Transcriptional Profile of Mycobacterium tuberculosis Replicating in Type II Alveolar Epithelial Cells

M. tb alters its transcriptional profile in type II AEC

To define the transcriptional profile of M. tb during infection of type II AEC, the post-infection (pi) time point of 72 hr (3 days) was chosen because earlier studies have demonstrated that M. tb is located in endosomes of A549 cells at this time [29]. Moreover, M. tb cultured in A549 cells is cytotoxic to these cells by days 4–5 pi [23]. Time points prior to day 3 would not provide adequate amounts of bacteria required to carry out the microarray experiments described in this study. At 72 hr of replication inside type II AEC, 261 M. tb genes were differentially expressed compared to M. tb grown in 7H9 laboratory broth (Fig 1). Of these, 186 genes were upregulated and 75 downregulated with fold changes ranging from +2 to +10 and -2 to -44 fold, respectively (Fig 1). The degree of correlation amongst replicates (biological and technical) is evident in the heat map (Fig 1), and the effects of dye flip are similar across the three experiments.

Numbers of differentially expressed genes were sorted by functional category as designated in the Tuberculist website (http://tuberculist.epfl.ch) (Table 1). By raw numbers, the most affected categories were “Intermediate metabolism and respiration” (68), “Cell wall and cell processes” (52), and “Conserved hypotheticals” (46). By percentage of representation in the M. tb genome, the most affected functional categories were “Information pathway” (11.2%), “virulence, detoxification, and adaptation” (9.6%), and “lipid metabolism” (8.5%). By number, the most upregulated categories were “Intermediate metabolism and respiration” (59), “Cell wall and cell processes” (38), and “Information pathways” (25) together accounting for two thirds of the upregulated genes, indicating an active metabolic and replicative state. The category “Conserved hypotheticals” solely accounted for more than one third (28) of the downregulated genes. Interestingly, of these 28 downregulated genes, nearly half (13) are upregulated (and the remainder not differentially expressed) in activated macrophages [35]. The functions of these genes are unknown but could represent adaptations that distinguish the type II AEC endosome from the macrophage phagosome. “Regulatory proteins” was one of the least represented categories (2%); however, by definition, changes in expression of individual genes in this category can have profound and widespread downstream effects on the expression of other genes. Three genes encoding “Regulatory proteins” were downregulated by M. tb in this study- Rv3133c (devR), Rv31332c (devS) and Rv0081. DevR (DosR) and DevS (DosS) are the response regulator and histidine kinase, respectively, of a two-component regulatory system associated with M. tb survival during a stress-induced, non-replicative, dormant state [36, 37] and Rv0081, a predicted regulatory protein, is itself a member of the DevR-DevS regulon [38]. In contrast, DevR (DosR), DevS (DosS) and Rv0081 as well as many other genes of the DevR (DosR) regulon are upregulated in activated macrophages [35].

The DevR (DosR) regulon is downregulated in the intra-AEC environment

The DevR (DosR) M. tb transcriptional regulator responds to several different stresses believed to exist within the intra-macrophage and/or intra-granuloma environments, e.g., starvation, hypoxia, NO, oxidative stress [36, 38–43]. This regulon consists of at least 48 known genes whose induction is associated with suppressed replication and inhibited aerobic respiration as M. tb transitions into a persistent state [44]. While the DevR (DosR) response does not drive the pathway to non-replicating “dormancy”, it plays a role in maintaining M. tb viability during the dormant state [45]. Examination of the differential expression of the DevR (DosR) regulon can provide insight into stresses encountered as well as the respiratory and replicative state of M. tb in a given environment at a given time.

Twenty six of the 48 DevR (DosR) regulon genes were downregulated by M. tb during residence in type II AEC; this representation among downregulated genes is statistically significant by Fisher’s Exact test (Table 2); based on fold change these genes were among the most downregulated genes in type II AEC (Table 2 and S1 Table). In contrast, during residence in differentiated human monocytic cell line THP-1 [46] and in IFN-γ activated murine bone marrow derived macrophages [35], 9 and 41 DevR (DosR) regulon genes were upregulated, respectively. These results suggest that the AEC endosome is distinct from the macrophage phagosome (naïve or activated), and may prepare M. tb for an active rather than a persistent state of infection.

Consistent with this supposition is the upregulation of 2/5 genes encoding “resuscitation promoting factors” (RPFs), Rv1009 (RpfB) and Rv2450c (RpfE), during replication in type II AEC (Table 2); their upregulation constitutes statistical enrichment of this category of genes (Table 2). Deletion of rpfB delays M. tb reactivation in mice [47]. Both RpfB and RpfE interact with RipA which is essential for proper M. tb cell division [48]. Based on the in vitro and in vivo phenotypes of sequential rpf deletion mutants and single gene complementation of multi-rpf mutants, it has been suggested that there is a hierarchy of function among RPFs, with RpfB and RpfE as most influential [49, 50]. Interestingly, no rpf genes are differentially expressed by M. tb in macrophages [35, 46].

Indicators of M. tb respiratory state in type II AEC

M. tb is an obligate aerobe, and utilizes an energy efficient aerobic respiratory pathway during acute infection [51]; however, during exposure to nitric oxide (NO), a product of macrophage activation, this aerobic pathway is inhibited and M. tb transitions to a non-replicating state [51]. There are two types of NADH dehydrogenase (DH) in M. tb whose activities contribute to the quinilone pool, NADH DH-1 whose subunits are encoded by genes nuoA-N (Rv3145-Rv3158) and two NADH DH-2 encoded by Rv1854c (ndh) and Rv0392c (ndhA), respectively. NADH DH-1 is utilized during aerobic respiration and the acute stage of infection while NADH DH-2 is utilized during anaerobic electron transfer [51, 52]. When M. tb is exposed to NO, the NADH DH-2 is used preferentially to drive an alternative electron pathway. Not unexpectedly, 7 (nuoA, B, D, G, H, I, and J) of the 14 M. tb NADH DH-1 subunit genes were upregulated in type II AEC while neither NADH DH-2 encoding gene were differentially expressed (Table 3). This pattern of regulation in type II AEC is statistically significant (Table 3). In contrast, in activated macrophages M. tb downregulates 6 NADH DH-1 genes and upregulates the NADH DH-2 gene, ndh [35].

Further downstream in the electron flow pathway, during aerobic respiration, cytochrome c reductase (encoded by genes qrcCAB) and aa3 cytochrome c oxidase (encoded by genes ctaCDE) activities lead to transfer of electrons to oxygen as the final receptor. In contrast, the alternative electron receptor pathway under NO stimulation utilizes nitrate reductase (encoded by genes narGHJI) to transfer electrons to nitrate as the final receptor. Consistent with the preferential use of NADH DH-1 in type II AEC, M. tb also upregulates qcrA and qcrB of the cytochrome c reductase and ctaC of the cytochrome c oxidase and downregulates narH and narI of the nitrate reductase as well as narK2 involved in nitrate transport (Table 3). In activated macrophages, qcrC and qcrA are downregulated as well as ctaC and ctaE while narK2 is upregulated [35]. Shi et al., analysed the M. tb respiratory transcriptome changes that occur over time in the aerosol-infected mouse lung and describe a model of three respiratory states- an acute replicating phase which is evident during the first couple weeks of infection followed by a transitional state lasting through day 50 characterized by inhibition of DH-1 and cytochrome c oxidase, accumulation of nitrate, upregulation of DevR (DosR) and subsequent upregulation of narK2, after which M. tb enters a chronic non-repicating phase [51]. Interestingly, the respiratory transcriptome of M. tb in type II AEC coincides with the acute replicating phase described by Shi L et al whereas the transcriptome of M. tb in activated macrophages [35] overlaps with the transitional respiratory state.

Transcriptional Evidence for Active Replication of M. tb in Type II AEC

It is well documented that M. tb invade and replicate in the type II AEC A549 cells [24, 29, 30, 53], and as described above the respiratory transcriptome of M. tb in these cells correlates with the early acute replicating phase in vivo. To further verify a state of active replication, other transcriptional “markers” of replication were evaluated, e.g., indicators of protein synthesis, cell wall synthesis, and a requirement for energy.

The 30S and 50S ribosomal subunits are components of the prokaryotic 70S ribosome which functions as translational machinery in the process of protein synthesis. Of the 23 genes encoding 30S ribosomal proteins, 5 (Rv0682-rpsL, Rv2412-rpsT, Rv3442c-rpsI, Rv3459c-rpsK, Rv3460c-rpsM) were upregulated in M. tb replicating in type II AEC (Table 4). In contrast, 13 genes are downregulated in activated macrophages [35]. In addition, 5/36 genes encoding 50S ribosomal proteins (Rv0714-rplN, Rv0716-rplE, Rv1298-rpmE, Rv2441c-rpmA, Rv3456c-rplQ) were upregulated by M. tb in type II AEC (Table 4) while 16/36 are downregulated by M. tb in activated macrophages [35]. The numbers of genes representing each set of bacterial ribosomal proteins upregulated in type II AEC are statistically enriched (Table 4). Furthermore, the gene encoding the elongation factor, Rv0685 (tuf) which facilitates translation, was upregulated during replication in type II AEC but was unaffected in the activated macrophage environment (S1 Table; [35]). This pattern of M. tb differential gene expression in type II AEC is consistent with a demand for increased protein synthesis.

Increased bacterial replication necessitates increased synthesis of cell-wall components. Mycolic acids are long chain fatty acids and are among the defining characteristics of the mycobacterial cell wall contributing to “cording” and “acid-fastness” [54, 55]. Genes encoding enzymes involved in mycolic acid synthesis are also statistically represented among M. tb genes upregulated in type II AEC (Table 4). Among the 21 genes designated as encoding either mycolic acid synthase or FAS II enzymes, 7 were upregulated during replication in type II AEC (Table 4) compared to 3 downregulated in activated macrophages [35].

Finally, active bacterial replication requires energy. All 8 genes that encode subunits of M. tb ATP synthase were upregulated during infection of type II AEC (Table 4). In stark contrast, during infection of activated macrophages, 7 of the 8 genes are downregulated [35]. Collectively, these transcriptional adaptations to the type II AEC cells are consistent with active M. tb replication.

M. tb Transcriptional Responses to Host Stresses in Type II AEC

M. tb alters its transcriptional profile in the presence of specific stresses in vitro, e.g., hypoxia, starvation, nitric oxide (NO) exposure, acidic pH, etc [39, 56–59]. These stresses and responses are believed to correlate to different environments encountered in vivo, within macrophages and granulomas [42, 43]. The heat shock proteins (HSPs) maintain proper folding of proteins, especially under stresses that induce protein misfolding, e.g., increased temperature. Expression of M. tb HSPs is also induced under stresses like hypoxia and starvation [60]. Of the 10 genes encoding defined HSPs or activators of HSPs in M. tb, 5 were upregulated while replicating in type II AEC giving statistical significance to this group of differentially expressed genes (Table 5). This is similar to the M. tb response observed in activated macrophages [35]. While hypoxia likely contributes to M. tb heat shock response in activated macrophages, the upregulation of HSPs in A549 cells is unlikely to be caused by hypoxia in view of the significant downregulation of hypoxia-induced genes.

The majority of DevR (DosR) regulon genes (39/48) are inducible by hypoxic conditions [56]. Sherman et al. identified 47 M. tb genes that are inducible under hypoxic conditions in vitro [56]. Of these, 15 were downregulated during M. tb replication in type II AEC indicating adequate oxygen availability to M. tb in these cells (Table 5). This is in contrast to the upregulation of 10 hypoxia-inducible genes in the naïve differentiated human monocytic cell line THP-1 [46] and the upregulation of nearly all (40) in activated murine bone marrow-derived macrophages [35]. Another indication of a paucity of oxygen availability is the universal stress response. Nine genes of the M. tb genome encode “Universal Stress Proteins” (USPs). These are expressed when oxygen deprivation leads to inhibition of growth [61]. Five of these genes (Rv1996, Rv2028c, Rv2623, Rv2005c, and Rv2026c) were downregulated by M. tb in type II AEC (Table 5). Again, this is in striking contrast to M. tb in activated macrophages where M. tb upregulates 7 USP genes [35]. Genes defined as hypoxia-induced as well as genes encoding USPs exhibit statistical representation among downregulated genes in this study (Table 5). These results suggest that, unlike the intra-phagosomal compartment of activated macrophages, the type II AEC M. tb-containing endosome is not hypoxic and therefore, hypoxia is not a likely cause for the upregulation of heat shock response genes described earlier. Possibly, the increase in protein synthesis in A549 cells may necessitate an increase in protein folding chaperone activity.

Another potential stress encountered within the bacteria-containing vacuole is low pH as the vacuole matures and acidifies. M. tb has mechanisms to inhibit the maturation of the macrophage phagosome by reducing acidification and preventing lysosome fusion; however, activation of the macrophages with IFN-γ allows the macrophage to overcome this M. tb mechanism and acidification can proceed [62]. Phagosomal fusion with lysosomes leads to a pH as low as 4.5. Fisher et al. have identified 23 M. tb genes that are induced by acid shock, i.e., 15 minute exposure pH 5.5 versus pH 6.9 in 7H9 broth [58]. Four of these genes were upregulated by M. tb giving statistical significance to this response by the bacteria in type II AEC (Table 5). At 72 hr, M. tb reside in endocytic vacuoles in A549 [29]. The pH of A549 endosomal compartments has been shown to drop to below 5.3 [63], and the M. tb-containing compartment in A549 progresses to the late endosomal stage (majority co-localization with Rab7 by 72 hr) [64]. Thus, some overlap between the genes upregulated during acid-shock and in M. tb residing in type II AEC is expected. During M. tb infection of macrophages 8/23 acid shock induced genes are upregulated [35, 46].

Nutrient Status of M. tb in Type II AEC

The M. tb-containing macrophage phagosome is known to be nutrient and iron limiting [3, 35]; however, the relative access to these essentials during infection of type II AEC is not defined. None of the genes induced by starvation in vitro were upregulated and four (Rv3288c, Rv3287c-rsbW, Rv2557, Rv3288c) were downregulated suggesting that M. tb experiences type II AEC as a nutrient rich environment [39] (S1 Table). Interestingly, many of the M. tb gene categories described above as upregulated in A549 cells (e.g., genes encoding ATP synthase, NADH DH1, and ribosomal proteins) are downregulated during starvation [39] indicating the type II AEC endosome is a generally nutrient-rich site for M. tb.

To assess the availability of iron in type II AEC, the expression of three groups of M. tb genes were examined. Genes involved in the synthesis of siderophores (exochelins, mycobactins), high-affinity iron chelators which capture iron sequestered by host iron-chelating molecules [65], are expected to be upregulated in an iron-limiting environment. Genes within the ESX-3 locus, required for the bacteria to then obtain the iron from the siderophores [66], are also expected to be upregulated in low iron conditions. In an environment where iron is in excess of need, the gene encoding iron storage protein, BfrB, is expected to be upregulated. The pattern of expression of these three groups of genes in M. tb in type II AEC is consistent with a moderately iron-limited environment. Two of the 14 siderophore biosynthesis genes and 6 of the 11 ESX-3 locus genes were upregulated while bfrB was downregulated (Table 6). While the upregulation of 2 siderophore biosynthesis genes does not generate a statistically significant representation, genes of the ESX-3 locus are statistically represented (Table 6). In comparison, M. tb residing in activated macrophages upregulate 10 siderophore biosynthesis genes indicating lower availability of free iron in the activated macrophage phagosome compared to the type II AEC endosome.

In response to infection, host cells attempt to limit nutrient availability to the intracellular pathogens [67]. Interestingly, three tryptophan synthesis genes, Rv1613 (trpA), Rv1612 (trpB) and Rv1611 (trpC) were upregulated in M. tb during residence in type II AEC cells (Table 6 and S1 Table) giving statistical significance to the upregulation of this pathway (Table 6). In fact, trpA and trpC were the two most highly upregulated of all M. tb genes in the type II AEC environment with 10.1 and 9.5 fold changes, respectively. Indoleamine 2,3-dioxygenase (IDO-1) is a host enzyme that breaks down tryptophan and plays a role in killing of intracellular pathogens [68], and IFN-γ-mediated IDO-1 activity in A549 has been associated with such antimicrobial effects [69]. While IDO-1 is induced in M. tb-infected human BAL macrophages [70], the effect of M. tb infection on IDO-1 production in type II AEC is not known. If IDO-1 is also induced in A549 cells during M. tb infection, the ability of M. tb to synthesize its own tryptophan may contribute to overcoming the tryptophan limitation in this environment. Interestingly, M. tb tryptophan auxotrophs are avirulent in mice [71]. These genes are not differentially expressed during infection of naïve or activated macrophages [35, 46].

Genes Encoding ESAT-6 and ESAT-6-like Proteins are Upregulated During Infection of Type II AEC

While virulent strains of M. tb are cytotoxic to A549 monolayers and exhibit cell-cell spreading in A549 monolayers, M. bovis BCG is attenuated for both these phenotypes [26, 28]. esat-6 transcription was upregulated ~4 fold in M. tb replicating in type II AEC (S1 Table). The RD1 genomic region which encompasses esat-6, and much of the ESX-1 locus which encodes the type VII secretion system required for secretion of ESAT-6 [72, 73] is missing in all BCG strains. M. bovis BCG, M. tb ΔRD1 and M. tb Δesat-6 mutants are attenuated for dissemination from the lungs of aerosol-infected animals [74–76]. ESAT-6 has been shown to be a laminin-binding adhesin as well as a lysin that binds to A549 cells and causes their lysis [32]. No differential regulation of esat-6 transcripts is observed in M. tb during replication in naïve macrophages, and esat-6 is downregulated in activated macrophages [35, 46]. Together these results suggest that invasion and replication in AEC enables M. tb to enhance its disseminative potential.

ESAT-6 belongs to a family of 22 additional secreted proteins designated “ESAT-6-like” based on their small size (~100 amino acids), a conserved helix-hairpin-helix structure and a central WXG motif [77]. Genes for 5 ESAT-6-like proteins, Rv0288 (EsxH), Rv1038c (EsxJ), Rv1197 (EsxK), Rv2347c (EsxP), and Rv3620c (EsxW), were also upregulated during M. tb replication in type II AEC (Table 7), making esat-6-like genes statistically represented among upregulated M. tb genes (Table 7). Of these, esxH was the most highly upregulated (>6 fold) (Table 7 and S1 Table). Like esat-6, this gene is not upregulated in macrophages [35, 46]. EsxH is encoded in the ESX-3 locus, is specific to the M. tb complex [78], and is expressed in virulent M. tb H37Rv but not in the avirulent H37Ra strain [79]. Importantly, EsxH is strongly recognized by PBMCs from TB patients indicating its expression during human infection [80]. Despite low sequence homology to ESAT-6, the structure of EsxH in complex with its binding partner EsxG is highly similar to the structure of ESAT-6/CFP10 complex [81, 82]. The EsxH/EsxG complex contributes to inhibition of phagosome maturation in macrophages [83]; the function of EsxH during infection of type II AEC is not known.

The other 4 upregulated ESAT-6-like family genes, esxJ, esxK, esxP, and esxW, lie outside of any ESX loci (Table 7). Interestingly, although these 4 proteins also have little sequence similarity to ESAT-6, they have ~98% identity amongst themselves and belong to a small sub-family of the ESAT-6-like family characterized by a C-terminal “QILSS” motif [84]. The gene products also belong to the “Mtb9.9 family” of secreted T cell antigens [80, 85, 86]. In the avirulent M. tb H37Ra, the four corresponding genes have multiple mutations including stop codons in esxP and esxW [87]. esxP and esxW are located within RD5 and RD8, respectively, regions present in M. tb but missing from most M. bovis BCG strains [84]. Interestingly, esxK and esxP are highly expressed in sputum from patients with active TB [88], and M. tb-infected AEC have been reported to be present in sputum [88]. The structures of these proteins have not been solved; however, in silico modeling of EsxW in complex with Rv3619c (EsxV) shows remarkable similarity to the ESAT-6:CFP-10 complex and the EsxH:EsxG complex [81, 82, 89]. None of these genes (esxJ,K,P,W) are differentially expressed during infection of THP-1 or activated mBMM [35, 46]. These results suggest that EsxJ, EsxK, EsxP and EsxW have as yet unknown function(s), which may be important to M. tb during replication in type II AEC and subsequent dissemination from the lungs.

Upregulation of Genes Encoding Adhesins and Other Potential Host Interaction Proteins

Adhesins are extracellular or surface exposed proteins that promote bacterial interactions with the host via binding to specific extracellular matrix proteins (ECM). ESAT-6 binds AEC and specifically to laminin, an ECM protein found both on AEC membranes and as the predominant component of the basement membrane (BM) underlying AEC [32]. We have postulated that ESAT-6 may contribute to M. tb dissemination from the lung by exposing the BM through lysis of AEC and anchoring M. tb to the BM via its interaction with laminin [32]. Whether EsxH or the other ESAT-6-like proteins described above have similar properties has yet to be investigated. Another M. tb adhesin, Rv1837c (glcB, malate synthase- MS) was also upregulated in M. tb replicating in type II AEC (2.88 fold) (S1 Table). M. tb MS has been shown to bind laminin and fibronectin, and to type II AEC in a laminin-dependent manner [33]. Rv1566c and Rv2190c, also upregulated in this study (5.22 fold and 3.90 fold, respectively) (S1 Table) have domains that belong to the NLPC_p60 family found in bacterial proteins involved in host cell invasion [90–92]. Also upregulated in A549 cells was Rv0129c (4.83 fold) (S1 Table) which encodes, Ag85C, one of three secreted fibronectin-binding and elastin-binding proteins of the immunogenic Ag85 complex [93–95]. Rv3922c, upregulated (2.79 fold) (S1 Table), encodes a protein of unknown function in M. tb; however, due to its high similarity to the Aeromonas hydrophila α-hemolysin, it is described as a possible hemolysin (tuberculist.epfl.ch). Upregulation of Rv1566c, Rv2190c, Rv0129c, and Rv3922c in type II AEC could contribute to the M. tb/host interaction. Interestingly, the recently described M. tb adhesin, Rv0931c (PknD), which is a secreted, laminin-binding protein that contributes to infection of the human brain microvascular endothelial cell (but not macrophages, AEC, or other endothelial cells) and to invasion of the blood-brain barrier by M. tb animal studies [96] was not upregulated by M. tb replicating in type II AEC. The absence of differential expression of pknD in type II AEC, suggests cellular specificity with regards to M. tb adhesin expression.

Verification of Select Genes by quantitative RT-PCR

To verify the differential expression of a few of these genes, unamplified RNA from M. tb H37Rv replicating in A549 (the same conditions as for microarray analysis) was analyzed by qRT-PCR as was RNA extracted from M. tb H37Rv grown logarithmically in laboratory broth (Table 8 and S1 Table). As expected, transcripts of the 23S rRNA (negative control) did not show differential expression by qRT-PCR. Transcripts for esat-6 were upregulated ~3.8 fold by microarray and ~4.8 fold by qRT-PCR. Similarly transcripts for esxH were upregulated ~6.3 fold by microarray and ~10.0 fold by qRT-PCR. Thus, in both assays esxH was more highly upregulated than esat-6. Among genes identified as downregulated by microarray, genes of the DevR (DosR) regulon were statistically enriched (Table 2); many of these genes have been found to be upregulated by M. tb in activated macrophages [35]. Two of these genes, Rv3132c (devS, dosS) and Rv3131, were tested by qRT-PCR for verifying their downregulation during infection of type II AEC. Specifically, Rv3132c (devS, dosS) was found to be down ~7.5 fold by microarray and ~2.4 fold by qRT-PCR. Rv3131, the most downregulated gene (~44 fold) observed by microarray, was found to be tremendously downregulated by qRT-PCR (10 fold greater than by microarray). The fold changes for each of the above described genes are statistically different from the negative control, 23S (Table 8). pknD was not differentially expressed by microarray, and there was no significant difference between the transcripts for PknD and the 23S negative control by qRT-PCR.

A549 culture and infection with M. tb

A549 cells (ATCC; CCL-185) were cultured in 37°C 5% CO₂ in RPMI supplemented with 7.5% FBS, 2 mM L-glutamine, 1% MEM nonessential amino acids in ~5–7 x 225cm² flasks (Corning) per experiment. Once confluent, cells of one flask were detached with RPMI/ 500 mM EDTA for counting. This number was used to determine a bacterial inoculum consistent with an MOI of ~5:1 (5 bacteria/cell). M. tb H37Rv was grown to mid-log phase in Middlebrook 7H9 broth as previously described [97], and single cell suspensions used to infect A549 monolayers in the remaining flasks for 1 hr. The infected cells were washed extensively with warm RPMI to remove extracellular bacteria and maintained in the supplemented RPMI containing 1% FBS, 2 mM L-glutamine and 1% non-essential amino acid for 3 days at 37°C 5% CO₂.

Isolation of M. tb from A549 and extraction of M. tb RNA

At 3 days pi, infected cells were washed twice with warm RPMI then lysed with GTC solution (4 M guanidinium thiocynate, 0.5% sodium N-lauryl sarcosine, 25 mM trisodium citrate, 0.1 M 2-mercaptoethanol, 0.5% Tween 80, pH 7.0) and centrifuged at 5000 g for 20 min at 4°C to pellet bacteria [118]. The pellets were washed in GTC solution and resuspended in TRI reagent containing polyacrylamide carrier (Molecular Research Center, Cincinnati, OH, USA). Total RNA was extracted using the protocol described previously [46]. Reference RNA from M. tb H37Rv grown to log phase in 7H9 broth was extracted similarly. The reference bacteria was cultured as follows. Briefly, 10 ml Middlebrook 7H9 media was inoculated with a 0.5 ml frozen aliquot of bacteria (OD600 = 0.7 at time of storage) yielding a starting culture of OD600 = 0.035. The culture was incubated at 37°C with shaking (110 rpm) for 7–9 days and harvested at an OD600 of 0.7–0.9 (mid to late log phase) [97].

DNA Microarrays

RNA from M. tb cultured in A549 cells for 3 days from two experiments (two biological replicates) was used for the DNA microarray studies with RNA from M. tb grown to log phase in Middlebrook 7H9 broth as the reference. RNA quality and quantity were verified by Agilent Bioanalyzer 2100, and RNA integrity numbers for each experimental RNA sample F1, F2, and 2F, were 8.7, 8.6, and 7.7, respectively, and 9.5 for reference RNA. To ensure adequate amounts of RNA for the microarray study, an RNA amplification strategy was used on similar amounts of all RNA samples from M. tb grown in A549 and Middlebrook 7H9 [97, 119, 120]. This strategy was verified by microarray previously in our laboratory to confirm that the expression profiles of amplified M. tb RNA (aRNA) are similar to those of unamplified RNA [97]. The M. tb microarray chips containing 70-mer oligonucleotides representing all open reading frames annotated in the M. tb H37Rv genome sequence were obtained from the Center for Applied Genomics (Public Health Research Institute; Newark, NJ). Labeled cDNA from aRNA of log phase M. tb grown in 7H9 broth was used as the reference control on each chip. For hybridization, labeled cDNA probes generated from aRNA obtained from M. tb grown in each A549-grown biological replicate was mixed with the labeled reference cDNA probe prior to purification with Microcon YM10 filter (Millipore). The M. tb arrays were hybridized overnight with the mixed labeled cDNA probes in duplicate to include a Cy3/Cy5 dye flip [46]. For one biological sample, four technical replicates were used (including two dye flips) and for the other biological sample, two technical replicates were used (one dye flip). The microarrays were scanned and processed with an Axon 4000B scanner and GenePix Pro 6.1 software, respectively. The chips were normalized by the print-tip Lowess method, and the Cy5/Cy3 intensity ratio was determined for each gene [121]. The intensity ratio data obtained from M. tb microarray chips was used to perform Significance Analysis of Microarrays (SAM) with Multiarray Viewer Software on the TMEV website for determination of differentially expressed genes of M. tb grown in A549 cells compared to 7H9 broth grown M. tb [122]. Among genes identified by SAM, only genes that showed a ≥ 2 fold change at a false discovery rate of <2% were considered significantly differentially expressed. Data were deposited in Gene Expression Omnibus repository (Accession Number: GSE58466). The general functional categorization for M. tb differentially expressed genes was determined as described on the TubercuList website (http://tuberculist.epfl.ch/).

Statistical Analysis

The heat map was generated using TMEV MeV_4_9_0 software. Fisher's Exact Test (Microsoft Fisher's Exact Test Calculator) was applied to the data to determine the hypergeometric probability of representation of specific functional gene categories among differentially expressed genes.

Quantitative RT-PCR

Expression of selected genes was determined in unamplified RNA from M. tb grown in A549 and from M. tb grown logarithmically in 7H9 Middlebrook laboratory broth by qRT-PCR [32]. M. tb gene specific primers were designed using Primer3 software (S2 Table). Briefly, M. tb RNA was subjected to synthesis of first-strand cDNA using Superscript II RNase H^- reverse transcriptase (RT). The real-time PCR was performed using iQ SYBR Green supermix, primers, and cDNA in a MyiQ2 two color Real Time PCR Detection System (Bio-Rad). For each RNA sample, a reaction without RT was performed as a negative control. For quantitation, a standard curve was generated for each gene using a serial dilution of M. tb genomic DNA and respective primers. Copies of 16S rRNA was used to normalize the transcript levels of the respective genes. The ratio of normalized copies of each selected gene from M. tb grown in A549 to normalized copies in 7H9 broth-grown M. tb was calculated to determine the fold change in expression of that gene in A549 compared to reference 7H9 broth. Prior to quantitation, the specificities of PCR products were verified by amplification of each gene target using respective primers with M. tb H37Rv genomic DNA as templates and sequencing of amplified products. Unpaired two-tailed t test was applied to determine statistical differences between fold changes of selected genes and the fold change of 23S (negative control) using GraphPad Prism 6 soft ware.