1.3 Cyclic-di-GMP signaling system
1.3.3 Role of c-di-GMP in virulence
C-di-GMP signaling is involved in regulating virulence of many bacteria which include E. coli, S. Typhimurium, Vibrio cholerae, Pseudomonas aeruginosa, Bordetella pertussis, Xanthomonas campestris, Legionella pneumophila, Brucella
melitensis and Anaplasma phagocytophilum [75,93,98,100-105]. In V. cholera,
downregulation of c-di-GMP levels by the PDE VieA leads to activation of cholera toxin [106]. P. aeruginosa c-di-GMP signaling is required for biofilm formation in chronic infection and also the acute infection phenotype [90,107,108]. The plant pathogen X. campestris pathovar campestris (Xcc) causes disease through a HD-GYP domain protein expression to regulate the production of extracellular enzymes and extracellular polysaccharide[109], and motility. In S. typhimurium, the EAL-domain like protein STM1344 causes resistance to oxidative stress, inhibits rapid macrophage killing and is required for virulence in the typhoid fever mouse model. Intriguingly, STM1344 has no PDE activity nor has bind activity to c-di-GMP, and hence the involvement of c-di-GMP signalling in these phenotypes remains to be investigated [110].
11
1.3.4 Domain structure of GGDEF- and EAL-protein encoding genes in K.
pneumoniae CG43S3
GGDEF and EAL domain proteins are widespread in bacterial genomes. Very often one bacterial genome contains more than one GGDEF and EAL domain protein.
The sequenced genome of S. typhimurium codes for 20 GGDEF/EAL domain proteins;
5 contain a GGDEF, 8 an EAL domain and 7 contain both. On the other hand, E. coli K-12 has 12 GGDEF, 12 EAL and 7 GGDEF-EAL domain proteins [111]. A recent study search for conserved GGDEF and EAL domains in three sequenced K.
pneumoniae genomes revealed multiple copies of GGDEF and EAL containing
proteins: 21 for K. pneumoniae Kp342, 18 for K. pneumoniae MGH 78578 and 17 for K. pneumoniae NTUH-K2044 [112]. The K. pneumoniae CG43 genome sequence has
recently been resolved and the annotation results show 24 GGDEF- and EAL-domain protein encoding genes. As shown in Fig. 1.1, 10 GGDEF-, 11 EAL- and 4 GGDEF-EAL-domain protein encoding genes are identified. The sensor domains identified include MASE, CHASE, CACHE and CSS motif [113]. Genome-wide approach to study these proteins might shed light on their functional roles in different environmental settings.
12
Fig. 1.1. Domain architecture of putative c-di-GMP signaling proteins encoded
by the K. pneumoniae CG43. The genome of K. pneumoniae CG43 encoding (A)
GGDEF, (B) EAL, and (C) GGDEF/EAL, the analysis of protein functional domain,
13
as indicated, was performed using the Pfam database provided online (http://www.sanger.ac.uk /Software/Pfam/). The identities of the proteins are also shown in blue: four EAL domain proteins (YjcC, FimK and MrkJ) that have been described in K. pneumoniae [58,91,114,115] and (chapter 2).
1.4 Oxidative stress
The most commonly discussed oxidants that cause damage to DNA, proteins, and cell membranes and often results in cell death are the reactive oxygen species (ROS) including superoxide anion (O2.-), hydrogen peroxide (H2O2), and the hydroxyl radical (HO.), and the reactive nitrogen species (RNS), which include nitric oxide (NO.) and peroxynitrite (ONOO-). During infection, pathogens have equipped to protect themselves from the oxidative burst of phagocytic cells and the challenging oxidative environments within cellular and extracellular compartments.
The oxygen species can be excluded from active sites by electron transfer to the redox cofactors. Reactions of this type occur by the formation of reactive oxygen species (ROS) and their subsequent inactivation of enzymes. Intracellular molecular oxygen can adventitiously abstract electrons from the exposed redox moieties of electron-transfer enzymes, thereby generating partially reduced oxygen species. A mixture of O2− and H2O2 is formed, reflecting the fact that either one or two electrons can be transferred in an oxidation event. Flavoenzymes, which is ubiquitous and
14
abundant, in particular are responsible for all aerobic organisms to experience a steady flux of endogenously generated oxidants [116]. The overall reaction rate is proportional to collision frequency; thus, O2− and H2O2 fluxes depend directly upon the ambient concentration of oxygen [117,118]. Several source of oxidative stress have been identified that include (a) intracellular enzyme autooxidation. In exponentially growing E.coli, both O2.-and H2O2 are generated by the auto-oxidation of components of the respiratory chain [119]. (b) envirmental redox reactions, (c) H2O2 released by competing microbes, (d) phagosomal NADPH oxidase and (e) redox cycling antibiotics , plants or microorganisms secrete redox-cycling antibiotics that diffuse into the competing bacteria, chemically oxidize redox enzymes and transfer the electrons to molecular oxygen [118].
1.4.1 Oxidative stress response in bacteria
The defense mechanisms, which play an important role in determining the bacterial virulence, include sensing, avoiding, and removing the oxidants. In Escherichia coli, superoxide is removed by SODs (SodA, SodB, SodC) [120],
generating hydrogen peroxide which is then removed by catalases (KatE, KatG) and peroxidases (AhpC). The transcriptome analysis of Pseudomonas aeruginosa or Staphylococcus aureus response to H2O2 revealed many more genes including the
15
virulence genes, the genes encoding products involved in DNA repair and anaerobic metabolism were induced. Nevertheless, many of the genome-wide analysis revealed the expression of more than 100 genes were induced in responding to either H2O2 or paraquat indicating the complexity of the antioxidant strategies [121-123].
Many of the defenses are controlled by regulators that respond to iron (e.g., Fur), oxygen tension (e.g., FNR and ArcAB), superoxide (e.g., SoxRS), and hydrogen peroxide (e.g., OxyR). In E. coli, OxyR is known to regulate more than nine genes including katG (hydroperoxidase I), ahpC (alkylhydroperoxide reductase), gorA (gluthione reductase), oxyS (regulatory RNA), and fur that are involved directly or indirectly in the oxidative stress response. The antioxidant genes katE, katG, and sodC have been reported to be components of RpoS regulon .While the expression of
superoxide dismutase SodA and SodB are respectively controlled by at least 3 global regulators including SoxRS, ArcAB, and Fur [124].
RpoS is a sigma factor which regulates expression of a variety of genes in both E.
coli and Salmonella spp. that allows the bacteria to adapt, resist, and survive under
stress condition. The deletion of rpoS in V. vulnificus was found to down-regulate the expression of fur, indicating a cascade regulation between the two global regulators.
Fur, which is complexed with iron as an iron responsive regulator, binds to multiple sites (a 19-bp Fur box) with differential affinities to repress the transcription of genes
16
required for iron acquisition, acid and oxidative stress responses. However, Fur has also been demonstrated to function as a transcriptional activator for several genes including fur itself (Hp and Vv), and virulence associated genes [125-127].
In K. pneumoniae CG43, the 2CS response regulator encoding gene kvgA deletion was found to reduce the expression of katG and sodC. KvgAS has also been shown to be induced expression in upon treatment with 0.2 mM paraquat, which suggesting an involvement of the 2CS in oxidative stress defense in the bacteria. Moreover, we have reported in K. pneumoniae CG43 that rpoS deletion reduced the kvgAS expression which also showing a cascade regulation on the expression of 2CS. This suggests complex signaling networks with inter-connection regulatory circuits are required for multiple stress signal integration [128].
1.5 Transcriptome analysis
Transcriptome analysis of gene expression can be profiled by high throughput techniques including SAGE, serial analysis of gene expression [129], microarray [130], and sequencing of clones from cDNA libraries [131]. Microarrays have been the method of choice providing high throughput and affordable costs. However, the microarray technology has limitations including insufficient sensitivity for quantifying lower abundant transcripts, narrow dynamic range and non-specific hybridizations.
17
Sequencing-based methods such as SAGE rely upon cloning and sequencing cDNA fragments. This approach allows quantification of mRNA abundance by counting the number of times cDNA fragments from a corresponding transcript are represented in a given sample, assuming that cDNA fragments sequenced contain sufficient information to identify a transcript. The currently used RNAseq is a SAGE-based technology that allows for high-throughput analysis of transcriptomes. Its decrease in the cost allows routinely used in research, increasing application and fast development.
In RNAseq method, complementary DNAs (cDNAs) generated from the mRNA of interest are directly sequenced using next-generation sequencing (NGS) technologies.
The obtained reads can then be aligned to a reference genome for a whole-genome transcriptome map.
The NGS technology enables massive parallel sequencing at a reasonable cost [132,133]. Current NGS platforms include the Roche 454 Genome Sequencer, Illumina's Genome Analyzer, and Applied Biosystems' SOLiD. These platforms can analyze tens to hundreds of millions of DNA fragments simultaneously, generate giga-bases of sequence information from a single run, and have revolutionized SAGE and cDNA sequencing technology [134].
18
1.5.1 RPKM measure
Measures of RNA abundance are important for many areas of biology and often obtained from RNAseq using Illumina sequence data. These measures need to be normalized to remove technical biases in the sequencing approach, such as the length of the RNA species and the sequencing depth of a sample. RPKM, which is the most frequently used measure of RNAseq data, calculates the number of reads mapped to a particular gene region (rg) and the feature length, flg [135]. The calculation is as RPKMg = rg x 109/flg x R, where R is the total number of reads from the sequencing run of that sample.
19
Chapter 2
YjcC, a c-di-GMP phosphodiesterase protein, regulates the oxidative stress response and virulence of Klebsiella
pneumoniae CG43
20
2.1. Abstract
This study shows that the expression of yjcC, an in vivo expression (IVE) gene, and the stress response regulatory genes soxR, soxS, and rpoS are paraquat inducible in Klebsiella pneumoniae CG43. The deletion of rpoS or soxRS decreased yjcC expression, implying an RpoS- or SoxRS-dependent control. After paraquat or H2O2
treatment, the deletion of yjcC reduced bacterial survival. These effects could be complemented by introducing the ΔyjcC mutant with the YjcC-expression plasmid pJR1. The recombinant protein containing only the YjcC-EAL domain exhibited phosphodiesterase (PDE) activity; overexpression of yjcC has lower levels of cyclic di-GMP. The yjcC deletion mutant also exhibited increased reactive oxygen species (ROS) formation, oxidation damage, and oxidative stress scavenging activity. In addition, the yjcC deletion reduced capsular polysaccharide production in the bacteria, but increased the LD50 in mice, biofilm formation, and type 3 fimbriae major pilin MrkA production. Finally, a comparative transcriptome analysis showed 34 upregulated and 29 downregulated genes with the increased production of YjcC. The activated gene products include glutaredoxin I, thioredoxin, heat shock proteins, chaperone, and MrkHI, and proteins for energy metabolism (transporters, cell surface structure, and transcriptional regulation). In conclusion, the results of this study suggest that YjcC positively regulates the oxidative stress response and mouse
21
virulence but negatively affects the biofilm formation and type 3 fimbriae expression by altering the c-di-GMP levels after receiving oxidative stress signaling inputs.
2.2. Introduction
During infection, pathogens protect themselves from the oxidative burst of phagocytic cells and the challenging oxidative environments within cellular and extracellular compartments. Upon exposure to oxidative stress such as tellurite, paraquat or hydrogen peroxide, E. coli exhibits an increase in the intracellular ROS and the content of protein carbonyl groups [136-138]. Reactive oxygen species (ROS), including superoxide anion (O2.-), hydrogen peroxide (H2O2), and hydroxyl radicals (HO.), may damage DNA, proteins, and cell membranes and often lead to cell death [117,118]. The bacterial defense mechanism includes sensing, avoiding, and removing the ROS [123]. In general, SodA, SodB, and SodC remove superoxide, whereas catalases (KatE and KatG) and peroxidases (AhpC and GST) remove hydrogen peroxide [139,140]. These various stress defenses are controlled by regulators that respond to superoxide and redox-cycling drugs (e.g., SoxRS), hydrogen peroxide (e.g., OxyR), iron (e.g., Fur), or oxygen tension (e.g., FNR and ArcAB) [140-143]. Diguanylate cyclases (DGCs) and phosphodiesterases (PDEs) regulate the levels of bacterial second messenger cyclic di-GMP (c-di-GMP) by
22
catalyzing molecular synthesis and hydrolysis, respectively [71,81]. The regulatory roles of c-di-GMP appear in numerous bacteria in various cellular functions, including cell surface remodeling [144], cellulose synthesis [111], virulence [145], motility [107], and biofilm formation [146-148]. E. coli YfgF, which exhibits PDE activity, regulates not only surface cell remodeling but also the oxidative stress response by modulating c-di-GMP levels [11]. The disruption of Salmonella enteric Var.
typhimurium cdgR, which encodes a PDE protein, also decreases bacterial resistance
to hydrogen peroxide and accelerates death by macrophages [149].
Klebsiella pneumoniae pyogenic liver abscess isolates often carry heavy capsular
polysaccharides (CPS) to avoid phagocytosis or death by serum factors [33,150]. This thick and viscous structure also helps regulate the bacterial colonization and biofilm formation at the infection site [151]. Several regulators, such as RcsB, RmpA, RmpA2, KvhR, KvgA, and KvhA, help control the CPS biosynthesis by regulating the cps transcriptions in K. pneumoniae [39,152]. An increase in CPS synthesis protects K. pneumoniae from oxidative stress [32,153,154]. However, whether the modulation of c-di-GMP affects CPS synthesis remains unclear.
The expression of yjcC, an IVE gene isolated from the liver abscess isolate K.
pneumoniae CG43, is inducible in the presence of 10
M paraquat [115]. Sequence
analysis of YjcC shows a signal peptide followed by 2 transmembrane domains and a
23
CSS motif at the N-terminal region, whereas the C-terminal contains a conserved EAL domain of the PDE enzyme [155]. In addition, the encoding gene yjcC is cluster-located with soxRS genes, suggesting that it plays a role in the oxidative stress response. This study investigates whether YjcC plays a role in oxidative stress defenses and if YjcC uses PDE activity to execute its regulation.
2.3. Results
2.3.1The YjcC expression is paraquat inducible, and SoxRS and RpoS dependent
To confirm the previously reported paraquat-induced expression phenotype [115],
the IVE DNA containing the 5’ non-coding region and part of the coding sequence of
yjcC was isolated from K. pneumoniae CG43S3 and cloned in front of the
promoterless lacZ gene of pLacZ15 [152]. The resulting plasmid was called pPyjcC1. The sequence analysis of PyjcC1 shows a conserved Fnr box TGTGA-N6-TCACA [156]
centered approximately 400-bp upstream of the yjcC start codon. This process also generated recombinant plasmids pPyjcC2 and pPyjcC0 carrying truncated forms of PyjcC1. These plasmids respectively removed the putative Fnr box and the small stem-loop sequence of the 33-bp coding region shown in Fig.2.1 (A). As Fig. 2.1 (B) shows, the bacteria containing pPyjcC1 exhibited the highest level of
-galactosidase activity,
whereas CG43S3[pPyjcC2] had the lowest activity. In addition, the activity of PyjcC1, but24
not PyjcC2 nor PyjcC0, increased after adding 10 μM of paraquat to the culture medium.
This paraquat-induced characteristic also appeared when the concentration increased to 30 μM, further enhancing the activity of PyjcC1.
As Fig. 2.1 (C) shows, the addition 30 μM paraquat to the bacterial culture significantly increased the yjcC mRNA level. Compared to the expression of the well-characterized stress response regulators SoxS, SoxR, RpoS, and Fnr, the yjcC gene expression was more responsive to paraquat than to hydrogen peroxide exposure.
This study also investigates whether yjcC is subjected to regulation by SoxRS or RpoS. As Fig.2.1 (D) shows, the deletion of soxR, soxS, or rpoS reduces the yjcC expression, implying that SoxRS and RpoS play a positive role in yjcC expression.
2.3.2 YjcC plays a positive role in the oxidative stress response
Paraquat is a superoxide anion generator. Thus, the paraquat-inducible expression suggests that YjcC plays a role in the oxidative stress response. To investigate this possibility, an yjcC deletion mutant was generated through an allelic exchange strategy. As Fig. 2.2 (A) shows, the yjcC deletion mutant was more sensitive to paraquat and hydrogen peroxide when compared to the wild type bacteria K.
pneumoniae CG43S3. The deletion effect could be complemented by transforming the
yjcC expression plasmid pJR1 into the mutant. However, introducing the mutant pJR2,
25
which expresses the mutant form of YjcC with the conserved E residue of the EAL domain replaced by A or pJR3 (which carries the coding region of the YjcC EAL domain), had no complementation effect. Neither of the two EAL-domain protein encoding plasmids pmrkJ and pfimK, which carry PDE activity, could complement the yjcC deletion effect. These results suggest that the stress response is YjcC dependent
and both the N-terminal signaling receiving region and the EAL domain of YjcC are required and specific for an oxidative stress response.
To determine if the YjcC-EAL domain exhibits PDE activity, the recombinant expression plasmid containing the DNA coding for the EAL domain of YjcC or the AAL coding region of pJR2 was constructed and overexpresed in E. coli, and the recombinant proteins were purified. Figure 2.2 (B) shows that the purified EAL domain protein exhibits PDE activity towards pNpp. This activity is lower than the level of the recombinant MrkJ [157], but considerably higher than the activity of the
recombinant protein AALyjcC. As Fig.2.2 (C) shows, the c-di-GMP level of CG43S3ΔyjcC[pJR1] was significantly lower than those of CG43S3[pRK415], CG43S3ΔyjcC[pJR2], or CG43S3ΔyjcC[pJR3]. This suggests that YjcC in vivo
functions as a PDE enzyme capable of reducing the intracellular c-di-GMP levels. The deletion of yjcC gene from CG43S3 increased the c-di-GMP amounts and the difference between the levels was much more apparent after the bacteria exposure to
26
30 µM paraquat Fig. 2.2(D). This also suggests that YjcC is able to degrade c-di-GMP and the catalytic activity could be enhanced by oxidative stress.
2.3.3 Deletion of yjcC places bacteria in an oxidative stress state
As Figs. 2.3 (A) and (B) show, the deletion of yjcC after treatment of H2O2 or paraquat significantly raised the levels of the fluorescent probe H2DCFDA (used to
monitor the formation of ROS) and carbonyl proteins. The introduction of pJR1 into CG43S3ΔyjcC mutant appeared to reduce the levels of ROS and the carbonyl proteins,
showing that YjcC is involved in the removal of ROS or damaged molecules. Thus, this study also investigates the anti-oxidant activity of YjcC. As Fig. 3C shows, the deletion of yjcC reduced the oxidant scavenging activity, as assessed by the absorbance change at 517 nm for the decolorization degree of the purple color, supporting the possibility that YjcC modulates anti-oxidant activity in a certain manner. Numerous studies have shown that Fur and RpoS affect and regulate numerous SODs and catalases [139,158-161]. Figure 2.3 (D) shows that zymogel analysis and total activity measurement exhibit significant changes in the SOD or catalase activity after the deletion of fur or rpoS. However, the deletion of yjcC has no apparent influence on SOD or catalase activity, suggesting that the YjcC-dependent anti-oxidant enzyme remains to be identified.
27
2.3.4 YjcC plays a regulatory role in the virulence, CPS production, biofilm
formation, and type 3 fimbriae expression
YjcC, previously identified as an IVE gene product, is likely involved in infection [115]. To investigate whether YjcC is a virulence factor for the bacteria to establish infection, a mouse peritonitis model was employed. As Table 1 shows, the LD50 to Balb/c mice increased approximately 10-fold after yjcC deletion; introducing ΔyjcC with pJR1, but not pJR2 or pJR3, could restore the LD50. This indicates that YjcC expression at a certain stage is required for mouse infection. It is interesting to note that the ΔyjcC colony is smaller and less mucoid, as determined by a string test [39], than its parental strain on LB agar plate. Therefore, sedimentation analysis and glucuronic acid content measurement are carried out to determine the CPS production.
As Fig. 2.4 (A) shows the deletion of yjcC reduces CPS production. The CPS
deficient phenotype can be fully complemented with the transformation of pJR1 into the ΔyjcC mutant. However, transforming the mutant with pJR2 or pJR3 partially
restores glucuronic acid production.
The second messenger c-di-GMP plays an important role in bacterial biofilm formation [14-16]. Figure 2.4 (B) shows that the biofilm formation activity of ΔyjcC appears to increase compared to the parental strain, whereas that transformed with pJR1 decreases biofilm formation. This can be attributed to the level changes of
28
c-di-GMP (Fig. 2.2 C), which indicates that the YjcC expression of pJR1 significantly reduces the c-di-GMP level, thus reducing biofilm forming activity. Moreover, type 3 fimbriae is a major determinant of biofilm formation in K. pneumoniae [64].
Therefore, this study also investigates the deletion effect on type 3 fimbriae expression. As Fig. 2.4 (C) shows, the western blot hybridization with anti-MrkA antibody shows that the yjcC deletion also increased the major pilin MrkA production of type 3 fimbriae.
2.3.5 Effects of YjcC overexpression assessed using a transcriptome study
This study uses comparative transcriptome analysis between CG43S3[pRK415] and CG43S3[pJR1] to gain further insights into how YjcC executes its regulation.
Analysis of the genome annotation of liver abscess isolate K. pneumoniae NTHU-K2044 [112] shows that the increased expression of yjcC significantly enhances the expression of 34 genes. As Table 2.2 shows, the YjcC-activated genes can be categorized into 12 functional groups. These include the oxidative stress response genes grxA, ybbN, dinI, priB, and stpA, which are involved in anti-oxidation [162,163] or DNA repair [163], the heat shock chaperone protein encoding genes ibpB, ibpA, htpG, and dnaK, which are generally induced in stress conditions [164],
and the genes coding for chaperone ClpB and PspB to protect protein from
29
aggregation and help maintain proton motive force (PMF) to counteract stress conditions [165,166]. Increasing the expression of yjcC also enhanced the expression of PilZ domain protein MrkH and the LuxR-type transcription factor MrkI.
aggregation and help maintain proton motive force (PMF) to counteract stress conditions [165,166]. Increasing the expression of yjcC also enhanced the expression of PilZ domain protein MrkH and the LuxR-type transcription factor MrkI.