Mycobacterium tuberculosis exploits the formation of new blood vessels for its dissemination

The mechanisms by which the airborne pathogen Mycobacterium tuberculosis spreads within the lung and leaves its primary niche to colonize other organs, thus inducing extrapulmonary forms of tuberculosis (TB) in humans, remains poorly understood. Herein, we used a transcriptomic approach to investigate the host cell gene expression profile in M. tuberculosis–infected human macrophages (ΜΦ). We identified 33 genes, encoding proteins involved in angiogenesis, for which the expression was significantly modified during infection, and we show that the potent angiogenic factor VEGF is secreted by M. tuberculosis-infected ΜΦ, in an RD1-dependent manner. In vivo these factors promote the formation of blood vessels in murine models of the disease. Inhibiting angiogenesis, via VEGF inactivation, abolished mycobacterial spread from the infection site. In accordance with our in vitro and in vivo results, we show that the level of VEGF in TB patients is elevated and that endothelial progenitor cells are mobilized from the bone marrow. These results strongly strengthen the most recent data suggesting that mycobacteria take advantage of the formation of new blood vessels to disseminate.

M. marinum-infected Μ Φ frequently leave the primary established granulomas, and migrate both hematogenously and through tissues to trigger new granuloma formation 8 . The formation of new blood vessels orchestrated by the Μ Φ seems to be important for the dissemination of M. marinum 10 . Indeed, pharmacological inhibition of the VEGF receptor reduces the infection burden and limits bacterial spreading 10 .
The ability of M. tuberculosis to disseminate via the bloodstream and the lymph is well established. As a consequence, TB has been described in virtually all tissues or organs 11,12 . Extrapulmonary TB represents about 20% of all TB cases in immuno-competent patients and over 50% of the cases in HIV-infected individuals 13,14 . Nearly a quarter of children infected in the first year of life will develop tuberculous meningitis, miliary spread, or bone disease within 2 years 15 . Some observations also support the idea that pulmonary TB may result also from a dissemination of M. tuberculosis. Among the individuals initially infected with M. tuberculosis, only 10% will suffer from the disease with half of these (mainly infants, children, and immuno-compromised individuals) developing active disease (or primary TB) within one year. Adults usually develop a so-called secondary TB (or post-primary TB) as result of a reactivation of a prior infection 16,17 . Secondary TB affects mainly the upper part of the lung 16 , whereas primary TB is likely to develop in any part of the organ. This differential localization may be caused by the early hematogenous spread of the bacteria during primary infection 18 .
Despite significant efforts to understand how M. tuberculosis colonizes its host, the cellular and molecular mechanisms involved in mycobacterial dissemination in humans remain poorly understood [19][20][21] . We report here a study of this important process in the physiopathology of TB by investigating the role of Μ Φ in mycobacterial spread. Using a transcriptomic approach, we identified an angiogenic signature in human monocyte-derived Μ Φ infected with M. tuberculosis. The expression of 33 genes, encoding proteins involved in the formation of new blood vessels, was significantly modulated during infection. Consistent with this finding, we show that the potent angiogenic factor VEGF is secreted by M. tuberculosis-infected Μ Φ , in an RD1-dependent manner. We next assessed the effect of inhibiting angiogenesis in mice engrafted with human M. tuberculosis-infected Μ Φ , and in mice infected with M. tuberculosis via the respiratory route. In both models, mycobacterial spread from the site of infection was strongly impaired by the presence of angiogenesis inhibitors. In accordance with our in vitro and in vivo results, we observed increased angiogenesis in patients with TB. The VEGF concentration in serum of TB patients was elevated, confirming previous studies 5,6,22,23 , and the level of circulating endothelial progenitor cells in blood was also increased when compared to healthy donors. Overall, our data support the idea that mycobacteria exploit Μ Φ for dissemination by inducing the formation of new blood vessels.

Expression of genes involved in angiogenesis is up-regulated in ΜΦ upon M. tuberculosis infection.
Since the early granuloma is a highly vascularized structure, we hypothesized that angiogenesis might be playing a foremost rule here. Consequently, temporal changes to the transcriptome of human monocyte-derived Μ Φ following M. tuberculosis infection were analyzed 24 . mRNAs encoding 31 molecules involved, directly or indirectly, in angiogenesis were strongly up-regulated following M. tuberculosis infection (Fig. 1a). The expression of VEGF-A, a key regulator of endothelial cell sprouting and angiogenesis 25 , was 21-fold higher in infected cells than in uninfected ones, suggesting that M. tuberculosis-infected Μ Φ promote the formation of blood vessels. Also, the expression of the ribonuclease/angiogenin inhibitor (RNH) gene was down-regulated in M. tuberculosis-infected cells. RNH is a powerful inhibitor of angiogenin (ANG) 26 , a major angiogenic factor induced in M. tuberculosis-infected Μ Φ (Fig. 1a). Angiogenesis is a stepwise process comprising remodeling of the extracellular matrix (ECM) 27,28 due to several enzymes, of which metalloproteinases (MMPs) are the most representative ones. Amongst the MMPs implicated in angiogenesis 29,30 , the gene expression of MMP1, 3, and 10 were up-regulated after M. tuberculosis infection, whereas that of TIMP2, an MMP inhibitor, was down-regulated ( Fig. 1a). Chemokine encoding genes were also induced (Fig. 1a). Chemokines promote the growth of blood vessels and are involved in the recruitment of circulating endothelial progenitor cells 25 . To validate our transcriptomic data, a selected panel of genes was examined in more detail. We used ELISA to confirm the upregulation of granulocyte-macrophage colony-stimulating factor (GM-CSF), VEGF-A (hereafter VEGF), oncostatin M (OSM), and interleukin 8 (CXCL8) secretion in M. tuberculosis-infected Μ Φ (Fig. 1b). The expression patterns of the genes for heparin-binding EGF-like growth factor (HB-EGF), bone morphogenetic protein 6 (BMP6), angiopoietin-like 4 (ANGPTL-4), and inhibin beta A (INHBA) were also validated by testing for the proteins by western blotting (data not shown).
VEGF secretion correlates with M. tuberculosis virulence. We next investigated whether VEGF secretion is restricted to M. tuberculosis-infected Μ Φ , or if other mycobacteria species can also induce the release of VEGF. We infected Μ Φ with either Mycobacterium smegmatis, which is non-pathogenic; bacillus Calmette-Guérin (BCG), which is an attenuated Mycobacterium bovis used for vaccinations; or heat-killed and live M. tuberculosis H37Rv. After 48h, Μ Φ infected with M. smegmatis, BCG, or heat inactivated M. tuberculosis H37Rv secreted 8.2, 2.4, and 2.1-fold less VEGF, respectively, than Μ Φ infected with the virulent strain H37Rv (Fig. 2a). Similar results were obtained 18 h post-infection, although the VEGF concentration was lower in cells infected with M. smegmatis or heat-killed M. tuberculosis.
We focused on BCG to understand why it is a poor inducer of VEGF. Attenuation of BCG is mainly due to the loss of the RD1 region, which encodes part of the virulent ESX-1 secretion system 31 . Reintroduction of RD1 into BCG results in a significant increase in virulence 32 . Similarly, the phenotype of H37Rv:Δ RD1 is comparable to that of BCG in human macrophages and mice 33 . ESX-1 represents one of the five type VII secretion systems of M. tuberculosis, and is responsible for the secretion of two major virulence factors also encoded by RD1, the 6-kDa early secreted antigenic target (ESAT-6) and the 10-kDa culture filtrate protein (CFP-10) 34 . Infection of Μ Φ with RD1-complemented BCG resulted in a similar level VEGF secretion as that observed with M. tuberculosis infection (Fig. 2b). Consistent with this observation, Μ Φ infected with RD1 deleted H37Rv secreted 8-fold less VEGF than Μ Φ infected with the parental strain (Fig. 2b). The existence of several M. tuberculosis strains, with Scientific RepoRts | 6:33162 | DOI: 10.1038/srep33162 different virulence potencies, led us to examine the link between VEGF expression by infected Μ Φ and the virulence of M. tuberculosis strains. Infection of Μ Φ with highly-virulent clinical strains of M. tuberculosis, belonging to the Beijing family, induced an even greater level of VEGF secretion than that observed for Μ Φ infected with the two laboratory strains H37Rv or Erdman (Fig. 2c). These results suggest that VEGF-mediated angiogenesis may be exploited by the bacillus to increase its virulence, or foster its persistence inside the host.

Human macrophage-controlled angiogenesis allows Mycobacterium tuberculosis to disseminate.
Recently it has been shown that M. marinum infection triggers angiogenesis in the zebrafish and favors bacterial dissemination 10 . The transcriptomic analysis of M. tuberculosis-infected human Μ Φ suggests that these cells also express factors which stimulate endothelial cells to form new blood vessel networks. To test this possibility, we adapted a well-established model used to evaluate the ability of human tumor cells to form metastases 35 . Briefly, Matrigel plugs containing uninfected or infected human Μ Φ were injected subcutaneously into the abdominal area of SCID mice. After one week, well-defined structures which evoke early granulomas, were visible (Fig. 3a). These structures were sites of bacterial multiplication, and some contained a central core of necrotic cells. Of particular interest for our study, the implants with M. tuberculosis-infected cells were highly vascularized, whereas unstimulated Μ Φ or bacteria alone were unable to trigger angiogenesis (Fig. 3b). Posterior histological examination of the Matrigel plugs containing infected Μ Φ showed several CD31-positive (CD31) blood vessels, both at the implant periphery and within the gel itself (Fig. 3c).
VEGF inhibitors are among the best-known anti-angiogenic agents 25 . Having identified VEGF as a putative target molecule for Μ Φ -induced angiogenesis, we assessed the effect of its inhibition using a humanized monoclonal antibody against VEGF. After adding the anti-VEGF antibody into the Matrigel implants, dissemination of the bacillus to the lungs and spleen was markedly hindered -a decrease of about 96% in the lung and 98% in the spleen, Fig. 4a,b). Bacterial spread to the draining lymph nodes was less affected by the anti-VEGF antibody -decrease of 60% though not statistically significant (Fig. 4c). The anti-VEGF antibody had no effect on the intracellular viability of the bacillus (Fig. 4d).
VEGFR-2 mediates M. tuberculosis spread from the lungs. VEGF binds to three receptors: VEGFR-1 and VEGFR-2 which are highly expressed on vascular endothelial cells, and VEGFR-3, for which the expression is confined to the lymphatic endothelium 36 . To confirm that inhibiting angiogenesis abolishes mycobacterial dissemination and to identify which VEGF receptor is involved; the role of VEGFR-1 and VEGFR-2-mediated signaling was evaluated. Mice infected intranasally with M. tuberculosis were injected with an anti-VEGFR-1 hexapeptide and an anti-VEGFR-2 neutralizing monoclonal antibody, three times a week, for a two-week period. As expected, M. tuberculosis infection induced the formation of new lung blood vessels, and this angiogenic trigger was prevented by injection of the anti-VEGFR-2 antibody (Fig. 5a). Blocking the VEGFR-1 itself had no effect on mycobacterial spread (data not shown). Inhibition of VEGFR-2 mycobacterial growth decreased slightly in the lung (about 57% on average, very close to statistical significance, P = 0.052; Fig. 5b), and strongly impaired mycobacterial spread to the spleen (mean decrease of 86%, Fig. 5c) and liver (mean decrease of 77%, Fig. 5d). It is likely that this effect of the anti-VEGFR-2 antibody is due to its action on endothelial cells. Indeed, repeated VEGFR-2 antibody injection did not affect the percentage or the absolute cell numbers (data not shown) of myeloid cells -namely Μ Φ , dendritic cells and neutrophils -as determined from the expression of CD11b and CD11c (Fig. 5e).

Angiogenesis occurs in vivo in TB patients. To strengthen evidence that TB induces angiogenesis in humans,
we quantified the level of VEGF in TB patients. The presence of VEGF was confirmed by ELISA in the serum of TB patients and it was not detected in the serum of control subjects (Fig. 6a), confirming previous results 5, 6,22,23 . Several studies show that circulating endothelial progenitor cells (EPCs) are incorporated into new and preexisting blood vessels during tumor vascularization 25 ; therefore, we hypothesized that M. tuberculosis-infected Μ Φ facilitate the recruitment and differentiation of EPCs. M. tuberculosis-infected Μ Φ express chemokines that may allow mobilization of the EPCs from the bone marrow to the site of infection and the growth factors necessary for their survival and differentiation (Fig. 1a). To test this possibility, we used flow cytometry to evaluate the number of circulating EPCs in the blood of controls, contacts, and TB patients. As shown in Fig. 6b,c, EPCs defined as CD31 + CD34 + VEGF-R2 + CD45 − were indeed mobilized from the bone marrow in TB patients.

Discussion
Infection with M. tuberculosis leads to the recruitment of mononuclear cells from neighboring blood vessels and to the formation of highly vascularized granulomas 3,4,37 . The mechanisms controlling the formation of new blood vessels and the role of neovascularization during TB remain nevertheless poorly understood. Here, we show in vitro that human Μ Φ rapidly initiate an angiogenic program upon M. tuberculosis infection allowing mycobacterial spread. We identified various genes of which the expression was profoundly modified during infection, and in particular 33 genes encoding growth factors or chemokines promoting recruitment, multiplication, and survival of endothelial cells and EPCs, including VEGF, the most potent angiogenic factor (Fig. 1). In vivo, VEGF was detected in the serum of pulmonary TB patients confirming previous studies 5, 6,22,23 (Fig. 6) and we found for the first time that EPCs are mobilized from the bone marrow (Fig. 6). VEGF is secreted by M. tuberculosis-infected human Μ Φ in an RD1 dependent manner. BCG, which lacks the RD1 virulence locus, and RD1 deleted M. tuberculosis strains are indeed poor inducers of VEGF relative to M. tuberculosis (Fig. 2). Also, M. tuberculosis strains belonging to the Beijing/W lineage induced more VEGF secretion than the laboratory strains H37Rv and Erdman. Interestingly, Beijing strains induce significantly less inflammatory cytokines than H37Rv, are particularly virulent in animal models, and are associated with extrapulmonary disease in humans 38 . Other mycobacterial factors might also be involved in the formation of new blood vessels. Injection of beads coated with trehalose 6,6′ -Dimycolate (TDM) from BCG into mice causes the formation of vascularized granulomas 39,40 . However, in our mouse model, injection of Matrigel mixed with M. tuberculosis alone was insufficient to induce neovascularization of the plugs. Further studies are needed to evaluate if TDM is exported from the mycobacterial phagosome in vivo and whether this can induce blood vessel formation. During cancer, angiogenesis promotes tumor progression and metastasis 25 . Angiogenesis may play a similar role during TB and may be implicated in the extrapulmonary forms of the disease. A recent study pleads in favour of this hypothesis. In the zebrafish model, the formation of new blood vessels facilitates M. marinum dissemination 10 . Although the basic vascular plan of the zebrafish embryos shows strong similarity to that of other vertebrates 41 , anatomical differences with humans (such as the lack of lungs) and the use of M. marinum instead of M. tuberculosis make it difficult to extrapolate these results to human TB. However, in our study, we show that inhibiting angiogenesis also impairs M. tuberculosis dissemination from the primary site of infection, including the lungs. Inhibiting VEGF signaling using blocking antibodies against VEGF or VEGFR-2 impaired dissemination of M. tuberculosis to the lungs, spleen, and liver, and to a lesser extent, to draining lymph nodes (Figs 4 and 5). Colonization of the lymph nodes by the bacillus is likely to involve various mechanisms and to occur via lymph vessels. M. tuberculosis may use migrating dendritic cells or Μ Φ as Trojan horses. Indeed, dendritic cells can transport live M. tuberculosis from the granuloma to the draining lymph nodes 42,43 . Recently, it has been shown that a subset of mycobacteria-infected inflammatory dendritic cells can leave granulomas and form new lesions associated with bacteria-specific T-cells 44 . The formation of new lymph vessels may also be involved. Lymphangiogenesis is regulated by other factors, such as VEGF-C and VEGF-D 45 . VEGF-C expression is up-regulated in mycobacterial granulomas and induces lymph vessel sprouting via VEGFR-3 46 . Lymphangiogenesis thus facilitates the emigration of CD11c + cells out of out of granulomas and into the lymph nodes 46 . Of note, human lymphatic endothelial cells are permissive to M. tuberculosis growth in an RD1-dependent manner 47 . Inhibiting VEGF may thus not be sufficient to prevent M. tuberculosis dissemination to draining lymph nodes.
Whether M. tuberculosis disseminates in humans from the site of infection as free bacteria or within Μ Φ remains to be determined. Using the zebrafish embryo-Mycobacterium marinum model of TB, it has been shown that hematogenous dissemination can occur soon after the establishment of the first lesion, within Μ Φ in an RD1-dependent manner 8 . Alveolar Μ Φ are considered to be resident cells but recent work has shown that these cells can migrate following exposure to Streptococcus pneumonia 48 . Also, extracellular bacteria released from necrotic Μ Φ may directly invade endothelial cells 49,50 and thereby freely enter the blood. Heparin-binding hemagglutinin adhesion (HBHA), a virulence factor of M. tuberculosis, may be involved in these processes. Disruption of the M. tuberculosis hbha gene significantly affects mycobacterial interaction with epithelial cells and impairs extrapulmonary dissemination of M. tuberculosis in the mouse 51 . M. tuberculosis escape from the granuloma is likely to be a critical factor in disease progression. In adults, most forms of TB occur at a site distant from that of the primary infection and years after the first encounter with the bacillus 16,17 . These observations suggest that M. tuberculosis in its primary niche is efficiently controlled by the immune system, but defense mechanisms are compromised once the organism gains access to the blood stream. M. tuberculosis spread following Μ Φ induced angiogenesis may thus favor mycobacterial latency in immunocompetent adults. Hematogenous dissemination might also allow the bacteria to colonize the thymus and to interfere with T cell differentiation, generating T cells tolerant to M. tuberculosis 52 .
Other human pathogens might exploit the blood/lymphatic system and the formation of new vessels. For example, lesions in patients suffering from lepromatous leprosy (which is characterized by formation of highly The data are representative of two independent experiments. Error bars represent the means ± the SEM. *P < 0.05, **P < 0.01, and ***P < 0.001; ns indicates not significant. bacilliferous granulomas) contain numerous CD31-positive micro-vessels not found in patients with paucibacilliary tuberculoid leprosy 53 . Anti-angiogenic agents may be used as adjunctive host-directed therapies in TB as they are for leprosy. VEGF antagonists are already used for the treatment of various diseases and recently, two independent studies have confirmed the feasibility of such an approach. In M. marinum infected-zebrafish, the targeting of VEGFR signaling increases the efficacy of the first-line antitubercular drug rifampicin 10 , and in M. tuberculosis infected-rabbits, treatment with an anti-VEGF antibody normalizes the vasculature of granulomas, leading to the penetration of small molecules 54 . Angiogenesis is not restricted to the pulmonary form of TB. VEGF levels are also high in patients with tuberculous pleural effusion 6,55 or tuberculous meningitis 56,57 . Further studies are needed to evaluate whether individuals who secrete high levels of angiogenic factors upon M. tuberculosis infection are more at risk to develop the extrapulmonary forms of the disease.

Methods
Ethics Statement. All animal experiments described in the present study were conducted at the Institut Pasteur according to European Union guidelines for handling of laboratory animals (http://ec.europa.eu/environment/chemicals/lab_animals/home_en.htm) and were approved by the Institut Pasteur Animal Care and Use Committee and the Direction Sanitaire et Vétérinaire de Paris under permit #A-75-1600. Sera and PBMC were obtained from healthy subjects (controls), contacts, and TB patients. Written informed consent was provided by all participants. The study was approved by the National Ethics Committee of the Ministry of Health in Madagascar (N°033-SANPF). All methods were performed in accordance with the relevant guidelines and regulations.
Bacteria, ΜΦ and infection. Human monocytes were purified from buffy coats and differentiated into Μ Φ according to a previously described procedure 58 . M. tuberculosis H37Rv was grown from a frozen stock to Figure 6. Angiogenesis is an active process in TB patients. (a) ELISA testing for VEGF in serum from healthy blood donors (control, n = 8), contacts (n = 7) and TB patients (n = 17). (b) PBMC isolated from contacts (n = 11) and from TB patients (n = 9) were stained for CD31 and CD45 (upper two panels). Expression of CD34 and VEGFR-2 was analyzed in cells from R2 (lower two panels). (c) Percentage of EPCs for each contact or patient was quantified by flow cytometry. *P < 0.05, **P < 0.01, and ***P < 0.001.
Scientific RepoRts | 6:33162 | DOI: 10.1038/srep33162 mid-log phase in 7H9 medium (Becton Dickinson) supplemented with albumin-dextrose-catalase (ADC, Difco). The virulence of bacteria in the frozen stock was checked by infecting C57BL/6 mice intranasally with 10 3 bacilli: after 21 and 42 days, the bacterial load in the lungs was approximately 10 7 bacteria. Before infection, bacteria were washed three times and resuspended in 1 ml PBS. Clumps were disassociated by 30 passages through a needle, and then allowed to sediment for 5 min. The density of bacteria in the supernatant was verified by measuring the OD600 and aliquot volumes defined to allow 1 bacterium-per-cell infections: cells were infected in six-well plates with each well containing 2 × 10 6 cells in 3 ml medium containing Macrophage Colony Stimulating Factor (R&D Systems). After 4 h of incubation at 37 °C, infected cells were washed twice in RPMI 1640 to remove extracellular bacteria and were incubated in fresh medium. In the experiments comparing different mycobacterial strains, Μ Φ were infected with Mycobacterium smegmatis mc 2 155, BCG Pasteur, or the M. tuberculosis strains H37Rv, Erdman, GC1237, 5750, 5757, 5777, or 5787. The bacterial counts were confirmed by counting the colony-forming units (CFUs). Only experiments with similar numbers of bacteria of the different strains were retained for the analyses.
In vivo Matrigel plug assay. SCID female mice (7-8 weeks old) were purchased from Charles River (France) and maintained according to Institut Pasteur guidelines for laboratory animal husbandry. The in vivo Matrigel plug assay was adapted from a protocol from Passaniti et al. 59 . Briefly, Matrigel plugs containing 2.5 × 10 6 untreated human Μ Φ or human Μ Φ infected with M. tuberculosis for 18 h, were injected subcutaneously into the abdominal area of anesthetized SCID mice. Anti-VEGF antibody (5 mg/kg; Avastin, Genentech) and isotype controls were included in the Matrigel implants as specified. After one week, the mice were sacrificed. The spleens, lungs, and lymph nodes of each animal were aseptically removed, homogenized, and the bacteria counted as described previously 60 . Matrigel plugs were photographed and prepared for histological examination.
Determination of bacterial counts. Μ Φ were lysed in cold distilled water with 0.1% Triton X-100. Bacteria were enumerated as previously described 58 and plated on 7H11. CFUs were scored after three weeks at 37 °C.
Mouse infection. SCID mice were used to avoid any effects on T cells, as many studies have suggested that VEGF suppresses T cell activation 61,62 . Briefly, mice were anesthetized with a cocktail of ketamine (100 mg/kg; Merial) and xylasine (15 mg/kg; Bayer). Mice were infected intranasally with 300 to 500 CFUs of M. tuberculosis. To block VEGF/VEGFR-1 and 2 interactions, mice were i.p. injected three times a week with 200 μ g/mouse anti-VEGFR-1 hexapeptide (Gly-Asn-Gln-Trp-Phe-Ile, Peptron) and with 25 μ g/mouse anti-VEGFR-2 mAb (R&D Systems) as previously described 61 . As negative controls, mice were injected with saline, with reverse peptide (Ile-Phe-Trp-Gln-Asn-Gly, Peptron), or with 25 μ g/mouse isotype control Ab (R&D Systems). After 14 days, mice were killed by CO 2 asphyxiation. Lungs, spleens and livers were harvested, homogenized, and plated on agar for colony counts.
Immunohistochemistry. Matrigel implants were fixed with zinc for 2 days at 4 °C. After fixation, the tissues were dehydrated in a series of ethyl alcohol concentrations and embedded in paraffin at 37 °C. For histological examination, 5 μ m-thick sections were cut and stained with hematoxylin and eosin (H&E) and using the Ziehl-Neelsen method. Vessels were immunostained with an anti-CD31 antibody (Abcam).
Visualization of the vasculature. To visualize blood vessels, SCID mice were first anesthetized and given intracardiac injections of 100 μ g DyLight 594 labeled tomato lectin (Lycopersicon esculentum; Vector Laboratories, Clinisciences). After 5 min, the heart was perfused with PBS-BSA 1% and with 4% paraformaldehyde. Lungs were then frozen and 60 μ m-thick sections cut. The slides were mounted with Fluoromount G (SouthernBiotech), and cells counterstained with DAPI to localize the nucleus. A laser-scanning microscope (Zeiss LSM 700) in the tile scan mode was used to capture a mosaic of images.
Flow cytometry analysis. Cells were isolated from the lungs using a GentleMACS dissociator (Miltenyi Biotec) according to the manufacturer's protocol. Briefly, lungs were placed in GentleMACS C tubes (Miltenyi Biotec), mixed with DNase I and collagenase D, and homogenized. The fragments were incubated at 37 °C for 30 min under slow mixing and homogenized again. The homogenized lung was passed through a 70-μ m pore-size cell strainer, and red blood cells were removed using a red blood cell lysis buffer. Cells were then stained as described previously using the following antibodies: anti-CD11c-PE (eBiosciences) and anti-CD11b-FITC (eBiosciences).
Patients, sample collection and processing. TB was diagnosed by smear observation and/or bacterial culture and/or from clinical symptoms. All individuals included in the study were HIV negative. 75% and 70% of the contacts and the controls, respectively, were PDD+ positive. Chest X-rays were performed only for contacts with a large PPD reaction. Contacts with abnormal chest X-ray results were considered to be TB patients.

Statistics.
Statistical analyses were performed with GraphPad Prism, using the Mann-Whitney test. A P value below 0.05 was considered to be significant.