M A J O R A R T I C L E
Genetic Determinants of Capsular Serotype K1
of Klebsiella pneumoniae Causing Primary
Pyogenic Liver Abscess
Yi-Ping Chuang,1Chi-Tai Fang,2Shau-Yan Lai,2Shan-Chwen Chang,2and Jin-Town Wang1,2
1Department of Microbiology, National Taiwan University College of Medicine, and2Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan
Background. Primary pyogenic liver abscess (PLA) caused by Klebsiella pneumoniae is an emerging infectious disease. Capsular serotype K1 and the magA gene have been reported to be associated with this disease.
Methods. The prevalence of magA was determined by polymerase chain reaction (PCR). The sequences of the
magA flanking region were completed by inverse PCR and direct sequencing. Serotyping was performed by double
immunodiffusion. Insertion mutations and trans-complementation were used to define the K1 genetic determi-nation region.
Results. Thirty-five of 42 strains from patients with PLA were magA positive, whereas only 1 of 32 non-PLA strains was magA positive. All 36 magA-positive strains were serotype K1, and the 38 magA-negative strains were not (36/36 vs. 0/38;P!.0001). Sequencing of the magA flanking region revealed a putative capsular polysaccharide synthesis (cps) region; this region was 25 kb in length and contained 20 open reading frames (ORFs); of these ORFs, 9 were cotranscribed as part of an operon and differed from both MGH78578 and the Chedid strain. Mutation of 4 genes in this region turned the mutant strains anti-K1 negative. Trans-complementation restored the K1 phenotype.
Conclusions. The operon containing magA is responsible for capsular serotype K1 of K. pneumoniae. Several loci in the operon are unique determinants of K1 strains.
Klebsiella pneumoniae is a common hospital-acquired,
gram-negative pathogen that causes urinary tract in-fections, nosocomial pneumonia, and intra-abdominal infections [1–3]. However, community-acquired pri-mary pyogenic liver abscess (PLA) caused by K.
pneu-moniae has become an emerging disease recently [4–
11]. There have been1900 cases of K. pneumoniae liver abscess, with or without septic metastatic complica-tions, reported from Taiwan during the past 15 years [8]. This disease has also been described in Asia, North America, and Europe [8].
Received 21 June 2005; accepted 20 September 2005; electronically published 20 January 2006.
Potential conflicts of interest: none reported.
Financial support: National Science Council (grant NSC 94-3112-B-002-024); National Health Research Institutes (grant NHRI-EX94-9435S); Liver Disease Prevention and Treatment Research Foundation in Taiwan.
Reprints or correspondence: Dr. Jin-Town Wang, Dept. of Microbiology, Nation-al Taiwan University College of Medicine, 1, Sec 1, Jen-Ai Rd., Taipei, Taiwan ([email protected]).
The Journal of Infectious Diseases 2006; 193:645–54
2006 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2006/19305-0005$15.00
Capsular serotypes K1 and K2 are considered to be the predominant virulent strains of K. pneumoniae; this has been confirmed by experiments in mouse models [12]. K1 has been further investigated as the most com-mon serotype isolated from patients with K.
pneumo-niae liver abscess and endophthalmitis [13].
Mucovis-cosity has also been documented as a virulence factor of K. pneumoniae [12, 14]. Using transposon muta-genesis, we identified a virulence gene, magA, that was involved in the hypermucoviscosity phenotype and also played an important role in resistance to serum and phagocytosis [7]. A high prevalence of the magA gene has also been found among PLA-associated K.
pneu-moniae strains [7, 15]. The function of magA is still
being explored. However, comparison of the magA flanking region between NTUH-K2044 and MGH78578 (a strain that resulted in pneumoniae infection in a 66-year-old patient; its complete genome sequence is avail-able at http://genomeold.wustl.edu/projects/bacterial/ kpneumoniae/) revealed a 31-kb magA flanking region that was absent from MGH78578 [7]. This region was replaced by a 26-kb fragment in MGH78578. The 5
646 • JID 2006:193 (1 March) • Chuang et al.
Table. 1. Bacterial strains and plasmids used in this study. Bacterial strain or plasmid
Genotype or relevant description
Reference or source Bacteria
Klebsiella pneumoniae strains
K. pneumoniae (74) Clinical isolates collected from National Taiwan University Hospital during 1997–2003
Present study
NTUH-K2044 Clinical isolate; the parent strain for generate isogenic mutants [7]
MGH78578 ATCC700721; the strain chosen for full genome sequence ATCC
A5054 K1 reference strain from Statens Serum Institute, Denmark [16]
Chedid Laboratory strain; O1:K2 [16]
ATCC8052 K2 strain used for the control of serotyping ATCC
Escherichia coli strains
DH10B F⫺mcrA D(mrr-hsdRMS-mcrBC) f80lacZDM15 DlacX74 recA1 endA1 araD139 D(ara, leu)7697 galU galK l⫺rpsL nupG
Invitrogen
CC118lpir D(ara-leu) araD DlacX74 galE galK phoA20 thi-1 rpsE rpoB argE(Am) recA1 lpir phage lysogen
[19]
S17-1lpir hsdR recA pro RP4-2 (Tc::Mu; Km::Tn7)(lpir) [20]
Plasmids
pUT-Km pUTKm1 [21] derived plasmid, with miniTn5 excised by EcoRI, tnp excised by SalI, and
bla removed by ApaLI, then with an insertion of Km resistance cassette from pUC4K
[22] into PstI site
Present study
pCRII-TOPO-CAT pCRII-TOPO was inserted with CAT [23] cassette into XbaI site. Present study NOTE. ATCC, American Type Culture Collection; Km, kanamycin.
sequence upstream of these 2 fragments could be aligned with capsular polysaccharide synthesis (cps) loci of the Chedid strain (GenBank accession number D21242; an O1:K2 strain) [16]. All 3 strains contained galF, ORF2, and 4 conserved genes—
orfX (wzi), wza, wzb, and wzc—that were considered to be
responsible for the translocation and surface assembly of cap-sular polysaccharide [17, 18]. Therefore, magA and its flanking regions might be associated with capsular polysaccharide bio-synthesis. In the present study, we determined the correlation between the magA gene and capsular serotype K1 and tried to identify the genetic determinants of capsular serotype K1. MATERIALS AND METHODS
Bacterial strains and plasmid vectors. Bacterial strains and plasmids used in this study are listed in table 1. Clinically isolated K. pneumoniae strains were collected at National Tai-wan University Hospital (NTUH) [15]. From 1997 to 2003, there were 42 consecutive tissue-invasive strains (i.e., strains isolated from the blood of patients with PLA with or without septic complications such as meningitis or endophthalmitis) and 32 non–tissue-invasive strains (i.e., strains isolated from the blood of patients with sepsis but without liver abscess or any tissue-invasive diseases such as endophthalmitis or men-ingitis) isolated [15]. K. pneumoniae and Escherichia coli were cultured in Luria-Bertani (LB) medium supplemented with ap-propriate antibiotics, including 100 mg/mL ampicillin, 50 mg/ mL kanamycin, or 100 mg/mL chloramphenicol, as described elsewhere [7].
Serotyping. Initially, antisera from the Laboratory of Health-Care Associated Infection, Health Protection Agency, were used [13]; however, cross-reactions between serotypes K1 and K2 were noted. Therefore, we used 2 other typing methods to document the capsular serotype of each K. pneumoniae strain. Capsular serotypes were screened using Klebsiella antisera SEI-KEN (Denka Seiken), in accordance with the manufacturer’s instructions. In brief, bacteria were cultured overnight in Wor-fel-Ferguson’s medium (0.2% yeast extract, 0.025% magnesium sulfate, 0.1% potassium sulfate, 0.2% sodium chloride, and 2% sucrose). The bacterial cells were harvested and resuspended in saline. One loopful of dense capsular suspension of the bacteria was then mixed with serotype-specific serum (K1–K6). Within 1 min, the agglutination results were read.
To reconfirm results for strains that initially tested as serotype K1, extracellular polysaccharides were isolated by a modified hot water–phenol extraction method [24, 25]. In brief, bacteria cultured overnight in a minimal medium (supplemented with 0.2% glucose instead of 0.4% glucose) [26] were harvested and resuspended in 150 mL of water. An equal volume of hot phenol (pH 6.6; Amresco) was added, and the mixture was vortexed vigorously. The mixture was then incubated at 65C for 20 min, followed by chloroform extraction and centrifugation. The ex-tracts should have included both capsule and lipopolysaccha-ride (LPS) [24]. However, since the antiserum specifically in-teracted with capsule antigen only, the LPS contamination did not affect the subsequent experiments. Extracted polysaccha-rides were reacted with a serotype K1–specific antiserum
(Sta-Table 2. Primers used in this study.
Primer Sequence Nucleotide position Purpose
1166F GGTGCTCTTTACATCATTGC 10 to 29 of magA magA prevalence
936R GCAATGGCCATTTGCGTTAG 1292 to 1273 of magA
18E-F613 GCTAACAGAAGACATCCCTA 105 to 124 of ORF8 ORF8mutant construct and ORF8prevalence 18E-R596 GAGTTACCCTCCACAAACTTC 470 to 450 of ORF8
wcaG+118-F GTTGGTTCAGCAATCGTAAG 118 to 137 of wcaG wcaG mutant construct wcaG+448-R CGCAGAGTTTTATACCGGC 448 to 430 of wcaG
wcaI+29-F ATGGTCTTGACGTTAGGGTC 29 to 48 of wcaI wcaI mutant construct wcaI+525-R GGATACTTTGTCTGCAGACG 525 to 506 of wcaI
114F GGATGTAGCAACGAAGCAAG ⫺302 to ⫺283 of ORF8 ORF8complementation
1166R GCAATGATGTAAAGAGCACC +60 to +41 after ORF8
10111F AGTGTTTCTGGTGAAGACGC ⫺187 to ⫺168 of wcaG wcaG complementation
wcaG-R TTAACTCCGGAAGCTTTGCTG 513 to 493 of wcaG
wcaH+383F GATGAGTGCGGATGAAA ⫺76 to ⫺60 of wcaI wcaI complementation wcaI-R CCATTCCACCTAACGTAATC +49 to +30 after wcaI
1040F CGTTACGAACTTGAACGAGC ⫺118 to ⫺99 of magA magA complementation
936R GCAATGGCCATTTGCGTTAG +65 to +46 after magA
orf3F GAAACGCAATCAGTGTAGCG 170 to 189 of wzx wzx prevalence
orf3R CTTGCTTCGTTGCTACATCC 1009 to 990 of wzx
gmd+72F AGATGGTTCATATCTCGCAG 72 to 91 of gmd gmd prevalence
wcaG+77R GTCATTTCTTTACTCCAGAG 77 to 58 of gmd
orf6F CCTTGGAGGACTGTCCTTAA 226 to 245 of ORF10 ORF10-ORF11prevalence 23KS-431F ATGTTGGCCCAGGTGCAAAT 137 to 156 of ORF11
wcaG+118-F GTTGGTTCAGCAATCGTAAG 118 to 137 of wcaG wcaG-wcaH prevalence wcaI-R2 CACCATTTCACCACTATACT +86 to +67 after wcaH
wcaI+29-F ATGGTCTTGACGTTAGGGTC 29 to 48 of wcaI wcaI prevalence wcaI-R CCATTCCACCTAACGTAATC +49 to +30 after wcaI
wcaJ+29F CGAATGCGAATGCTTCCTTG 29 to 48 of rfbP rfbP prevalence rfbP-R CAATGCTTATCTTAAGCAGC +26 to +7 after rfbP
Figure 1. Constructs of the insertion mutation in NTUH-K2044. The orientations of open reading frames are represented by arrows. The “X” denotes homologous recombination. The primers shown here indicate the relative positions in different insertion mutant strains. See figure 3 for target genes. KmR, kanamycin resistance cassette.
tens Serum Institute), using a double immunodiffusion assay. Each assay was performed with 1.5% Nobel agar (Difco) in nor-mal saline. Ten microliters of anti-K1 serum (75% dilution) was loaded into the central well, and 20 mL of each polysaccharide extract was loaded into peripheral wells. After an overnight in-cubation at 37C, the precipitation lines were read. The agars
were then soaked in 50% normal saline for 6 h, covered with a piece of filter, and allowed to air-dry overnight. Thoroughly dried agars were treated with 1% azocarmine (dissolved in 2% glacial acid) (Chroma) for 2 h, followed by a 5% glacial acid destain. The precipitation lines were generally visible before azocarmine staining; however, the pictures were clearer after staining.
648 • JID 2006:193 (1 March) • Chuang et al. Table 3. Prevalence of the magA gene and results of capsular serotyping of Klebsiella pneumoniae strains.
Serotype
No. of magA-positive strains/ total no. of strains Tissue invasive Non–tissue invasive K1 35/35 1/1 K2 0/1 0/1 K3 0/0 0/1 K4 0/0 0/0 K5 0/0 0/0 K6 0/1 0/0 NT 0/5 0/29 Total 35/42 1/32
NOTE. There are 3 additional control strains: A5054 (K1) is magA positive; ATCC8052 (K2) and MGH78578 (K52) are magA negative. NT, nontypeable (i.e., serotype other than K1–K6). The difference between serotype K1 and the other strains was significant (36/36 K1 strains magA positive, vs. 0/38 other strains;
, x2 test).
P!.0001
Figure 2. The prevalence of the magA gene in K1 and non-K1 Klebsiella pneumoniae strains. A, Polymerase chain reaction (PCR) for magA. B, PCR for 23srDNA. Lanes 1–8, K1 strains (lanes 1–7 are tissue-invasive strains, and lane 8 is a non–tissue-invasive strain); lanes 9–16, non-K1 strains (lanes 9–11 are tissue-invasive strains, and lanes 12–16 are non–tissue-invasive strains); lane 17, A5054 (K1); lane 18, ATCC8052 (K2); lane 19, MGH78578 (K52); and lane 20, size marker.
magA prevalence. To determine the prevalence of magA, polymerase chain reaction (PCR) using a primer pair specific for magA (1166F and 936R) was performed [7]. Primers used in this study are listed in table 2. PCR was performed as de-scribed elsewhere [7]. Briefly, 3 mL of overnight-cultured bac-terial broth was added to 10 mL of water and boiled for 15 min to release DNA template. The reaction mixtures, which con-tained primers (0.4 mmol/L each), dNTP (0.1 mmol/L each),
Taq polymerase (2.5 U; Bioman), and 13 mL of the above DNA
template was then kept at 96C for 3 min, followed by 30 temperature cycles of 96C for 30 s, 56C for 15 s, and 72C for 1 min. The expected PCR product was 1282 bp in length. Sequence analysis of the magA flanking region. The se-quence of the magA flanking region was initially obtained by inverse PCR, and the cps loci were finally assembled by chro-mosomal walking, using the NTUH-K2044 phagemid library [27], until sequences of both ends matched those of MGH78578. Construction of K. pneumoniae mutant strains. Primer pairs were designed to amplify partial regions of the following genes: ORF8 (primers 18E-F613 and 18E-R596), wcaG (primers wcaG+118-F and wcaG+448-R), and wcaI (primers wcaI+92-F and
wcaI+525-R) (table 2). PCR-amplified fragments were blunted by
a T4 DNA polymerase (New England Biolabs) and ligated into a blunted EcoRI site of a pUT-Km vector (table 1) that contained a kanamycin-resistance gene. The resultant plasmid constructs were transformed into an E. coli S17-1lpir, followed by conju-gation with the parent K. pneumoniae strain NTUH-K2044, as described elsewhere [7]. Transconjugants were selected using a minimal medium [26, 27] supplemented with 50 mg/mL kana-mycin. The genotype was confirmed by PCR, using
alignment-specific primer pairs (i.e., strains found to be positive by use of primer pair 1 and 3 and primer pair 2 and 4 but negative by use of primer pair 1 and 2, as shown in figure 1).
Other mutants—magA⫺, wza⫺, wzc⫺, rmpA2⫺, and wzm⫺— from a mutant library of NTUH-K2044 were screened for de-creased mucoviscosity by a string test and were identified by inverse PCR and sequencing [7]. wza⫺and wzc⫺mutant strains had impaired capsular antigen translocation ability [17]. rmpA2⫺ and wzm⫺mutants were defective in their regulation of capsule production and O-antigen biosynthesis [12, 28], respectively.
Trans-complementation. Intact ORF8, wcaG, wcaI, and magA genes were amplified by PCR and cloned into a
pCRII-TOPO-CAT plasmid (table 1). These plasmids were trans-formed into their corresponding isogenic mutant strains by electroporation [29]; for selection of complementation strains, LB agar plates were supplemented with 50 mg/mL kanamycin and 100 mg/mL chloramphenicol. To confirm the role of this predicted cps region in K1 capsular synthesis by phenotype conversion, the K1-specific regions from wzx to rfbP (wzx to
wcaI was unique for NTUH-K2044, whereas rfbP was present
in both NTUH-K2044 and Chedid strains but was absent in MGH78578) were cloned into pBK-CMV vectors (Stratagene) and transformed into non-K1 K. pneumoniae strains.
Identifying operons by reverse-transcription PCR (RT-PCR) analysis. Total RNA was extracted from NTUH-K2044 as de-scribed elsewhere [27]. Forty micrograms of total RNA was
Figure 3. Comparison of capsular polysaccharide synthesis (cps) gene clusters between Chedid (K2), NTUH-K2044 (K1), and MGH78578 (K52). The locations and orientations of the open reading frames (ORFs) are indicated by arrows. ORFs are cited in order, starting from galF; ORFs with homologs are cited by putative gene names, and those without homologs are cited as ORFs and are numbered, with galF considered to be no. 1 (ORF in Chedid strain, ORFin NTUH-K2044, and ORFin MGH78578). Solid arrows refer to ORFs with significant similarity (the aligned genes with scores1200 and
expect values!e⫺50in the BLASTP search), and unshaded arrows indicate the ORFs with low similarity between any 2 strains (the homolog could not be found even when the expect threshold was
set to be 10 in the BLASTP search). The first nucleotide of galF defines position 1 in the sequence; nos. beneath each axis denote positions in kilobases. Dashed lines indicate the replaced region be-tween NTUH-K2044 and MGH78578.
Table 4. Annotation of open reading frames (ORFs) of capsular polysaccharide synthesis (cps) loci in NTUH-K2044. ORF no. ORF name Product size, aa
ORF location,ant (position
within GenBank sequence) ORF homolog characteristics Source
Sequence identity,b% (no. of matching amino acids/total no. of amino
acids)
GenBank accession no. 1 galF 296 1–891 (7144–8034) UTP–glucose-1-phosphate uridylyltransferase K. pneumoniae (Chedid) 100 (296/296) Q48447 2 ORF2 209 1284–1913 (8427–9056) acidPPc, acid phosphatase homologsc(ORF2) K. pneumoniae (Chedid) 88 (184/209) Q48448 3 wzi 477 2873–4306 (10016–11449) Capsule assembly, orfX E. coli 92 (443/477) AAD21561 4 wza 377 4449–5582 (11592–12725) Putative capsule polysaccharide export protein precursor (ORF4) K. pneumoniae (Chedid) 92 (348/377) Q48450 5 wzb 144 5647–6021 (12790–13164) Probable protein-tyrosine-phosphatase E. carotovora 62 (89/142) YP_049524 6 wzc 717 6056–8188 (13199–15331) Putative transmembrane protein, tyrosine-protein kinase E. coli 53 (371/692) AAD21564.1 7 wzx 429 8277–9566 (15420–16709) O–antigen flippase F. tularensis tularensis 20 (89/443) YP_170390 8 ORF8 356 9568–10638 (16711–17781) Polysaccharide pyruvyl transferasec … … … 9 magA 408 10670–11896 (17813–19039) Putative O-antigen ligase, waaLc V. cholerae 20 (65/319)d AF443846.1
10 ORF10 366 11930–13030 (19073–20173) Putative glycosyltransferase B. fragilis 28 (71/250) AAD40719
11 ORF11 170 13011–13523 (20154–20666) Galactoside O-acetyltransferase M. acetivorans 33 (42/127) NP_617091
12 gmd 383 13520–14671 (20663–21814) GDP-D-mannose dehydratase S. flexneri 88 (329/370) NP_837675 13 wcaG 346 14604–15644 (21747–22787) GDP-4-keto-6-L-galactose reductase; GDP-fucose synthetase E. coli 77 (248/318) NP_754466 14 wcaH 152 15646–16104 (22789–23247) GDP-mannose mannosylhydrolase E. coli 48 (72/149) NP_754465 15 wcaI 404 16116–17330 (23259–24473) Putative colonic biosynthesis glycosyl transferase E. coli 57 (233/406) AAG57110 16 rfbP 467 17632–19035 (24775–26178) Probable CPS biosynthesis glycosyltransferase (ORF14); rfbP
homolog
K. pneumoniae (Chedid) 73 (345/467) Q48460 17 gnd 468 19199–20605 (26342–27748) Gluconate-6-phosphate dehydrogenase E. coli 99 (464/468) BAA28321 18 manC 471 20841–22256 (27984–29399) Mannose-1-phosphate guanylyltransferase (ORF16); cpsB
homolog
E. coli 97 (459/471) BAA07745
19 manB 456 22279–23649 (29422–30792) Phosphomannomutase E. coli 96 (441/456) P37755 20 ugd 388 23813–24979 (30956–32122) Putative UDP-glucose dehydrogenase E. coli 99 (385/388) AAC45348.1
NOTE. B. fragilis, Bacteroides fragilis; CPS, capsular polysaccharide synthesis; E. carotovora, Erwinia carotovora; E. coli, Escherichia coli; F. tularensis, Francisella tularensis; K. pneumoniae, Klebsiella pneumoniae; M. acetivorans, Methanosarcina acetivorans; S. flexneri, Shigella flexneri; V. cholerae, Vibrio cholerae.
a
The first nucleotide of galF (GenBank accession no. Q48447) defines position 1.
b
Determined by BLAST-P.
c
Position-specific score matrix producing significant alignments.
d
Figure 4. Double immunodiffusion with rabbit anti-K1 serum from Sta-tens Serum Institute (center wells) and capsular extracts (20 mL) from 1 mL (A and B) or 5 mL (C and D) of overnight-cultured Klebsiella pneu-moniae strains in the peripheral wells. A, Well 1, wild-type NTUH-K2044; well 2, magA⫺mutant; well 3, magA⫺mutant strain with magA trans-complementation; well 4, wcaI⫺mutant; well 5, wcaI⫺mutant strain with wcaI trans-complementation; well 6, A5054. Wells 1, 3, 5, and 6 showed a positive immunoprecipitation line, whereas wells 2 and 4 did not. B, Well 1, NTUH-K2044; well 2, wza⫺mutant; well 3, wzc⫺mutant; well
4, wzm⫺mutant; well 5, rmpA2⫺mutant; and well 6, ATCC8052. Wells 1, 4, and 5 showed a positive immunoprecipitation line, whereas wells 2, 3, and 6 did not. C and D, Trans-complementation of the wzx–rfbP gene fragment (C) and magA gene (D). Well 1, NTUH-K2044 as a positive control; wells 2–6, non-K1 K. pneumoniae strains (well 2, NTUH-N3529; well 3, NTUH-N3932; well 4, NTUH-N5906; well 5, NTUH-N2059; well 6, ATCC8052). Two (wells 4 and 5 in panel C) of the 5 non-K1 strains complemented with the wzx–rfbP fragment turned K1 positive, whereas magA complementation converted none of the 5 strains (wells 2–6 in panel D).
reverse transcribed using 0.1 mmol of random primer (Roche) and 600 U of Moloney murine leukemia virus reverse tran-scriptase (Invitrogen) in a reaction mixture of 60 mL. Reaction mixtures without reverse transcriptase were included as nega-tive controls. PCR was performed with 1 mL of each reverse-transcription reaction mixture as template and subjected to the above-mentioned temperature cycles.
RESULTS
Capsular serotype and prevalence of magA. Thirty-five of the 42 tissue-invasive strains were magA positive, whereas only 1 of the 32 non–tissue-invasive strains was found to be positive for magA by PCR (35/42 vs. 1/32;P!.001, x2test) (table 3).
The capsular serotypes of 74 clinical isolates were initially
screened with Klebsiella antisera SEIKEN. Of 42 tissue-invasive strains, 35 were K1, 1 was K2, 1 was K6, and 5 were nontypeable (i.e., a serotype that was unable to react with the K1–K6 antisera in the kit). However, of 32 non–tissue-invasive strains, only 1 was K1, 1 was K2, and 1 was K3; the other 29 strains were nontypeable (table 3). The results were further confirmed with anti-K1 serum from Statens Serum Institute, using double im-munodiffusion analysis. The results obtained with these 2 assays were all consistent. This showed that all 37 K1 K. pneumoniae strains (35 tissue-invasive strains, 1 non–tissue-invasive strain, and the reference strain A5054) were magA positive, whereas all 40 non-K1 K. pneumoniae strains (7 tissue-invasive strains, 31 non–tissue-invasive isolates, MGH78578, and ATCC8052) were magA negative (37/37 vs. 0/40;P!.0001) (table 3 and fig-ure 2).
Comparison of cps loci in NTUH-K2044, MGH78578, and the Chedid strain. Because there was no sequence match be-tween the magA and MGH78578 genomes, we used inverse PCR and the NTUH-K2044 phagemid library [23] to extend the upstream and downstream regions of the magA gene in NTUH-K2044 until matches were found. DNA sequences were then compared with MGH78578. A 31-kb fragment in NTUH-K2044 replaced a 26-kb region in MGH78578; they were both downstream of galF, ORF2, wzi (orfX), and the wza-wzb-wzc cluster, as shown in figure 3. The ORFs between galF and ugd in NTUH-K2044 (GenBank accession number AB198423) were thus considered to be cps loci by comparison with the K2 Chedid strain [16, 18].
The percentage of GC content in the cps region was 42.5%, in contrast to 57.7% of the full genome in MGH78578. Genetic loci in the cps region of NTUH-K2044 (K1), MGH78578 (K52), and the Chedid (K2) strain, which have different capsular se-rotypes, had similar functions, such as glycosyltransferase (table 4). However, sequences from wzx to wcaI of NTUH-K2044 (K1) revealed a low similarity, compared with ORF7–ORF13 of Chedid (K2) and wbaP–wzx in MGH78578 (K52) (figure 3), as determined by National Center for Biotechnology Informa-tion BLAST (version 2.2.11; available at: http://www.ncbi.nlm .nih.gov/blast/; (expect, 10; word size, 3; matrix, BLOSUM62; gap costs: existence, 11 extension, 1). These comparisons sug-gested that the ORFs from wzx to wcaI of NTUH-K2044 might be unique for the K1 capsule, while rfbP was relatively specific in some serotypes.
Characterization of the K1 determinant region of cps loci in NTUH-K2044. Insertion mutants, including ORF8⫺, wcaG⫺, and wcaI⫺, as well as magA⫺, were constructed in the
NTUH-K2044 parent strain. These mutants turned negative for anti-K1 serum. Complementation experiments showed that anti- K1-neg-ative magA⫺, ORF8⫺, wcaG⫺, and wcaI⫺ mutants turned K1 positive by transformation of a pCRII-TOPO-CAT carrying their corresponding intact genes (figure 4A). wza⫺and wzc⫺
652 • JID 2006:193 (1 March) • Chuang et al.
Figure 5. The transcription units of capsular polysaccharide synthesis (cps) loci in NTUH-K2044 defined by reverse-transcription (RT) polymerase chain reaction (PCR). The genetic organization of cps loci is shown in the upper part of the figure. Panels a–o show the corresponding PCR products for primers located at each open reading frame (ORF) junction. The bars denote RT-PCR–positive junctions, and bars with an X denote RT-PCR–negative junctions. The lower part of the figure shows the RT-PCR results by ethidium bromide–stained agarose gel. Lane 1, RT-PCR products of each junction; lane 2, RT-PCR without reverse transcriptase, as a negative control; lane 3, PCR with genomic DNA as a template, as a positive control. Arrowheads indicate the expected sizes of RT-PCR products.
mutants that were defective in capsular antigen translocation were used as negative controls (figure 4B).
The prevalence of genes other than magA in the region from
wzx to rfbP was also tested between 20 K1 strains (table 2) (of
the first 20 of 36 K1 strains, 19 were tissue invasive and 1 was non–tissue invasive) and 20 non-K1 strains (of the first 20 of 38 non-K1 strains, 3 were tissue invasive and 17 were non– tissue invasive). The 9 loci (wzx, ORF8, ORF10-ORF11, gmd, wcaG-wcaH, wcaI, and rfbP) were all present in K1 strains (20/
20). Like magA, wzx, ORF8, ORF10-ORF11, gmd, and wcaI
were totally absent in non-K1 strains (0/20). However,
wcaG-wcaH and rfbP could still be found in 25% (5/20) and 20%
(4/20) of non-K1 strains, respectively. Transformation of the plasmid containing the wzx to rfbP region were able to turn 2 of the 5 non-K1 K. pneumoniae strains K1 positive, whereas plasmids containing only magA were unable to turn any of the 5 strains K1 positive (figure 4C and 4D).
Analysis of transcriptional units by RT-PCR. The pre-dicted promoter regions in cps loci of NTUH-K2044 were first analyzed using software from the Berkeley Drosophila Genome Project (available at: http://www.fruitfly.org/seq_tools/promoter .html). It revealed hundreds of putative promoters in cps loci. We thus determined the transcriptional unit by RT-PCR anal-ysis. Fifteen primer pairs were designed to amplify the junction of ORFs, using DNaseI-treated total RNA as an RT-PCR
tem-plate (figure 5). RT-PCR results for units a, b, d, e, f, g, h, i, j, k, l, m, and n were positive, whereas those for c and o were negative (figure 5, lane 1). The results suggested that the tran-scriptional units were wzi to wzc and wzx to ugd.
DISCUSSION
Capsule—but not LPS—of K. pneumoniae has been determined to be the important virulence factor against complement and phagocyte attack in vitro [30]. Serological tests have revealed that K. pneumoniae of capsular serotypes K1 and K2 were the predominant virulent strains among those isolated from pa-tients with either bacteremia or liver abscess (85/134 were K1; 19/134 were K2) [13, 14, 31]. The serotype of PLA-causing K.
pneumoniae has not yet been identified. K. pneumoniae PLA
differs from traditional liver abscess in 3 aspects: first, it is community acquired; second, patients with K. pneumoniae liver abscess have no history of hepatobiliary diseases; and third, 11%–12% of patients have other septic metastatic lesions [8, 32]. K. pneumoniae PLA has become an important emerging infectious disease in Asia, especially in Taiwan [8]. The reason for the increased incidence of K. pneumoniae PLA is still not well understood. In our study, 83.3% (35/42) of bacteria caus-ing primary liver abscess were serotype K1; however, only 2.4% (1/42) were serotype K2. The difference might be due to
cross-reactions between K1 and K2 resulting from the use of antisera from the Laboratory of HealthCare Associated Infection, Health Protection Agency, since such cross-reactions were found in our initial tests. Although the structures of capsular K1 and K2 lack mannose-a2/3-mannose and l-rhamnose-a2/3-rhamnose, which easily prompt lectinophagocytosis [33, 34, 35], it seems that the biological function of capsular K1 antigen may play a more important role in the pathogenesis of PLA or its derived metastatic complications. Actually, we found that all 9 K.
pneu-moniae strains that caused complications with metastatic
le-sions in our series were serotype K1 (8 caused endophthalmitis, and 1 caused meningitis).
Serotyping and studies of the bacterial genome provide clues for tracing the origin of this disease. The association of serotype K1 strains with PLA and the prevalence of magA prompted us to study the correlation between capsular serotype K1 and
magA. In our clinical isolates, as well as in the reference strains,
we found that all magA-positive strains were capsular serotype K1, whereas none of the magA-negative strains was serotype K1. The strong correlation suggested that magA was associated with K1 strains.
By comparison with sequences of the K2 Chedid strain and the K52 MGH78578 strain, we identified a cps region specific for capsular serotype K1. Mutant strains defective in several genes located in this region turned negative for anti-K1 serum. In addition, complementation of such mutant strains restored a positive K1 phenotype. The wzm⫺ and rmpA2⫺ mutants— which lacked O-antigen biosynthesis and regulation of capsule production, respectively—were isolated by decreased mucovis-cosity, using a string test [7]. RmpA2 has been shown to be a transcriptional activator of cps loci, and mutation in rmpA2 would result in 13%–29% lower capsule production [36]. How-ever, these 2 mutants retained the K1-positive phenotype, sug-gesting that these 2 ORFs, located outside of the regions be-tween wzx and wcaI, did not affect antigenic structure and were not K1 determinants. RT-PCR results showed that there was an operon (from wzx to ugd) specific for K1. We determined the prevalence of genes from wzx to rfbP, because they seemed to be unique on the basis of sequence comparison. The prev-alences of several genes from wzx to rfbP indicated that they, like magA, were also very unique for K1 (wzx, ORF8, ORF10 -ORF11, gmd, and wcaI were present in 20/20 vs. 0/20 in K1
and non-K1 strains, respectively). The other 2 gene fragments,
wcaG-wcaH and rfbP, were relatively specific but not unique
for K1 (wcaG-wcaH was present in 20/20 vs. 5/20, and rfbP was present in 20/20 vs. 4/20 of K1 and non-K1 strains, re-spectively). We did not test the remaining ORFs in the operon, since sequence comparison showed that they were relatively conserved (figure 3). Although we could not rule out the pos-sibility that some ORFs in this region could also play a role in LPS biosynthesis, the fact that trans-complementation of the
wzx-rfbP fragment turned some (2/5) strains K1 positive
con-firmed that these genes were involved in capsular synthesis. These results confirmed that the operon from wzx to wcaI was responsible for capsular serotype K1. Failure of K1 conversion in the other 3 strains was very likely due to the difference in other genes of the cps region, although comparison of available
cps sequences in the 3 strains suggested that they were less
variable.
Capsular serotyping is a convenient method; however, some antisera, such as those from the Laboratory of HealthCare As-sociated Infection, Health Protection Agency, may yield cross-reactions between K1 and K2. Therefore, sequences of this cps region would help to confirm serotyping results that are in doubt. As shown in our study, PCR using primer pairs derived from magA or other genes unique for K1 is a rapid and accurate method to detect K1 strains. Serotype K1 has been shown to have a more close relationship to PLA with metastatic lesions, such as endophthalmitis [13]. Metastatic endophthalmitis can result in blindness or evisceration unless treated within 24 h with appropriate systemic and intravitreous antibiotics [37]. Rapid molecular diagnosis using the genetic determinants will help these patients.
The structural unit of capsular K1 has been resolved as r4)- [2,3-(S)-pyruvate]-b-d-GlcA-(1r4)-a-l-Fuc-(1r3)-b-d-Glc-(1r [38]. This is in agreement with the function of some genes in cps loci: ORF8is a putative polysaccharide pyruvyl trans-ferase, and wcaG is a putative GDP-fucose synthetase. There are other sugar metabolism–associated loci, such as gmd and
manC, that are involved in mannose metabolism; however, they
are not included in the structural unit. Polysaccharide biosyn-thesis is a complex process, and many intermediates and en-zyme substrates could be involved; however, only structural units remain at the end of this process. For example, Gmd dehydrates GDP-d-mannose into GDP-4-keto-6-deoxy-d-man-nose, which is subsequently converted into GDP-l-fucose by WcaG, a GDP-fucose synthetase [39]. Therefore, the structural unit could not be predicted directly from the annotation of genes in cps loci.
In conclusion, we have identified a magA-containing chro-mosomal region that encodes capsular structure and is specific for the K1 serotype. This cps region contains an operon from
wzx to ugd in NTUH-K2044. Among these ORFs, magA, wzx, gmd, wcaI, and ORF10-ORF11 seem unique for K1 and can be used for rapid molecular typing. These results may help in the early detection and tracing the origin of this emerging infectious disease.
Acknowledgments
We thank Shih-Feng Tsai of National Health Research Institutes, for assistance in DNA sequencing of clones of a phagemid library of NTUH-K2044, and Mei-Chuan Chang, for laboratory work assistance.
654 • JID 2006:193 (1 March) • Chuang et al. References
1. Abbott SL. Klebsiella, Enterobacter, Citrobacter, Serratia, Plesiomonas, and other Enterobacteriaceae. In: Murray PR, Baron EJ, Jorgensen JH, Pfaller MA, Yolken RH, eds. Manual of clinical microbiology. 8th ed. Washington DC: American Society for Microbiology Press, 2003: 684–700.
2. Eisenstein BI, Zaleznik DF. Enterobacteriaceae. In: Mandell JEB, Dolin R, eds. Principles and practice of infectious diseases. 5th ed. Phila-delphia: Churchill-Livingstone, 2000:2294–309.
3. Podschun R, Ullmann U. Klebsiella spp. as nosocomial pathogens: epidemiology, taxonomy, typing methods, and pathogenicity factors. Clin Microbiol Rev 1998; 11:589–603.
4. Chang SC, Fang CT, Hsueh PR, Chen YC, Luh KT. Klebsiella
pneu-moniae isolates causing liver abscess in Taiwan. Diagn Microbiol Infect
Dis 2000; 37:279–84.
5. Cheng DL, Liu YC, Yen MY, Liu CY, Wang RS. Septic metastatic lesions of pyogenic liver abscess: their association with Klebsiella pneumoniae bacteremia in diabetic patients. Arch Intern Med 1991; 151:1557–9. 6. Chiu CT, Lin DY, Liaw YF. Metastatic septic endophthalmitis in
py-ogenic liver abscess. J Clin Gastroenterol 1988; 10:524–7.
7. Fang CT, Chuang YP, Shun CT, Chang SC, Wang JT. A novel virulence gene in Klebsiella pneumoniae strains causing primary liver abscess and septic metastatic complications. J Exp Med 2004; 199:697–705. 8. Ko WC, Paterson DL, Sagnimeni AJ, et al. Community-acquired
Kleb-siella pneumoniae bacteremia: global differences in clinical patterns.
Emerg Infect Dis 2002; 8:160–6.
9. Liu YC, Cheng DL, Lin CL. Klebsiella pneumoniae liver abscess asso-ciated with septic endophthalmitis. Arch Intern Med 1986; 146:1913–6. 10. Wang JH, Liu YC, Lee SS, et al. Primary liver abscess due to Klebsiella
pneumoniae in Taiwan. Clin Infect Dis 1998; 26:1434–8.
11. Yang CC, Yen CH, Ho MW, Wang JH. Comparison of pyogenic liver abscess caused by non-Klebsiella pneumoniae and Klebsiella
pneumo-niae. J Microbiol Immunol Infect 2004; 37:176–84.
12. Wacharotayankun R, Arakawa Y, Ohta M, et al. Enhancement of ex-tracapsular polysaccharide synthesis in Klebsiella pneumoniae by RmpA2, which shows homology to NtrC and FixJ. Infect Immun 1993; 61: 3164–74.
13. Fung CP, Chang FY, Lee SC, et al. A global emerging disease of Klebsiella
pneumoniae liver abscess: is serotype K1 an important factor for
com-plicated endophthalmitis? Gut 2002; 50:420–4.
14. Mizuta K, Ohta M, Mori M, Hasegawa T, Nakashima I, Kato N. Vir-ulence for mice of Klebsiella strains belonging to the O1 group: rela-tionship to their capsular (K) types. Infect Immun 1983; 40:56–61. 15. Ma LC, Fang CT, Lee CZ, Shun CT, Wang JT. Genomic heterogeneity
in Klebsiella pneumoniae strains is associated with primary pyogenic liver abscess and metastatic infection. J Infect Dis 2005; 192:117–28. 16. Arakawa Y, Wacharotayankun R, Nagatsuka T, Ito H, Kato N, Ohta
M. Genomic organization of the Klebsiella pneumoniae cps region re-sponsible for serotype K2 capsular polysaccharide synthesis in the vir-ulent strain Chedid. J Bacteriol 1995; 177:1788–96.
17. Drummelsmith J, Whitfield C. Gene products required for surface expression of the capsular form of the group 1 K antigen in Escherichia
coli (O9a:K30). Mol Microbiol 1999; 31:1321–32.
18. Rahn A, Drummelsmith J, Whitfield C. Conserved organization in the cps gene clusters for expression of Escherichia coli group 1 K antigens: relationship to the colanic acid biosynthesis locus and the cps genes from Klebsiella pneumoniae. J Bacteriol 1999; 181:2307–13.
19. Herrero M, de Lorenzo V, Timmis KN. Transposon vectors containing non-antibiotic resistance selection markers for cloning and stable chro-mosomal insertion of foreign genes in gram-negative bacteria. J Bac-teriol 1990; 172:6557–67.
20. Parke D. Construction of mobilizable vectors derived from plasmids RP4, pUC18 and pUC19. Gene 1990; 93:135–7.
21. de Lorenzo V, Herrero M, Jakubzik U, Timmis KN. Mini-Tn5 trans-poson derivatives for insertion mutagenesis, promoter probing, and chromosomal insertion of cloned DNA in gram-negative eubacteria. J Bacteriol 1990; 172:6568–72.
22. Taylor LA, Rose RE. A correction in the nucleotide sequence of the Tn903 kanamycin resistance determinant in pUC4K. Nucleic Acids Res 1988; 16:358.
23. Chang KC, Yeh YC, Lin TL, Wang JT. Identification of genes associated with natural competence in Helicobacter pylori by transposon shuttle random mutagenesis. Biochem Biophys Res Commun 2001; 288:961–8. 24. Westphal O, Jann K. Bacterial lipopolysaccharide extraction with phe-nol-water and further applications of the procedure. In: Whistler RL, ed. Methods in carbohydrate chemistry. New York: Academic Press, 1965:83–91.
25. Arakawa Y, Ohta M, Wacharotayankun R, et al. Biosynthesis of
Kleb-siella K2 capsular polysaccharide in Escherichia coli HB101 requires the
functions of rmpA and the chromosomal cps gene cluster of the vir-ulent strain Klebsiella pneumoniae Chedid (O1:K2). Infect Immun 1991; 59:2043–50.
26. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: a laboratory manual. New York: Cold Spring Harbor Laboratory Press, 1989. 27. Chou HC, Lee CZ, Ma LC, Fang CT, Chang SC, Wang JT. Isolation of
a chromosomal region of Klebsiella pneumoniae associated with allantoin metabolism and liver infection. Infect Immun 2004; 72:3783–92. 28. Bronner D, Clarke BR, Whitfield C. Identification of an ATP-binding
cassette transport system required for translocation of lipopolysaccha-ride O-antigen side-chains across the cytoplasmic membrane of
Kleb-siella pneumoniae serotype O1. Mol Microbiol 1994; 14:505–19.
29. Chang KC, Ho SW, Yang JC, Wang JT. Isolation of a genetic locus associated with metronidazole resistance in Helicobacter pylori. Bio-chem Biophys Res Commun 1997; 236:785–8.
30. Cortes G, Borrell N, de Astorza B, Gomez C, Sauleda J, Alberti S. Molecular analysis of the contribution of the capsular polysaccharide and the lipopolysaccharide O side chain to the virulence of Klebsiella
pneumoniae in a murine model of pneumonia. Infect Immun 2002;
70:2583–90.
31. Nassif X, Sansonetti PJ. Correlation of the virulence of Klebsiella
pneu-moniae K1 and K2 with the presence of a plasmid encoding aerobactin.
Infect Immun 1986; 54:603–8.
32. Seeto RK, Rockey DC. Pyogenic liver abscess: changes in etiology, management, and outcome. Medicine (Baltimore) 1996; 75:99–113. 33. Kabha K, Nissimov L, Athamna A, et al. Relationships among capsular
structure, phagocytosis, and mouse virulence in Klebsiella pneumoniae. Infect Immun 1995; 63:847–52.
34. Athamna A, Ofek I, Keisari Y, Markowitz S, Dutton GG, Sharon N. Lectinophagocytosis of encapsulated Klebsiella pneumoniae mediated by surface lectins of guinea pig alveolar macrophages and human monocyte-derived macrophages. Infect Immun 1991; 59:1673–82. 35. Ofek I, Sharon N. Lectinophagocytosis: a molecular mechanism of
recognition between cell surface sugars and lectins in the phagocytosis of bacteria. Infect Immun 1988; 56:539–47.
36. Lai YC, Peng HL, Chang HY. RmpA2, an activator of capsule biosyn-thesis in Klebsiella pneumoniae CG43, regulates K2 cps gene expression at the transcriptional level. J Bacteriol 2003; 185:788–800.
37. Chou FF, Kou HK. Endogenous endophthalmitis associated with pyo-genic hepatic abscess. J Am Coll Surg 1996; 182:33–6.
38. Kenne L, Lindberg B. Bacterial polysaccharides. In: Aspinall GO, ed. The polysaccharides. Vol 2. New York: Academic Press, 1983:287–363. 39. Coyne MJ, Reinap B, Lee MM, Comstock LE. Human symbionts use a host-like pathway for surface fucosylation. Science 2005; 307:1778–81.
