Deletion of fur led to overproduction of CPS

CHAPTER 4 RmpA Regulation of Capsular Polysaccharide Biosynthesis in Klebsiella

4.3.11 Deletion of fur led to overproduction of CPS

If rmpA expression is negatively regulated by Fur, then the CPS level in the fur deletion mutant should increase. As shown in Fig. 4.12A, the Δfur mutant formed more mucoid and glistening colonies when compared with the parental strain. In a string test, the Δfur mutant could form a string at least three-fold longer than its parental strain (data not shown).

Introduction of pRK415-Fur carrying a functional fur allele in Δfur mutant resulted in small colonies (Fig. 4.12A) and readily precipitated phenotype (Fig. 4.12B). The changes in CPS production were also evident in the glucuronic acid content measurement. As shown in Fig.

4.12C, the fur mutant exhibited more than two-fold increase in the glucuronic acid production while transformation with pRK415-Fur resulted in an approximately 50% reduction of the glucuronic acid.

4.4 Discussion

The presence of rmpA and rmpA2 on pLVPK as two independent loci 29 kb apart has been demonstrated (37). As shown in Fig. 4.1, the DNA fragments containing rmpA2 and rmpA also harbored an iron-acquisition gene cluster and insertion sequence, which are characteristics of a pathogenicity island, and hence named PAI-1 and PAI-2, respectively.

Why and how the evolutionary convergence of the similar gene organization occured remained to be explained. In this study, we show that rmpA and rmpA2 genes are present in one bacterial strain and both encode a mucoid factor contributing to K2 CPS biosynthesis.

Presumably, differential control on the expression of these two mucoid factors is required for an efficient regulatory function with no redundant activity in the bacteria.

Sequence analysis revealed that RmpA and RmpA2 belong to the UhpA-LuxR family of transcription factors, which also include RcsA and RcsB (224). The involvement of RcsB in Klebsiella K2 capsule biosynthesis of the recombinant E. coli K-12 has been demonstrated (245). Previous studies have found that RcsB must interact with RcsA to form a heterodimer in order to bind specifically to the cps promoter for transcription initiation (122). In E. coli, the cellular level of RcsA was limited at 37°C due to its degradation by the Lon protease, and thus the colonic acid capsule was overproduced only at lower temperatures (87). In K.

pneumoniae CG43, however, a profound production of CPS was observed at 37°C. We assumed that RmpA or RmpA2 could take the place of RcsA to regulate the cps expression at 37°C. This hypothesis was supported by the fact that the deletion of rcsA in K. pneumoniae CG43 did not affect the mucoid phenotype or the CPS amount at 37°C (data not shown).

Moreover, the two-hybrid analysis and co-immunoprecipitation assay further demonstrated that RmpA or RmpA2, in addition to RcsA, was able to interact with RcsB. Therefore in K.

pneumoniae, multiple accessory factors may be employed in order to increase the CPS biosynthesis in response to different environmental stimuli.

Though bacterial two-hybrid analysis has clearly indicated an interaction between RcsB and its auxiliary factors, the bacterial growth on the indicator plate was similar between the positive control strain and the strain carrying pBT-RcsB and pTRG-RcsA while the results from the β-galactosidase activity measurements showed that the level of reporter gene expression in the positive control strain was approximately four times as the strain carrying pBT-RcsB and pTRG-RcsA. The discrepancy may be due to the concentration of antibiotics administrated in the indicator plate, in which all the bacteria strains expressing the resistance

gene above a threshold level could grow on the plate. The hypothesis was supported by the relatively weaker growth of the strains carrying pBT-RcsB along with pTRG-RmpA, pTRG-RmpA2 or pTRG-RmpAN, which also revealed a lower level of β-galactosidase activity, in comparison with the positive control strain or the strain harboring pBT-RcsB and pTRG-RcsA. Nevertheless, the levels of bacterial growth could be further quantified as the relative survival of strains challenged with different doses of antibiotics, in order to correlate the phenotype observations to the expression levels of the reporter gene and to differentiate between weak interactions.

Previously over-expression of rmpA by pET30 system resulted in a significant reduction of bacterial growth after IPTG induction, and most of the protein produced was found in the insoluble fraction, which could not be solubilized by 6N or 8N urea after prolonged incubations (16 to 24 h). The over-expression by using pET32 system in combination with E.

coli strain carrying a trxB mutation also failed to produce soluble recombinant RmpA proteins though the growth retardation after IPTG induction was reduced. As a result, the pGEX system with a GST fusion tag was utilized. Although the growth retardation was no longer observed after IPTG induction, the majority of GST-RmpA remained insoluble and rapidly degraded during purification processes. To improve the solubility of the recombinant RmpA protein, strategies including addition of ion cofactors, using a larger fusion tag such as mannose-binding protein or co-expression with a chaperone protein or with an interaction partner such as RcsB could be considered.

In a typical 2CS, the response regulator could undergo a conformational change to form a homo-dimer upon phosphorylation. In the case of RcsB, which interacted with more than one factor, it remained unclear if phosphorylation state of RcsB would be in favor of homo-dimerization. Since we have shown that RcsB protein could bind specifically to RcsA, RmpA or RmpA2, whether RcsB proteins with alterations on the Aspartic acid at position 56 to Alanine or Glutamate mimicking un-phosphorylated or constitutively-phosphorylated status respectively would require further analysis.

A conserved RcsAB box sequence has been identified in Klebsiella K2 cps Porf1-2 (250), and analysis of the Porf16-17 sequence also revealed a semi-conserved RcsAB box. This suggested that the decrease of the CPS production in the ΔrmpA mutant may have been due to a reduction in the expression of cps-orf1-2 and cps-orf16-17. The existence of an rmpA gene has been correlated to the hypermucoviscosity phenotype in K. pneumoniae clinical isolates (273). It is interesting to note that no rmpA homologue in any other bacterial genomes could

be identified. The BLAST (http://www.ncbi.nlm.nih.gov) search in the released genome sequences of K. pneumoniae NTUH-K2044 (262), MGH78578 (http://genome.wustl.edu/), and 342 (74) revealed the presence of rmpA only in NTUH-K2044, which is a heavy encapsulated clinical isolate of K1 serotype (65, 273). We have also observed that rmpA was more prevalent in the strains belonging to K1/K2 serotypes (36 out of 36 isolates, 100%) than in the non-K1/K2 clinical isolates (28 out of 83 isolates, 33.73%) collected in our laboratory.

The presence of rmpA in K. pneumoniae K1 or K2 isolates has also been associated with virulence in mice (24). These findings suggest that the profound expression of K. pneumoniae CPS, which is a major virulence factor, could be attributed to the regulation by RmpA or its homologue.

Similar to rmpA2, rmpA harbored a poly(G) tract in the coding sequence. The occurrence of DNA slip-strand synthesis may result in a truncated RmpA of abnormal function. Conceivably, the frequency of the mutation caused by DNA slip-strand synthesis in rmpA or rmpA2 may play a role in the differential expression of the two regulatory genes.

Although the modified LacZα-complementation assay in this study was feasible in detecting the variation of poly(G) tract, no significant differences in the rate of change in the poly(G) tracts of rmpA and rmpA2 during bacterial replication was noted. However, it remained to be clarified if the rate of change would differ when bacterial replication took place in a different growth medium or when cells were grown under environmental stress. Moreover, changes in the poly(G) tract would result in the biosynthesis of either a full-length/functional or a truncated/non-functional RmpA and RmpA2 protein. Since bacterial growth would be impaired upon the accumulation of RmpA, it would be more appropriate and practical to investigate phase variation of the poly(G) tract when the full-length coding region of rmpA and rmpA2 were taken into consideration.

It was noted that the white colony retrieved in the modified LacZα-complementation assay did not turn to blue compared with the blue-to-white frequency investigated in this study, which was approximately 0.3% in both rmpA and rmpA2 poly(G). The possible explanation could be that the once the poly(G) tract was altered to result in a stop codon, the chance to restore the correct reading frame was at least one-thirds lower given the same probability of adding/deleting a guanidine during DNA replications.

The importance of the poly(G) tract would be more apparent in clinical surveillance of rmpA/rmpA2 genes as not all rmpA or rmpA2 gene detected by using PCR would inevitably encode a functional/full-length protein. We have also found that the presence of a poly(G)

tract would severely interfere the validity of DNA sequencing, and thus the sequencing process could only be completed by using a primer complementary to the non-coding region of rmpA or rmpA2. It is therefore strongly recommended that the reading frames of the poly(G) tract be determined by DNA sequencing analysis in clinical prevalence studies of rmpA/rmpA2 gene.

A Fur-box like sequence was identified within -146 and -104 upstream of the transcription start site (Fig. 4.11B). This site was not uncommon compared with a genome-wide study in Yersinia pestis (77), in which 17 out of 34 predicted Fur binding sites were located at -100 relative to each transcriptional start site. Also as shown in Fig. 4.11C, the rmpA transcript was increased upon the deletion of fur indicating a negative regulation of Fur on rmpA expression in K. pneumoniae.

Fur governs iron uptake by repressing the transcription of the genes involved in the biosynthesis of siderophore, iron storage, iron sparing and respiration (10, 139, 244). In addition , it is also involved in virulence properties (247, 263), acid stress response (73), osmotic shock (104) and chemotaxis (60). In this study, a negative role of Fur in CPS biosynthesis that is achieved by the reduction of the RmpA expression is reported for the first time. The findings imply that some K. pneumoniae strains may face iron shortage during infection; accordingly, the genes involved in both the biosynthesis of the iron acquisition system and the production of CPS are up-regulated. Therefore, coordination between CPS production and iron uptake may play an important role in the pathogenesis of the bacteria.

Nevertheless, cps expression was also subjected to regulation by the 2CS response regulators KvgA, KvhA, and KvhR (142). How the interplay between Fur, RmpA and the 2CS determines the control of the cps expression remains to be elucidated.

In summary, the overall scene could be depicted as shown in Fig. 4.13. In K.

pneumoniae CG43, the mucoid factor RmpA exerts a regulatory role on cps expression in an RcsB-dependent manner. The interaction between RcsB and its auxiliary factors RcsA, RmpA or RmpA2 leads to the elevated biosynthesis of K2 CPS. A differential regulation of rmpA and rmpA2 expression is also noted as fur deletion increased rmpA expression but had no effect on rmpA2 expression through a direct binding of Fur to the putative promoter of rmpA in addition to the promoters of siderophore biosynthetic genes. The findings not only connect iron uptake to capsule biosynthesis but also report the first time that the Klebsiella Fur is involved in the regulation of K2 CPS biosynthesis.

4.5 Table

Table 4.1. Virulence properties of K. pneumoniae strains.

Strain LD50 (CFU) Survival rate in human serum (%)^a

CG43S3 1 × 10⁴ 95.6 ± 3.6

CG43S3ΔrmpA 5 × 10⁵ 71.2 ± 5.6

CG43S3ΔrmpA[pRK415-RmpA] ND^b > 99

CG43S3ΔgalU (control) 1 × 10⁶ 0

a Percent survival rate in human serum is expressed as 100 × (the number of viable bacteria after treatment with human serum/ the number of viable bacteria after treatment with PBS).

b ND, not determined.

4.6 Figure

PAI-1 (25,734 bp)

PAI-2 (28,549 bp)

1 kb

yebB yebA vagC vagD rmpA2 iutA iucD iucC iucB shiF

50113 bp IS3 75847 bp

IS630

lacZ lacZ

iucA

P

_rmpA2

::lacZ P

_iucA

::lacZ

82504 bp 111053 bp

insA insB rmpA pagO iroN iroD iroC iroB fecI fecA insA insB

IS1 IS1

lacZ lacZ

P

_rmpA

::lacZ P

_iroB

::lacZ

fecR

PAI-1 (25,734 bp)

PAI-2 (28,549 bp)

1 kb

yebB yebA vagC vagD rmpA2 iutA iucD iucC iucB shiF

50113 bp IS3 75847 bp

IS630

lacZ lacZ

iucA

P

_rmpA2

::lacZ P

_iucA

::lacZ

82504 bp 111053 bp

insA insB rmpA pagO iroN iroD iroC iroB fecI fecA insA insB

IS1 IS1

lacZ lacZ

P

_rmpA

::lacZ P

_iroB

::lacZ

fecR

Fig. 4.1. Comparison of rmpA and rmpA2 containing PAI-like regions

The arrows indicate predicted open reading frames and insertion sequences. The reporter constructs used for promoter activity measurement are shown below.

(A)

(B)

Glucuronic acid content (μg per 109 CFU) 0

Glucuronic acid content (μg per 1010 CFU) 0

Fig. 4.2. Comparison of precipitation speeds and K2 CPS production in K. pneumoniae strains

(A) The strains tested were grown overnight in LB broth at 37°C and subjected to centrifugation at 4,000 ×g for 5 min. (B) The glucuronic acid content, which served as an indicator of K2 CPS, was determined from overnight K. pneumoniae cultures. The results are expressed as the average of the triplicate samples. Error bars indicate standard deviations. *, P < 0.01 compared with CG43S3. **, P < 0.001 compared with CG43S3. (C) Quantification of Klebsiella K2 CPS production. The results are expressed as an average of triplicate samples. Error bars indicate standard deviations. #, P < 0.001 compared with ΔrmpA mutant carrying pRK415 or pRK415-RcsB. &, P < 0.001 compared with ΔrcsB mutant carrying pRK415 or pRK415-RmpA.

(A)

Fig. 4.3. Expression of K2 cps genes in various genetic backgrounds

The β-galactosidase activities of K2 cps Porf1-2::lacZ, Porf3-15::lacZ and Porf16-17::lacZ in K.

pneumoniae CG43S3ΔlacZ (wild-type) and its isogenic strains (ΔrmpAΔlacZ, ΔrmpA2ΔlacZ, and ΔrcsBΔlacZ) harboring each of the reporter plasmids pOrf12, pOrf315 or pOrf1617 were determined from log-phased cultures grown in LB broth (A) or M9-glucose medium (B). The results are shown as an average of triplicate samples. Error bars indicate standard deviations. *, P < 0.01 compared with CG43S3ΔlacZ. **, P < 0.001 compared with CG43S3ΔlacZ.

(A)

(B)

(C)

(D)

1 2 3

4 5 6

Fig. 4.4. Effect of poly(G) tract variation on RmpA and RmpA2 coding sequence

The DNA sequences and the encoded amino acids encompassing the poly(G) tract (boxed guanidine sequences) in rmpA (A) and rmpA2 (B) genes are shown. The numbers denote the nucleotide position relative to the tri-nucleotide encoding the first amino acid in full-length RmpA or RmpA2. Different numbers of guanidine resulted in the translation of full-length RmpA (rmpA-10Gs) and RmpA2 (rmpA2-11Gs) or the truncated forms of RmpA (rmpA-11Gs and rmpA-9Gs) and RmpA2 (rmpA2-10Gs and rmpA2-9Gs) due to the occurrence of an ochre or an opal stop codon, which are boxed and indicated by an arrow. (C) Overnight cultures of E.

coli JM109 transformants carrying yT&A (1 and 4), pHY290 (2), pHY291 (3), pHY305 (5) or pHY306 (6) were spotted onto an LB agar plate supplemented with ampicillin, IPTG and X-gal.

After incubation at 37°C for 24 h, the plate was photographed. (D) Log-phased culture of E.

coli JM109 harboring pHY291 was diluted serially and plated onto LB agar plates supplemented with ampicillin, IPTG and X-gal. After incubation at 37°C for 24 h, the plate was photographed. The arrow indicated a white bacterial colony, which harbored a pHY291 carrying alteration in the poly(G) tract, which is confirmed by DNA sequencing.

P_orf1-2

Fig. 4.5. RmpA and RcsB activated the expression of K2 cps genes in a coordinated manner

The K. pneumoniae CG43S3ΔlacZ (wild-type) and its isogenic strains (ΔrmpAΔlacZ and ΔrcsBΔlacZ), each carrying a chromosomally-integrated K2 cps Porf1-2::lacZ cassette, were transformed individually with pRK415 and its derived plasmids. The β-galactosidase activities were determined from log-phased (OD600 of 0.7) cultures grown in LB broth. The results are shown as the average of the triplicate samples. Error bars indicate standard deviations. **, P <

0.001 compared with each strain carrying pRK415 or pRK415-RmpAN. #, P < 0.001 compared with ΔrmpAΔlacZ[pRK415-RmpA]. &, P < 0.001 compared with ΔrcsBΔlacZ[pRK415-RcsB].

(A)

Fig. 4.6. Bacterial two-hybrid analysis of the interaction between RcsA/RcsB, RcsB/RmpA, and RcsB/RmpA2 proteins

(A) The growth of serially-diluted cultures of E. coli reporter strains co-transformed with pTRG and pBT or the derived plasmids was investigated on the indicator plate. (B) The E. coli reporter strains co-transformed with pTRG and pBT or the derived plasmids were grown to log-phase (OD600 of 0.5) in LB broth, induced with IPTG, and the β-galactosidase activities were determined. The results are shown as the average of the triplicate samples. Error bars indicate standard deviations. *, P < 0.001 compared with negative control strain carrying the pBT and pTRG. #, P < 0.001 compared with the reporter strain carrying either pBT/pTRG-RcsA or pBT-RcsB/pTRG. &, P < 0.001 compared with the reporter strain carrying the same pTRG-derived plasmid and the bait vector pBT.