T cells home to the thymus and control infection

Abstract

The thymus is a target of multiple pathogens. How the immune system responds to thymic infection is largely unknown. Despite being considered an immune privileged organ, we detect a mycobacteria-specific T cell response in the thymus following dissemination of Mycobacterium avium or Mycobacterium tuberculosis. This response includes pro-inflammatory cytokine production by mycobacteria-specific CD4⁺ and CD8⁺ T cells, which stimulates infected cells and controls bacterial growth in the thymus. Importantly, the responding T cells are mature peripheral T cells that recirculate back to the thymus. The recruitment of these cells is associated with an increased expression of T_h1 chemokines and an enrichment of CXCR3⁺ mycobacteria-specific T cells in the thymus. Finally, we demonstrate it is the mature T cells that home to the thymus that most efficiently control mycobacterial infection. Although the presence of mature T cells in the thymus has been recognized for some time, these data are the first to show that T cell recirculation from the periphery to the thymus is a mechanism that allows the immune system to respond to thymic infection. Maintaining a functional thymic environment is essential to maintain T cell differentiation and prevent the emergence of central tolerance to the invading pathogens.

Introduction

The appearance of adaptive immunity in jawed vertebrates is considered a major step in the evolution of the immune system, as it provides the host with a crucial tool to combat microbial pathogens: namely B and T lymphocytes [reviewed in (1, 2)]. B and T lymphocytes express antigen receptors that result from the recombination of germline encoded gene segments, which establishes a diverse repertoire of antigen receptors. This allows B and T cell antigen receptors to recognize antigens the host has not previously encountered. The TCRs, which recognize short linear peptide fragments in the context of antigen presenting molecules (class I or II MHC), allow the host to survey both the intra- and extracellular environment.

The diverse repertoire of TCRs, which are generated by genetic recombination, creates specific challenges for the immune system. First, not all TCRs generated will recognize self-MHC/peptide complexes. Second, some TCRs will recognize self-MHC/peptide complexes with such a high affinity/avidity that will render them potentially autoreactive. To avoid full differentiation of T cells with these unwanted characteristics, thymocytes go through a complex process of selection during their differentiation in the thymus. T cell precursors emerge from the bone marrow and home to the thymus, a primary lymphoid organ that exists for the purpose of supporting T cell differentiation. Two fundamental selection processes occur during T cell differentiation in the thymus – positive and negative selection [reviewed in (3)]. The outcome of thymic selection depends on the organ microenvironment including which antigens are present within the thymus and the cytokine milieu surrounding the differentiating cells. Positive selection creates a pool of thymocytes that bind self-MHC/peptide complexes while T cells expressing TCRs that fail to bind these complexes undergo cell death by apoptosis. T cells expressing TCRs that bind self-MHC/peptide complexes with high affinity/avidity are also eliminated through negative selection. This process of central tolerance is essential to generate a repertoire of T cells that is self-restricted and self-tolerant, and which limits the development of potentially autoreactive T cells. Thus, proper thymic activity is crucial for the development and maintenance of a repertoire of functional T cells. Infants born with genetic mutations that abrogate T cell differentiation are profoundly susceptible to infection, which confirms the essentiality of T cells for cell-mediated immunity (4).

Thymic activity decreases naturally with age, although the adult thymus still supports T cell differentiation and T cells continue to be exported into the peripheral T cell pool (5, 6). Coincident with T cell differentiation and thymic activity, the developing immune system is assaulted by various microbial pathogens. In fact, thymic function is affected during infection by several pathogens, including species of bacteria, virus, fungi and parasites (7, 8). Both in humans and animal models the thymus is itself a target of infection (7–19). Infection of the thymus by mycobacterial species has been reported both in humans (16) and in animal models (17–19). The incidence of thymic infection during active tuberculosis is unknown although there are several case reports in the clinical literature (16). However, it is relevant that vertical transmission of tuberculosis (mother to child) frequently occurs during early childhood, at the height of thymic activity (20).

Although the thymus has been formerly considered an immune privileged site (21), this idea is being reconsidered. In fact, it is clear that protection of the thymus from infection, particularly during childhood, a time of diverse and recurrent infection, should be a function of the immune system. Since negative selection of T cells depends on the antigens encountered in the thymus during differentiation, infection of this organ could theoretically lead to the development of immune tolerance to pathogens. How the immune system prevents, or combats thymic infection, to maintain the organ's integrity and function, and keep it free from microbial antigens is still unclear. Interestingly, the thymus does contain several populations of mature T cells. These include innate lymphocytes such as iNKT cells (22), γδ T cells (23), and MR1-restricted MAIT cells (24), all of which have been implicated in host defense against infection. In addition, there are recirculating conventional CD4⁺ and CD8⁺ T cells (25). Why mature T cells should recirculate back to the thymus is unknown, although others have hypothesized that these T cells play a role in surveying the thymus for infection [reviewed in (25–27)].

Despite the introduction of BCG vaccination a century ago and the development of multiple pharmacological drugs that are active against mycobacteria, these bacteria are still one of the most prevalent infectious agents worldwide (28). Among these, Mycobacterium tuberculosis alone is estimated to cause approximately 8.8 million new infections and 1.4 million deaths per year (28). In addition to M. tuberculosis, other members of the Mycobacterium genus including Mycobacterium marinum, Mycobacterium ulcerans, Mycobacterium leprae and Mycobacterium avium cause disease in humans (29). Some of these are opportunistic infections and mainly affect immunocompromised individuals (30). In fact, the spread of HIV dramatically increased the prevalence of active mycobacterial infection, which is the main cause of death in patients with AIDS (31).

We previously showed that experimental infection with M. tuberculosis and M. avium leads to the establishment of thymic infection in chronically infected mice (18, 19). Now we report that infection of the thymus is followed by the establishment of protective immunity within the thymus, characterized by the appearance of both CD4⁺ and CD8⁺ T cells specific for mycobacterial antigens. These antigen-specific T cells do not originate from the pool of differentiating T cells in the thymus, but are instead T cells that recirculate from peripheral organs back to the infected thymus to control infection. Their recruitment to the thymus correlates with the expression of the chemokine receptor CXCR3 and the production of CXCL9 and CXCL10 by the infected thymus. These data are the first to show that T cell recirculation to the thymus is a mechanism used by the immune system to survey and protect the thymus from infection and maintain thymic integrity.

Materials and Methods

Mice and infection

C57BL/6 (WT) mice were purchased from Charles River Laboratories (Barcelona, Spain) or from Jackson Laboratories (Bar Harbor, ME) and CD45.1 mice (B6.SJL-Ptprca Pepcb/BoyJ)(32) and TCRα KO (B6.129S2-Tcra^tm1Mom/J) from The Jackson Laboratory (Bar Harbor, ME)(33). RAG-GFP mice (34) were kindly provided by Dr. António Bandeira (Pasteur Institute, Paris, France). Both TCRα KO and RAG-GFP mice were bred in our facilities. Mice were 7 to 10 weeks old at the start of the experiments. All animal experiments were performed in accordance with National and European Commission guidelines for the care and handling of laboratory animals and were approved by the National Veterinary Directorate and by the local Animal Ethical Committee or by the Dana Farber Cancer Institute Animal Care and Use Committee (Animal Welfare Assurance no. A3023-01), under Public Health Service assurance of Office of Laboratory Animal Welfare guidelines. Mice infected with M. tuberculosis were housed in a biosafety level 3 facility under specific pathogen-free conditions at the Animal Biohazard Containment Suite (Dana Farber Cancer Institute, Boston, MA).

Experimental infection

M. avium (strain 2447, provided by Dr. F. Portaels, Institute of Tropical Medicine, Antwerp, Belgium) infection was performed intravenously through the lateral tail vein delivering 10⁶ CFU per mouse. For each M. tuberculosis (Erdman strain) infection, a bacterial aliquot was thawed, sonicated twice for 10 s in a cup horn sonicator, and then diluted in 0.9% NaCl–0.02% Tween 80. A 15 ml suspension of M. tuberculosis was loaded into a nebulizer (MiniHEART nebulizer; Vortran Medical Technologies) and mice were infected via the aerosol route using a nose-only exposure unit (Intox Products) and received 100–200 CFU/mouse. At different times post-infection, mice were euthanized by carbon dioxide inhalation or by decapitation and organs were aseptically removed, individually homogenized and viable bacteria were enumerated by plating 10-fold serial dilutions of organ homogenates onto 7H10 or 7H11 agar plates for M. avium and M. tuberculosis, respectively. Plates were incubated at 37°C and M. avium and M. tuberculosis colonies were counted after 7 and 21 d, respectively.

Gene expression analysis

Total RNA was isolated from thymi, spleens and lungs using TRIZOL reagent or TRIZOL Plus RNA purification system (Invitrogen, CA, USA). Five hundred nanograms of total RNA were amplified using the Superscript RNA amplification system (Invitrogen CA, USA) according to the manufacturer's instructions. mRNA transcripts were assessed by quantitative real-time PCR (qPCR) using SsoFast™ EvaGreen Supermix® (BioRad, CA, USA) in a BioRad CFX96™ Real-Time System with a C1000™ Thermal Cycler or a Stratagene Mx3005P Thermal Cycler. The hypoxanthine guanine phosphoribosyl transferase (HPRT) was used as reference gene. Specific oligonucleotides were used for Hprt (sense: 5'-GCT GGT GAA AAG GAC CTC T-3'; antisense: 5'-CAC AGG ACT AGA ACA CCT GC-3'), Ifnγ (sense: 5'-CAA CAG CAA GGC GAA AAA GG-3'; antisense: 5'-GGA CCA CTC GGA TGA GCT CA-3'), Tnf (sense: 5'-TGC CTA TGT CTC AGC CTC TTC-3'; antisense: 5'-GAG GCC ATT TGG GAA CTT CT-3'), Cxcl9 (sense: 5'-CTT TTC CTC TTG GGC ATC AT-3'; antisense: 5'-GCA TCG TGC ATT CCT TAT CA-3'), Cxcl10 (sense: 5'-GCT GCC GTC ATT TTC TGC-3'; antisense: 5'-TCT CAC TGG CCC GTC ATC-3'), Ccl4 (sense: 5'-AGC ACC AAT GGG CTC TGA-3'; antisense: 5'-TTT GGT CAG GAA TAC CAC AGC -3') and inducible nitric oxide synthase (Inos; sense: 5'-CTC GGA GGT TCA CCT CAC TGT-3'; antisense: 5'-GCT GGA AGC CAC TGA CAC TT-3'). The cDNA was denatured for 1 min at 95 °C, followed by 40 cycles of 95 °C for 15 s, incubation at the optimized melting temperature for 20 s, and 72 °C for 20 s. Optimized melting temperatures were 57 °C for Cxcl9 and Ccl4, 58 °C for Hprt, Ifnγ and Cxcl10 and 59 °C for Inos. The expression level of each gene was determined using the ΔΔCt method taking into account the efficiency of the PCR reaction (35). Data are represented as the ratio of the expression level of the gene for each infected mice over the mean expression level of the gene in uninfected mice.

Protein quantification in tissue homogenates

Protein was extracted from spleen, lung and thymus using the Bio-Plex Cell Lysis kit (Bio-Rad, CA, USA) and the concentration of IFNγ, TNF and CCL4 were measured using a mouse Bio-Plex cytokine assay (Bio-Rad, CA, USA). In the case of M. tuberculosis-infected tissues, the concentrations of CXCL9 and CXCL10 were measured using the mouse CXCL9/MIG Quantikine ELISA Kit or mouse CXCL10/IP-10/CRG-2 Quantikine ELISA Kit (R&D Systems, MN, USA).

Immunohistochemistry

Detection of iNOS was performed by immunohistochemistry in paraformaldehyde fixed, paraffin embedded tissues. Briefly, 5 μm thymic sections were dehydrated and antigens were `unmasked' by incubation at 96 °C for 30 min in 1 mM EDTA 0.05% Tween pH 8. Non-specific binding was blocked using 4% BSA in PBS with 0.05% Tween 20 and endogenous peroxidases were blocked by incubation with 3% hydrogen peroxide for 30 min. Tissues were incubated overnight, at 4 °C, with purified rabbit anti-mouse iNOS (clone M-19, Santa Cruz Biotechnology, CA, USA) and detection was performed using a peroxidase goat anti-rabbit IgG (Vector Labs, CA, USA) followed by incubation with DAB until color development. Mycobacteria were detected by Ziehl-Neelsen, using standard procedures, after the iNOS staining. Slides were visualized using a BX61 microscope with an Olympus DP70 camera. No significant signal was observed when iNOS stain was performed in iNOS KO mice.

In vitro stimulation and IFNγ measurement by ELISA

Cell suspensions from thymus and spleen were prepared by gentle disruption of the organs between two notched slide glasses or by forcing organs through a 70 μm nylon strainer (Fisher). For lung preparations, tissue was digested for 1 h at 37 °C in 1 mg/mL collagenase (Sigma) prior to straining. Erythrocytes were lysed using a hemolytic solution (155 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM sodium EDTA pH 7.2) and, after washing, cells were resuspended in supplemented DMEM or RPMI (10% heat inactivated FCS, 10 mM HEPES, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 mg/ml streptomycin and 50 U/ml penicillin, all from Invitrogen). Cells were enumerated in 4% trypan blue on a hemocytometer or on a Countess Automatic Cell Counter (Life Technologies). 5×10⁵ cells were plated in each well of a 96-well plate and incubated, in triplicate, in the presence M. avium total extract proteins (4 μg/ml), Ag85_280-294 peptide (4 μg/ml; Metabiom, Germany) (36), TB10.4_4-11 peptide (10 μM; New England Peptide) (37) or ESAT-6_1-20 peptide (10 μM; New England Peptide) (38). Incubation in the presence of Concanavalin A (4 μg/ml; Sigma) or αCD3/αCD28 (1 mg/mL; BioLegend) or in the absence of stimuli were used as positive and negative controls, respectively. Supernatants were collected after 72 h of culture and the concentration of IFNγ was determined by ELISA (R4-6A2 and biotinylated AN18 were used as capture and detection Abs, respectively, from eBiosciences; or using the Mouse IFNγ ELISA MAX Standard kit, from Biolegend).

Flow cytometry

Surface staining was performed with antibodies specific for mouse CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD24 (M1/69), CD44 (IM7) and CXCR3 (CXCR3-173) (from Biolegend, CA, USA, or from BD Pharmingen, CA, USA). The tetramers Ag85_280-294-loaded I-A^b and TB10.4_4-11-loaded H-2 K^b were obtained from the National Institutes of Health Tetramer Core Facility (Emory University Vaccine Center, Atlanta, GA, USA); staining was performed for 30 min at 37 °C before incubation with the antibodies mix for the Ag85_280-294-loaded I-A^b tetramer or for 20 min on ice during the incubation with the surface antibodies for the TB10.4_4-11-loaded H-2 K^b tetramer. All stainings, except the one with the Ag85_280-294-loaded I-A^b tetramer, were fixed before acquisition with 2% formaldehyde in PBS for 30 min.

Cell analysis was performed on a LSRII flow cytometer or on a FACS Canto using FACS Diva Software (Becton Dickinson, NJ, USA). Data were analyzed using FlowJo Software (Tree Star, OR, USA). Single-lymphocyte events were gated by forward scatter versus height and side scatter for size and granularity.

Thymic transplant

Thymic lobes were aseptically removed from WT mice and kept in cold supplemented DMEM until being transplanted under the kidney capsule of anesthetized TCRα KO mice (200 μg xylazine hydrochloride and 200 μg ketamine hydrochloride, administered i.v.).

T cell chimeras

Single cell suspensions of pools of spleen and thymus from M. avium infected RAG-GFP mice (20 to 22 wpi) were prepared. CD4⁺ T cells were purified from each suspension using the CD4⁺ T cell isolation kit for spleenocytes or CD8a (Ly-2) microbeads (both from Miltenyi Biotec, Germany) for thymocytes. Magnetic separation was performed with an autoMACS separator (Miltenyi Biotec, Germany). After purification, cells were counted and stained with antibodies for CD3, CD4, CD8 and CD24 for sorting. Splenic CD4⁺CD8⁻CD3⁺ cells, thymic CD4⁺CD8⁻CD3⁺RAG⁻CD24^lo cells and thymic CD4⁺CD8⁻CD3⁺RAG^int cells were sorted using a FACSAria cell sorter (Becton Dickinson, NJ, USA) with a purity ranging from 98 to 100%, depending on the experiment and on the population. 4 to 4.2×10⁵ cells were transferred per TCRα KO receptor mouse.

Adoptive transfer of CXCR3-expressing T cells

Single cell suspensions of pools of spleens from M. tuberculosis infected B6 mice (20 to 26 wpi) were prepared. Total T cells were purified from each suspension using the T cell isolation kit II (from Miltenyi Biotec, Germany). Magnetic separation was performed with an autoMACS separator (Miltenyi Biotec, Germany). After purification, cells were counted and 7×10⁷ cells were transferred intravenously to each recipient. Recipient mice (CD45.1) had been infected for 12 weeks, and were analyzed 24 h following transfer. After sacrifice, magnetic bead-based enrichment was performed on single cell suspensions of thymocytes, as previously described (39). Briefly, cells were stained with a PE-labeled antibody specific for mouse CD45.2 (104) (from Biolegend, CA, USA), followed by incubation with anti-PE beads (from Miltenyi Biotec, Germany). Magnetic separation was performed with an autoMACS separator (Miltenyi Biotec, Germany). The bound fraction, enriched for donor cells, was analyzed by FACS (as described above).

Statistical analysis

All data are represented as mean + SEM. Data were verified for Gaussian distribution or Mann-Whitney U test were performed to compare two groups. To compare more than 2 groups, one-way ANOVA, followed by Bonferroni post-hoc test was performed. Differences with a p<0.05 were considered significant and represented by *.

Results

Thymic integrity is maintained following mycobacterial infection

The ability of mycobacteria to disseminate to the thymus has been previously described upon aerosol challenge with M. tuberculosis [Figure 1A and (18)] and following intravenous infection with M. avium (18, 19). In both cases the bacterial load in the thymus is initially low or undetectable, but increases profoundly until it plateaus, in a pattern similar to the one observed in the lung and spleen, indicative of immunological restriction of bacterial replication. However, bacterial dissemination to the thymus is delayed compared to other organs, and peaks late during the course of infection – 16 weeks post infection (wpi) for M. avium (18, 19) and 12 wpi for M. tuberculosis (Figure 1A). In addition, thymic infection by M. tuberculosis is also more heterogeneous, with a small fraction of mice having undetectable bacterial load within the thymus even at 24 wpi (Figure 1A). The fact that we did not observe this heterogeneity previously (18) might be a consequence of the greater number of animals studied in the present experiments or of slight differences in the infection protocol (different M. tuberculosis strains and/or different apparatus used for aerosol infection).

Figure 1. M. tuberculosis infection of the thymus.

A. Bacterial burden in the lung, spleen and thymus after aerosol infection with M. tuberculosis. Data is pooled from eleven independent experiments each with 3 to 7 subjects per time point, for a total 97 subjects. Each dot represents one mouse, the solid line is the mean, and the dashed line is the lower detection limit. Number of thymocytes (B) and percentage of the main thymocyte populations (based on CD3, CD4 and CD8 expression) (C) after M. tuberculosis infection. Both infected (closed bars) and age-matched uninfected controls (open bars) are represented. *, p<0.05 by Mann-Whitney U test. Bars represent the mean ± SEM (n = 5 mice/group). Data is representative of two independent experiments.

As previously observed for M. avium infection with a low virulence strain (19), infection with M. tuberculosis does not induce loss of thymic cellularity even at 24 wpi, as compared to uninfected age-matched controls (Figure 1B). Percentages of the four main thymic populations - assessed by CD3, CD4 and CD8 expression [CD4⁻CD8⁻CD3⁻ double negative (DN); CD4⁺CD8⁺CD3^low/−; double positive (DP); CD4⁺CD8⁻CD3⁺ single positive CD4 (CD4SP); CD4⁻CD8⁺CD3⁺ single positive CD8 (CD8SP)] – are also maintained following infection, although a small increase in the percentages of SP cells, both CD4SP and CD8SP, and decrease in DP, is observed from 12 wpi on (Figure 1C). These data indicate that the presence of M. tuberculosis within the thymus does not induce premature thymic atrophy – a common consequence of systemic infection (8) – nor major alterations in thymic cell populations, as has been shown previously for M. avium (19).

An immune response develops in the thymus following mycobacterial infection

The stabilization of bacterial growth in the primary infected organ, spleen for M. avium and lung for M. tuberculosis, is associated with the establishment of an effective acquired immune response (40–42). To evaluate whether an immune response is established in the thymus, the expression of key cytokines required for immunity to mycobacteria were measured in whole organ homogenates of mice infected with M. avium (intravenously) or with M. tuberculosis (by aerosol), and compared to uninfected controls. The selected time-points reflect periods when: 1) the peak in the immune response is observed in the primary infected organs upon M. tuberculosis (lung) (43) and M. avium (spleen) infection (40) (4 wpi for both infections); 2) the bacterial burden in the thymus is still increasing (4 wpi for M. tuberculosis, 4 and 12 wpi for M. avium); 3) the bacterial burden plateaus in the thymus (12 wpi for M. tuberculosis, 16 wpi for M. avium); 4) bacterial growth in the thymus has been sustained for a long period (24 wpi) (Figure 1) (18, 19).

During both infections, IFNγ expression is significantly increased as early as 4 wpi in the primary infected organ (spleen for M. avium – Figure 2A; lung for M. tuberculosis – Figure 2B). In the thymus, IFNγ expression is increased only at later time points consistent with the delayed dissemination and establishment of bacterial control (16 wpi for M. avium – Figure 2A; 12 wpi for M. tuberculosis – Figure 2B). The kinetics of TNF expression is similar to that of IFNγ in mice infected with M. tuberculosis (Figure 2B). In contrast, in mice infected with M. avium TNF is not upregulated in the thymus and only transiently upregulated in the spleen (Figure 2A). In agreement with the observed changes in gene expression, elevated levels of IFNγ protein are detected in the thymus 16 wpi with M. avium (Figure 2C), and both IFNγ and TNF protein are increased in the thymus 12 wpi with M. tuberculosis (Figure 2D). These data show that the thymus is the site of an ongoing immune response that involves the production of pro-inflammatory cytokines after infection, and the cytokine levels peak concordantly with the control of bacterial replication.

Figure 2. An immune response is established in the thymus after mycobacteria infection.

RNA expression levels were determined by qPCR (A, B) and protein concentration were determined by multiplex (C, D) in tissues of M. avium (A, C) and M. tuberculosis (B, D) infected mice. Bars refer to fold increase of infected mice in comparison to the average of uninfected mice and represent mean ± SEM (n = 4 to 8 mice/group). Data is representative of 2 to 3 independent experiments. *, p<0.05 by Mann-Whitney U test (statistics were performed by comparing uninfected with infected mice, before performing the ratio).

Mycobacteria-specific T cells are detected in the thymus after infection

To determine whether a mycobacteria-specific T cell response accompanies the inflammatory changes observed in the infected thymus, IFNγ production by thymus cells obtained from infected mice after stimulation with defined mycobacterial antigens was assessed. IFNγ production was specifically induced when spleen or thymus cells from M. avium infected mice were stimulated with M. avium protein extract or with Ag85_280-294 peptide [primarily recognized by CD4⁺ T cells (36)]. These antigens led to maximal IFNγ production by splenocytes at 4 wpi (Figure 3A), as previously described (40). In contrast, these antigens did not trigger IFNγ production by thymic cells until 16 wpi and even then elicited much lower amounts of IFNγ (Figure 3A). Similarly, large amounts of IFNγ were produced by lung cells obtained from M. tuberculosis-infected mice after stimulation with the immunodominant epitope ESAT6_1-20 [recognized by CD4⁺ T cells (38)] or TB10.4_4-11 [recognized by CD8⁺ T cells (37)] as early as 4 wpi. Although a M. tuberculosis-specific response could be detected in the thymus as early as 4 wpi, a more substantial response was detected 12 wpi (Figure 3B) and, as in the case of M. avium, in quite lower amounts and associated with control of the infection and a plateau of the bacterial burden in the thymus (Figure 3B).

Figure 3. Mycobacteria-specific T cell responses are detected in the thymus after M. avium and M. tuberculosis infection.

Cells from M. avium (A) or M. tuberculosis (B) infected mice were stimulated in vitro in the presence of M. avium protein extract or Ag85_280-294 peptide (A) and ESAT6_1-20 or TB10.4_4-11 peptides (B). Age-matched uninfected mice were used as controls. Unstimulated and Concanavalin A or αCD3/αCD28 stimulated cultures were used as negative and positive controls of the in vitro stimulation, respectively (data not shown). IFNγ quantification in cell supernatants was performed by ELISA. C. TB10.4-specific CD8⁺ T cells were detected in lung and thymus of M. tuberculosis infected mice using the K^b/TB10.4_4-11 tetramer. Closed symbols represent infected mice and open symbols uninfected mice. Data points represent the mean ± SEM (n = 4 to 6 mice/group). Data is representative of two to four independent experiments. *, p<0.05 by Mann-Whitney U test.

We also determined the frequency of TB10.4_4-11-specific CD8⁺ T cells during M. tuberculosis infection using K^b/TB10.4_4-11 tetramers, which can identify antigen-specific CD8⁺ T cells independently of their function (Figure 3C and Supplementary Figure 1). Similar to our other data, the frequency of antigen-specific T cells closely correlated with the production of IFNγ in the infected organs. Taken together, these data indicate that mycobacteria-specific T cells are present within the thymus following infection. Although the magnitude of the response differs from that observed in the dominant target organs, the kinetics in the thymus resembles other tissue-specific responses to mycobacterial infection.

Mycobacteria-infected cells in the thymus express iNOS

The production of IFNγ by antigen-specific T cells is a central feature of protective immunity against mycobacterial infection (44, 45). Among the important antibacterial actions of IFNγ is upregulation of inducible nitric oxide synthase (iNOS) expression by macrophages. iNOS catalyzes the production of nitric oxide (NO), which has a significant role in controlling M. tuberculosis infection (46). Despite the fact that NO plays no role in the protective immunity to M. avium (47, 48), iNOS expression represents a suitable marker of macrophage activation in this scenario. The elevated IFNγ level in the infected primary organs was associated with increased iNOS expression by 4 wpi. An increase in iNOS expression in the thymus was only observed at later time points: 24 wpi for M. avium (Figure 4A) and 12 wpi for M. tuberculosis (Figure 4B). These data are consistent with a protective T cell response in the thymus leading to iNOS induction, NO production, and control of bacterial replication.

Figure 4. Mycobacterial infection in the thymus is associated with increased iNOS expression.

iNOS RNA expression levels were quantified by qPCR in tissues of M. avium (A) and M. tuberculosis (B) infected mice. Bars refer to fold increase of infected mice in comparison to the average of uninfected mice and represents the mean ± SEM (n=4 to 8 mice/group). Data is representative of 2 to 3 independent experiments. *, p<0.05 by Mann-Whitney U test (statistics were performed by comparing uninfected with infected mice, before performing the ratio). C, D. Representative medullary thymic sections of M. avium (C) and M. tuberculosis (D) infected thymi stained for iNOS (brown). Bacilli were detected by Ziehl-Neelsen staining. Shown are representative images obtained from the analysis of 3 to 5 thymi per time-point from of 2 to 3 independent experiments. Bar = 10μm.

In addition to measuring iNOS gene expression, we detected iNOS protein by immunohistochemical staining. Although iNOS is expressed in the thymus even in the absence of infection (49), we detected an increase in the number of cells containing M. avium (Figure 4C) or M. tuberculosis (Figure 4D) and expressing iNOS throughout the course of infection, with most of the infected cells expressing iNOS at later time points. For M. avium and M. tuberculosis, most infected cells in the thymus were located in the medulla or corticomedullary region, although occasional bacteria were observed in the cortex. M. tuberculosis-infected cells were also infrequently observed in the subcapsular zone. Thus, our data showing augmented iNOS expression and its colocalization with infected cells indicates the establishment of a bona fide protective immune response within the thymus.

Mycobacteria-infected thymi contain cells able to transfer protection against infection

Having shown that the thymus is a site of infection where antigen-specific immune responses are detected, we next asked whether cells within infected thymi could confer protection against subsequent infection. To that end, we transplanted M. avium-infected or uninfected thymic lobes from WT mice under the kidney capsule of TCRα KO mice (33), that lack peripheral αβ T cells (Figure 5A).

Figure 5. M. avium-infected thymi contain T cells able to confer protection during infection.

A. Schematic representation of the experiment. Thymic lobes from M. avium infected mice (24 wpi) or from uninfected WT mice were transplanted under the kidney capsule of TCRα KO receptor mice. Transplanted mice were infected 2 to 3 days post-transplant and sacrificed 4 and 8 weeks later. Non-transplanted TCRα KO mice were used as controls. Number of CD4⁺ T cells (B) and bacterial load (C) were assessed in the spleen. Each column represents mean ± SEM (n = 4 mice/group) from one of two experiments. *, p<0.05 by one-way ANOVA.

The transplanted mice were challenged with M. avium and sacrificed at 4 or 8 wpi, at which time thymic engraftment was confirmed macroscopically. At both time points, the spleens of mice transplanted with infected or uninfected thymic lobes had similar numbers of CD4⁺ T cells, indicating that cells leaving the infected or uninfected thymus were equally able to reconstitute the peripheral T cell pool of the recipient TCRα KO mice (Figure 5B). Both groups receiving thymic transplants had lower bacterial burdens than the untransplanted TCRα KO mice (Figure 5C). This demonstrates that the cells emerging from the thymic grafts are functional and can mediate protection against microbial pathogens. Importantly, mice receiving infected thymic grafts were significantly more protected than those receiving uninfected grafts, despite similar T cell reconstitution (Figure 5C). These data support the hypothesis that the greater protection conferred by the infected thymic grafts is mediated by M. avium-specific T cells contained within the grafts, that efficiently confer protection against infectious challenge.

Mycobacteria-specific T cells within the thymus are recirculating cells

After establishing that antigen-specific T cells within the infected thymus produce protective cytokines, induce NO production, and transfer protection, we next sought to determine the origin of these cells. In particular, we wished to determine whether differentiating T cells were primed in the thymus, or alternately, whether mature mycobacteria-specific T cells, primed in the periphery, traffic to the thymus.

Our previous data demonstrated that T cells that differentiate within M. avium-infected thymi are tolerant to the invading pathogen (19). Moreover, newly differentiated T cells arising from the thymus are not fully mature (50) and need additional signals within secondary lymphoid organs to achieve full differentiation (51, 52). Therefore, we used a genetic model in which GFP is expressed under the control of the RAG2 promoter (34), for the purpose of identifying newly differentiated T cells in the thymus. The RAG2 enzyme is essential for genetic recombination of B and T cell antigen receptors and is down-regulated afterwards. Even though RAG2 is down-regulated after T cell differentiation in the thymus, cells remain GFP⁺ for about 2 weeks (50). As such, it is highly expressed in differentiating T cells and its expression is a useful marker to distinguish between newly differentiated T cells (high to intermediate GFP expression) and mature peripheral T cells (no GFP expression).

When RAG-GFP mice were infected with M. avium there was no difference on the percentage of GFP⁻ among CD4SP cells, suggesting that there is no difference on the amount of thymic recirculating T cells (Figure 6A and Supplementary Figure 1). However, upon infection most of the Ag85_280-294-specific CD4SP cells in the thymus were GFP⁻, showing a modification in the repertoire of recirculating T cells (Figure 6A). In contrast with what was observed for M. avium, thymi from M. tuberculosis infected mice showed an increase in the percentage of GFP⁻ within CD8SP cells, suggesting an increase in T cell trafficking from the periphery back to the thymus (Figure 6B). A similar increase was detected when analyzing mycobacteria-specific T cells (TB10.4_4-11-specific CD8SP T cells) during M. tuberculosis infection, in agreement with what was observed for antigen-specific cells in the thymus of M. avium infected mice (Figure 6B). In both cases, no tetramer positive cells were found in the thymus of control uninfected mice, which is in agreement with the small precursor frequency we observe for T cells reactive against these mycobacterial antigens in naïve mice (data not shown). In parallel, we show that the majority of TB10.4_4-11-specific CD8SP T cells within the thymus of WT mice infected with M. tuberculosis are CD44^hiCD24^lo (Figure 6C), a phenotype typical of re-circulating CD8⁺ T cells (25, 53).

Figure 6. Mycobacteria-specific T cells recirculate to the thymus upon infection.

A. Total recirculating CD4SP cells from the thymi of naïve or M. avium-infected RAG-GFP mice and recirculating CD4SP Ag85-specific cells from the thymi of infected mice were enumerated. Represented are density plots of the expression of the GFP marker in total CD4SP cells (left) and in CD4SP Ag85⁺ cells (right). B. Total recirculating CD8SP cells from the thymi of naïve or M. tuberculosis-infected RAG-GFP mice and recirculating CD8SP TB10.4-specific cells from the thymi of infected mice were enumerated. Represented are density plots of the expression of the GFP marker in total CD8SP cells (left) and in CD8SP TB10.4⁺ cells (right). C. In WT mice, using the surface markers CD44 and CD24, total recirculating CD8SP cells and recirculating CD8SP TB10.4-specific cells from the thymi of infected mice were enumerated. Represented are density plots of the expression of CD44 and CD24 in total CD8SP cells (left) and in CD8SP TB10.4⁺ cells (right). All represented dot plots show concatenated data from all animals of the group. Bars represent mean ± SEM. Data is pooled from 2 to 3 independent experiments, with 5 to 8 mice/group. *, p<0.05 by one-way ANOVA.

T cell chemokines are increased in the thymus during infection

The above results indicate that antigen-specific T cells from the periphery respond to infection in the thymus by trafficking back to this organ. To determine whether increased chemokine production is associated with the recruitment of mycobacteria-specific T cells to the thymus, we measured the expression of chemokines that recruit T_h1 cells (54), including CXCL9, CXCL10 and CCL4, after M. avium and M. tuberculosis infection. We found increased mRNA expression of all three chemokines in the primary infected organs during M. tuberculosis infection and at least at the initial phase of M. avium infection (Figure 7A and B). These chemokines were also detected in the thymus though at lower levels and at later time points – after 16 wpi for M. avium (Figure 7A) and 12 wpi for M. tuberculosis (Figure 7B). At the protein level, increased CCL4 was detected in the spleen but not in the thymus during M. avium infection, possibly because the expression levels were low (data not shown). During M. tuberculosis infection, the expression of these chemokines was strongly induced in the lungs of mice infected for 4 weeks, and the responses appeared maximal for all three chemokines by 12 wpi (Figure 7C). These chemokines were also highly expressed in the thymus, but their expression was not detected until 12 wpi (Figure 7C).

Figure 7. Recirculation of mycobacteria-specific T cells into infected thymi correlates with increased Th1 recruiting chemokines and their cognate receptors.

RNA expression (A, B) and protein concentration (C) were determined in tissues of M. avium (A) and M. tuberculosis (B, C) infected mice. Bars refer to fold increase of infected mice in comparison to the average of uninfected mice and represents mean ± SEM (n = 4 to 8 mice/group). Data is representative of 2 to 3 independent experiments. *, p<0.05 by Mann-Whitney U test (statistics were performed by comparing uninfected with infected mice, before performing the ratio). D. FACS plot of CXCR3 expression in thymic CD8SP TB10.4⁻ and CD8SP TB10.4⁺ cells from M. tuberculosis-infected mice. Represented is the concatenated data from 3 mice analyzed. Gating strategy is depicted in Supplementary figure 1. E. Percentage of CXCR3 expressing cells between CD8SP TB10.4⁻ and CD8SP TB10.4⁺ cells in M. tuberculosis-infected mice. Each bar represents mean ± SEM (n = 4 to 8 mice/group). Data is representative of 2 to 3 independent experiments. F. CXCR3 expression by donor CD4SP and CD8SP in the spleen and thymus. T cells from M. tuberculosis-infected CD45.2 mice were transferred to M. tuberculosis-infected CD45.1 recipient mice, and their CXCR3 expression was analyzed the next day. See Supplementary Figure 2 for scheme and gating strategy. Each bar represents mean ± SEM (n = 3 mice). Data is representative of 3 independent experiments. *, p<0.05 by one-way ANOVA.

The high levels of CXCL9 and CXCL10 detected in the infected thymus led us to investigate whether mycobacteria-specific T cells express CXCR3, the receptor for these chemokines (54). We found the majority of TB10.4_4-11-specific CD8SP T cells in M. tuberculosis infected thymi to express high levels of the chemokine receptor CXCR3 (Figure 7D and E).

To support a role for CXCR3 in the recruitment of antigen-specific T cells to the infected thymus, we adoptively transferred purified splenic T cells obtained from M. tuberculosis infected donors into congenically marked infected recipients (Supplementary Figure 2). Analysis of the donor cells immediately before transfer found that 30–40% of the CD4⁺ and CD8⁺ T cells expressed CXCR3. 24 hrs after injection, the proportion of donor CD4⁺ and CD8⁺ donor T cells recruited to the thymus that expressed CXCR3 was increased to 65–85%. The increased CXCR3 expression was limited to the infected thymus, and was not observed in the spleen or the lung of recipient mice (Figure 7F and Supplementary Figure 2). This data suggests that CXCR3 participates in the recruitment of antigen-specific cells to the infected thymus. Thus, although the thymus is the key organ for T cell differentiation, mycobacteria-specific T cells from the periphery are recruited to the thymus following infection. Our data is consistent with the chemokines CXCL9 and CXCL10 playing an important role in the recruitment of these antigen-specific T cells from peripheral organs to the thymus in order to fight infection.

Peripheral recirculating T cells within infected thymi efficiently confer protection against infection

To confirm that peripheral T cells recirculating to infected thymi are the ones that confer protection to infection, RAG-GFP mice were infected with M. avium and 20 wpi sacrificed and their thymi and spleens collected. Highly purified T cell populations were sorted accordingly to the expression of GFP and of surface markers: 1) total CD4⁺ T cells from the spleen (CD3⁺CD4⁺CD8⁻); 2) thymic recirculating CD4SP cells (CD3⁺CD4⁺CD8⁻GFP⁻CD24^lo); 3) thymic newly differentiated CD4SP cells (CD3⁺CD4⁺CD8⁻GFP^int). A GFP^int population was sorted to the detriment of the total GFP⁺ population as we wished to exclude the most immature GFP^hi CD4SP cells (Figure 8A). Each cell subset was transferred to TCRα KO mice that had been infected the previous day with M. avium. Eight weeks later, the mice were analyzed (Figure 8B).

Figure 8. Thymic recirculating T cells from M. avium-infected mice efficiently confer protection to infection.

20 weeks after M. avium infection, RAG-GFP mice were sacrificed and their thymi and spleens removed. Spleen CD4⁺CD3⁺ cells, thymic CD3⁺CD4⁺CD8⁻GFP⁻CD24^lo(recirculating) cells and thymus CD3⁺CD4⁺CD8⁻GFP^int cells were sorted and 4 to 4.2×10⁵ cells (depending on the experiment) were transferred to TCRα KO mice one day after M. avium infection. Mice were sacrificed 8 weeks post-transfer. Non-transplanted TCRα KO mice were used as controls. A. Purity of the transferred thymic populations. Thymic cell suspensions were depleted of CD8⁺ cells (top panels); and CD3⁺CD4⁺CD8⁻GFP^int (middle panels) and CD3⁺CD4⁺CD8⁻GFP⁻CD24^lo (bottom panels) cells populations were sorted. B. Schematic representation of the experiment. C. Number of spleen CD4⁺ T cells (left panel) and bacterial load in the spleen (middle panel) and liver (right panel) were assessed. Each column represents mean ± SEM (n = 4 to 9 mice/group) from two combined independent experiments. *, p<0.05 by one-way ANOVA.

Despite receiving the same number of cells, mice that received recirculating thymic CD4SP cells (GFP⁻) had lower numbers of CD4⁺ T cells than animals that received either newly differentiated T cells from the thymus (GFP^int) or total splenic T cells (Figure 8C, left panel). Nevertheless, administration of recirculating thymic CD4SP cells provided significantly more protection from M. avium infection in the spleen and liver compared to animals that received newly differentiated thymic T cells (Figure 8C, middle and right panels). In addition, the protection provided by recirculating thymic CD4SP T cells is comparable to that of total splenic CD4⁺ T cells. Although transfer of newly differentiated T cells from the thymus provided partial protection against M. avium infection in comparison to TCRα KO mice, this was significantly reduced compared to protection conferred by recirculating thymic CD4SP cells. Taken together, these results indicate that T cells recirculating from the periphery back into the thymus are capable of protecting mice against infection.

Discussion

Our understanding of the role of T cells in resistance to microbial infection is generally limited to the effector or memory phases of the immune response, or their role in vaccine-induced immunity. Despite the key role of the thymus in generating such T cells, the consequences of infection of the thymus itself have been rarely studied. Our previous studies have shown that mycobacteria disseminate to the thymus (18, 19) and alter the process of T cell differentiation leading to tolerance against the invading pathogen (19). Here, we address whether and how the immune system is able to defend the thymus against infection, to preserve its structure, and its ability to generate T cells.

Our results show that like other tissues that are the target of mycobacterial infection, mycobacteria-specific T cells appear in the thymus following the dissemination and subsequent growth of bacteria. Furthermore, the appearance of antigen-specific T cells roughly correlates with the host's ability to control bacterial growth in the thymus. Antigen-specific T cells in the thymus secrete IFNγ and stimulate antimicrobial functions of infected cells, as manifested by the upregulation of iNOS. Thus, T cell responses in the thymus resemble ones occurring in the lung or spleen although they differ in magnitude. In fact, both the number of antigen-specific T cells and related immune molecules detected in the thymus is significantly lower than found in the lung or the spleen. Although the thymic immune response is less pronounced, its ability to control bacterial growth seems similarly efficient as the one established in other peripheral organs. Additionally, our previous data shows that cellular infiltrates and granuloma-like lesions, characteristic of inflammation in other tissues, are not detected in the infected thymus (18). The reasons for this difference are not clear, but we speculate that avoiding a massive pro-inflammatory state and preventing disruption of the cellular architecture is required for the thymus to remain functional. The smaller magnitude of the thymic-associated immune response might contribute to the maintenance of thymic integrity during the course of infection. It has previously been shown that infection with a highly virulent M. avium strain, which is associated with a vigorous immune response, causes robust thymic atrophy (48). Recent results from our laboratory (55) show that this depends on a synergistic effect between IFNγ-induced iNOS induction and corticosteroid production, as blocking the effect of each one individually prevents thymic atrophy. Of notice, infection with the low virulence M. avium strain used in the present work is not accompanied by increased serum levels of corticosteroids (55), which may contribute to the maintenance of thymic integrity reported here.

An important question that our data address is the origin of antigen-specific T cells in the thymus that mediate protection against mycobacterial infection. In our model, T cells from the periphery recirculate back to the thymus and are responsible for bacterial control. Thus, although the thymus maintains the differentiation of T cells during infection, these do not appear to be the cells that defend the thymus against infection. These data are consistent with our previous results showing that newly differentiated T cells that mature in a thymus infected by M. avium are tolerant to mycobacterial antigens, and consequently are not optimally able to establish protective immunity(19). Interestingly, T cells that differentiate in an infected thymus were not completely impaired in their ability to proliferate, to traffic to infected organs, and to confer a small amount of protection against infection, albeit less efficiently than T cells that develop within an uninfected thymus (19). In agreement with those findings, we now report that newly differentiated T cells are much less effective at conferring protection than peripheral T cells that recirculate back to thymus. Therefore, we conclude that recirculation of peripheral activated T cells back to the thymus is a mechanism to survey and protect this organ from invading pathogens.

These data are also in agreement with other work that finds newly differentiated T cells defective in T_h1 commitment (dampened cytokine production and transcriptional factor expression) and biased toward the T_h2 lineage (56), and a requirement for newly differentiated T cells to exit the thymus and reside in the periphery before they become fully functional (50–52).

Another interesting question arising from this study concerns whether the antigen-specific cells found in the thymus are primed in peripheral organs and traffic back to the thymus because of thymic infection, or are recirculating T cells that traffic back and are then primed within the infected thymus. Although we have not addressed this question directly, our data on the production of T_h1 chemokines within the infected thymus, the expression of the cognate chemokine receptors by the antigen-specific T cells found within this organ, and the preferential recruitment of CXCR3⁺ cells to the infected thymus, strongly supports the notion that the antigen-specific T cells found in the thymus are primed in other tissues during infection and then traffic to the thymus in response to a chemokine gradient.

Thymic recirculating T cells, the focus of multiple studies, have had numerous functions attributed to them including thymic surveillance, tolerance induction, and even modulation of both negative and positive selection (25–27). To our knowledge, the results presented here are the first evidence that recirculating T cells fight infection in the thymus, a function that has been previously hypothesized (25, 57, 58). The observation that mycobacteria-specific T cell trafficking to the thymus is associated with an increase in the levels of T_h1-related chemokines in this organ, and the expression of specific chemokine receptors such as CXCR3 by the recirculating T cells, lead us to propose that the immune system has evolved mechanisms to recruit peripheral T cells to the thymus during infection. This might ensure that invading pathogens are unable to disrupt the thymus, the primary lymphoid organ where T cells are generated. Such mechanisms must be particularly relevant for pre-natal and early childhood infections. In both scenarios, pathogen dissemination to the thymus could lead to the presentation of microbial antigens in a “self” context, and lead to the deletion of developing T cells that recognize pathogen-specific antigens. Exclusion of pathogen reactive T cells from the peripheral pool could have severe effects during childhood, when the peripheral T cell pool is still developing, but the consequences could persist throughout adulthood.

Childhood infections are a great cause of morbidity and mortality, particularly in the developing world (59). Interestingly, HIV and M. tuberculosis, two of the most important pathogens in infants and young children, are both able to infect the thymus (16, 60) and thymic infection alters the output and selection of T cells in experimental models using these pathogens (9, 19). The evidence that thymic infection can compromise thymic function implies that the recruitment of protective immune responses to the thymus is essential to prevent detrimental alterations in immunological function.

The implications of our study extend beyond childhood infection. Skepticism about the potential immunological consequences of thymic infection would not be surprising if one were to believe the dogma the adult thymus becomes immunologically useless as it ages and progressively loses function. However, our understanding of the role of the thymus during adulthood is rapidly evolving. Interestingly, the control of persistent infections has been suggested to benefit from the continuous recruitment of naïve T cells generated in the thymus (61, 62). In this case, alterations in the development of these T cells could be detrimental to the ongoing immune response. The integrity of the adult thymus is required for other aspects of ongoing immune responses such as antibody generation (63). This observation raises the hypothesis that thymic disruption secondary to infection can impair T cell responses to infection as well as other components of the immune system. Additionally, decreased thymic output contributes to the progression from HIV infection to AIDS (7, 64), indicating that thymic export is important in the context of chronic infection in adults, especially in the context of M. tuberculosis/HIV co-infection. Combined, these data show that thymic infection has an impact on the development of the immune system and ongoing immune responses to pathogens. Thus, it is not surprising that mechanisms have evolved to specifically respond to the invasion of the thymus by microbial pathogens. We propose that the recirculation of T cells from the periphery to the thymus, and the presence of mature T cells in this organ, represent strategies that evolved to protect the thymus and sustain its activity during and after infectious episodes.

Abbreviations used in this manuscript

BCG

bacillus Calmette-Guérin

DN

double negative

DP

double positive

iNOS

inducible NO synthase

SP

single positive

WPI

weeks post infection

WT

wild-type