Mycobacterium tuberculosis WhiB4 regulates oxidative stress response to modulate survival and dissemination in vivo

Host-generated oxidative stress is considered one of the main mechanisms constraining Mycobacterium tuberculosis (Mtb) growth. The redox-sensing mechanisms in Mtb are not completely understood. Here we show that WhiB4 responds to oxygen (O2) and nitric oxide (NO) via its 4Fe-4S cluster and controls the oxidative stress response in Mtb. The WhiB4 mutant (MtbΔwhiB4) displayed an altered redox balance and a reduced membrane potential. Microarray analysis demonstrated that MtbΔwhiB4 overexpresses the antioxidant systems including alkyl hydroperoxidase (ahpC-ahpD) and rubredoxins (rubA-rubB). DNA binding assays showed that WhiB4 [4Fe-4S] cluster is dispensable for DNA binding. However, oxidation of the apo-WhiB4 Cys thiols induced disulphide-linked oligomerization, DNA binding and transcriptional repression, whereas reduction reversed the effect. Furthermore, WhiB4 binds DNA with a preference for GC-rich sequences. Expression analysis showed that oxidative stress repressed whiB4 and induced antioxidants in Mtb, while their hyper-induction was observed in MtbΔwhiB4. MtbΔwhiB4 showed increased resistance to oxidative stress in vitro and enhanced survival inside the macrophages. Lastly, MtbΔwhiB4 displayed hypervirulence in the lungs of guinea pigs, but showed a defect in dissemination to their spleen. These findings suggest that WhiB4 systematically calibrates the activation of oxidative stress response in Mtb to maintain redox balance, and to modulate virulence.

with aliquots of freshly prepared proline NONOate (Cayman Chemicals, Ann Arbor, MI). Half-life of proline NONOate is 1.8 sec at pH 7.4. Aliquots of WhiB4 were placed in an anaerobic cuvette and titrated by injection with aliquots of a 2.5 mM stock solution of proline NONOate. Samples were then transferred to EPR tubes and immediately frozen in liquid nitrogen.

Construction of MtbwhiB4
Complementary oligonucleotides were used to amplify the DNA fragment containing whiB4 from the Mtb genome. An internal fragment of the whiB4 ORF was replaced by a loxP-chloramphenicol:hygromycin-loxP cassette and cloned into pYUB572 (Bardarov et al., 2002) to create pYUBwhiB4. The allelic replacement of whiB4 was carried out as described (Bardarov et al., 2002), and disruption was confirmed by PCR and Southern blot analysis. The disruption eliminated amino acids 33 to 86 of the 118 amino acid WhiB4 protein. To construct the whiB4 complemented strain, the whiB4 ORF, along with its native promoter (~500 bp sequence upstream to ATG) and the ribosomal binding site were cloned into the E.coli-mycobacterial shuttle integrative vector pCV125 (Alland et al., 2000) to generate pCV125:whiB4. The MtbwhiB4 strain was transformed with pCV125:whiB4 and the expression of whiB4 was confirmed in the transformants by qRT-PCR (Table S2).

Microarray hybridization and data analysis
Microarrays used in this study were produced and processed at the Center for Applied Genomics at Public Health Research Institute, New Jersey. These microarrays consist of 4,295 70-mer oligonucleotides representing 3,924 ORFs from Mtb H37Rv (http://www.sanger.ac.uk) and 371 unique ORFs from strain CDC1551 (http://www.tigr.org) that are absent in the Mtb H37Rv strain. For microarray analysis, total RNA was extracted from the three biological replicates of wild type Mtb H37Rv and MtbΔwhiB4 at an OD 600 of 0.4 as described (Kumar et al., 2008). Briefly, 1.5 g of RNA was reverse transcribed with random hexamers (Invitrogen) and the resulting complementary DNA was labeled with Cy3-dUTP or Cy5-dUTP (PerkinElmer), and competitively hybridized to whole genome arrays. Hybridization was performed overnight. After washing the arrays were dried by centrifugation (100 x g, 2 min) and scanned. The detailed labeling and hybridization protocol can be obtained at http://www.cag.icph.org/downloads_page.htm#LabelingProtocols.

Scan Protocol:
The arrays were scanned with a GenePix4200AL scanner (Molecular Devices, Sunnyvale, CA). The images were processed using GenePix 6.1 and the resulting text files were exported to Microsoft Excel.

Data Process:
The chips were normalized by the print-tip Lowess method (Pang et al., 2007).
The data were filtered by removing all the spots that were below the background noise or flagged as 'bad'. Also, spots were flagged if they were not present in two out of three replicates. The images were processed using GenePix pro 6.1 software. The ratio of the mean median intensity of Cy5 over the mean median intensity of Cy3 was determined for each spot and the fold change values were calculated. The data was further processed through a modified t-test (SAM; significance analysis of microarrays) (Tusher et al., 2001) using MEV software (Saeed et al., 2003). We set the false discovery rate to *zero* for obtaining tightest possible data. The microarray data discussed in this manuscript have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO series accession number GSE37840.

qRT-PCR analysis
Mtb cells were grown till an OD 600 of 0.4 and RNA was isolated as described (Kumar et al., 2008). First-strand cDNA synthesis was performed using 500 ng of the total RNA with iScript Select cDNA Synthesis Kit (Bio-Rad) using random oligonucleotide primers. PCR was performed using gene specific primers (Table   S4). Gene expression was analyzed with real-time PCR using iQ TM SYBR Green Supermix (Bio-Rad) and a CFX96 RT-PCR system (Bio-Rad). Data analysis was performed with the CFX Manager TM software (Bio-Rad). PCR efficiencies were normalized to obtain accurate expression levels. For comparison between wt Mtb, MtbΔwhiB4, and the whiB4 complemented strains the induction ratio for each gene was normalized to Mtb 16S rRNA expression.

Estimation of NAD + /NADH and detection of membrane potential
Briefly, Mtb cells were rapidly harvested and resuspended in 0. μl of cycling cocktail and 90 μl of cell extract. The mixture was mixed with 10 μl alcohol dehydrogenase (Sigma-Aldrich; 500 U/ml). The reduction of MTT by alcohol dehydrogenase using NADH was monitored at 565 nm each minute for 10 min. The intensity of the reaction product color is directly proportional to the NAD + /NADH concentrations in the sample. NAD + and NADH standards were included in separate wells to determine concentrations. The data reported is the result of 3 independent experiments assayed in triplicate. Assaying for membrane potential was performed on similarly cultured cells using the BacLight bacterial membrane potential kit (Invitrogen, Life Technologies) as described (Rao et al., 2008). Briefly, cells were diluted to an OD 600 of 0.05 in Dubos medium and 1 ml of cells were stained using 3 M of DiOC 2 for 15 min in the presence or absence of 25 M CCCP. The cells were fixed using 2% paraformaldehyde and analyzed for fluorescein and texas red dye using flow cytometry.

Preparation of redox modified forms of WhiB4
The redox modified forms of holo-WhiB4 for the DNA binding reactions were generated by anaerobically reassembling the Fe-S cluster as described in the earlier section. Holo-WhiB4 was separated from low molecular weight polymers by size exclusion chromatography and DNA binding was performed in the anaerobic glove box as described (Singh et al., 2009). Apo-WhiB4 was generated as previously described (Alam et al., 2007). Reduced apo-WhiB4 was generated by addition of 300 mM DTT. Oxidized apo-WhiB4 was generated by removal of DTT from the apo-WhiB4 by size exclusion chromatography followed by treating the samples either with 50 mM diamide or exposure to air for 2 h.

B) UV-visible spectroscopy or absorption spectroscopy is based on
detecting colors in the ultra-violet spectral region. The Fe-S clusters give strong, distinctive spectra because they often display intense yellowish/brownish color.
This technique provides limited information about the redox state of an Fe-S cluster. UV-vis spectroscopy is most often used in cluster chemistry as a means to determine reaction kinetics, such as the rate of Fe-S cluster reconstitution (observed by the development of yellowish brown color), degradation of Fe-S cluster by O 2 (loss of yellowish brown color), and reaction between NO donor and Fe-S cluster protein (change of color to yellowish green).

SI note 2:
In vitro transcription assays are widely used to understand the mechanisms of gene regulation. Until now, in vitro transcription assays for studying gene regulation in Mtb were generally performed using holo-RNA polymerase (holo-RNAP) prepared by the reconstitution of E. coli core RNAP (i.e., RNAP lacking a sigma factor) with excess of purified sigma factor derived from Mtb (Song et al., 2008) The holo-RNAP containing both E. coli and mycobacterial components may have altered transcription properties, making these assays prone to artifacts. Recently, native holo-RNAP purified from Mycobacterium smegmatis (Msm) was extensively used to perform in vitro transcription assays (Smith et al., 2010). However, since Msm contains 26 putative sigma factors, native holo-RNAP was found to exist as a complex mixture of different holoenzymes harboring distinct sigma factors and showed poor activity and low promoter specificity (China and Nagaraja, 2010). The recently described in vivo reconstitution of holo-RNAP with a specific mycobacterial sigma factor promises to overcome these limitations (China and Nagaraja, 2010). This in vivo expression technology allowed precise assembly of core RNAP with the overexpressed sigma factor inside Msm in stoichiometric amounts. The RNAP- A purified by the in vivo expression technology showed high activity and specificity towards  A -dependent promoters (China and Nagaraja, 2010). Using the highly efficient in vivo reconstituted RNAP- A , we have demonstrated the influence of apo-WhiB4 on transcription from whiB4 and rrnA promoters by performing singleround transcription assays. Our selection of RNAP- A for transcription assays was guided by the fact that another WhiB family member (WhiB3) interacts specifically with the  A (Steyn et al., 2002).  , 1996, , Verma et al., 1994. A transcript of 206 bp is expected upon transcription of the rrnA amplicon, which corroborates to the size observed in figure 6B.
The ANNPP is based on error-back-propagation (EBP) algorithm and has been successfully utilized with high efficiency (~97% prediction capability) to predict promoters and TSPs of several mycobacterial genes (Kalate et al., 2003). The          (Cole et al., 1998) or on the website http://www.tigr.org.   BP-whiB4 primers contain attB recombination sites (indicated by the lower case letters) required for the GATEWAY TM Cloning strategy.