M. tuberculosis Gene Expression during Transition to the "Non-Culturable" State.

We analyzed the gene expression profile under specific conditions during reversible transition of M. tuberculosis cells to the "non-culturable" (NC) state in a prolonged stationary phase. More than 500 genes were differentially regulated, while 238 genes were upregulated over all time points during NC cell formation. Approximately a quarter of these upregulated genes belong to insertion and phage sequences indicating a possible high intensity of genome modification processes taking place under transition to the NC state. Besides the high proportion of hypothetical/conserved hypothetical genes in the cohort of upregulated genes, there was a significant number of genes belonging to intermediary metabolism, respiration, information pathways, cell wall and cell processes, and genes encoding regulatory proteins. We conclude that NC cell formation is an active process involved in the regulation of many genes of different pathways. A more detailed analysis of the experimental data will help to understand the precise molecular mechanisms of dormancy/latency/persistence of M. tuberculosis in the future. The list of upregulated genes obtained in this study includes many genes found to be upregulated in other models of M. tuberculosis persistence. Thirteen upregulated genes, which are common for different models, can be considered as potential targets for the development of new anti-tuberculosis drugs directed mainly against latent tuberculosis.


INTRODuCTION
Mycobacterium tuberculosis -the causative agent of tuberculosis -can persist in the human host for decades after infection. Such a latent M. tuberculosis state is traditionally connected with its transition to the dormant state, accompanied by loss of culturability [1]. this makes it practically impossible to reveal latent infection by traditional biochemical and microbiological means and attempt to cure it by antibiotic therapy. to study latent infection in live organisms, several modifications of the experimental model of dormancy during hypoxia in vitro are used [2,3]. However, none of them imitates such an important state of bacteria as their "non-culturability" in the dormant state. We have established an experimental model where dormant M. tuberculosis cells are "non-culturable" (nc) and can reactivate under special conditions [4].
to reveal the biochemical processes accompanying the transition of bacteria to the nc state and to understand the mechanisms of this phenomenon, we analyzed M. tuberculosis gene expression profile during the formation of nc cells.

METhODS
M. tuberculosis total rnA samples were extracted from cells in the late logarithmic phase (5 days of cultivation) and during the transition of cells to the nc state under incubation in the stationary phase at different time points (21, 30, 41 and 62 days of cultivation) as described previously [5]. cDnA was generated from 1μg rnA using random hexamers and reverse transcriptase (Superscript III, Invitrogen, Karlsruhe, Germany) according to the manufacturer's instructions. reverse transcribed samples were purified with the QIAquick Pcr purification kit (Qiagen, Hilden, Germany) and labeled with cy3-and cy5-uLS according the suppliers' recommendations (Kreatech Diagnostics, Amsterdam, the netherlands). Finally, labeled samples were purified with KreApure spin columns. Microarray experiments were performed as dual-color hybridizations. In order to compensate for the specific effects of the dyes and to ensure statistically relevant data, a color-swap dye-reversal analysis was performed. cy3-labeled cDnA (250ng) corresponding to cells from different time points in the stationary phase was competitively hybridized with the same amount of cy5-labeled cDnA of the control sample as color-swap technical replicates onto self-printed microarrays comprising a collection of 4,325 M. tuberculosis-specific "Array-ready" 70mer DnA oligonucleotide capture probes and 25 control sequences (Operon Biotechnologies, Koeln, Germany) at 42°c for 20 h. Arrays were washed 3 times using a SSc wash protocol followed by scanning at 10 μm (Microarray Scanner BA, Agilent, technologies, Waldbronn, Germany). Image analysis was carried out with Agilent's feature extraction software version (Agilent, technologies, Waldbronn, Germany). the extracted MAGe-ML files were further analyzed with the rosetta resolver Biosoftware, Build 7.1 (rosetta Biosoftware, Seattle, uSA). ratio profiles comprising color-swap hybridizations were combined in an error-weighted fashion to create ratio experiments. Anticorrelation of dye-reversals was determined by the compare function of resolver. next we applied a Student's t-test. Finally, by combining a 1.5-fold change cutoff to ratio experiments and the anticorrelation criterion together with the signatures from the Student's t-test, all valid data points had a P-value < 0.01, rendering the analysis highly robust and reproducible.

RESuLTS AND DISCuSSION
We found earlier that M. tuberculosis cultivation in the modified Sauton medium without K + supplemented by dextrose, BSA, and sodium chloride led to a decrease in colony forming units (cFu) on the solid medium in the stationary phase [4]. After 60 days of cultivation, the cFu count dropped to 10 5 per ml (Fig. 1), which meant a transition of 99.9% of cells to the nc state. During further cultivation of cells, spontaneous recovery of nc cells was observed. It is important that the nc state was reversible, and that cells with a minimum cFu count could be reactivated after regrowing them in fresh medium.
comparison of the gene expression profile at different time points from the stationary to the logarithmic phase (5-day cultivation) revealed a different expression (at least 1.5-fold) for a significant number of genes (566), which corresponds to 14% of the M. tuberculosis genome. Some 238 genes are upregulated and 237 downregulated over all time points during the entire culture period. Table 1 shows the functional category of differentially regulated genes during the transition of cells to the nc state.
Besides the significant amount of conserved hypotheticals/unknown genes, many genes involved in the intermediary metabolism and respiration, virulence, detoxification and adaptation, lipid metabolism, information pathways, cell wall and cell processes were downregulated.
A considerable amount of genes coding hypothetical proteins were also found to be upregulated in the nc state: remarkably, genes encoding insertion sequences and phages represented about a quarter of the genes upregulated in the nc state, whereas their proportion in the genome was smaller -only 3.7%. this fact is a possible illustration of the high intensity of genome modification processes during the transition of cells to the nc state. A significant proportion of upregulated genes belonged to the intermediary metabolism and respiration category, in par-ticular, gcvB and ald, coding, respectively, glycine dehydrogenase and L-alanine dehydrogenase, proteases pepR and clpC2. icl1 -one of the genes coding isocitrate lyase, anaplerotic enzyme, existing in the M. tuberculosis cells in two isoforms icl 1 and icl 2 -was found upregulated. Isocitrate lyase is the key enzyme of the glyoxylate cycle -a metabolic pathway, which is an alternative for the tricarboxylic acid cycle and allows the synthesis of carbohydrates from simple precursors. In particular, it plays an important role in seed germination, where fatty acids are used as the main storage of carbon and energy. the induction of some genes involved in lipid degradation, such as fadD9, fadE24, fadE26, and fatty acid degradation, scoA, is indicative of the active role of the glyoxylate cycle in nc cells already found for the Wayne persistence model [2].   During transition to the nc state, some genes used as markers of stress conditions were induced: the heat-shock protein hsp, the chaperones rv0440 and rv3417с, as well as sigma-factors: sigG -regulating genes which are necessary for survival inside the macrophages and sigB, which can control stationary phase regulons and general resistance to stress. Induction of ccsA, whose product takes part in the cytochrome biosynthesis at the step of heme attachment, and cyp132, coding one of the cytochrome's P450 oxidizing different xenobiotics, could evidently reflect accumulation of toxic components in cultures during transition. enzymes of the non-mevalonate pathway of isoprenoid biosynthesis ispF and ispD were also induced in the nc cells. there are data indicating that some of the metabolites of this pathway can affect the immune response of the host [6]. A number of induced genes are involved in the information pathways and those encoding regulatory proteins; in particular, the transcriptional regulator furA, which acts as a global negative control element, employing Fe 2+ as a cofactor to bind the operator of the repressed genes. It seems to regulate the transcription of katG, which is induced in the nc state. katG encodes a multifunctional enzyme, exhibiting both catalase, a broad-spectrum peroxidase and peroxynitritase activities and is believed to play a role in the intracellular survival of mycobacteria within macrophages, protecting them against the aggressive components produced by phagocytic cells. Some genes taking part in the cell wall and cell processes, in particular the transporters ctpG and ctpC encoding AtPases-transporting metal cations and the transporter rv2688с involved in antibiotic resistance and export of antibiotics across the membrane, are activated.
to identify the genes that were significantly upregulated during transition to the nc state, we used stringent criteria: the expression level during the whole time course in the stationary phase was upregulated at least 3-fold in comparison to the expression in the logarithmic phase. Fifty-one genes met this criterion (Table 2).
Among the genes with a substantially high level of expression, genes encoding insertion sequences and phages -20 genes out of the 51-are prime candidates, while 13 genes encode hypothetical proteins with unknown function. It is remarkable that the significantly upregulated genes belonged to intermediary metabolism and the respiration category. Moreover, these genes mainly encode proteins involved in degradation processes; namely bglS -beta-glycosidase (hydrolyzes the terminal, non-reducing beta-D-glucose residue); pip -proline iminopeptidase (specifically catalyses the removal of n-terminal proline residues from small peptides); clpC2 AtP-dependent protease; and ald -L-alanine dehydrogenase (catalyses alanine hydrolyze -an important constituent of the peptidoglycan layer). In addition, the pdhA coding the alpha subunit of pyruvate dehydrogenase and taking part in the energetic metabolism and catalyzing the conversion of pyruvate to acetyl-coA was highly expressed. Significant upregulation of sthA, a soluble pyridine nucleotide transhydrogenase that catalyses the conversion of nADPH generated by catabolic pathways to nADH, which is oxidized by the respiratory chain for energy generation, is a sign of the prevalence of catabolic reactions in cell metabolism in the nc state.
Analysis of the global gene expression profile has been published for several M. tuberculosis persistence models, in particular the Wayne model of the non-replicating state during hypoxia [5,7,8], the gradual depletion of the carbon source under decreased oxygen tension [9], the adaptation of M. tuberculosis within macrophages [10], and in vivo within artificial granulomas in mice [11]. considering the results of these studies, the gene expression profile in our model of "non-culturability" in the stationary phase has, evidently, some overlaps with the above-mentioned models of persistence ( Table 3).
Little in common was found between the genes induced in our model of "non-culturability" and the Wayne dormancy model during hypoxia (Table 3). the Wayne model is characterized by the induction of genes of the dormancy survival regulon (Dos-regulon), a group of 49 genes under the control of devR which codes the regulatory part of the two-component system. upregulation of the Dos-regulon was found not only for dormant cells under hypoxia in vitro, but also for M. tuberculosis cells within macrophages [10], and in the artificial granulomas in mice [11]. In our model of M. tuberculosis transition to the nc state in the stationary phase, only two genes from Dos-regulon -rv0571c and rv2631 -were found upregulated. Dos-regulon induction was not found in the persisting cells during starvation [12], and only two genes of Dos-regulon were activated during persistence at gradual depletion of the carbon source [9].
A recently published paper [13] demonstrated that the role of Dos-regulon is apparently overestimated not only as a universal regulator of the dormant state of mycobacteria, but also as a general response on hypoxia. Genes of the Dosregulon were shown to be activated only 2 hours after hypoxia. thereafter expression of at least half of these returned to the baseline [13]. the authors observed a significant induction of another 230 genes after further cultivation during hypoxia, and hereafter their expression level was stable. thus, the authors refer to this group of genes as enduring hypoxia response (eHr) genes. considering the gene expression profile for our model of transition to the nc state, we found significant overlap with this group of genes (Table 3), which was rather unexpected because the conditions for nc cell formation developed in our laboratory did not imply any oxygen limitation. Some overlap with eHr [13] was found for the persistence model of gradual depletion of the carbon source [9] and the transcriptional response to multiple stresses [14]. therefore, it is possible to conclude that eHr genes may not only play a role as hypoxia markers, but may also be a general regulon of the dormant state of M. tuberculosis, independent of its induction.
thus, the data presented here indicate that cell transition to dormant state is an active process and that numerous genes are involved in it. the future task is to investigate this process in detail in order to understand the molecular mechanisms in the cells during the transition to the dormant state.
Based on the results of the transcriptome analysis of the nc cells obtained in our model and those obtained in several models of persistence, it is possible to pinpoint some shared genes that are upregulated in these models (Table 4). the genes presented in Table 4 and their products are believed to be important for further study, because some of them could represent new targets for anti-tuberculosis drug candidates directed mainly against latent tuberculosis.