Limited role of the Toll-like receptor-2 in resistance to Mycobacterium avium

Abstract

The role of the Toll-like receptor (TLR)-2 in the generation of protective immunity to Mycobacterium avium was evaluated using gene-disrupted mice. TLR-2^−/− mice were more susceptible than wild-type C57Bl/6 mice to M. avium strains that were able to proliferate in vivo before the development of protective immunity and mycobacteriostasis. In contrast, the elimination of non-virulent strains was not affected by the mutation. The generation of interferon-γ (IFN-γ)-producing T cells and the expression of the interleukin-12 p40 gene were reduced in TLR-2-deficient mice as compared to C57Bl/6 mice early during infection with M. avium strain 2447. The generation of protective CD4⁺ T cells was also compromised in the mutated mice as compared with the controls. Our data show that TLR-2 is required for optimal immunity against certain virulent M. avium strains.

Introduction

The initiation of an immune response to a pathogen requires that conserved molecular motifs of the microbe or consequences of infection are recognized by receptors expressed on cells responsible for innate immunity. The latter then influence the emerging adaptive immune response by expressing selected cytokines and accessory molecules.^1,2 Among the receptors used for the early detection of infecting microbes, Toll-like receptors (TLR) are believed to play a major role.³ TLRs belong to a phylogenetically ancient family of receptors⁴ and recognize many different microbial components, ranging from lipopolysaccharide, peptidoglycan, lipopeptides and glycolipids to flagellin or DNA containing unmethylated CpG motifs.⁵ However, each particular TLR seems to recognize a limited range of molecules.^6,7

TLRs signal through the adapter molecule, MyD88, and lead to activation of the transcription factor, nuclear factor (NF)-κB. Additional pathways exist that fine-tune the response to better fit it to the microbe, triggering responses such as activation of the mitogen-activated protein (MAP) kinases cascade, of the double-stranded RNA-binding protein kinase (PKR) and of the activator protein (AP)-1.⁶ Thus, the engagement of distinct TLRs by different microbes may lead to different responses.^6,8–10

TLR-2 has been described as the main receptor for mycobacterial constituents. It recognizes lipoarabinomannan (LAM), its precursor phosphatidylinositol mannoside (PIM), and the 19 000-molecular weight lipoprotein.^8,11–14 In addition, TLR-2 has been shown to be involved in the response to whole mycobacteria^15–17 and to affect the fate of these microbes inside macrophages (Mφs).¹⁸ Furthermore, the expression of TLR-2 itself is induced by mycobacteria.¹⁹ TLR-2 was found to be recruited into the phagosomal envelope in Mφs, therefore playing its role as an early sensor of the invading micro-organisms.²⁰

The in vivo role of TLR-2 in infections has not yet been documented in detail. In one study, 10 out of 45 humans with lepromatous leprosy were shown to have a mutation in the intracellular region of TLR-2; this led to the substitution of arginine, at position 677, by tryptophan.²¹ This mutation was not found among patients with the tuberculoid form of leprosy or among healthy humans. Another study has shown a higher level of expression of TLR-2 and TLR-1 in the lesions of patients with the tuberculoid form of leprosy compared with the lepromatous form of the disease.¹⁷

Mice genetically deficient in TLR-2 have been obtained^22,23 and shown to be more susceptible to the Gram-positive bacterium Staphylococcus aureus²⁴ and to the Gram-negative bacterium Borrelia burgdorferi.²⁵ The consequences of TLR-2 deficiency in the resistance to Mycobacterium tuberculosis have been reported to depend on the dose of mycobacteria used for infection.²⁶ Here we assessed the role of TLR-2 in the development of protective immunity to Mycobacterium avium, an opportunistic pathogen that predominantly infects immunocompromised hosts.

Materials and methods

Animals

TLR-2-deficient mice, on a C57BL/6 background,²² were bred at the IBMC animal facility. C57BL/6 mice were purchased from Harlan (Barcelona, Spain).

Mycobacteria

All strains of M. avium used in this study were clinical isolates. Strains 2447 and 1983, with a smooth transparent (SmT) morphotype, were provided by Dr F. Portaels (Institute of Tropical Medicine, Antwerp, Belgium); both SmT and SmD (opaque) variants of strain 2-151 were provided by Dr J. Belisle (Colorado State University, Fort Collins, CO). All bacteria were grown in Middlebrook 7H9 broth (BD Biosciences, Pals Alto, CA), collected during the exponential growth phase, resuspended in saline with 0·05% Tween-80 and stored at −80° until used.

Mouse infections and quantification of bacterial loads in the infected organs

Mice were infected by injection of 10⁶ colony-forming units (CFU) of mycobacteria into one of the lateral tail veins. At the time-points indicated in the figures and text, four or five mice of each strain were killed to evaluate bacterial loads in their organs. Livers, spleens and lungs were aseptically collected and homogenized in water containing 0·05% Tween-80. The homogenates were serially diluted in water containing 0·05% Tween-80 and plated in Middlebrook 7H10 Agar medium (Difco). The colonies were cultured at 37° until ready to count.

In vitro stimulation of splenic cells and interferon-γ (IFN-γ) measurement

Cells obtained from the spleen of each mouse were washed with Hank's balanced salt solution (HBSS; Gibco Invitrogen, Carlsbad, CA) and the erythrocytes lysed with a haemolytic solution (155 mm NH₄Cl, 10 mm KHCO₃, pH 7·2). Cells were then distributed in 96-well plates, at a concentration of 2·5 × 10⁵ cells/well, and incubated in triplicate in Dulbecco's modified Eagle's medium (DMEM; Gibco), supplemented with 10% fetal bovine serum (FBS; Gibco) and either with no further stimulus or stimulated with mycobacterial antigens (4 µg/ml) or concanavalin A (5 µg/ml; Sigma, St Louis, MO). Supernatants were collected after 3 days of culture and the concentration of IFN-γ present was measured by enzyme-linked immunosorbent assay (ELISA).

Mycobacterial envelope antigens were prepared as described previously.²⁷ Briefly, bacteria were grown in Middlebrook 7H9 medium until late log phase, collected by centrifugation and disrupted by ultrasound. The sonicate was centrifuged to discard intact bacteria and the supernatant was then dialysed against phosphate-buffered saline (PBS). The suspension was ultracentrifuged and the pellet, consisting of the envelope antigens, was resuspended in PBS.

To measure the IFN-γ present in spleen-cell culture supernatants, an ELISA was performed, using R4-6A2 and biotinylated AN18 as the capture and detection antibodies, respectively.

Measurement of interleukin-12 (IL-12) p40 expression by semiquantitative reverse transcription–polymerase chain reaction (RT-PCR)

Total RNA was extracted from a small portion of liver and lung tissue of individual mice, after lysis in guanidinium isothiocyanate buffer. Reverse transcription was performed using a kit from Invitrogen. The message for hypoxanthine phosphoribosyltransferase (HPRT) was amplified using specific primers. All samples were standardized for the same HPRT expression level, and this level was verified to be below saturation point, by comparison with a curve generated by serial dilutions of one of the cDNAs. The same amounts of cDNA were then used to PCR amplify the interleukin IL-12 p40 message, using primers with the following sequences: CGTGCTCATGGCTGGTGCAAA (sense) and CTTCATCTGCAAGTTCTTGGG (antisense). The PCR products of both messages were run in an agarose gel containing ethidium bromide. The gel was photographed and the image analysed to evaluate band intensity, using imagequant software (Amersham Biosceinces, Uppsala, Sweden).

Adoptive transfer of T cells

Donor CD4⁺ T cells were purified from the spleens of donor mice. Either non-infected mice (non-immune cells) or mice that had been infected 1 month previously with 10⁶ CFU of M. avium 2447 SmT (immune cells) were killed, their spleens were aseptically collected and a single-cell suspension was prepared. Red blood cells were lysed as described above. CD4⁺ T cells were purified using antibody-coated magnetic beads (Miltenyi Biotec, Cologne, Germany), using the procedure recommended by the manufacturer. Recipient mice were sublethally gamma-irradiated, with a 500 rads dose, 24 hr before infection with 10⁶ CFU of M. avium 2447 SmT. Two hours after infection, they were injected intravenously with either immune or non-immune CD4⁺ T cells, obtained as described above. Each mouse received the number of cells equivalent to one donor mouse (≈ 7 × 10⁶ non-immune cells or 15 × 10⁶ immune cells). One month after the infection and adoptive transfer process, the animals were killed and the bacterial loads in their livers were evaluated as CFUs (see above). The protection index was calculated by subtracting the mean log₁₀ CFU per organ, obtained from mice that received immune T cells, from the value obtained in mice transfused with non-immune cells of the same background.

Infection and treatment of bone marrow-derived Mφs

Mφs were derived from mouse bone marrow and cultured in 24-well plates, as previously described.²⁸ On day 10 of culture, when the cells had completely differentiated into Mφs, they were infected with M. avium. Approximately 10⁶ CFU of M. avium 2447 SmT were added to each well (≈ 10 bacteria per Mφ), in 0·2 ml of DMEM. Cells were incubated for 4 hr at 37° in a CO₂ atmosphere, washed with warm HBSS to remove non-internalized bacteria and then reincubated in DMEM containing 10% FBS and 10% L929 cell-conditioned medium (LCCM). In some wells, the Mφs were immediately lysed and the number of viable intracellular bacteria counted as CFUs. Other cells were incubated for 7 days to measure intracellular growth of the bacteria. Some Mφs were treated, just after infection with M. avium, with 1 µg/ml M. avium 2447-derived cell envelope antigens (prepared as described above) or they were given daily doses of 100 U/ml IFN-γ and 50 U/ml tumour necrosis factor-α (TNF-α), between day 0 and day 4 postinfection. The difference in terms of log₁₀ CFU per well between day 0 and day 7 was termed ‘log₁₀ increase’. For each condition tested, three culture wells were used. The results presented correspond to the mean and standard deviation (SD) of these three wells.

Results

Role of TLR-2 in the control of in vivo infections with M. avium

In order to directly evaluate the role of TLR-2 during the in vivo infection with M. avium, we infected TLR-2 gene-disrupted mice with different strains of M. avium and compared the growth of the bacteria in their organs with the growth in wild-type animals (Fig. 1). The effect of a lack of TLR-2 was distinct for each strain of M. avium tested. The strains of M. avium (2-151 SmT and 2447 SmT) that are able to grow reached higher bacterial loads in the absence of TLR-2. Strains that have a limited capacity to grow in the mouse, namely 1983 SmT and 2–151 SmD, were eliminated, even in TLR-2-deficient mice, although the clearance was somewhat delayed in the spleens of mice infected with 1983 SmT. In the case of the M. avium strains where the loss of TLR-2 had an impact on resistance to infection, the differences in bacterial loads were seen at late time-points of infection, after the emergence of the adaptive immune response.

Figure 1.

Role of Toll-like receptor-2 (TLR-2) in the control of in vivo infections with Mycobacterium avium. TLR-2-deficient (closed squares) and C57BL/6 wild-type (open circles) mice were infected intravenously with 10⁶ colony-forming units (CFU) of the indicated strains of M. avium. At the indicated time-points, groups of four or five mice were killed and the bacterial loads of their organs were quantified as colony-forming units (CFU). The graphics show the geometric means ± 1 standard deviation (SD) of the log₁₀ CFU per organ of each group. Statistical analysis was performed using the Mann–Whitney U-test. *P < 0·05. A: The number of bacteria is below the detection level. SmD, smooth opaque morphotype; SmT, smooth transparent morphotype.

Effect of the lack of TLR-2 in the production of IFN-γ by splenocytes

IFNγ-producing T cells are a key element in the control of M. avium infections in the mouse.²⁹ To assess how the absence of TLR-2 affected the development of this population, we measured the capacity of total spleen cells, obtained from infected animals, to produce IFN-γ when stimulated with M. avium-derived antigens. As shown in Table 1, the production of IFN-γ on day 15 of infection was lower in TLR-2-deficient mice than in wild-type animals, for all M. avium strains tested. On day 30, the levels of IFN-γ were still lower in the TLR-2-deficient mice infected with the 1983 SmT strain, but differences were not apparent with the other strains.

Table 1.

Interferon-γ (IFN-γ) produced in vitro by splenocytes collected on days 15 and 30 after infection with the indicated Mycobacterium avium strains

			Control	Antigen	Concanavalin A
2–151 SmT	Day 15	Wild type	50·1 ± 11·5	1144·9 ± 531·1	2137·8 ± 624·6
		TLR-2−/−	44·4 ± 13·7	585·4 ± 171·7*	2729·7 ± 782·4
	Day 30	Wild type	93·4 ± 10·1	3044·3 ± 825·7	4411·4 ± 419·5
		TLR-2−/−	81·3 ± 27·5	3889·5 ± 594·6	4429·8 ± 375·7
2447 SmT	Day 15	Wild type	163·5 ± 5·8	3149·8 ± 844·1	6090·0 ± 577·4
		TLR-2−/−	156·0 ± 5·4	377·3 ± 153·8*	4796·5 ± 2005·8
	Day 30	Wild type	180·5 ± 11·7	6095·5 ± 718·9	5417·5 ± 668·1
		TLR-2−/−	156·5 ± 3·7	6235·0 ± 249·8	6725·3 ± 517·4
1983 SmT	Day 15	Wild type	109·4 ± 2·3	2898·5 ± 446·2	3442·3 ± 186·6
		TLR-2−/−	84·3 ± 3·3	1674·1 ± 755·8*	4796·5 ± 945·1
	Day 30	Wild type	140·2 ± 2·4	5615·8 ± 133·3	6510·4 ± 1790·8
		TLR-2−/−	140·2 ± 5·4	458·8 ± 161·2*	5445·2 ± 716·0
2–151 SmD	Day 15	Wild type	78·5 ± 2·2	1693·3 ± 873·0	8096·2 ± 1477·6
		TLR-2−/−	74·9 ± 1·4	845·8 ± 1445·5	1599·9 ± 2385·3
	Day 30	Wild type	97·5 ± 4·0	149·2 ± 42·6	1260·0 ± 1118
		TLR-2−/−	94·5 ± 15·7	108·5 ± 3·1	782·5 ± 1259

Data are expressed in pg/ml.

SmD, smooth opaque morphotype; SmT, smooth transparent morphotype; TLR-2, Toll-like receptor-2.

^*P < 0·05 using the Mann–Whitney U-test.

Decreased IL-12 expression in the absence of TLR-2

The differentiation of IFN-γ-producing T cells during M. avium infection is highly dependent on the early production of IL-12.^30,31 The activation of TLR-2 on the cells of the innate immune system can lead directly to the production of IL-12.1–5 To investigate whether the decrease in production of IFN-γ we had seen was correlated with a low level of early IL-12 production, we measured (by RT–PCR) IL-12 p40 expression in the livers and lungs of mice infected with M. avium 2447 SmT, 15 and 30 days after infection. In spleen, there is a high basal level of IL-12 production by uninfected mice, which hampers the comparison between groups. Figure 2 shows that TLR-2^−/− mice had, in fact, a lower level of expression of IL-12 p40 on day 15 of infection as compared with C57BL/6 animals. The average intensity of the bands from the livers was 3578 ± 1870 for wild-type mice, as compared to 1403 ± 918 for TLR-2^−/− mice. As for the lungs, the values were 1364 ± 382 and 912 ± 206, respectively. Thirty days after infection, however, the deficient IL-12 p40 expression had already been recovered in the TLR-2^−/− mice.

Figure 2.

Decreased expression of interleukin-12 (IL-12) p40 in the absence of Toll-like receptor-2 (TLR-2). RNA was isolated from mouse livers and lungs, 15 and 30 days after infection with Mycobacterium avium 2447, and amplified by reverse transcription–polymerase chain reaction (RT–PCR) using IL-12 p40-specific primers. The products were run in an agarose gel, photographed following incubation with ethidium bromide and analysed using imagequant software.

Role of TLR-2 in the generation of protective immune T cells

To directly assess the role of TLR-2 in the development of a protective CD4⁺ T-cell-mediated immune response, adoptive transfer experiments were performed. To generate the immune cells, groups of TLR-2^−/− and wild-type animals were infected intravenously with 10⁶ CFU of M. avium 2447 SmT. One month later, these animals were killed, their spleens were collected and CD4⁺ T cells were purified using antibody-coated magnetic beads. Recipient wild-type and TLR-2^−/− mice were sublethally irradiated, infected intravenously with 10⁶ CFU of M. avium 2447 SmT the following day and, 2 hr after infection, were injected with the immune CD4⁺ T cells described above. Each mouse received the number of cells equivalent to one donor mouse spleen. One month later, the mice were killed, the bacterial loads of their organs determined and the protection conferred by the immune cells was evaluated. The results are shown in Fig. 3. When wild-type mice were transfused with immune cells generated in normal donors, they had a 0·8 log decrease in the bacterial burden found in the liver, as compared to mice receiving non-immune T cells. By contrast, when wild-type animals received immune cells from TLR-2-deficient mice, they had a log₁₀ protection of only 0·2. Interestingly, when TLR-2-deficient mice were transfused with immune cells from normal mice, they were protected to almost the same extent as wild-type mice.

Figure 3.

Failure to generate protective CD4⁺ T cells in the absence of Toll-like receptor-2 (TLR-2). Sublethally irradiated wild-type or TLR-2-deficient mice were infected with Mycobacterium avium 2447 [with a smooth transparent (SmT) morphotype] and transfused with immune CD4⁺ T cells from either wild-type or TLR-2^−/− donors. Control animals were transfused with non-immune CD4⁺ T cells of the same background. More details of this can be found in the Materials and methods. One month later, the mice were killed and the bacterial loads in their livers were quantified as colony forming units (CFUs). The difference in terms of log₁₀ CFU per organ between mice that received immune cells and mice that received non-immune cells of the same background is considered the Protection Index. The average Protection Index ± 1 standard deviation (SD) of groups of five mice is shown.

Direct role of TLR-2 activation in the intra-Mφ growth of M. avium

Although the results, described immediately above, indicate an impairment of T-cell activation as one of the key factors for the increased susceptibility of TLR-2-deficient mice to infection with M. avium 2447 SmT, we could not exclude the possibility that TLR-2 activation in the infected Mφs had a direct effect on the capacity of those Mφs to control the bacterial growth. To test this hypothesis, we infected either wild-type or TLR-2-deficient bone marrow-derived Mφs with M. avium 2447 SmT and measured the intra-Mφ growth of the bacterium for 7 days. The growth inside TLR-2^−/− Mφs was similar to or slightly higher than that of wild-type Mφs (Fig. 4). Moreover, when we treated infected Mφs with mycobacterial cell envelope antigens, presumably containing TLR-2 agonists,¹⁸ the growth of M. avium was inhibited to a greater extent in wild-type mice as compared to TLR-2-deficient Mφs. When treated with IFN-γ and TNF, Mφs from both strains were equally capable of controlling M. avium growth.

Figure 4.

Inhibition of mycobacterial growth, induced by bacterial envelope components, but not by interferon-γ (IFN-γ) and tumour necrosis factor (TNF), is decreased in the absence of Toll-like receptor-2 (TLR-2). Bone-marrow-derived macrophages (Mφs) were infected with Mycobacterium avium 2447 [with a smooth transparent (SmT) morphotype] and incubated for 7 days to measure the intracellular growth of bacteria. In one set of experiments, Mφs were treated, just after infection, with 1 µg/ml M. avium 2447-derived cell envelope antigens. In the other set of experiments, Mφs were given daily doses of 100 U/ml interferon-γ (IFN-γ) and 50 U/ml tumour necrosis factor (TNF). The difference, in terms of log₁₀ colony-forming units (CFU) per well, between days 0 and 7 was termed ‘log₁₀ increase’. For each condition tested, three culture wells were used. The results presented correspond to the mean and standard deviation of these three wells. Statistical analysis was performed using the Mann–Whitney U-test. *P < 0·05. Results shown are from one representative experiment of at least three.

Discussion

Here we report on the involvement of TLR-2 in the control of infection of mice with certain strains of M. avium, providing in vivo evidence to support the notion that TLR-2 is a pattern-recognition receptor for M. avium.^15,32 Although mice deficient in the TLR-2 gene were as capable of clearing an infection with the low-virulence M. avium strain 2-151 SmD as wild-type mice, they showed delayed clearance of the strain 1983 SmT and impaired control of the strains 2447 SmT and 2-151 SmT. Differences in bacterial loads between TLR-2-deficient and wild-type mice were apparent only at late time-points of the infection, suggesting that either they were the result of an impairment in the development of the adaptive immune response, or that the differences in bacterial loads during the early phase were very small and only reached statistical significance at later time-points. Our data suggest that, in fact, a combination of the two effects can take place. On the one hand, we showed here that TLR-2 was important for the early priming of T cells for the production of IFN-γ. Spleen cells from TLR-2-deficient mice infected with M. avium produced less IFN-γ in response to M. avium antigens than wild-type controls, although those levels became similar at later time-points. Concordantly, mice infected with strain 2447SmT, the one that led to the highest differences in susceptibility, showed a delay in the production of IL-12 p40 in TLR-2^−/− mice as compared to the control animals. Moreover, the ability of immune CD4⁺ T cells from TLR-2-deficient mice to protect against infection was severely hampered. These data show that TLR-2 is required for the optimal generation of IFN-γ-producing protective CD4⁺ T cells. In addition, the fact that immune CD4⁺ T cells from control C57BL/6 mice protected the TLR-2-deficient animals suggested that the defect in the mutant mice was related to the induction of the CD4⁺ T cells, and not to the expression of immunity. However, the data obtained with bone marrow-derived Mφs infected with M. avium and treated with M. avium envelope components, indicate that TLR-2 can also play a role in the direct activation of the infected Mφ, to increase its antimicrobial capacity.

The lack of a detectable effect of the deficiency in TLR-2 in the clearance of non-virulent strains of M. avium might be explained by two types of factors. On the one hand, those bacteria might lack the necessary mechanisms to survive in a mammalian host, even in the absence of effector antimicrobial mechanisms. However, we have shown, for one of these strains –M. avium 1983 – that the elimination is dependent on TNF production and utilization of the type I TNF receptor, TNF-R p55.^33,34 A second and more probable interpretation of the results presented here is that additional receptors recognize components of these strains and convey the necessary signals for the induction of TNF secretion or other effector pathways. The putative alternative receptors could be other TLRs, and future work should address this, namely by studying MyD88-deficient mice, which fail to be activated by many of these receptors. They could also be other types of receptors, such as C-type lectins.³⁵ TLR-2, TLR-4 and perhaps even additional receptors may recognize the tubercle bacilli,^13,36–38 while TLR-1/TLR-2 heterodimers are important for the recognition of M. leprae,¹⁷ and it is not clear if the same applies to M. avium. The limited role of individual TLRs in the control of mycobacterial infection is well illustrated by the somewhat conflicting results, recently described by two different groups, on the infection of TLR-deficient mice with M. tuberculosis. While Abel et al. reported that TLR-4 is required for the control of chronic M. tuberculosis infection,³⁹ Reiling et al. found no differences in susceptibility to infection, using a very similar model of low-dose aerosol infection.²⁶ Interestingly, Reiling et al. reported that TLR-2 deficient mice also did not show an increase in susceptibility to M. tuberculosis when infected with a low dose, but were found to have increased susceptibility when infected with a higher dose. Furthermore, we have observed that the M. bovis strain bacille Calmette–Guérin (BCG) and M. smegmatis are controlled as effectively in TLR-2-deficient as in wild-type animals (M. S. Gomes, unpublished observations).

It is therefore fair to speculate that, for some microbes, innate immunity-sensing mechanisms are redundant. Even for the strains of M. avium where the deficiency in TLR-2 resulted in an exacerbation of mycobacterial growth, the effect was transient and control of infection eventually took place. Thus, other sensing mechanisms clearly operate in these infections. although they appear to be less sensitive than the TLR-2 pathway.

In summary, we report evidence for a limited role of TLR-2 in inducing T helper 1 (Th1) protective immune responses and activating Mφs for anti-mycobacterial activity to a limited set of mycobacterial isolates and suggest the existence of redundant pathways leading to such responses during infection with other mycobacteria.