Background
Tuberculosis of the central nervous system (CNS-TB) is associated with significant mortality and high distressing levels of neurological morbidity, the majority of survivors suffering permanent neurologic complications [
1‐
3]. It disproportionately affects children, especially in the developing world and immunocompromised individuals despite the availability of anti-tuberculosis treatment [
3‐
5]. The lengthy duration of therapy for CNS-TB makes adherence difficult. In Sub-Saharan African and some Asian countries, tuberculosis is now one of the most common forms of bacterial meningitis due to the effects of HIV [
6‐
8].
CNS-TB results from a rupture of subpial or subependymal foci which has been deposited earlier following lympho-haematogenous dissemination of
Mycobacterium tuberculosis from the primary pulmonary infection, or rupture of an adjacent parenchymal focus [
1,
9]. It is widely accepted that bacilli reach the CNS by a haematogenous route secondary to disease elsewhere in the body. Since the brain parenchyma and the meninges are physiologically and anatomically protected from the systemic circulation system by the blood–brain barrier (BBB), the mechanisms by which the bacilli initially invade this barrier are not clear. It is believed that
M. tuberculosis can cross the BBB as a free organism or via infected peripheral myeloid cells [
10]. However, the latter hypothesis seems controversial, as myeloid cellular traffic is severely restricted into the CNS prior to invasion by
M. tuberculosis [
11], however lymphoid cells in physiological state can bypass the BBB by entering the subarachnoid space via the meningeal veins or the choroid plexus [
12,
13].
Tumour necrosis factor (TNF) has been shown to be critical in the pathogenesis of
M. tuberculosis in the central nervous system [
14‐
16]. TNF is synthesised by several cell types of haematopoietic origin, such as microglia/macrophages, neutrophils, dendritic cells and lymphocytes, and non-haematopoietic origin such as astrocytes and neurons [
17‐
20]. The production of TNF in the CNS alters the BBB permeability and CSF leukocytosis in experimental bacterial meningitis [
21,
22] and has been implicated in fostering the progression of CNS-TB in a rabbit model [
16]. However, Tobin et al. [
15] have shown using human and zebrafish models that either state, TNF deficiency or excessive TNF production, causes macrophage lysis, therefore placing
M. tuberculosis in a permissive extracellular niche [
15,
23,
24]. Within this context, it is important to understand the role of TNF in host immunity during
M. tuberculosis infection in the CNS. Here, we generated neuron-specific TNF-deficient (NsTNF
−/−) mice and investigated outcomes after intracerebral
M. tuberculosis infection in comparative studies with TNF
f/f control and global TNF
−/− mice.
Methods
Mice
All mouse strains, TNF
f/f (TNF-floxed wild-type [
25]), Syn1-Cre (synapsin I promoter-driven Cre-transgenic, obtained from the Jackson Laboratory,
http://jaxmice.jax.org/strain/003966.html), NsTNF
−/− (neuron-derived TNF-deficient) and TNF
−/− (TNF-deficient [
25]) mice, were bred and maintained under specific pathogen-free conditions at the animal facility of the University of Cape Town (South Africa). Adult mice between 8 and 12 weeks of age were used, and infected mice were maintained under biosafety level 3 conditions. All animal procedures were approved by the Animal Research Ethics Committee (AEC reference number: 010/018), University of Cape Town, in accordance with the South African National Standard.
Genotyping PCR analysis
Genotyping of mouse strains was performed by PCR analysis of DNA extracted from tail biopsies. Genetic analysis of TNF
f/f mice was previously described [
25]. Syn1-Cre and NsTNF
−/− mouse were genotyped for Cre expression using the primer sets IMR1084 (5′ GCG GTC TGG CAG TAA AAA CTA TC 3′) and IMR1085 (5′ GTG AAA CAG CAT TGC TGT CAC TT 3′), and the internal controls IMR7338 (5′ CTA GGC CAC AGA ATT GAA AGA TCT 3′) and IMR7339 (5′ GTA GGT GGA AAT TCT AGC ATC ATC C 3′). For the presence of the TNF loxP gene in NsTNF
−/− mice, primers KO41 (5′ TGA GTC TGT CTT AAC TAA CC 3′) and KO42 (5′ CCC TTC ATT CTC AAG GCA CA 3′) were used [
25].
Aerosol infection and immunohistochemistry of brain tissue
M. tuberculosis with green fluorescent expressing protein (M. tuberculosis H37Rv-GFP, provided by Joel Ernst, New York University School of Medicine, USA) was grown in Middlebrook 7H9 medium (Difco™ Laboratories) containing 25 μg/ml kanamycin and 0.5 % glycerol and enriched with 10 % OADC. Bacterial cultures were incubated at 37 °C and grown until log phase, aliquoted and stored at −80 °C. A frozen aliquot of M. tuberculosis was thawed, passed 30 times through a 29-G needle and diluted in sterile saline. Mice were infected by aerosol inhalation at a dose of 200–500 colony-forming units (CFUs)/lung under biosafety level 3 conditions using a Glas-Col Inhalation Exposure System Model A4224. For analysis, infected mice were deeply anaesthetised and transcardially perfused with 4 % paraformaldehyde. Brains were sectioned at 40 μm using a vibratome and processed for immunohistochemical staining. Neurons were labelled with polyclonal anti-MAP2 (1 μg/ml, Abcam) or anti-β-III-tubulin antibody (1 μg/ml, Abcam), and anti-rabbit Cy3-conjugated secondary antibody (1.5 μg/ml, Jackson ImmunoResearch Laboratories). The microglia/monocytes were labelled with monoclonal anti-CD11b antibody (Clone M1/70, 1 μg/ml, Abcam) and anti-rat Cy3-conjugated secondary antibodies (1.5 μg/ml, Jackson ImmunoResearch Laboratories). The sections were incubated with nuclear marker DAPI (Sigma) and then mounted in fluorescent mounting medium and images captured using a Zeiss 510LSM unit.
Intracerebral infection and clinical scoring
M. tuberculosis strain H37Rv was grown at 37 °C in Middlebrook 7H9 broth containing 10 % OADC and 0.5 % Tween 80 until log phase, then aliquoted and stored at −80 °C. A frozen aliquot was thawed, passed 30 times through a 29-G needle and diluted in sterile saline. Intracerebral infection was performed using a stereotaxic approach of directly injecting
M. tuberculosis H37Rv into the cerebral cortex. Prior to inoculation, a small burr hole was constructed anterior to the bregma and to the left of the midline in the skull exposing the dura mater. Five mice per strain were inoculated intracerebrally with 1 × 10
3–1 × 10
4 CFUs of
M. tuberculosis H37Rv using a Hamilton syringe (Gastight no. 1701, Switzerland). The burr hole was sealed with bone wax and the skin sutured. Animals received a prophylactic pain killer for 3 days at 24-h intervals. The clinical scoring system was adapted from previous reports [
26,
27]. Here, mice were weighed and scored daily for neurologic manifestations during the course of infection as follows: normal (no detectable signs) = 0; head tilt = 1; motility or decrease activity = 2; behaviour depression = 3; and moribund state = 4. Organs of infected mice were harvested and processed at 1, 2, 3 and 15 weeks post-infection. Haematoxylin and eosin (H&E)-stained brain sections at 3 weeks post-infection were reviewed and analysed by a pathologist who was blinded to the study.
Colony enumeration assay
Bacterial burdens in the brains, lungs and spleens of infected mice were determined at specific time points after infection with M. tuberculosis. Organs were weighed and homogenised in 0.04 % Tween 80 saline. Tenfold serial dilutions of organ homogenates were plated in duplicates on Middlebrook 7H10 (Becton, Dickinson and Company) agar plates containing 10 % OADC (Life Technologies, Gaithersburg, MD) and were incubated at 37 °C for 19–21 days. The concentration of M. tuberculosis was then determined by counting the CFUs.
Flow cytometry
To determine the expression level of TNF by neurons, TNFf/f, NsTNF−/− and TNF−/− mice were intracerebrally stimulated with 5 μg/ml of lipopolysaccharide (LPS) for 90 min. Mice were transcardially perfused with 4 % paraformaldehyde. For single-cell suspensions, brains were isolated and tissue passed through a 70-μm nylon cell strainer (Beckton and Dickinson) and washed 2× with phosphate-buffered saline (PBS) and the cell concentration was determined by counting in the presence of trypan blue. TNF expression in neurons was measured by intracellular staining through the addition of saponin buffer to permeabilise the cells, which were then labelled with polyclonal anti-β-III-tubulin antibody (20 μg/ml, Abcam) and an anti-rabbit PE-conjugated secondary antibody (1.5 μg/ml, Jackson ImmunoResearch Laboratories), and anti-TNF antibody (TNF:Alexa 647, Clone MP6-XT22, BD Pharmingen™).
To analyse surface marker expression in microglia, macrophages and dendritic cells, the following antibodies were used: CD11b:PerCP-Cy5-5 (Clone M1/70, BD Pharmingen™ [2 μg/ml]); CD11c:Alexa 700 (Clone HL3, BD Pharmingen™ [2.5 μg/ml]); CD45:APC (Clone 30-F11, BD Pharmingen™ [2 μg/ml]); CD80:FITC (16-10A1, BD Pharmingen™ [4 μg/ml]); CD86:V450 (Clone GL1, BD Horizon™ [4 μg/ml]) and MHCII/(I-A/I-E):PE (M5/114.15.2, BD Pharmingen™ [2 μg/ml]). Cells were washed with PBS/0.1 % bovine serum albumin (BSA)/0.01 % NaN
3 and incubated with the appropriate antibodies for 20 min in the dark. Excess antibodies were removed by washing cells 2× with PBS/0.1 % BSA/0.01 % NaN
3. Pelleted cells were fixed for 18–24 h in 2 % phosphate buffered formaldehyde and analysed on a FACSCalibur (Beckton Dickinson) flow cytometer using CellQuest software. Microglia were defined as CD11b
+CD45
low, macrophages as CD11b
+CD45
high and dendritic cells as CD11c
+CD45
high as previously described [
28,
29].
Quantification of chemokines and cytokines
Supernatants from brain homogenates were prepared for cytokine and chemokine measurement by enzyme-linked immunosorbent assay (ELISA) after 3 weeks subsequent to intracerebral M. tuberculosis infection. The chemokines MCP-1, MIP-1α and RANTES, and the cytokines IL-1β, IL-12p70 and TNF (R&D Systems, Germany) were measured using commercially available ELISA reagents according to the manufacturer’s instructions. Chemokine and cytokine concentrations were measured by absorbance using a Versamax Microplate Reader (Molecular Devices, LLC, CA) with SoftMax software.
Statistical analysis
The data are presented as the mean ± standard error of the mean (SD). Statistical analysis was performed by one-way ANOVA and one-tailed t test for comparisons amongst the time points. For all tests, a p value of ≤0.05 was considered significant.
Discussion
TNF concentration forms a pivotal point of balance that determines outcome of disease during
M. tuberculosis infection. Susceptibility is promoted either under conditions where excess TNF prevails or low TNF concentrations are present [
15,
16,
23,
24]. Under such polar conditions, either macrophage/cell effector function is impaired which leads to intracellular mycobacterial replication and eventual cell lysis or excessive TNF levels promote macrophage necrosis, in so doing enrich conditions for bacilli growth. Diverse cell types can act as contributory sources of TNF (during pathogenic challenge) and may therefore be differential contributors to the TNF environment required for optimum mycobacterial control. It is known that microglia, as resident immune cells of the CNS, express TNF and that recruited dendritic cells and macrophages will similarly induce TNF during pathogen challenge [
40,
52,
53]. Neurons, as the single most abundant cell type within the CNS, are capable of regulating immune responses during CNS insult [
54,
55]. We have reported
M. tuberculosis targeting of neurons and its capability to generate reactive immune responses [
42]; however, its role in
M. tuberculosis control remains largely untested. In the presence of pathogens, neurons induce specific cytokine signatures, TNF comprising a key component of many such signature responses. The synthesis of TNF by neurons was observed after systemic immune challenge [
56]. Upon viral infection, both in vitro and clinical studies have shown TNF expression in neurons [
57‐
60]. In other neurological diseases, upregulation of neuronal TNF expression has provided an indication to the severity of disease progression and inflammatory neuropathology [
43,
61‐
63]. In view of these findings, we postulated that neuron-derived TNF may have a contributory role to regulate CNS immunity during
M. tuberculosis infection. We therefore investigated the overall contribution of TNF and in particular the involvement of neuron-derived TNF in host-mediated immunity directed against
M. tuberculosis during central nervous system infection. Several converging lines of study/evidence have indicated a neuroprotective role for TNF and its induction demonstrated to be a key determinant of disease outcome during cerebral infection [
64]. In this study, we initially demonstrated that global TNF is critical to control
M. tuberculosis dissemination to the CNS and regulates immune-mediated brain pathology during infection. Moreover, we demonstrate for the first time that TNF is an absolute requirement to mediate protective immunity against CNS-TB; however, neuron-derived TNF is considered redundant. Thus, host susceptibility to CNS-TB in the absence of TNF corroborates its significance in neuroprotective immunity. Although the findings in this study appear contrasting to those of Tsenova et al., where TNF was reported to promote mycobacterial onset of disease and pathogenesis [
16], it is not unexpected and may rather reflect opposite ends of the TNF functionality spectrum as previously reported [
23,
24]. TNF concentration in dynamic homeostasis is absolutely essential for effective control of
M. tuberculosis infection, a shift in balance that favours either excessive or reduced levels being detrimental to the host that promotes conditions for mycobacterial persistence.
Mycobacterial infection of the CNS generates a robust immune response characterised by both pro-inflammatory and anti-inflammatory immune cellular and cytokine profiles [
65]. While microglia were reported to be the primary source of TNF during CNS infection [
65], we and others have shown that different cell types have hierarchical importance as sources of TNF in determining the outcome of tuberculosis disease [
37]. TNF derived from T cells critically defines resolution of pulmonary infection, while TNF from macrophages has a transient role to control
M. tuberculosis replication [
37]. Our findings however ascribe neurons as a redundant cellular source of TNF to control
M. tuberculosis infection or regulate immune pathology and suggest a far greater role for other resident or recruited cells to produce TNF. Interestingly, others have reported that TNF produced by neurons is redundant for mediating inflammation but that membrane TNF expressed on astrocytes has a functional role [
66].
Nonetheless, the overall requirement of TNF for control of both mycobacterial replication and immune-mediated pathology was evident and the rapid rate of host susceptibility suggested immune deficiency during early infection. Brain pathology, particularly at end-stage disease, was characterised by increased inflammation of dendritic cells and macrophages, and microglial proliferation which was underpinned by an enhanced pro-inflammatory chemokine environment in the absence of TNF. These observations align with earlier reports of a regulatory role for TNF in inflammation during
M. tuberculosis infection [
67]. In earlier studies, we reported on a deregulated overall cytokine and chemokine response in the absence of TNF during pulmonary tuberculosis [
37]. We show here a neutrophil-dominant pathology in TNF
−/− mice also associates with a similar unregulated increase in chemokine synthesis during CNS-TB. Moreover, a lethal neutrophil-driven inflammatory response associated with, amongst others, MIP-1α in the absence of microRNA 223 in pulmonary tuberculosis was reported [
68]. Taken together, the data suggest that perturbation of controlled host immune regulation which promotes a dominant neutrophilic response leads to increase susceptibility with fatal consequences. Selective inhibition of neutrophil recruitment under conditions of immune suppression during tuberculosis should therefore be considered.
Microglia, macrophages and dendritic cells as the main facilitators of early innate immune protection are activated in a TNF-dependent manner [
69,
70]. We therefore postulated that early host susceptibility of TNF
−/− mice was the result of defective activation of innate antigen-presenting immune cells during early infection. Activation of macrophages, dendritic cells and microglia within the first 2 weeks of infection appeared to be independent of TNF with equivalent levels of activated cells evident in TNF
f/f, NsTNF
−/− and TNF
−/− mice; however, the ability to subsequently sustain the required measures of activated cells was TNF dependent. These cells which are critical effectors of early innate immunity were rendered functionally defective in the absence of TNF. Initial cytokine signature responses by innate immune cells include synthesis of IL-12 and IL-1 [
71,
72], both of which are critical to control tuberculosis infection [
14,
73‐
75]. Mice deficient for either cytokine develop rapid onset of disease and succumb to infection, while patients with defective IL-12 signalling display susceptibility to tuberculosis infection. Maturation of antigen-presenting cells is characterised by the expression of, amongst others, MHCII and CD80 [
76] during
M. tuberculosis infection and is mediated largely by the presence of IL-1 and TNF [
77,
78]. Moreover, IL-12 synthesis by mature activated dendritic cells is dependent on TNF-mediated signalling [
78] and is a key cytokine required to restrict bacilli replication. Here, we observed that maturation of microglia, dendritic cells and macrophages initially occurs independent of TNF; however, sustained cell activation was TNF dependent. The deficiency in IL-12 and IL-1β synthesis in the CNS when TNF is absent argues strongly for functional failure of innate immune cells and inability to effectively participate in control of
M. tuberculosis. Moreover, the reduction of MHCII- and CD80-expressing cells indicated reduced capacity to initiate
M. tuberculosis-specific Th1 protective immune responses required for control of intracellular bacilli, and hence predisposed the host to a susceptible phenotype.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
NMF, NJH, RK, BS and NA were involved in the experiments on the intracerebral infection and the subsequent analysis. NJH and PR performed the experiments on aerosol infection and neural immunohistochemistry. DG contributed to the analysis of histopathology. VQ, BR and LK contributed to the experimental design. MJ oversaw the project. NMF, NJH and MJ prepared the figures and wrote the manuscript. All authors read and approved the final manuscript.