Correspondence: Noemí Yokobori, Instituto de Medicina Experimental (IMEX) – CONICET, Academia Nacional de Medicina, Pacheco de Melo 3081 (C1425ASU), Buenos Aires, Argentina. Tel.: +54 11 4805 5695; fax: +54 11 4803 9475; e-mail: email@example.com
Some multidrug-resistant (MDR) Mycobacterium tuberculosis (Mtb) genotypes are the cause of large outbreaks, including strain M identified in Argentina. In contrast, its kin strain 410 has only caused a single case to date. Cell wall antigens from Mtb were associated with the modulation of macrophage (MΦ) cell death, and the ability to inhibit of MΦ apoptosis is considered a virulence mechanism. In this study, the ability these two clinical isolates with divergent epidemiology to induce MΦ cell death was evaluated using whole inactivated bacteria. We showed that gamma-irradiated (I-) strains induced MΦ necrosis, the strongest inducer being I-410. Cell death biased towards apoptosis with the heat-killed (hk) strains, both hk-MDR strains being poorer inducers of MΦ apoptosis than was H37Rv. These effects were partly due to their ability to induce anti-apoptotic mechanisms which were not related to the lack of tumor necrosis factor alpha induction or a compensatory effect of interleukin-10. The most noticeable difference between strain M and strain 410 was the ability shown by hk-M to interfere with apoptosis induced by hk-H37Rv. Thus, heat-stable and heat-labile antigens from these epidemiologically divergent Mtb strains differ in their ability to manipulate MΦ death.
The reemergence of tuberculosis (TB) is mainly associated with its synergy with the HIV/AIDS pandemic and the emergence of multidrug-resistant (MDR) organisms (World Health Organization, 2010). Due to its clonal replication, Mycobacterium tuberculosis (Mtb) has long been considered a species with low genetic variability, but in the last decades molecular epidemiology has revealed the existence of several geographically distributed families (Hershberg et al., 2008). Moreover, it has been shown that some genotypes are more prone to cause outbreaks and/or acquire drug-resistance, e.g. the Beijing family originating in Asia (Parwati et al., 2010). Although mutations that confer drug resistance can imply a fitness cost, some MDR Mtb genotypes are able to overcome this disadvantage and are as virulent as fully drug-sensitive genotypes (Cohen & Murray, 2004). The virulence of a certain Mtb genotype can be defined as its ability to cause active disease in humans. Thus, the level of clustering, revealed by molecular genotyping techniques, can be regarded as a measurable parameter of virulence (Zhang et al., 1999; Brites & Gagneux, 2012).
In the early 1990s, there was a large outbreak of MDR-TB which started in the Muñiz Hospital in Buenos Aires, Argentina, for which the so-called strain M has been identified as responsible (Ritacco et al., 1997; Palmero et al., 2003). This highly successful strain has a characteristic RFLP pattern and still causes about 25% of all MDR-TB cases in the country nowadays (López et al., 2008; Ritacco et al., 2012). Among several variants of strain M, strain 410 was identified during the early epidemic as the cause of a single MDR-TB case that has remained unique, suggesting that this strain has an impaired ability to cause disease in new hosts. In line with this, it has been suggested that the epidemiological success or failure of a certain genotype is related to the immune response the genotype elicits upon infection (López et al., 2003; Theus et al., 2005).
Mtb has developed numerous mechanisms to evade host immune response, including the inhibition of host cell apoptosis. Apoptosis is a well characterized cell death program that can be initiated by the so-called death receptors, leading to the activation of caspase-8 (extrinsic pathway) or by mitochondrial membrane damage and caspase-9 activation (intrinsic pathway), both cascades converging in the cleavage and activation of the effector caspase-3 (Lee et al., 2009). Apoptosis of infected MΦ serves as a defense mechanism because apoptotic death of MΦ infected with avirulent mycobacteria is paralleled by the activation of microbicidal mechanisms that are yet not defined (Molloy et al., 1994; Lee et al., 2006; Arcila et al., 2007) and acts as a bridge between innate and adaptive immunity due to antigen cross-presentation to T cells (Winau et al., 2006; Alemán et al., 2007; Hinchey et al., 2007). On the other hand, if infected MΦ undergo necrosis, Mtb remains viable (Molloy et al., 1994). In this context, the ability of H37Rv to inhibit MΦ apoptosis has been considered a virulence mechanism (Briken & Miller, 2008; Lee et al., 2009).
The complex cell wall of mycobacteria is particularly rich in lipid and glycolipid pathogen-associated molecular patterns with immunomodulatory properties, which are considered to be determinants of virulence of the pathogenic species (Guenin-Macé et al., 2009). Moreover, the hypervirulent phenotype of selected Beijing strains has been attributed to the presence of a particular phenolic glycolipid (Reed et al., 2004), suggesting that variations in cell wall components are relevant to the epidemiological performance. Furthermore, several non-protein mycobacterial components were associated with the modulation of cell death (Ciaramella et al., 2004; Kaku et al., 2007; Velmurugan et al., 2007), including the mannose-capped lipoarabinomannan (Dao et al., 2004), but their exact relationship with apoptosis and bacterial virulence remains to be established.
In the light of the above, we investigated in the present study whether inactivated whole bacilli of the epidemiologically successful strain M and the non-clustered strain 410 differ in their ability to induce MΦ apoptosis.
Materials and methods
Mtb strains and epidemiological background
Three Mtb strains were used in the study: MDR Mtb strains M and 410, and the laboratory strain H37Rv. Isolate #6548, representative of strain M, had been obtained in 1997 from a patient with AIDS hospitalized at the Muñiz Hospital. The so-called strain 410 is a variant of strain M presenting a single-band difference in the IS6110 RFLP fingerprint pattern (Geffner et al., 2009) and was isolated in 1992 from an HIV− patient with extensive pulmonary TB. The patient remained highly infectious for 7 years and was assisted and hospitalized in several health centers in the outbreak area. To date, strain 410 has not produced any secondary MDR-TB case. Both isolates belong to the Haarlem-2 family of Mtb on the basis of their spoligotype.
The strains belonged to the collection kept at the Reference Laboratory for Mycobacteria at the INEI-ANLIS ‘Carlos G. Malbrán’ in Buenos Aires, Argentina. Laboratory reference strain H37Rv was kindly provided by I.N. de Kantor (former head of TB laboratory, INPPAZ PAHO/WHO). Reculture was kept to a minimum to avoid loss of virulence. Cultures were grown at 37 °C on Lowenstein–Jensen slants until day 15, and thereafter harvested, suspended in saline solution (SS) and disaggregated by vortexing. All three strains were submitted to standard IS6110 DNA fingerprinting (van Embden et al., 1993) and spoligotyping (Kamerbeek et al., 1997). For biosafety reasons, bacterial stocks were inactivated by gamma-irradiation (2.4 Mrads from a 137Cs source; I-Mtb). To determine whether there were differences in heat-stable antigens among the strains, the strains were inactivated by autoclaving at 121 ºC for 20 min prior to use [heat-killed (hk) bacilli]. Both inactivated whole bacterial cell suspensions were diluted with pyrogen-free SS at final concentration of 1 OD600 nm (approximately 108 bacilli mL−1; Geffner et al., 2009), and were aliquoted and kept at −20 °C until use. To separate hk-Mtb strains from the soluble fraction, aliquots were centrifuged twice at 8000 g for 10 min. Pellets were washed and suspended in SS, and adjusted to 1 OD600 nm.
Human macrophage cultures
Human monocyte-derived macrophages (MΦ) were obtained from buffy coats from peripheral blood of 15 healthy volunteers (Servicio de Hemoterapia, Hospital Fernández, Buenos Aires, Argentina). All donors were seronegative for HIV, hepatitis B, syphilis and Chagas disease; purified protein derivative (PPD) skin test status was unknown. The Ethics Committee of the Academia Nacional de Medicina approved all experimental procedures. The mononuclear cell fraction was purified by centrifugation in Ficoll-Hypaque gradient. Monocytes were isolated by plastic adherence and were allowed to differentiate into MΦ for 5 days at 37 °C in 5% CO2 in RPMI 1640 (HyClone; Thermo Scientific) plus 10% fetal calf serum (Natocor, Argentina), 100 U mL−1 of penicillin and 100 μg mL−1 streptomycin in 24-well plates (Corning). Approximately 2 × 105 MΦ per well were recovered. The viability and purity of cells were controlled by Trypan blue exclusion and flow cytometry (FACS) analysis, respectively.
In vitro stimulation of MΦ
MΦ were stimulated with hk- or I-Mtb at several bacteria-to-cell ratios and incubation times depending on the experiment. In some experiments, MΦ were stimulated with hk-Mtb strains combined with the following: recombinant human tumor necrosis factor alpha (rTNF-α; 10 or 50 ng mL−1; Peprotech), 1 μg mL−1 staurosporine (ST; Sigma) to specifically induce the intrinsic apoptotic pathway, and 50 μM ammonium pyrrolidine dithiocarbamate (PDTC; Sigma) to block nuclear factor kappa B (NFκB) activation. Preliminary experiments were performed to determine the optimal concentration of PDTC, assessed as increased MΦ apoptosis induction by TNF-α (Liu et al., 2004).
Flow cytometric analysis
The following anti-human monoclonal antibodies were used: fluorescein isothiocyanate (FITC)-conjugated anti-caspase-3 (BD), FITC-anti-TNF-α (eBioscience), phycoerythrin (PE)-conjugated anti-interleukin (IL)-10 (BD), PE-anti-TNFR1, PE-anti-TNFR2 (Invitrogen), PE-anti-Bcl-2 (Ancell), and their corresponding isotype-matched antibodies. After stimulation, MΦ were detached by gently scrapping the cell culture well after 20 min at 4 ºC. For surface markers, harvested cells were stained for 20 min at 4 ºC, washed, fixed with 0.5% paraformaldehyde and suspended in Isoflow™ (BD) for acquisition. For intracellular cytokine staining, brefeldin A (5 μg mL−1; Sigma) was added to the culture for the final 4 h of stimulation to block secretion; cells were fixed and permeabilized according to the manufacturer's instructions (Perm2; BD). Fluorescence-labeled cells were measured in a FACScan cytometer (BD) and data were analyzed with cellquest (BD) or fcs express software (de novo Software). In all, 20 000 specific events were acquired on the basis of forward and side scatter properties.
Apoptosis and mitochondrial membrane potential loss analysis
Induction of apoptosis was determined by annexin V(AV)/PI staining (Annexin V-FITC Apoptosis Detection kit; Sigma) following the manufacturer's instructions. AV+PI− cells were considered early apoptotic and AV+PI+ and AV−PI+ cells as late apoptotic/necrotic cells. Results were expressed as percentage of positive MΦ. To determine the optimal incubation time, MΦ were stimulated with the inactivated strains for 2, 5 and 18 h. Differences in cell death were maximal at 5 h, thus this time point was chosen for the subsequent experiments (data not shown). For the detection of mitochondrial membrane potential (ΔΨmt) loss, MΦ were stained with DIOC6(3) (Invitrogen; Rojas et al., 2000). Briefly, stimulated MΦ were incubated in a 2-nM DIOC6(3) for 15 min at 37 °C and then washed in phosphate-buffered saline. The sample was analyzed immediately by FACS. MΦ that lost ΔΨmt were defined as those with DIOC6(3)low staining.
Cytokine detection in supernatants by enzyme-linked immunosorbent assay
MΦ were incubated with hk-Mtb strains for 5 h. Culture supernatants were harvested, centrifuged to remove cellular debris, and kept at −80 ºC until assayed. TNF-α was detected by enzyme-linked immunosorbent assay (ELISA, Ready-SET-Go!; eBioscience) following manufacturer's instructions.
Data were expressed as medians and 25th–75th percentiles. The nonparametric Friedman test was used to compare the different treatments within each group, followed by the paired Wilcoxon signed rank test. A value of P < 0.05 was considered significant.
Inactivated MDR clinical isolates and the laboratory strain H37Rv differ in their ability to induce macrophage cell death
To determine whether the selected Mtb strains differ in their ability to induce MΦ cell death, MΦ were incubated at different ratios of gamma irradiated (I-) or hk H37Rv, M and 410 strains and analyzed by AV/PI staining. No significant induction of apoptosis or necrosis was observed in unstimulated MΦ (Fig. 1a). I-Mtb strains predominantly induced MΦ necrosis, whereas hk-Mtb strains mainly induced apoptosis in a dose-dependent fashion (Fig. 1a and b). Necrosis was significantly higher for I-410 strain, particularly at a high MΦ to I-Mtb ratio, and this trend was partially conserved with hk-Mtb (Fig. 1a, lower panels). hk-H37Rv induced a strong apoptotic response at 1 : 10 and most notably at 1 : 50 MΦ : Mtb ratio, compared with hk-MDR strains, allowing the laboratory strain to be discriminated from the clinical isolates. Consistent differences between strains were best observed at a MΦ-to-bacteria ratio of 1 : 20, and this ratio was therefore chosen for all subsequent experiments. Dot plots from a single representative experiment are shown in Fig. 1(b). Next, we examined whether the ability of the hk-strains to induce apoptosis was dependent on heat-stable bacterial cell wall components or on soluble antigens that might have been released to the supernatant during inactivation. As can be observed in Fig. 1(c), only hk-bacterial pellets showed the pro-apoptotic activity. The differential induction of apoptosis by hk-strains was confirmed by intracellular staining of the effector caspase-3. hk-H37Rv induced a significantly higher increase in the percentage of caspase-3+ MΦ than either of the hk-MDR strains, the activation that was induced by strain 410 being even lower than that induced by the outbreak strain M (Fig. 1d).
Disruption of mitochondrial membrane potential by hk-Mtb strains
Because mitochondrial inner membrane damage has been associated with MΦ death induced by virulent strains of Mtb, induction of ΔΨmt loss was evaluated. After 5 h of stimulation all three hk-strains induced low but significant levels of ΔΨmt loss compared with untreated controls, and no differences were observed among them (Fig. 2).
Hk-strain M inhibits hk-H37Rv-induced apoptosis
To establish whether the lower level of apoptosis induced by the hk-MDR strains involved an inhibitory mechanism, MΦ were pre-incubated with hk-strains M or 410 for 1 h and then exposed to hk-H37Rv for the following 4 h. Whereas preincubation of MΦ with strain M inhibited the percentage of AV+PI− MΦ induced by hk-H37Rv (Fig. 3) without a significant increase in necrotic cells (data not shown), hk-410 did not modify the levels of AV+PI− MΦ induced by hk-H37Rv.
The lower induction of apoptosis by hk-MDR strains could not be explained by the lack TNF-α production
To evaluate whether TNF-α was involved in the MΦ apoptosis induced by the hk-strains, intracellular TNF-α expression was determined in MΦ after 5 h of stimulation by FACS. Although 5 h after stimulation the three hk-strains induced TNF-α, its expression was significantly higher in hk-H37Rv- than in hk-M- or 410-treated MΦ (Fig. 4a). These results were confirmed by determining the concentration of TNF-α by ELISA at the same time point: median values (25th–75th percentiles) in C, 1.25 (0.2–3.9) pg mL−1; hk-H37Rv, 44.3 (31.6–51.0); hk-M, 17.3 (7.5–20.9); hk-410, 23.5 (8.7–35.6); statistical significance: hk-H37Rv vs. hk-M or hk-410 (P < 0.05, n = 6). Notably, when MΦ were pretreated with strain M for 1 h and then exposed to H37Rv for the following 4 h of the assay, the hk-M strain did not interfere with H37Rv-induced TNF-α expression, suggesting that different mechanisms are involved in TNF-α production and apoptosis induction by hk-strains (Fig. 4a).
To determine whether the lower production of TNF-α could be involved in the lower levels of MΦ apoptosis observed with hk-MDR strains, MΦ were pulsed for 1 h with hk-H37Rv, M or 410, and rTNF-α was added for the next 4 h. Although the addition of high amounts of rTNF-α (50 ng mL−1) alone induced apoptosis, this cytokine did not modify the percentage of AV+PI− MΦ induced by the hk-strains (Fig. 4b). Furthermore, none of the hk-strains induced differences in the surface expression of TNFR1 or TNFR2 at the same time points (data not shown).
As it has been reported that IL-10 is able to antagonize the TNF-α mediated apoptosis induced by Mtb, the induction of this cytokine was determined. The low induction of apoptosis by the hk-MDR strains was not paralleled by a higher IL-10 expression, as its induction was marginal: Median (25th–75th percentiles) %IL-10+ MΦ in C, 1.4 (0.7–1.7); hk-H37Rv, 4.1 (1.6–6.6); hk-M, 2.4 (1.6–6.6); hk-410, 2 (1.4–3.2); Friedman test (P > 0.05).
Inhibition of staurosporine-induced apoptosis is greater with hk-MDR strains
To evaluate whether the hk-MDR strains were able to interfere with the intrinsic apoptotic pathway, MΦ were pre-incubated for 1 h with a low MΦ to hk-Mtb strains ratio before addition of ST. All three strains reduced the levels of ST-induced %AV+PI− MΦ with this decrease being more pronounced for hk-M and hk-410 (Fig. 5a). Dot plots from a single representative experiment are shown in Fig. 5(b). The lower apoptosis levels observed with MDR strains could not be explained by the induction of the anti-apoptotic molecule Bcl-2 because no differences in its intracellular expression were found in MΦ stimulated with the hk-strains (data not shown).
Hk-MDR strains induce low levels of apoptosis even in the absence of NFκB activity
Having observed that hk-MDR strains differed in their ability to interfere with pro-apoptotic signaling, we evaluated whether NFκB, a transcription factor related to MΦ survival, was involved. Inhibition of NFκB with PDTC did not modify the percentage of AV+PI− in non-stimulated or in hk-H37Rv- or 410-stimulated MΦ, but slightly increased hk-M-induced apoptosis (Fig. 6). Remarkably, the apoptosis levels of neither hk-strain M nor strain 410 reached those of hk-H37Rv after NFκB inhibition.
Herein, we have investigated the ability of antigens from two genetically related MDR isolates of Mtb with different epidemiological characteristics to induce MΦ cell death. Our results showing a predominantly necrotic MΦ death induced by gamma-irradiated Mtb are in accordance with evidence obtained with live Mtb at high multiplicity of infection (Chen et al., 2006; O'Sullivan et al., 2007; Divangahi et al., 2009). However, when hk-strains were used, all the strains mainly induced apoptosis and were the highest levels observed with hk-H37Rv. Thus, our results show that necrosis is induced by heat-labile components of whole I-Mtb strains and that the heat-stable cell wall components shift MΦ cell death towards apoptosis. Taking into account that the mycobacterial cell wall is highly enriched in lipid and glycolipid antigens relevant in the modulation of the immune response (Guenin-Macé et al., 2009) and MΦ cell death, we considered that heat inactivation would unveil differences in these non-protein antigens between the selected Mtb strains. In this line, it has been found that while some cell wall glycolipids can induce apoptosis (Dao et al., 2004), others have an anti-apoptotic effect on MΦ (Rojas et al., 2000; Nuzzo et al., 2002; Torrelles et al., 2009) being their proportion and surface availability (Torrelles et al., 2008) critical for mycobacterial virulence (Briken et al., 2004). We hypothesize that the differences in these and/or other heat-stable cell wall components might have determined the observed differential capacity to modulate MΦ cell death by hk-Mtb strains.
At a first glance, hk-MDR strains were indistinguishable in terms of induction of MΦ apoptosis, but the lower apoptosis induced by hk-M and hk-410 relied on different mechanisms. The most noticeable difference between the clustered and the non-clustered MDR Mtb strains was observed in their ability to interfere with the apoptosis induced by hk-H37Rv. At a high bacteria to MΦ ratio, hk-H37Rv mainly elicited MΦ death independently of ΔΨmt loss, and only strain M was able to block partially this pathway. This characteristic was not linked to a higher induction of Bcl-2, although the participation of other anti-apoptotic members of this molecular family cannot be ruled out (Sly et al., 2003). Virulence of H37Rv has been related to its ability to interfere with TNF-α-induced apoptosis (Balcewicz-Sablinska et al., 1998) and a correlation between lower induction of TNF-α and Mtb epidemiological success has been reported (Manca et al., 1999, Theus et al., 2006, Wong et al., 2007). However, we found no differences in TNF-α induction between the hk-clustered and the non-clustered strains, although this might be a consequence of the inactivation process. Furthermore, even in the absence of NFκB activity, a transcription factor related to resistance to apoptotic cell death (Loeuillet et al., 2006), both hk-M and hk-410 induce a very weak pro-apoptotic signal, which could not be ascribed to the lack of TNF-α production or a reduced expression in the TNFRs. Consistently, pretreatment of MΦ with hk-M inhibited hk-H37Rv-induced apoptosis but did not interfere with TNF-α expression, demonstrating that in our experimental conditions apoptosis was not dependent on this cytokine. Moreover, the low MΦ apoptosis induced by both MDR strains could not be ascribed to a higher IL-10 production either (Rojas et al., 1999). In addition, although all three strains tested were able to interfere with the ST-induced intrinsic apoptotic pathway, the inhibition of this pathway was stronger with hk-M and 410. Thus, the low induction of MΦ apoptosis by these hk-strains could be the result of a balance between a weak pro-apoptotic signal and anti-apoptotic signal (Briken & Miller, 2008).
I-410 was characterized by a stronger ability to induce necrosis, which was partially conserved after heat inactivation with a marginal activation of caspase-3. This suggests that rather than a faster progression to secondary necrosis, induction of non-classical pathways might be involved (Lee et al., 2006). It is widely accepted that Mtb attenuation is associated with a pro-apoptotic phenotype (Sly et al., 2003; Briken & Miller, 2008) and virulence with the induction of necrosis (Keane et al., 2000; Chen et al., 2006), but the results obtained with I-410 did not fit this pattern. However, this association between virulence and cell death induction was found when employing laboratory strains which did not meet the epidemiological definition of virulence adopted to select our MDR clinical isolates (Brites & Gagneux, 2012).
Although we were able to reproduce some features of live Mtb, like the predominance of necrosis induction by I-strains, this pathogen relies on active mechanisms like the ESX-1 secretion system to interfere with the effector functions of MΦ, including interference with MΦ cell death (Abdallah et al., 2011). As the major limitation of the current work is that active mechanisms are abolished in inactivated strains, it is difficult to extrapolate our results to the pathogenic process that would take place in vivo. Further studies using live organisms are required to conclusively link the type of cell death induction with the epidemiological success or failure of these MDR-Mtb strains. However, at least three main conclusions can be drawn from the current study:
induction of MΦ necrosis relies on heat-labile antigens;
both heat-stable and heat-labile antigens of these epidemiologically divergent Mtb strains differ in their ability to manipulate MΦ death;
heat-stable antigens of the highly virulent strain M, but not those of the non-clustered strain 410, interfere with hk-H37Rv-induced MΦ cell death.
As we have previously demonstrated that strain M induced less CD8+ cytotoxic T-cell activity against MΦ than strain 410 (Geffner et al., 2009), it was of interest to discover in the current study that the clustered isolate also enhances another mechanism that preserves MΦ integrity. Although other isolates related to strain M should be tested, the similar effect of hk-M and hk-410 observed in several parameters evaluated in our study could be due to family- or subfamily-specific heat-stable antigens.
This work was supported by grants from ANPCyT (PAE-PICT 2007-2329, PAE-PICT 2007-2323 and PAE-PICT 2007-2328), and Fundación Alberto J. Roemmers. We acknowledge Dr. Oscar Bottasso for helpful discussion. None of the authors has any potential financial conflict of interest related to this manuscript.