strain KPN1411 and CG258 strain KPN113 5 4 (mean, 1.22  10 CFU for KPN1411 compared to 8.98  10 CFU for KPN113; data not shown). Similarly, there was no significant difference in dissemination to the May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 11 ® Long et al. FIG 7 Host immune response to ESBL-producing K. pneumoniae in a mouse model of pneumonia. Flow cytometry analysis of leukocytes recovered from infected lungs, including neutrophils (A), monocytes (B), T cells (C), B cells (D), alveolar macrophages (E), CD4 T cells (F),  T cells (G), IL-17-producing CD4 T cells (H), and IL-17-producing  T cells (I). (J) IL-17-producing  T cells were also recovered from the spleen. spleen (four of five mice for KPN1411 compared to three of five mice for KPN113; data not shown). To further investigate host-pathogen interactions, the host immune response was examined in infected mice. Mice were inoculated by intranasal instillation, monitored for 7 days, and sacrificed, and flow cytometry analysis of immune cells recovered from their lungs and spleens was conducted (Fig. 7). Compared with uninfected control mice, animals infected with ESBL-producing K. pneumoniae had significantly more total CD45 leukocytes in their lungs, including neutrophils (9.04-fold; Fig. 7A), monocytes (4.06-fold; Fig. 7B), T cells (2.63-fold; Fig. 7C), and  T cells (25.21-fold; Fig. 7D). There was no significant difference in the numbers of B lymphocytes or alveolar macrophages (Fig. 7E and F). K. pneumoniae-infected lungs also had significantly more  and CD4 T cells expressing IL-17 (9.50- and 12.07-fold, respectively, Fig. 7G and H). However, compared to the CG258 strain, the CG307 strain elicited a significantly smaller fraction of IL-17-producing  T cells (Fig. 7I). IL-17-expressing CD4 T cells were also increased in the spleens of K. pneumoniae-infected animals, likely representing a reservoir of proliferating reactivated memory cells (4.78-fold; Fig. 7J). Taken together, the mouse pneumonia data demonstrate that the CG307 reference strain was as virulent as the epidemic CG258 reference strain in this disease model. Prediction of resistance to antimicrobial agents based on whole-genome se- quence data. The large number of genome sequences generated for this study of drug-resistant organisms gave us the opportunity to use machine learning-based classifiers to predict resistance to antimicrobial agents. To test the ability to use whole-genome sequence data to predict antimicrobial resistance phenotypes, we used the AdaBoost algorithm (57, 58) to generate classifiers for 16 agents commonly used to treat human K. pneumoniae infections. Results of the classifier prediction strongly correlated with the clinical phenotype, as measured by the area under the curve (AUC) and the F1 score (Fig. 8). Importantly, the prediction data generated the three highest scores for the carbapenem antibiotics. DISCUSSION In the aggregate, our data demonstrate that ESBL-producing K. pneumoniae strains causing infections in our large hospital system over several years have substantial genomic diversity. Analysis of large, comprehensive, population-based strain samples May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 12 ® ESBL K. pneumoniae FIG 8 Receiver operating characteristic curves of 16 AdaBoost-based classifiers predicting antimicrobial resistance. Antibiotic agent abbreviations: AMK, amikacin; ATM, aztreonam; FEP, cefepime; FOX, cefoxitin; CIP, ciprofloxacin; ETP, ertapenem; FOF, fosfomycin; GEN, gentamicin; IPM, imipenem; LVX, levofloxacin; MEM, meropenem; TZP, piperacillin-tazobactam; TET, tetracycline; TGC, tigecycline; TOB, tobramycin; SXT, trimethoprim-sulfamethoxazole. AUC is area under the concentration-time curve. The F1 score, the harmonic mean of precision and recall, is commonly used to compare classification methods. Data are the results of a 10-fold cross validation. can assist our understanding of the molecular pathogenesis of these organisms and thereby contribute to enhanced translational research. We identified the heretofore unrecognized numerical success of CG307 strains in the Houston region (Fig. 1). CG307 strains caused more infections than the pandemic CG258 strains throughout the study period (Fig. 3). We confirmed that CG307 strains continue to be an abundant cause of infections in our hospital system by sequencing an additional 96 ESBL-producing K. pneumoniae strains recovered in 2017 (S. W. Long, unpublished data). Although CG307 strains have been previously identified in our area, the extent to which this clonal type has successfully coexisted with CG258 to achieve a state of codominance in our region was a very unexpected discovery. The reason(s) the CG307 clone has become abundant in Houston but not other parts of North America is not understood. CG307 strains have not been reported previously to be a prominent cause of infections in a large, unbiased population over a multiyear period. The relatively recent success of CG307 progeny in Italy and Korea has caused alarm (33, 34, 36, 47). By all measures, including the in-hospital mortality rate, the CG307 strains causing infections in Houston appear to be as virulent as CG258 strains. Over the 45-month course of our study, the incidence of ESBL-producing K. pneumoniae steadily increased from approximately 18 to 28 strains recovered per month (Fig. 3). There was no preferential increase in CG258 or CG307 strains compared to other ST types during the study period. This increasing incidence mirrors a trend reported elsewhere and stresses the detrimental impact of multidrug-resistant K. pneumoniae on human health. The use of whole-genome sequence data to infer antimicrobial resistance pheno- types is an area of intense interest (59, 60). Two very different approaches have been pursued. The most common strategy attempts to predict phenotypes by identifying the presence or absence of genes and gene mutations conferring antimicrobial resistance (61). Although modestly successful, this strategy has two major disadvantages. First, it requires the curation of a comprehensive database that is updated as new discoveries are made. Second, this strategy is likely to fail to accurately predict resistance if the molecular mechanism is multifactorial or unknown. Since this study provides a very large, well-annotated data set, we sought an alternative approach using machine learning-based classifiers to predict antimicrobial resistance (Fig. 8). We have recently used this method to build accurate predictors of antimicrobial resistance phenotypes for several important human pathogens, including Acinetobacter baumannii, Mycobac- terium tuberculosis, Staphylococcus aureus, and Streptococcus pneumoniae (62). Machine May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 13 ® Long et al. learning approaches are advantageous because they require no a priori knowledge of the underlying molecular mechanism of resistance and may potentially identify new antimicrobial resistance-related genomic features. The results of this study suggest that for many antibiotics, it may be possible to build diagnostically valid classifiers for the prediction of antimicrobial resistance directly from the genome sequence data. Paired with new rapid sequencing technologies, these classifiers could provide clini- cally actionable data (63). We also attempted to build classifiers for the prediction of other clinical phenotypes. No classifiers could be generated to accurately predict the site of infection, length of stay, chronic or repeated infection, or death. The lack of predictive classifiers for these clinical phenotypes may reflect the extensive genomic and transcriptomic diversity present in the Houston strains or the complex comorbidi- ties present in patients with K. pneumoniae infections. Our data from the mouse model of pneumonia demonstrated that the host immune response to K. pneumoniae is multifaceted. We found a significant infiltration of neutrophils, monocytes, and T cells in infected lungs. We also detected IL-17 produc- tion by two different T-cell subsets, CD4 and  T cells. IL-17 is well known to play a key role in innate immunity to extracellular pathogens, particularly in patients with chronic disease or immune compromise. Chen et al. recently reported that IL-17 secretion by CD4 T cells provided protection against eight strains of K. pneumoniae belonging to three different capsular serotypes (STs not reported) recovered from patients in Southeast Asia, an area where very few CG307 strains have been reported (64). Herein, we are the first to suggest that IL-17 production by  T cells may also contribute to immunity to K. pneumoniae. Compared to the globally pandemic CG258 strain, the locally prominent CG307 strain elicited a significantly diminished IL-17 response by  T cells. In addition, the lack of a yersiniabactin locus in the great majority of CG307 strains may, in part, be related to this diminished IL-17 production. Nearly all K. pneumoniae produce their own siderophore, enterobactin, that binds iron in the body (52). Enterobactin is susceptible to inactivation by lipocalin 2 (Lnc2), which also stimulates an inflammatory response (65). K. pneumoniae may acquire the yersini- abactin locus to produce a siderophore that does not bind Lnc2, providing a survival benefit in exchange for the energy required to express yersiniabactin (52, 66). By limiting IL-17 production in the immune response, CG307 K. pneumoniae may be enhancing the effectiveness of the enterobactin siderophore system, negating the need for yersiniabactin loci in CG307 organisms for survival and success in the host. These discoveries provide one potential explanation for the numerical success of CG307 strains among patients in Houston. That is, the CG307 clone circulating in Houston, through an undetermined genomic or transcriptomic mechanism, may elicit a different innate immune response than CG258 strains. Additional studies are under way to investigate this idea. MATERIALS AND METHODS Strain sample. Strains were cultured from patient specimens submitted to the Diagnostic Microbi- ology Laboratory at Houston Methodist Hospital between September 2011 and May 2015. The Diag- nostic Microbiology Laboratory serves as the central diagnostic laboratory for the 2,000-bed Houston Methodist Hospital system. Strains were collected by the research team after routine microbiology diagnostics were completed. All of the strains collected were ESBL-producing K. pneumoniae, as deter- mined by automated methods (Phoenix ESBL test; BD). A summary of the isolate collection dates and sources is provided in Table S1A. Strains were collected and patient data were examined under human subjects protocol IRB1010-0199 as approved by the Institutional Review Board, Houston Methodist Research Institute. DNA extraction and whole-genome sequencing. Genomic and plasmid DNA was extracted in accordance with the Gram-negative protocol of the DNeasy Blood and Tissue kit (Qiagen). DNA was quantitated with the Qubit (Invitrogen) instrument, and whole-genome sequencing libraries were prepared with NexteraXT reagents (Illumina). Libraries were sequenced with Illumina MiSeq and NextSeq 500 instruments. Bioinformatic analysis. Initial bioinformatic analysis was performed on the high-performance computer cluster in the Houston Methodist Research Institute, a Beowulf cluster with 336 CPU cores and 1.8 TB of RAM. FASTQ files were processed with Trimmomatic (v 0.35) to remove adapter sequence 67, 68). SMALT (v 0.7.6) was used contamination and Musket (v 1.1) to correct short-read sequence errors ( to align reads, and FreeBayes (v 0.9.20) was used to identify variants (69). The in-house-developed scripts May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 14 ® ESBL K. pneumoniae prephix and phrecon were used to generate multi-FASTA files of SNPs for phylogenetic tree generation (https://github.com/codinghedgehog/). Phylogenetic trees were generated with FastTreeMP (v 2.1.8) using double precision and annotated with CLC Genomics workbench 9.5.2 (Qiagen) (70). Short-read sequence typing for bacterial pathogens (SRST2) (71) was used to identify the MLST assignment of each strain with the K. pneumoniae database of 2,470 STs available on 4 November 2016 curated at the Institut Pasteur (http://bigsdb.web.pasteur.fr/klebsiella/). SRST2 was also used to identify antibiotic resistance gene content and plasmid replicon content by using the SRST2 gene database files ARGannot.r1.fasta and PlasmidFinder.fasta. Reads were assembled into contigs with SPAdes v.3.6.2 (72). Kaptive was used to identify the capsule locus genotype and capsule locus gene content from assembled contigs (50). Kleborate was used to identify yersiniabactin, colibactin, and other siderophore locus content from assembled contigs (52). MEGA7 and MEGA-CC were used to generate distance matrices for pairwise distance analysis (73, 74). Gene content analysis was performed with the Proteome Comparison tool in PATRIC (75). Pacific Biosciences genome sequence data. Strains were sequenced to closure by the Pacific Biosciences SMRT long-read strategy in the laboratory of C. Mason at Weill Cornell Medical College. This technology permitted closed assembly of the core chromosome and plasmid DNA sequences from the target strains. Polished contigs were generated with HGAP. The plasmid sequences were compared to the NT database at NCBI with BLAST. Genome-wide RNA transcript analysis. RNA-Seq analysis was conducted as previously described (76). Briefly, 10 strains were selected to be broadly representative of our population, on the basis of the year of collection, phage type, CRE type, bla type, and numerical dominance of CGs (see Table S6). NDM-1 Two of these 10 strains were previously sequenced to closure (17), and 1 was sequenced to closure in this study with PacBio as described above. Duplicate cultures of each strain were inoculated into prewarmed LB medium and grown at 37°C to the mid-exponential phase (optical density at 600 nm [OD ] of 0.60 to 0.65). Total RNA was extracted with an RNeasy kit (Qiagen) according to the manufacturer’s instructions, and RNA-seq libraries were prepared with an Epicenter ScriptSeq kit (Illumina). The libraries were sequenced by using a 75-bp single-end protocol with a high-output v 3 reagent kit (Illumina). To generate a sufficient number of sequence reads for transcript analysis (average of 12 million reads per library), two runs were performed with one MiSeq instrument at the default software settings for the Generate FASTQ on instrument workflow. The two FASTQ files generated for each library were concatenated, and the number and quality of sequence reads were determined with FastQC (Babraham Bioscience Technologies, Cambridge, United Kingdom). The RNA transcripts were quantified from the concatenated FASTQ files with the default software settings relative to the chromosome of strain BK30684 with CLC Genomics Workbench v 9.5.2 (Qiagen). Differences of 1.5-fold by Baggerly’s test and a P value of 0.05 after applying Bonferroni’s correction for multiple comparisons were considered significant. Strain BK30684 was selected as the chromosome reference because the genome has been sequenced to closure, it is genetically representative of the globally disseminated CG258 clone, and it has been used in previous pathogenomic studies (17). To confirm that the many differences in gene expression observed between strains were due to differences in gene content, RNA-Seq reads were also mapped to CG307 reference strain KPN11 and CG14 reference strain KPN528 (see Table S3). To confirm expression of the KPC gene, reads were mapped to reference plasmid pNFST258C2 (GenBank accession no. CP006919.1). To confirm expression of the bla and OXA48 genes, reads were mapped to reference plasmids pPKPN1 NDM-1 (GenBank accession no. CP006799.1) and pKP112 (GenBank accession no. LN864819.1). Mouse model of pneumonia. A mouse model of pneumonia was used to compare the virulence of reference CG307 and CG258 strains KPN1411 and KPN113, respectively. C57BL/6 mice were inoculated by intranasal instillation of 1  10 CFU in 30 l (15 l in each naris). At 24 h postinoculation, mice were sacrificed. The lungs and spleens were removed, homogenized (Omni), serially diluted in sterile phosphate- buffered saline, plated on tryptic soy agar supplemented with 5% sheep blood, and incubated overnight at 37 C. CFU were counted, graphed as the mean  the standard error of the mean, and statistically compared by using the Mann-Whitney test (GraphPad). Similarly, mice were infected and sacrificed after 7 days and their lungs were harvested for flow cytometry analysis by standard methods (77). Preparation of genomic data for antimicrobial resistance prediction. Antimicrobial susceptibility testing for each strain was performed as part of the routine clinical workup. Machine learning-based classifiers for antimicrobial resistance phenotype prediction were created as previously described (62), by using the genomic data and clinical antimicrobial susceptibility data generated in this study. Briefly, the 1,777 genomes were assembled by using the PATRIC Full SPAdes assembly strategy and annotated by using the PATRIC annotation service (75). For each antimicrobial agent, the genomes were evenly divided into resistant and susceptible sets, matrices were built with 15-bp k-mers, and 10 iterative rounds of the AdaBoost algorithm were performed. Tenfold cross validation was applied to the data set for each classifier (62). The final set of classifiers generated for each antimicrobial agent is accessible in the PATRIC and RAST annotation services. Accession number(s). The whole genome sequence FASTQ files, RNAseq FASTQ files, and genomes of the strains sequenced by PacBio (see Table S3) have been deposited in the NCBI database under BioProject no. PRJNA376414. SUPPLEMENTAL MATERIAL Supplemental material for this article may be found at https://doi.org/10.1128/mBio .00489-17. FIG S1, PDF file, 0.4 MB. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 15 ® Long et al. FIG S2, PDF file, 0.04 MB. FIG S3, TIF file, 1.3 MB. TABLE S1, XLSX file, 0.3 MB. TABLE S2, PDF file, 0.1 MB. TABLE S3, PDF file, 0.04 MB. TABLE S4, PDF file, 0.1 MB. TABLE S5, PDF file, 0.1 MB. TABLE S6, PDF file, 0.1 MB. ACKNOWLEDGMENTS This work was supported by funds from the Fondren Foundation and the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Service (contract HHSN272201400027C). We thank Frank DeLeo and Barry Kreiswirth for providing reference strains BK30660 and BK30684. We thank Concepcion Cantu, Sarah Linson, Waleed Nasser, and Matthew Ojeda Saavedra for technical assistance. We thank Frank DeLeo, Timothy Palzkill, Katherine Perez, and Kathryn Stockbauer for suggestions to improve the manuscript. REFERENCES 1. Munoz-Price LS, Poirel L, Bonomo RA, Schwaber MJ, Daikos GL, Cormican CA, Doi Y. 2011. Colistin-resistant, Klebsiella pneumoniae carbapenemase M, Cornaglia G, Garau J, Gniadkowski M, Hayden MK, Kumarasamy K, (KPC)-producing Klebsiella pneumoniae belonging to the international Livermore DM, Maya JJ, Nordmann P, Patel JB, Paterson DL, Pitout J, epidemic clone ST258. Clin Infect Dis 53:373–376. https://doi.org/10 Villegas MV, Wang H, Woodford N, Quinn JP. 2013. Clinical epidemiology .1093/cid/cir401. of the global expansion of Klebsiella pneumoniae carbapenemases. 10. Castanheira M, Farrell SE, Wanger A, Rolston KV, Jones RN, Mendes RE. Lancet Infect Dis 13:785–796. https://doi.org/10.1016/S1473-3099(13) 2013. Rapid expansion of KPC-2-producing Klebsiella pneumoniae iso- 70190-7. lates in two Texas hospitals due to clonal spread of ST258 and ST307 2. Endimiani A, Choudhary Y, Bonomo RA. 2009. In vitro activity of NXL104 lineages. Microb Drug Resist 19:295–297. https://doi.org/10.1089/mdr in combination with beta-lactams against Klebsiella pneumoniae isolates .2012.0238. producing KPC carbapenemases. Antimicrob Agents Chemother 53: 11. Chen L, Chavda KD, Mediavilla JR, Zhao Y, Fraimow HS, Jenkins SG, Levi 3599–3601. https://doi.org/10.1128/AAC.00641-09. MH, Hong T, Rojtman AD, Ginocchio CC, Bonomo RA, Kreiswirth BN. 2012. 3. Snitkin ES, Zelazny AM, Thomas PJ, Stock F, NISC Comparative Sequenc- Multiplex real-time PCR for detection of an epidemic KPC-producing Kleb- ing Program Group, Henderson DK, Palmore TN, Segre JA. 2012. Tracking siella pneumoniae ST258 clone. Antimicrob Agents Chemother 56: a hospital outbreak of carbapenem-resistant Klebsiella pneumoniae with 3444–3447. https://doi.org/10.1128/AAC.00316-12. whole-genome sequencing. Sci Transl Med 4:148ra116. https://doi.org/ 12. Fowler RC, Scharn CR, Ali MA, Cavalieri SJ, Goering RV, Hanson ND. 2014. 10.1126/scitranslmed.3004129. Whole genome mapping of the first reported case of KPC-2–positive 4. Centers for Disease Control and Prevention (CDC). 2013. Notes from the Klebsiella pneumoniae ST258 in Nebraska. Diagn Microbiol Infect Dis field: hospital outbreak of carbapenem-resistant Klebsiella pneumoniae 79:384–386. https://doi.org/10.1016/j.diagmicrobio.2014.03.023. producing New Delhi metallo-beta-lactamase—Denver, Colorado, 2012. 13. Kitchel B, Rasheed JK, Patel JB, Srinivasan A, Navon-Venezia S, Carmeli Y, MMWR Morb Mortal Wkly Rep 62:108. https://www.cdc.gov/mmwr/ Brolund A, Giske CG. 2009. Molecular epidemiology of KPC-producing preview/mmwrhtml/mm6206a5.htm. Klebsiella pneumoniae isolates in the United States: clonal expansion of 5. Mathers AJ, Stoesser N, Sheppard AE, Pankhurst L, Giess A, Yeh AJ, multilocus sequence type 258. Antimicrob Agents Chemother 53: Didelot X, Turner SD, Sebra R, Kasarskis A, Peto T, Crook D, Sifri CD. 2015. 3365–3370. https://doi.org/10.1128/AAC.00126-09. Klebsiella pneumoniae carbapenemase (KPC)-producing K. pneumoniae 14. Bowers JR, Kitchel B, Driebe EM, MacCannell DR, Roe C, Lemmer D, de at a single institution: insights into endemicity from whole-genome se- Man T, Rasheed JK, Engelthaler DM, Keim P, Limbago BM. 2015. Genomic quencing. Antimicrob Agents Chemother 59:1656–1663. https://doi.org/10 analysis of the emergence and rapid global dissemination of the clonal .1128/AAC.04292-14. group 258 Klebsiella pneumoniae pandemic. PLoS One 10:e0133727. 6. Stoesser N, Giess A, Batty EM, Sheppard AE, Walker AS, Wilson DJ, https://doi.org/10.1371/journal.pone.0133727. Didelot X, Bashir A, Sebra R, Kasarskis A, Sthapit B, Shakya M, Kelly D, 15. Yigit H, Queenan AM, Anderson GJ, Domenech-Sanchez A, Biddle JW, Pollard AJ, Peto TEA, Crook DW, Donnelly P, Thorson S, Amatya P, Joshi Steward CD, Alberti S, Bush K, Tenover FC. 2001. Novel carbapenem- S. 2014. Genome sequencing of an extended series of NDM-producing hydrolyzing beta-lactamase, KPC-1, from a carbapenem-resistant strain Klebsiella pneumoniae isolates from neonatal infections in a Nepali hos- of Klebsiella pneumoniae. Antimicrob Agents Chemother 45:1151–1161. pital characterizes the extent of community- versus hospital-associated https://doi.org/10.1128/AAC.45.4.1151-1161.2001. transmission in an endemic setting. Antimicrob Agents Chemother 58: 16. Nordmann P, Cuzon G, Naas T. 2009. The real threat of Klebsiella pneu- 7347–7357. https://doi.org/10.1128/AAC.03900-14. moniae carbapenemase-producing bacteria. Lancet Infect Dis 9:228–236. 7. Wright MS, Perez F, Brinkac L, Jacobs MR, Kaye K, Cober E, van Duin D, https://doi.org/10.1016/S1473-3099(09)70054-4. Marshall SH, Hujer AM, Rudin SD, Hujer KM, Bonomo RA, Adams MD. 17. Deleo FR, Chen L, Porcella SF, Martens CA, Kobayashi SD, Porter AR, Chavda 2014. Population structure of KPC-producing Klebsiella pneumoniae iso- KD, Jacobs MR, Mathema B, Olsen RJ, Bonomo RA, Musser JM, Kreiswirth BN. lates from midwestern U.S. hospitals. Antimicrob Agents Chemother 2014. Molecular dissection of the evolution of carbapenem-resistant multi- 58:4961–4965. https://doi.org/10.1128/AAC.00125-14. locus sequence type 258 Klebsiella pneumoniae. Proc Natl Acad SciUSA 8. Chung The H, Karkey A, Pham Thanh D, Boinett CJ, Cain AK, Ellington M, 111:4988–4993. https://doi.org/10.1073/pnas.1321364111. Baker KS, Dongol S, Thompson C, Harris SR, Jombart T, Le Thi Phuong T, 18. Rodrigues C, Machado E, Ramos H, Peixe L, Novais Â. 2014. Expansion of Tran Do Hoang N, Ha Thanh T, Shretha S, Joshi S, Basnyat B, Thwaites G, ESBL-producing Klebsiella pneumoniae in hospitalized patients: a suc- Thomson NR, Rabaa MA, Baker S. 2015. A high-resolution genomic cessful story of international clones (ST15, ST147, ST336) and epidemic analysis of multidrug-resistant hospital outbreaks of Klebsiella pneu- plasmids (IncR, IncFIIK). Int J Med Microbiol 304:1100–1108. https://doi moniae. EMBO Mol Med 7:227–239. https://doi.org/10.15252/emmm .201404767. .org/10.1016/j.ijmm.2014.08.003. 9. Bogdanovich T, Adams-Haduch JM, Tian G-B, Nguyen MH, Kwak EJ, Muto 19. Munoz-Price LS, Quinn JP. 2009. The spread of Klebsiella pneumoniae May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 16 ® ESBL K. pneumoniae carbapenemases: a tale of strains, plasmids, and transposons. Clin Infect tertiary care hospitals reveals high frequency of non-CG258 clones of Dis 49:1739–1741. https://doi.org/10.1086/648078. carbapenem-resistant Klebsiella pneumoniae with distinct clinical char- 20. Richter SN, Frasson I, Franchin E, Bergo C, Lavezzo E, Barzon L, Cavallaro acteristics. Antimicrob Agents Chemother 60:332–342. https://doi.org/ A, Palù G. 2012. KPC-mediated resistance in Klebsiella pneumoniae in two 10.1128/AAC.01775-15. hospitals in Padua, Italy, June 2009–December 2011: massive spreading 36. Ahn C, Yoon SS, Yong TS, Jeong SH, Lee K. 2016. The resistance mech- of a KPC-3-encoding plasmid and involvement of non-intensive care anism and clonal distribution of tigecycline-nonsusceptible Klebsiella units. Gut Pathog 4:7. https://doi.org/10.1186/1757-4749-4-7. pneumoniae isolates in Korea. Yonsei Med J 57:641–646. https://doi.org/ 21. Pitout JD, Laupland KB. 2008. Extended-spectrum beta-lactamase- 10.3349/ymj.2016.57.3.641. producing Enterobacteriaceae: an emerging public-health concern. 37. Ruiz-Garbajosa P, Hernández-García M, Beatobe L, Tato M, Méndez MI, Lancet Infect Dis 8:159–166. https://doi.org/10.1016/S1473-3099(08) Grandal M, Aranzábal L, Alonso S, Lópaz MÁ, Astray J, Cantón R. 2016. A 70041-0. single-day point-prevalence study of faecal carriers in long-term care 22. President’s Council of Advisors on Science and Technology. 2014. Report hospitals in Madrid (Spain) depicts a complex clonal and polyclonal to the president on combating antibiotic resistance. Centers for Disease dissemination of carbapenemase-producing Enterobacteriaceae. J Anti- Control and Prevention, Atlanta, GA. https://www.cdc.gov/drugresistance/ microb Chemother 71:348–352. https://doi.org/10.1093/jac/dkv355. pdf/report-to-the-president-on-combating-antibiotic-resistance.pdf. 38. Mansour W, Grami R, Ben Haj Khalifa A, Dahmen S, Châtre P, Haenni M, 23. Emerson MO, Bratter J, Howell J, Jeanty PW, Cline M. 2012. Houston Aouni M, Madec JY. 2015. Dissemination of multidrug-resistant blaCTX- region grows more racially/ethnically diverse, with small declines in M-15/IncFIIk plasmids in Klebsiella pneumoniae isolates from hospital- segregation. A joint report analyzing census data from 1990, 2000, and and community-acquired human infections in Tunisia. Diagn Microbiol 2010. Rice University, Houston, TX. http://kinder.rice.edu/uploadedFiles/ Infect Dis 83:298–304. https://doi.org/10.1016/j.diagmicrobio.2015.07 Urban_Research_Center/Media/Houston%20Region%20Grows%20More .023. %20Ethnically%20Diverse%202-13.pdf. 39. Gona F, Barbera F, Pasquariello AC, Grossi P, Gridelli B, Mezzatesta ML, 24. Gomez SA, Pasteran FG, Faccone D, Tijet N, Rapoport M, Lucero C, Caio C, Stefani S, Conaldi PG. 2014. In vivo multiclonal transfer of Lastovetska O, Albornoz E, Galas M, Melano RG, Corso A, Petroni A. 2011. blaKPC-3 from Klebsiella pneumoniae to Escherichia coli in surgery Clonal dissemination of Klebsiella pneumoniae ST258 harbouring KPC-2 patients. Clin Microbiol Infect 20:O633–O635. https://doi.org/10 in Argentina. Clin Microbiol Infect 17:1520–1524. https://doi.org/10 .1111/1469-0691.12577. .1111/j.1469-0691.2011.03600.x. 40. Habeeb MA, Haque A, Nematzadeh S, Iversen A, Giske CG. 2013. High 25. Di Carlo P, Gulotta G, Casuccio A, Pantuso G, Raineri M, Farulla CA, prevalence of 16S rRNA methylase RmtB among CTX-M extended- Bonventre S, Guadagnino G, Ingrassia D, Cocorullo G, Mammina C, spectrum beta-lactamase-producing Klebsiella pneumoniae from Islam- Giarratano A. 2013. KPC-3 Klebsiella pneumoniae ST258 clone infection in abad, Pakistan. Int J Antimicrob Agents 41:524–526. https://doi.org/10 postoperative abdominal surgery patients in an intensive care setting: .1016/j.ijantimicag.2013.02.017. analysis of a case series of 30 patients. BMC Anesthesiol 13. https://doi 41. Doi Y, Hazen TH, Boitano M, Tsai Y-C, Clark TA, Korlach J, Rasko DA. 2014. .org/10.1186/1471-2253-13-13. Whole-genome assembly of Klebsiella pneumoniae coproducing NDM-1 26. Valentín-Martín A, Valverde-De Francisco A, Bosque-Vall M, Cantón- and OXA-232 carbapenemases using single-molecule, real-time se- Moreno R. 2013. First report of colistin-resistant KPC-2 producing ST258- quencing. Antimicrob Agents Chemother 58:5947–5953. https://doi.org/ Klebsiella pneumoniae in Spain. Enferm Infecc Microbiol Clin 31:489–491. 10.1128/AAC.03180-14. https://doi.org/10.1016/j.eimc.2012.12.002. 42. Villa L, Poirel L, Nordmann P, Carta C, Carattoli A. 2012. Complete 27. Rodríguez-Zulueta P, Silva-Sanchez J, Barrios H, Reyes-Mar J, Velez-Perez sequencing of an IncH plasmid carrying the blaNDM-1, blaCTX-M-15 and F, Arroyo-Escalante S, Ochoa-Carrera L, Delgado-Sapien G, Morales- qnrB1 genes. J Antimicrob Chemother 67:1645–1650. https://doi.org/10 Espinoza MdR, Tamayo-Legorreta E, Hernandez-Castro R, Garza-Ramos .1093/jac/dks114. U. 2013. First outbreak of KPC-3-producing Klebsiella pneumoniae 43. Hirsch EB, Chang KT, Lasco TM, Caeiro JP, Tam VH. 2011. Emergence of (ST258) clinical isolates in a Mexican medical center. Antimicrob Agents KPC-producing Klebsiella pneumoniae in Texas. Diagn Microbiol Infect Chemother 57:4086–4088. https://doi.org/10.1128/AAC.02530-12. Dis 69:234–235. https://doi.org/10.1016/j.diagmicrobio.2010.09.003. 28. Marsh JW, Krauland MG, Nelson JS, Schlackman JL, Brooks AM, Pasculle 44. Paltansing S, Kraakman MEM, Ras JMC, Wessels E, Bernards AT. 2013. AW, Shutt KA, Doi Y, Querry AM, Muto CA, Harrison LH. 2015. Genomic Characterization of fluoroquinolone and cephalosporin resistance mech- epidemiology of an endoscope-associated outbreak of Klebsiella pneu- anisms in Enterobacteriaceae isolated in a Dutch teaching hospital moniae carbapenemase (KPC)-producing K. pneumoniae. PLoS One 10: reveals the presence of an Escherichia coli ST131 clone with a specific e0144310. https://doi.org/10.1371/journal.pone.0144310. mutation in parE. J Antimicrob Chemother 68:40–45. https://doi.org/10 29. Tedja R, Nowacki A, Fraser T, Fatica C, Griffiths L, Gordon S, Isada C, van .1093/jac/dks365. Duin D. 2014. The impact of multidrug resistance on outcomes in 45. Gomez-Simmonds A, Greenman M, Sullivan SB, Tanner JP, Sowash MG, ventilator-associated pneumonia. Am J Infect Control 42:542–545. Whittier S, Uhlemann A-C. 2015. Population structure of Klebsiella pneu- https://doi.org/10.1016/j.ajic.2013.12.009. moniae causing bloodstream infections at a New York City tertiary care 30. Chen L, Mathema B, Chavda KD, DeLeo FR, Bonomo RA, Kreiswirth BN. hospital: diversification of multidrug-resistant isolates. J Clin Microbiol 2014. Carbapenemase-producing Klebsiella pneumoniae: molecular and 53:2060–2067. https://doi.org/10.1128/JCM.03455-14. genetic decoding. Trends Microbiol 22:686 – 696. https://doi.org/10 46. Dropa M, Lincopan N, Balsalobre LC, Oliveira DE, Moura RA, Fernandes .1016/j.tim.2014.09.003. MR, da Silva QM, Matté GR, Sato MIZ, Matté MH. 2016. Genetic back- 31. Lee CR, Lee JH, Park KS, Kim YB, Jeong BC, Lee SH. 2016. Global dissemi- ground of novel sequence types of CTX-M-8- and CTX-M-15-producing nation of carbapenemase-producing Klebsiella pneumoniae: epidemiology, Escherichia coli and Klebsiella pneumoniae from public wastewater treat- genetic context, treatment options, and detection methods. Front Mi- ment plants in Sao Paulo, Brazil. Environ Sci Pollut Res Int 23:4953–4958. crobiol 7:895. https://doi.org/10.3389/fmicb.2016.00895. https://doi.org/10.1007/s11356-016-6079-5. 32. Pitout JDD, Nordmann P, Poirel L. 2015. Carbapenemase-producing 47. Geraci DM, Bonura C, Giuffrè M, Saporito L, Graziano G, Aleo A, Fasciana Klebsiella pneumoniae, a key pathogen set for global nosocomial dom- T, Di Bernardo F, Stampone T, Palma DM, Mammina C. 2015. Is the inance. Antimicrob Agents Chemother 59:5873–5884. https://doi.org/10 monoclonal spread of the ST258, KPC-3-producing clone being replaced .1128/AAC.01019-15 in southern Italy by the dissemination of multiple clones of carbapenem- 33. Bonura C, Giuffrè M, Aleo A, Fasciana T, Di Bernardo F, Stampone T, nonsusceptible, KPC-3-producing Klebsiella pneumoniae? Clin Microbiol Giammanco A, Palma DM, Mammina C. 2015. An update of the evolving Infect 21:e15–e17. https://doi.org/10.1016/j.cmi.2014.08.022. epidemic of blaKPC carrying Klebsiella pneumoniae in Sicily, Italy, 2014: 48. Kim JO, Song SA, Yoon EJ, Shin JH, Lee H, Jeong SH, Lee K. 2017. emergence of multiple non-ST258 clones. PLoS One 10:e0132936. Outbreak of KPC-2-producing Enterobacteriaceae caused by clonal dis- https://doi.org/10.1371/journal.pone.0132936. semination of Klebsiella pneumoniae ST307 carrying an IncX3-type plas- 34. Park DJ, Yu JK, Park KG, Park YJ. 2015. Genotypes of ciprofloxacin- mid harboring a truncated Tn4401a. Diagn Microbiol Infect Dis 87: resistant Klebsiella pneumoniae in Korea and their characteristics accord- 343–348. https://doi.org/10.1016/j.diagmicrobio.2016.12.012. ing to the genetic lineages. Microb Drug Resist 21:622–630. https://doi .org/10.1089/mdr.2015.0001. 49. Wyres KL, Gorrie C, Edwards DJ, Wertheim HFL, Hsu LY, Van Kinh N, 35. Ocampo AM, Chen L, Cienfuegos AV, Roncancio G, Chavda KD, Kre- Zadoks R, Baker S, Holt KE. 2015. Extensive capsule locus variation and iswirth BN, Jiménez JN. 2015. A two-year surveillance in five Colombian large-scale genomic recombination within the Klebsiella pneumoniae May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 17 ® Long et al. clonal group 258. Genome Biol Evol 7:1267–1279. https://doi.org/10 Sengupta A, Mohades Z, Dery S, Deng W, Earl E, Demeter DV, Mills K, .1093/gbe/evv062. Mihai G, Ruzic L, Ketz N, Reineberg A, Reddan MC, Goddings A, Gonzalez- 50. Wyres KL, Wick RR, Gorrie C, Jenney A, Follador R, Thomson NR, Holt KE. Castillo J, Gorgolewski KJ, Froehlich C, Dekel G, Margulies DS, Craddock 2016. Identification of Klebsiella capsule synthesis loci from whole ge- RC, Fulcher BD, Glatard T, Das S, Madjar C, Sengupta A, Mohades Z, Dery nome data. Microb Genom 2:e000102. https://doi.org/10.1099/mgen.0 S, Deng W, Earl E, Demeter DV, Mills K, Mihai G, Ruzik L, et al. 2016. 2015 .000102. Brainhack proceedings. Gigascience 5(Suppl 1):46. https://doi.org/10 51. Holt KE, Wertheim H, Zadoks RN, Baker S, Whitehouse CA, Dance D, .1186/s13742-016-0147-0. Jenney A, Connor TR, Hsu LY, Severin J, Brisse S, Cao H, Wilksch J, Gorrie 64. Chen K, Eddens T, Trevejo-Nunez G, Way EE, Elsegeiny W, Ricks DM, Garg C, Schultz MB, Edwards DJ, Nguyen KV, Nguyen TV, Dao TT, Mensink M, AV, Erb CJ, Bo M, Wang T, Chen W, Lee JS, Gaffen SL, Kolls JK. 2016. IL-17 Minh VL, Nhu NTK, Schultsz C, Kuntaman K, Newton PN, Moore CE, receptor signaling in the lung epithelium is required for mucosal chemo- Strugnell RA, Thomson NR. 2015. Genomic analysis of diversity, popula- kine gradients and pulmonary host defense against K. pneumoniae. Cell tion structure, virulence, and antimicrobial resistance in Klebsiella pneu- https://doi.org/10.1016/j.chom.2016.10.003. Host Microbe 20:596–605. moniae, an urgent threat to public health. Proc Natl Acad SciUSA 65. Bachman MA, Miller VL, Weiser JN. 2009. Mucosal lipocalin 2 has pro- 112:E3574–E3581. https://doi.org/10.1073/pnas.1501049112. inflammatory and iron-sequestering effects in response to bacterial 52. Lam MMC, Wick RR, Wyres KL, Gorrie C, Judd LM, Brisse S, Jenney A, Holt enterobactin. PLoS Pathog 5:e1000622. https://doi.org/10.1371/journal KE. 2017. Frequent emergence of pathogenic lineages of Klebsiella pneu- .ppat.1000622. moniae via mobilisation of yersiniabactin and colibactin. bioRxiv https:// 66. Bachman MA, Oyler JE, Burns SH, Caza M, Lepine F, Dozois CM, Weiser doi.org/10.1101/098178. JN. 2011. Klebsiella pneumoniae yersiniabactin promotes respiratory tract 53. Diago-Navarro E, Chen L, Passet V, Burack S, Ulacia-Hernando A, Kodi- infection through evasion of lipocalin 2. Infect Immun 79:3309–3316. yanplakkal RP, Levi MH, Brisse S, Kreiswirth BN, Fries BC. 2014. https://doi.org/10.1128/IAI.05114-11. Carbapenem-resistant Klebsiella pneumoniae exhibit variability in capsu- 67. Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for lar polysaccharide and capsule associated virulence traits. J Infect Dis Illumina sequence data. Bioinformatics 30:2114–2120. https://doi.org/10 210:803–813. https://doi.org/10.1093/infdis/jiu157. .1093/bioinformatics/btu170. 54. Mehta SC, Rice K, Palzkill T. 2015. Natural variants of the KPC-2 carbap- 68. Liu Y, Schroder J, Schmidt B. 2013. Musket: a multistage k-mer spectrum- enemase have evolved increased catalytic efficiency for ceftazidime based error corrector for Illumina sequence data. Bioinformatics 29: hydrolysis at the cost of enzyme stability. PLoS Pathog 11:e1004949. 308–315. https://doi.org/10.1093/bioinformatics/bts690. https://doi.org/10.1371/journal.ppat.1004949. 69. Garrison E, Marth G. 2012. Haplotype-based variant detection from 55. Bushnell G, Mitrani-Gold F, Mundy LM. 2013. Emergence of New Delhi short-read sequencing. arXiv arXiv:1207.3907v2 [q-bio.GN]. http://arXiv metallo-beta-lactamase type 1-producing enterobacteriaceae and non- .org/abs/1207.3907v2. enterobacteriaceae: global case detection and bacterial surveillance. Int 70. Price MN, Dehal PS, Arkin AP. 2010. FastTree 2—approximately J Infect Dis 17:e325–e333. https://doi.org/10.1016/j.ijid.2012.11.025. maximum-likelihood trees for large alignments. PLoS One 5:e9490. 56. Gupta K, Hooton TM, Naber KG, Wullt B, Colgan R, Miller LG, Moran GJ, https://doi.org/10.1371/journal.pone.0009490. Nicolle LE, Raz R, Schaeffer AJ, Soper DE. 2011. International clinical 71. Inouye M, Dashnow H, Raven LA, Schultz MB, Pope BJ, Tomita T, Zobel practice guidelines for the treatment of acute uncomplicated cystitis J, Holt KE. 2014. SRST2: rapid genomic surveillance for public health and and pyelonephritis in women: a 2010 update by the Infectious Diseases hospital microbiology labs. Genome Med 6:90. https://doi.org/10.1186/ Society of America and the European Society for Microbiology and s13073-014-0090-6. Infectious Diseases. Clin Infect Dis 52:e103– e120. https://doi.org/10 72. Nurk S, Bankevich A, Antipov D, Gurevich A, Korobeynikov A, Lapidus A, .1093/cid/ciq257. Prjibelsky A, Pyshkin A, Sirotkin A, Sirotkin Y, Stepanauskas R, McLean J, 57. Freund Y, Schapire RE. 1997. A decision-theoretic generalization of on-line Lasken R, Clingenpeel SR, Woyke T, Tesler G, Alekseyev MA, Pevzner PA. learning and an application to boosting. J Comput Syst Sci 55:119–139. 2013. Assembling genomes and mini-metagenomes from highly chi- https://doi.org/10.1006/jcss.1997.1504. meric reads, p 158 –170. In Deng M, Jiang R, Sun F, Zhang X (ed), 58. Hastie T, Rosset S, Zhu J, Zou H. 2009. Multi-class AdaBoost. Stat Inter- Research in computational molecular biology: 17th Annual Interna- face 2:349–360. https://doi.org/10.4310/SII.2009.v2.n3.a8. tional Conference, RECOMB 2013, Beijing, China, April 7–10, 2013. 59. Ellington MJ, Ekelund O, Aarestrup FM, Canton R, Doumith M, Giske C, Springer, Berlin, Germany. http://link.springer.com/chapter/10.1007/ Grundman H, Hasman H, Holden MTG, Hopkins KL, Iredell J, Kahlmeter 978-3-642-37195-0_13. G, Köser CU, MacGowan A, Mevius D, Mulvey M, Naas T, Peto T, Rolain 73. Kumar S, Stecher G, Tamura K. 2016. MEGA7: molecular evolutionary J-M, Samuelsen Ø, Woodford N. 2017. The role of whole genome se- genetics analysis version 7.0 for bigger datasets. Mol. Biol. Evol. quencing in antimicrobial susceptibility testing of bacteria: report from 33:1870 –1874. https://doi.org/10.1093/molbev/msw054. the EUCAST subcommittee. Clin Microbiol Infect 23:2–22. https://doi 74. Kumar S, Stecher G, Peterson D, Tamura K. 2012. MEGA-CC: computing .org/10.1016/j.cmi.2016.11.012. core of molecular evolutionary genetics analysis program for automated 60. Metcalf BJ, Chochua S, Gertz RE, Jr., Li Z, Walker H, Tran T, Hawkins PA, and iterative data analysis. Bioinformatics 28:2685–2686. https://doi.org/ Glennen A, Lynfield R, Li Y, McGee L, Beall B, Active Bacterial Core 10.1093/bioinformatics/bts507. Surveillance Team. 2016. Using whole genome sequencing to identify 75. Wattam AR, Abraham D, Dalay O, Disz TL, Driscoll T, Gabbard JL, Gillespie resistance determinants and predict antimicrobial resistance pheno- JJ, Gough R, Hix D, Kenyon R, Machi D, Mao C, Nordberg EK, Olson R, types for year 2015 invasive pneumococcal disease isolates recovered in Overbeek R, Pusch GD, Shukla M, Schulman J, Stevens RL, Sullivan DE, the United States. Clin Microbiol Infect 22:1002.e1–1002.e8. https://doi Vonstein V, Warren A, Will R, Wilson MJC, Yoo HS, Zhang C, Zhang Y, .org/10.1016/j.cmi.2016.08.001. Sobral BW. 2014. PATRIC, the bacterial bioinformatics database and 61. Moran RA, Anantham S, Holt KE, Hall RM. 2017. Prediction of antibiotic analysis resource. Nucleic Acids Res 42:D581–D591. https://doi.org/10 resistance from antibiotic resistance genes detected in antibiotic- .1093/nar/gkt1099. resistant commensal Escherichia coli using PCR or WGS. J Antimicrob 76. Olsen RJ, Fittipaldi N, Kachroo P, Sanson MA, Long SW, Como-Sabetti KJ, Chemother 72:700–704. https://doi.org/10.1093/jac/dkw511. Valson C, Cantu C, Lynfield R, Van Beneden C, Beres SB, Musser JM. 2014. 62. Davis JJ, Boisvert S, Brettin T, Kenyon RW, Mao C, Olson R, Overbeek R, Clinical laboratory response to a mock outbreak of invasive bacterial Santerre J, Shukla M, Wattam AR, Will R, Xia F, Stevens R. 2016. Antimi- infections: a preparedness study. J Clin Microbiol 52:4210–4216. https:// crobial resistance prediction in PATRIC and RAST. Sci Rep 6:27930. https://doi.org/10.1038/srep27930. doi.org/10.1128/JCM.02164-14. 63. Craddock RC, Bellec P, Margules DS, Nichols BN, Pfannmöller JP, Badh- 77. Eagar TN, Tang Q, Wolfe M, He Y, Pear WS, Bluestone JA. 2004. Notch 1 war A, Kennedy D, Poline J, Toro R, Cipollini B, Rokem A, Clark D, signaling regulates peripheral T cell activation. Immunity 20:407–415. Gorgolewski KJ, Craddock RC, Craddock RC, Clark DJ, Das S, Madjar C, https://doi.org/10.1016/S1074-7613(04)00081-0. May/June 2017 Volume 8 Issue 3 e00489-17 mbio.asm.org 18

Journal

mBioPubmed Central

Published: May 16, 2017

There are no references for this article.