Table

Table 3.1. RstA-activated genes identified by subtractive cDNA hybridization

Predicted function Homologous gene (Accession no.) Predicted protein

% Identity (length of comparable amino acid sequence, no. of residues) Iron transporter K. pneumoniae fecE (BAH63871) iron(III) dicitrate transport ATP-binding protein 100% (198)

Ion channel K. pneumoniae yjeP (BAH61311) mechanosensitive ion channel family protein 97% (242) Cyclic di-GMP

metabolism K. pneumoniae KPN_00268 (ABR75721) cyclic diguanylate phosphodiesterase (EAL)

protein 40% (117)

K. pneumoniae KP1_3652 (BAH64237) hypothetical protein (GGDEF domain) 97% (44) Transport proteins K. pneumoniae gabP (BAH61898) RpoS-dependent gamma-aminobutyrate

transport protein 92% (36)

Lipid metabolism K. pneumoniae yjfP (BAH61336) putative esterase 94% (89) Regulatory proteins K. pneumoniae ybeF (BAH62363) putative LysR family transcriptional regulator 97% (131) tRNA modification K. pneumoniae mnmA (BAH62849) tRNA (5-methylaminomethyl-2-thiouridylate)-

methyltransferase 95% (85)

DNA metabolism K. pneumoniae tldD (BAH65428) Putative modulator of DNA gyrase 92% (158) Hypothetical protein K. pneumoniae yjcD (BAH61222) hypothetical protein 100%(98)

K. pneumoniae KP1_4995 (BAH65452) hypothetical protein 92% (38)

48

Table 3.2. RstA-repressed genes identified by subtractive cDNA hybridization

Predicted function Homologous gene (Accession no.) Predicted protein

% Identity (length of comparable amino acid sequence, no. of residues) Iron acquisition K. pneumoniae iroN (BAH64199) enterochelin and dihydrobenzoic acid receptor 92% (231)

Metal resistance K. pneumoniae pbrR (BAH66039) metal ion-sensing regulatory protein 99% (132) Transporter proteins K. pneumoniae KP1_0566 (BAH61430)) putative ABC transporter 98% (177) K. pneumoniae ybeX (BAH62385) putative integral membrane protein 99% (207) K. pneumoniae KP1_1781 (BAH62516) putative general substrate transporter 97% (163) K. pneumoniae mgtE (BAH64426) putative divalent cation transport protein 99% (139) Receptor proteins K. pneumoniae hmuR (BAF76156) TonB-dependent outer membrane receptor 100% (42) Fimbrial proteins K. pneumoniae KP1_4251 (BAH64781) putative fimbrial-like protein 98% (116) Antibiotic resistance K. pneumoniae KP1_5391 (BAH65810) putative beta-lactamase-like protein 98% (218) Energy metabolism K. pneumoniae nuoI (BAH64467) NADH dehydrogenase subunit I 97% (132) Sugar metabolism K. pneumoniae KPN_00998 (ABR76434) putative glycosyl transferase, group I 93% (103) K. pneumoniae gpmB (BAH61648) phosphoglycerate mutase 97% (146)

Glutathione biosynthesis K. pneumoniae gshB (BAH65148) glutathione synthetase 95% (132) Protein modification K. pneumoniae KP1_1405 (BAH62174) putative GNAT-family acetyltransferase 100% (68)

Protein folding K. pneumoniae dnaK (BAH61668) molecular chaperone 99% (248) Regulatory proteins K. pneumoniae ppk (BAH64622) polyphosphate kinase 99% (119) K. pneumoniae yeaG (ABR76626) RpoS-dependent stress kinase 95% (178) Hypothetical protein K. pneumoniae KP1_0493 (BAH61369) hypothetical protein 69% (41)

K. pneumoniae KP1_4582 (BAH65079) hypothetical protein 97% (67)

49

3.6 Figure

Fig. 3.1. Schematic representation of K. pneumoniae rstA locus and PrstA::lacZ activity measurements

(A) Diagrammatic representation of the K. pneumoniae rstA locus. Open reading frames are shown in large arrows. The PCR primers used to amplify DNA fragments harboring different PrstA regions are depicted, and the numbers refer to the positions relative to the translational start site. The dashed boxes indicate the PhoP and RstA binding sequences aligned with the rstA upstream region, and identical nucleotides are underlined. (B) The β-galactosidase activities of log-phased cultures of K. pneumoniae strains carrying placZ15 (vector control), pHY048, pHY050 or pHY053 grown in LB medium were determined and expressed as Miller units. The data shown were the average ± standard deviations from triplicate samples. *, P < 0.01 compared with the same strain carrying pHY048 or pHY050. #, P < 0.01 compared with CG43S3ΔlacZ harboring the same reporter plasmid.

(A)

Fig. 3.2. Identification of RstA-regulated genes by subtractive cDNA hybridization (A) The β-galactosidase activities of log-phased cultures of K. pneumoniae CG43S3ΔlacZ carrying pHY048 grown in LB or M9 medium were determined and expressed as Miller units.

The data shown were the average ± standard deviations from triplicate samples. (B) Gel electrophoresis of the PCR products from the subtractive cDNA hybridization. Lane 1, the RstA-repressed DNA amplicon. Lane 2, the RstA-activated DNA amplicon. M, DNA ladder.

(A)

Fig. 3.3. Growth of K. pneumoniae strains under iron-depletion/repletion

The Optical density of K. pneumoniae strains grown in (A) LB or LB supplemented with 200 μM 2’,2’-dipyridyl (LB+Dip) and (B) M9 or M9 supplemented with 20 μM ferrous sulfate (M9+Fe²⁺) was recorded at the indicated time points and plotted. Error bars, standard deviations from duplicate or triplicate samples.

(A)

Pb2+ (μM)

0.1 1 10 100

Relative growth

0.00 0.25 0.50 0.75 1.00

1.25 CG43

CG43-101

Growth inhibition assay (lead nitrate)

(B)

0.1 1 10 100

Relative growth

0.00 0.25 0.50 0.75 1.00

1.25 CG43S3

ΔrstA ΔrstB ΔrstAΔrstB

Pb2+ (μM)

Growth inhibition assay (lead nitrate)

Fig. 3.4. Effect of lead on the relative growth of K. pneumoniae strains

A growth inhibition assay was performed to investigate bacterial resistance to lead. Poor lead resistance is demonstrated in strains with reduced relative growths at low lead nitrate concentrations. Error bars, standard deviations from triplicate samples.

(A)

CG43S3 Δfur

CG43 CG43-101

(B)

CG43S3

ΔrstA

ΔrstB

ΔrstAΔrstB

Fig. 3.5. Siderophore production of K. pneumoniae strains on CAS agar plates

One microliter of K. pneumoniae overnight cultures was spotted onto CAS agar plates (162), incubated at room temperature for 16 h and photographed.

(A)

Fig. 3.6. Survival of K. pneumoniae strains after acid challenge

(A) The survival of K. pneumoniae CG43S3 grown in LB or M9 mediums, either with (adapted) or without (unadapted) pre-adaptation at pH 4.5 was determined at 30, 60 or 90 min after acid challenge at pH 3.0. (B) Strains were grown in M9 medium, adapted at pH 4.5, and the bacterial survival was determined at 30, 60 or 90 min post-challenge at pH 3.0. The survival rates were expressed as the CFU of viable bacteria divided by the CFU before acid challenge.

Error bars, standard deviations from triplicate samples.

LB LB (0.3% bile salt) LB (1% bile salt) LB (3% bile salt)

Fig. 3.7. Growth of K. pneumoniae strains on bile salt-containing medium

Overnight cultures of K. pneumoniae strains, as indicated above the figure, were serially diluted and spotted onto LB or M9 agar plates supplemented without or with different concentrations of bile salts. Bacterial growth after overnight incubation at 37°C was shown.

57

CHAPTER 4

RmpA Regulation of Capsular Polysaccharide

Biosynthesis in Klebsiella pneumoniae CG43

4.1 Abstract

Sequence analysis of the large virulence plasmid pLVPK in Klebsiella pneumoniae CG43 revealed the presence of another mucoid factor encoding gene rmpA besides rmpA2.

Promoter activity measurement indicated that the deletion of rmpA reduced K2 capsular polysaccharide (CPS) biosynthesis resulting in decreased colony mucoidy and virulence in mice. Introduction of a multicopy plasmid carrying rmpA restored CPS production in the rmpA or rmpA2 mutant but not in the rcsB mutant. Transformation of the rmpA deletion mutant with an rcsB-carrying plasmid also failed to enhance CPS production, suggesting that a cooperation of RmpA with RcsB is required for regulatory activity. This was further corroborated by the demonstration of in vivo interaction between RmpA and RcsB using bacterial two-hybrid analysis and co-immunoprecipitation analysis. A putative Fur binding box was only found at the 5’ non-coding region of rmpA. The promoter activity analysis indicated that the deletion of fur increased the rmpA promoter activity. Using EMSA, we further demonstrated that Fur exerts its regulatory activity by binding directly to the promoter.

As a result, the fur deletion mutant exhibited an increase in colony mucoidy, CPS production, and virulence in mice. In summary, our results suggested that RmpA activates CPS biosynthesis in K. pneumoniae CG43 via an RcsB-dependent manner. The expression of rmpA is regulated by the availability of iron and is negatively controlled by Fur.^a

a A part of this chapter has been published:

Cheng, H. Y., Y. S. Chen, C. Y. Wu, H. Y. Chang, Y. C. Lai, and H. L. Peng. 2010.

RmpA regulation of capsular polysaccharide biosynthesis in Klebsiella pneumoniae CG43. J Bacteriol 192:3144-58.

4.2 Introduction

Klebsiella pneumoniae, an important nosocomial pathogen, causes a wide range of infections, including pneumonia, bacteremia, urinary tract infection, and life-threatening septic shock (196). Clinically isolated K. pneumoniae strains usually produce a large amount of capsular polysaccharide (CPS), which confers not only a mucoid phenotype to the bacteria but also resistance to engulfment by professional phagocytes or to serum bactericidal factors (143, 203). CPS also plays a role in hindering fimbrial binding (209) and bactericidal effects resulting from antimicrobial peptides (30). The degree of mucoidy has been positively correlated with successful establishment of infection (176, 177). Most recently, the hypermucoviscosity of K. pneumoniae isolates has also been associated with the development of invasive syndrome (137). Among the identified serotypes, K. pneumoniae strains of K1 or K2 CPS are highly virulent in the mouse peritonitis model (167).

Klebsiella CPS resembles the E. coli group I CPS in primary structure and the mechanisms of biosynthesis (254). The chemical composition of Klebsiella K2 CPS, which contains uronic acid as the major component, has been determined as [→4-Glc-(1→3)-α-Glc-(1→4)-β-Man-(3←1)-α-GlcA)-(1→]n (246). Sequencing of the region responsible for K2 CPS biosynthesis in the K. pneumoniae Chedid strain revealed a total of 17 open reading frames organized into three transcriptional units (5). The two-component system (2CS) RcsCDB, which is the key regulatory system for E. coli colonic acid synthesis, often serves as a model for group I CPS biosynthesis (106, 155). Upon receiving environmental stimuli, the transmembrane sensor kinase RcsC undergoes autophosphorylation, the signal is subsequently relayed to the inner membrane Hpt (histidine-containing phosphotransfer) module RcsD and eventually to the cytoplasmic response regulator RcsB. The phosphorylated RcsB then interacts with RcsA, an auxiliary transcriptional regulator, and the heterodimer binds to the cps promoters, which in turn activates the biosynthesis of colanic acid capsule. RcsA is highly susceptible to degradation by the Lon protease; hence, lon mutation often leads to the accumulation of colanic acid (87).

However, the introduction of multicopy rcsB has been shown to suppress the rcsA-negative phenotype (23).

K. pneumoniae CG43, with a LD50 of 10 CFU for laboratory mice, is a highly mucoid clinical isolate of K2 serotype (36). Its mucoid phenotype has been correlated with the presence of the large virulence plasmid pLVPK; curing of this plasmid has rendered an

approximately 1000-fold decrease in mice virulence (12). We have also shown that rmpA2 on pLVPK encodes a transcriptional activator for the cps expression by binding directly to the putative promoters, Porf1-2 and Porf3-15 (130). Interestingly, sequencing of the large virulence plasmid pLVPK revealed an rmpA gene 29 kb away from rmpA2 (37). Excluding the extended 15 amino acids at the N-terminus of RmpA2, the deduced RmpA sequence shares an overall 71.4% identity and a conserved C-terminal DNA binding motif with the RmpA2 protein.

An rmpA gene on the 180-kb virulence plasmid pKP100 of K. pneumoniae 52145 has been reported twenty years ago (29). The encoding protein RmpA was subsequently demonstrated to be able to enhance colonic acid biosynthesis in E. coli HB101 (176). Later, the rmpA2 gene, carrying an extended 5’ sequences of the rmpA, was isolated and shown to be able to activate K2 capsule production in the recombinant E. coli K-12 harboring Klebsiella K2 cps genes (246). Herein, we report the characterization of RmpA in K2 CPS biosynthesis and comparative analysis of the expression of the two mucoid factor encoding genes. We have demonstrated the interaction between RmpA and RcsB on the regulation of the CPS biosynthesis and the involvement of Fur on the expression of rmpA has also been studied.

4.3 Results

4.3.1 Comparison of rmpA/rmpA2 containing regions in K. pneumoniae CG43

On the basis of the sequence analysis, the DNA fragment containing the rmpA2 gene with upstream vagC and vagD and downstream iucABCDiutA genes was labeled as PAI (pathogenicity island)-1, while the rmpA-, fecIRA- and iroBCD-containing region was named PAI-2 (Fig. 4.1). It was noted that both regulator genes were located downstream of the siderophore biosynthesis genes flanked by insertion sequences. The positive regulatory role of RmpA2 in the expression of the major virulence factor, CPS, has been demonstrated (20).

As a result, whether RmpA plays a role in the regulation of CPS biosynthesis was investigated.

4.3.2 Deletion of rmpA reduced CPS production and virulence

To assess the functional role of RmpA, the rmpA deletion mutant was generated using the allelic-exchange strategy. The colony of ΔrmpA mutant on the LB agar plate was found to be smaller than its parental strain, and the degree of mucoidy was reduced significantly as determined by a string test (130), which refers to the ability to form a string when the bacterial colony was picked with toothpick. As shown in the sedimentation test in Fig. 4.2A, the ΔrmpA mutant as well as the ΔrcsB mutant could be rapidly precipitated by low-speed centrifugation. The loss of mucoid phenotype in ΔrmpA mutant could be complemented with the transformation of pRK415-RmpA, or pRK415-RmpA2. Interestingly, the mucoid phenotype could not be restored by introducing the rcsB-expression plasmid pRK415-RcsB.

The sedimentation analysis also revealed that the introduction of pRK415-RmpA or pRK415-RmpA2 was able to increase the mucoviscosity of the rmpArmpA2 double mutant (Fig. 4.2A), indicating the independent regulatory activity of RmpA and RmpA2. The effect of rcsB deletion could only be complemented by transformation of the ΔrcsB mutant with pRK415-RcsB but not pRK415-RmpA. As assessed by measuring the glucuronic acid content, which served as an indicator for Klebsiella K2 CPS (187), deletion of rmpA or rcsB caused approximately 25% reduction in the amount of CPS compared with that of CG43S3 (Fig.

4.2B). The effect of rmpA or rcsB deletion could only be restored by transformation of ΔrmpA with pRK415-RmpA or transformation of ΔrcsB with pRK415-RcsB (Fig. 4.2C), which is consistent with the findings in Fig. 4.2A. Furthermore, the rmpA deletion appeared to increase LD50 from 1 × 10⁴ CFU to 5 × 10⁵ CFU in the mouse peritonitis model and

reduced the resistance to human serum from > 95% to > 70% (Table 4.1). The deficiency in serum resistance could be reverted by the introduction of pRK415-RmpA, suggesting a role of RmpA in bacterial virulence.

4.3.3 RmpA acted as an activator of cps expression

To investigate whether the CPS-deficient phenotype of ΔrmpA mutant was a result of altered expression of the cps genes, three reporter plasmids pOrf12 (Porf1-2::lacZ), pOrf315 (Porf3-15::lacZ) and pOrf1617 (Porf16-17::lacZ), each carrying a lacZ reporter gene transcriptionally fused to the putative promoter region of the K2 cps gene cluster (142), were used to transform K. pneumoniae strains CG43S3ΔlacZ, ΔrmpAΔlacZ, ΔrmpA2ΔlacZ or ΔrcsBΔlacZ individually. The promoter activity measurements shown in Fig. 4.3A reveal that the deletion of rmpA reduced the activity of Porf1-2::lacZ and Porf16-17::lacZ. A reduction in the activity P_orf1-2::lacZ or P_orf16-17::lacZ was also observed in the ΔrcsB mutant. Interestingly, the deletion of rmpA2 had less effect on the activity of Porf16-17::lacZ compared with the rmpA deletion. As shown in Fig. 4.3B, the deleting effect of rmpA, rmpA2 or rcsB on Porf3-15::lacZ activity was only apparent when the strains were grown in M9-glocuse minimal medium.

Compared with that of rmpA2, the deletion of rmpA resulted in a more drastic reduction in the activity of P_orf1-2::lacZ and P_orf16-17::lacZ implying a differential regulation of RmpA and RmpA2 on the cps promoters. Nevertheless, the results suggested that the expression of RmpA, RcsB, and RmpA2 is required for cps expression.

4.3.4 Effect of poly(G) tract variation on rmpA/rmpA2 expression

A close inspection of the rmpA and rmpA2 nucleotide sequences has revealed a poly(G) tract in both genes. Previously it has been reported that different K. pneumoniae clinical isolates harbored rmpA2 genes with various length of poly(G) tract encoding either a full-length or a truncated form of RmpA2, which lost its DNA-binding ability as well as its trans-activation on K2 cps expression (130). Similarly, only the 10-Gs version of the poly(G) tract in rmpA would allow the synthesis of full-length RmpA while other lengths of the poly(G) tract would result in the occurrence of an ochre stop codon and render the synthesis of a truncated RmpA (Fig. 4.4AB). It remained unknown, however, whether the length of poly(G) tracts in rmpA and rmpA2 would be altered in a different rate during bacterial growth.

To address this, a modified assay from the insertional restoration of LacZα-complementation (46) was designed. Firstly, the DNA fragments encompassing the translational start site and the latter stop codon encoding region in rmpA and rmpA2 genes were cloned in-frame with

the lacZα gene in yT&A to generate pHY291 and pHY306. Since the rmpA and rmpA2 genes harboring respectively the 10-Gs and 11-Gs versions of poly(G) tracts were introduced, E.

coli JM109 transformants harboring either pHY291 or pHY306 would form blue colonies on LB agar plates supplemented with IPTG and X-gal if the association between the recombinant LacZα peptides and the ω peptides synthesized from the chromosomal lacZω gene was successful. As a control, the same DNA fragments were cloned in the opposite direction to lacZα to generate pHY290 and pHY305. Transformants carrying pHY290 or pHY305 would fail to synthesize full-length LacZα peptides due to the presence of multiple stop codons in the inserted sequences. It was later found that transformants carrying pHY291 or pHY306 could form pale blue colonies on LB agar plates supplemented with IPTG and X-gal while strains harboring pHY290 or pHY305 formed white colonies. As a result, cultures of E. coli JM109 transformants harboring yT&A or each of the recombinant plasmids were either grown overnight and spotted onto LB agar plates supplemented with IPTG and X-gal or grown to log phase, diluted serially and plated onto the same plates.

As shown in Fig. 4.4C, strains carrying yT&A (no. 1 and 4) formed blue colonies while those carrying pHY290 (no. 2) or pHY305 (no. 5) formed white colonies as expected.

Interestingly, E. coli cells harboring pHY291 (no. 3) or pHY306 (no. 6) formed blue colonies during the initial period of incubation, and several white colonies were subsequently observed after a longer incubation duration. The white colonies were also found in the refreshed cultures carrying pHY291 or pHY306 as shown in Fig. 4.4D, although in approximately the same frequency (about 1 per 300 colonies in both transformants).

Sequence analysis of these white colonies has revealed various lengths of poly(G) tracts in the inserted sequence, implying the event of slip-strand synthesis during DNA replication.

The results indicated the presence of the similar mechanism governing the variation of RmpA and RmpA2 coding sequences.

4.3.5 RmpA regulates cps expression in an RcsB-dependent manner

To investigate the possibility of an interaction between RmpA and RcsB for cps expression, the lacZ reporter cassette on pOrf12 was cloned into a suicide vector, and the plasmid was mobilized into K. pneumoniae CG43S3ΔlacZ and its isogenic ΔrmpA and ΔrcsB mutant. The resulting strain harboring a chromosomally integrated Porf1-2::lacZ cassette was then transformed with different complementation plasmids, and the β-galactosidase activities were determined. As shown in Fig. 4.5, the introduction of pRK415-RmpA or pRK415-RcsB could enhance Porf1-2::lacZ activity in the parental strain, suggesting the functional activity of

RmpA or RcsB. Functional RmpA could enhance cps expression in the rmpA deletion strain but not the rcsB deletion strain. Consistent with the phenotype observed in Fig. 4.2B, the functional RcsB carried by pRK415-RcsB could not restore cps expression in the rmpA deletion strain. The introduction of pRK415-RmpAN, which encoded a truncated RmpA without the carboxyl terminal DNA binding region, into the rmpA deletion strain also failed to restore cps expression. The results suggest that RmpA activates cps expression in an RcsB-dependent manner and that its DNA binding motif is required for regulation.

4.3.6 Interaction between RmpA and RcsB using two-hybrid analysis

Since the cooperation of RcsA and RcsB for regulation on K2 cps expression has been demonstrated (250), we thought of using EMSA to investigate if the RmpA exerts RcsA-like activity to interact with RcsB in order to bind to the Porf1-2 region cooperatively. However, the overproduction of RmpA using the pET expression system appeared to impair cell growth significantly; hence, the bacterial two-hybrid assay was employed instead. Therefore plasmids pTRG-RcsA, pTRG-RmpA, pTRG-RmpA2, pTRG-RmpAN and pBT-RcsB which harbored λ-cI-RcsA, λ-cI-RmpA, λ-cI-RmpA2, λ-cI-RmpAN and α-RNAP-RcsB coding sequences were constructed. The interaction between α-RNAP and λ-cI fusion proteins would allow binding of λ-cI to the operator sequence and recruitment of α-RNAP to initiate the transcription of ampR as well as lacZ genes in the reporter cassette harbored in the E. coli reporter strain. The interaction between the recombinant proteins could be verified by bacterial growth on the X-gal indicator plate supplemented with carbenicillin and the level of interaction could be quantified by measuring the activation of the LacZ reporter. As shown in Fig. 4.6A, the strain carrying pBT-RcsB/pTRG-RcsA or the positive control plasmids grew well on the indicator plate. Substitution of pBT-RcsB and pTRG-RcsA with pBT or pTRG resulted in poor or no growth. Those strains carrying pBT-RcsB/ pTRG-RmpA, pBT-RcsB/

pTRG-RmpA2, or pBT-RcsB/ pTRG-RmpAN also grew on the indicator plate. As shown in Fig. 4.6B, the strain pBT-RcsB/pTRG-RcsA exhibited relatively higher activity than the bacteria carrying pBT-RcsB/pTRG-RmpA, pTRG-RmpA2 or pTRG-RmpAN. The transformants harboring pBT-RcsB and pTRG-RmpA, pTRG-RmpA2 or pTRG-RmpAN also exhibited higher activity compared with strains of which pBT-RcsB was replaced by pBT (Fig. 4.6B, right panel). The results suggested an in vivo interaction between RcsB and RmpA, and the N-terminal peptide (residues 1 to 84) of RmpA may play an important role in the interaction.

4.3.7 Co-immunoprecipitation analysis of the interaction between RmpA and RcsB To confirm the results from the two-hybrid analysis, co-IP was also performed. The

4.3.7 Co-immunoprecipitation analysis of the interaction between RmpA and RcsB To confirm the results from the two-hybrid analysis, co-IP was also performed. The