Background
The most severe form of
Mycobacterium tuberculosis (
M.tb) infection is central nervous system tuberculosis (CNS-TB) which has high mortality and serious long-term neurological sequelae even with effective anti-tuberculous treatment [
1‐
3]. Common manifestations of human CNS-TB are tuberculous meningitis (TBM), tuberculomas and tuberculous brain abscesses [
4]. Cerebral vasculitis and inflammation resulting in infarcts is the primary cause of permanent brain tissue damage in TBM and is among the worst consequences of CNS-TB [
5,
6]. Despite effective TB treatment with antibiotics and adjunctive corticosteroids, CNS-TB remains one of the more challenging clinical syndromes to manage.
To advance our understanding of CNS-TB, we need an appropriate animal model that recapitulates the neurobehavioral, immunopathological and histopathological changes in human CNS-TB to dissect pathogenesis and aid drug discovery. Several animal models of CNS-TB have been described, including guinea pigs, rabbits, mice, pigs, and zebrafish. The rabbit model closely mimics human disease, developing clinical and histological changes [
7‐
13]. However, a number of immunological tools profiling protein secretion and gene expression are unavailable for rabbits [
1] and, therefore, preclude their suitability for in-depth immunological studies.
The mouse model has many advantages over other animals, including the availability of genetic and molecular tools as well as cost-effectiveness for large studies. However, existing murine CNS-TB models do not display the clinical features and immunological phenotypes of CNS-TB observed in humans. C57BL/6 mice are generally resistant to CNS-TB infection, with no pathological abnormalities detected and no observed mortality over 24 weeks of study [
14]. BALB/c mice infected through the intracerebral route directly into the brain parenchyma with
Mycobacterium bovis BCG (BCG) had infiltration of inflammatory cells, but no granulomas were observed [
10]. This contrasts with human CNS-TB, where tuberculomas occur in approximately 30% of TBM patients [
15]. Intravenous inoculation of BALB/c mice with
M.tb strain CDC1551 successfully infected the CNS but did not produce granulomas in the brain and had low expression of brain chemokines and cytokines IL-1β, IL-6, TNF-α and IFN-γ, in contrast to the increased expression of these cytokines in the cerebrospinal fluid (CSF) of human TBM patients [
16,
17]. While some murine CNS-TB models have meningitis and/or brain granulomas, they do not demonstrate neurological signs of disease and mortality, unlike human CNS-TB [
14,
18]. Given the varying susceptibility and pathology of CNS-TB infection in different mouse strains, genetic predisposition is likely to play a crucial role. C3HeB/FeJ “Kramnik” mice were found to be hyper susceptible to
M.tb infection due to the presence of an allele, termed the “super susceptibility to tuberculosis 1” (
sst1) locus, and developed a more human-like lung pathology in contrast to C57BL/6 mice [
19,
20]. However, the ability of C3HeB/FeJ mice to develop CNS-TB remains to be explored.
Intracerebral-infection with
M.tb H37Rv directly into the brain parenchyma of inducible nitric oxide synthase (iNOS)-knockout mice resulted in neurological signs with meningitis, and mice exhibited 63% mortality post-infection (p.i.) [
21]. However, the development of intracerebral tuberculomas and immunological profile were not phenotyped in this mouse model. Cytokine-induced upregulation of iNOS or NOS2 by murine macrophages have been implicated in the killing of intracellular pathogens, such as
M.tb, but the role of this antimicrobial system in human macrophages remains unclear [
22,
23]. Studies have shown that activated human microglia, the brain resident macrophages, do not express iNOS [
24,
25] or reactive nitrogen intermediate (RNI) nitric oxide (NO) [
26], whereas murine microglia produced substantial amounts of NO on activation [
26]. Given the well-established role of macrophages in TB, the inter-species difference in microglia expression of iNOS may explain the species tropism barrier to the development of CNS-TB in mice.
To address the limitations of existing murine CNS-TB models, we explored the effects of mouse strains, M.tb strains and routes of infection on the development of CNS-TB disease. First, we compared two mouse strains, C3HeB/FeJ and Nos2−/− mice, to investigate whether the sst1 locus or Nos2 gene plays a more important role in CNS-TB infection. After selecting the suitable mouse strain, we investigated the effects of two different M.tb strains, H37Rv and CDC1551, and two routes of infection: intra-cerebroventricular (i.c.vent.) into the third ventricle and intravenous (i.v.), on the development of a murine CNS-TB model with human-like pathology. The i.c.vent. route of infection mimics the rupture of meningeal tuberculous lesions and the subsequent release of M.tb into the CSF, whereas the i.v. route mimics the hematogenous spread of M.tb. In this study, we showed that i.c.vent. infection of Nos2−/− mice with M.tb H37Rv developed the severe neurological symptoms and induced a high expression of adhesion molecules, chemokines, and inflammatory cytokines in the brain, consistent with the infiltration of inflammatory cells and pathological changes. This pre-clinical model can be used to understand the mechanisms in host immunopathology and evaluate treatment for CNS-TB.
Methods
Human CNS-TB brain specimen processing
The paraffin blocks of three surgical samples from patients with histological features indicative of CNS-TB infection were retrieved from the files of the Department of Pathology at Tan Tock Seng Hospital, Singapore. The specimens included leptomeninges and brain parenchyma, and demonstrated granulomatous inflammation typical of CNS-TB. Acid-fast bacilli were demonstrated on Ziehl–Neelsen histochemical stain in two out of three samples. Control brain sections were from the non-neoplastic brain parenchyma of three surgical pathology brain resection specimens. 4 µm thick sections were cut from each block for H&E staining according to the manufacturer’s instructions.
Bacterial strains and growth conditions for infection
Mycobacterium tuberculosis (M.tb) strains H37Rv and CDC1551 were kindly provided by Professor Nick Paton and Associate Professor Sylvie Alonso (both NUS, Singapore), respectively. For infection experiments, a frozen vial of M.tb was thawed and cultured to mid-logarithmic phase at an optical density of 0.6–0.8. Prior to infection, the M.tb was centrifuged at 3200g for 10 min and resuspended in 1 mL sterile 0.9% NaCl. The M.tb inoculum was then plated to determine the dose of infection.
Mouse cannula implantation and infection
Six-to-eight-week-old male C57BL/6
Nos2−/− (Stock No. 002609) and C3HeB/FeJ (Stock No. 000658) mice (Jackson Laboratory, Bar Harbor, Maine) were used for intra-cerebroventricular (i.c.vent.) or intravenous (i.v.) infection. Mice in the i.c.vent. group were cannulated 1 week before infection. An illustration of the stereotaxic coordinates of site of injection and the positioning of guide cannula is shown in Additional file
2: Fig. S1a. A motorized stereotaxic instrument (Neurostar, Tübingen, Germany) was used to implant a 26-gauge guide cannula (RWD, Shenzhen, China) into the third ventricle (coordinates from the bregma: − 1.6 mm posterior, 0 mm lateral, − 2.5 mm ventral). The same coordinates were used for both C57BL/6
Nos2−/− and C3HeB/FeJ Kramnik mice as the size of mice were similar at the time of cannulation.
Nos2−/− mice were 23.5 g (± 1.1) (mean ± s.d.) and C3HeB/FeJ Kramnik mice were 24.7 g (± 1.6) (
p = NS). Mice were injected with 0.5 µL of sterile 0.9% NaCl or
\(2\times {10}^{8}\) CFU/mL
M.tb through the brain cannula (over 5 min) using the syringe pump (Harvard Apparatus, Holliston, Massachusetts). Mice in the i.v. group were injected with 50 µL of sterile 0.9% NaCl or
\(2\times {10}^{6}\) CFU/mL
M.tb via the retro-orbital sinus.
M.tb was administered at a dose of 10
5 CFU to each animal, irrespective of the route of infection. This dose was chosen as previous CNS-TB murine models have administered
M.tb within the range of 10
5 to 10
6 CFU [
14,
16,
21,
27]. However, different infection routes have different recommended administration volumes (0.5 µL for i.c.vent. and 50 µL for i.v.) and the concentration of
M.tb for i.c.vent. route was 100-fold more concentrated than the i.v. route. All mice were observed for mortality and weight change. Humane endpoints included ≥ 20% weight loss, complete hind limb paralysis and repeated seizures. Infected mice were also monitored daily for 56 days after infection for clinical signs indicative of CNS-TB, such as limb weakness, tremors, and twitches.
30 µL of trypan blue was administered into four cannulated
Nos2−/− mice and the brains harvested 15 min post-administration to allow for distribution of the dye in both right and left cerebral hemispheres. A sagittal illustration of the ventricular system in the mouse brain, which include the lateral ventricles, third ventricle and aqueduct that leads to the fourth ventricle, is depicted in Additional file
2: Fig. S1b. Coronal sections of each brain verifies that the dye is in the ventricular system (Additional file
2: Fig. S1c), indicating successful brain cannulation into the third ventricle. Given the similar sizes of both strains of mice at cannulation, trypan blue was not instilled into the ventricles of the C3HeB/FeJ Kramnik mice, but H&E staining of i.c.vent.-infected Kramnik mice showed more marked meningeal inflammation than the brain parenchyma (Additional file
2: Fig. S1e), indicating the accurate placement of the cannula into the cerebral ventricles.
Nos2−/− or C3HeB/FeJ mice were infected with
M.tb 7 days after brain cannulation, and the blood, brain, lungs, liver and spleen were harvested 56 day post-infection (p.i.) for enumeration of mycobacterial load, histopathological analysis and immunological marker analysis (Additional file
2: Fig. S1d).
Neurobehavioral scoring
Neurobehavioral scoring was performed by a researcher (P.X.Y.) blinded to the route of infection and strain of
M.tb according to a scoring list for CNS-TB mouse model (Table
1), adapted from Tucker et al. [
12]. Each scoring parameter ranged from one, corresponding to no abnormalities, to a variable maximum score. The minimum total score is 3 indicating a normal mouse. Higher neurological scores reflect an increasing severity of neurological deficits with a maximum total score of 7.
Table 1
Composite neurobehavioral score criteria for CNS-TB mouse model
Tremors | |
Absent | 1 |
Present | 2 |
Twitch/jerk | |
Absent | 1 |
Mild (< 3 in 10 s) | 2 |
Severe (≥ 3 in 10 s) | 3 |
Eyes | |
Normal | 1 |
Closed eyelids | 2 |
Organ harvesting and processing
Eight week post-infection, mice were deeply anesthetized before cardiac puncture was performed for blood collection. The brain, lungs, liver and spleen were harvested and the gross pathology examined before tissue processing. Half of each organ was fixed in 10% neutral buffered formalin for histology, while the other half was homogenized for bacterial enumeration and characterization of immunological markers. Organs were homogenized by high-speed shaking in 2 mL microcentrifuge tubes filled with sterile PBS and 5/7 mm stainless steel beads using TissueLyser LT (Qiagen, Hilden, Germany).
Histopathological analysis
Histopathology was performed on the left hemisphere of the murine brain. The murine brain was fixed in 10% neutral buffered formalin, paraffin embedded and sectioned to 4 µm thickness. Brain slices were stained with hematoxylin–eosin (H&E) (Vector Laboratories, Burlingame, California) to characterise pathological lesions and Ziehl–Neelsen staining (Sigma-Aldrich, St. Louis, Missouri) to detect mycobacterium according to manufacturer’s instructions. Histopathological examination was carried out in a blinded fashion by a histopathologist (R.R.) based on the presence of pathological changes including inflammation, perivascular cuffing, gliosis, neuronal necrosis, granuloma, pyogranuloma and necrosis. Definition of granulomatous lesions in this study includes both granulomas and pyogranulomas. Grading of severity was assigned on the following scale: 0: no abnormalities detected; 1—minimal; 2—mild; 3—moderate; 4—marked and 5—severe. The total number and area of granulomatous lesions were measured from 6 different sections of 5–6 mice. To quantify the area of granuloma, we utilized the free-hand tool in ImageJ (NIH, Bethesda, Maryland) and manually demarcated the granuloma as a region of interest for area measurement.
Immunological marker analysis
Adhesion molecules, cytokines and chemokines were analysed by Fluorokine multianalyte profiling kit according to the manufacturer’s protocol (R&D Systems, Minneapolis, Minnesota) on the Bio-Plex 200 platform (Bio-Rad, Hercules, California). The minimum detection limit for the ICAM-1 and p-selectin were 52.7 pg/ml and 2.6 pg/ml, respectively. The minimum detection limit for the cytokines and chemokines were CCL-2/MCP-1 134 pg/ml, CCL-3/MIP-1α 0.452 pg/ml, CCL-4/ MIP-1β 77.4 pg/ml, CCL-5/ RANTES 19.1 pg/ml, CCL-7/ MCP-3 1.69 pg/ml, CCL-8/ MCP-2 0.283 pg/ml, CCL-11/Eotaxin 1.46 pg/m, CCL-12/ MCP-5 0.613 pg/ml, CCL-19/ MIP-3β 2.28 pg/ml, CCL-20/ MIP-3α 3.95 pg/ml, CCL-22/ MDC 0.965 pg/ml, CXCL-1/ KC 32.9 pg/ml, CXCL-2/ MIP-2 1.97 pg/ml, CXCL-10/ IP-10 6.85 pg/ml, CXCL-13/ BLC 19.3 pg/ml, IL-1α 8.17 pg/ml, IL-1β 41.8 pg/ml, IL-6 2.30 pg/ml, IL-12 p70 12.8 pg/ml, IL-17A 7.08 pg/ml, IL-27 1.84 pg/ml, LIX 2.02 pg/ml, TNF-α 1.47 pg/ml, IFN-γ 1.85 pg/ml. Brain homogenates were assayed at neat for all analytes and results were normalised to their total protein concentrations (Bio-Rad, Hercules, California).
Statistics
Continuous variables are presented as medians and interquartile range. Neurobehavior scores and body weight change between groups were compared using two-way ANOVA with post-hoc Tukey’s multiple comparisons test. Levels of adhesion molecules, cytokines and chemokines, and CFU counts between groups were compared using Mann–Whitney test. Comparison of survival curves between groups was calculated using the log-rank test. A two-sided value of p < 0.05 was considered significant. All analyses were performed using GraphPad Prism version 7.05 (Graphpad, San Diego, California).
Discussion
Human CNS TB is severe and progress is limited by a lack of good animal model systems that reflect immunopathology in human CNS TB. Our study determined the effects of mouse strain,
M.tb strain and route of infection on the development of a murine CNS-TB model with human-like pathology. Here, we show that i.c.vent. infection of
Nos2−/− mice with
M.tb H37Rv makes a CNS-TB model that shares similar clinical features of human CNS-TB, including neurological morbidity, high mortality, and increased CNS expression of inflammatory mediators. Importantly, our model demonstrated histological evidence of parenchymal granulomas in the cerebral cortex, hippocampus and the presence of necrotizing granulomas similar to human CNS-TB tuberculomas [
31,
32]. The presence of a central area of liquefactive necrosis in pyogranulomas of H37Rv i.c.vent.-infected mice resembled human caseating tuberculomas with central liquefaction, a clinical feature that has not yet been replicated in existing murine CNS-TB models. Other features of human CNS-TB include perivascular infiltration with immune cells and a microglial reaction [
31,
33]. Similar to that observed in humans, our CNS-TB model showed the presence of gliosis and perivascular cuffing throughout the brain parenchyma.
We evaluate the simultaneous expression of adhesion molecules, chemokines, and cytokines in an attempt to elucidate the mechanism underlying the chronic inflammatory state in human CNS-TB. While several clinical studies have unanimously demonstrated an increased CSF expression of inflammatory cytokines TNF-α, IFN-γ, IL-1β and IL-6 in TBM patients [
17,
29,
30], current murine CNS-TB models have failed to recapitulate this immunological profile [
14,
16]. Through immunological analysis, we showed that H37Rv i.c.vent.-infected
Nos2−/− mice had significantly increased expression of TNF-α, IFN-γ, IL-1β and IL-6, similar to human TBM patients [
29,
30], indicating that our pre-clinical model mirrors human CNS-TB. In addition, we demonstrated H37Rv i.c.vent.-infected mice exhibited upregulation of adhesion molecules p-selectin and ICAM-1 compared to saline controls, in keeping with the increased leukocyte infiltration in the brain and extends previous in vitro observations [
34,
35], and that
M.tb increases expression of endothelial adhesion molecules in a co-culture BBB model [
36].
While i.c.vent. infection of
Nos2−/− mice with either
M.tb H37Rv or CDC1551 resulted in a high mortality (67% and 60%, respectively), similar to human CNS-TB [
37,
38], H37Rv is superior to CDC1551 as the murine CNS-TB model for two reasons. First, H37Rv infection resulted in the development of more severe neurological deficits with a worse neurobehavioral score and earlier mortality than CDC1551 infection, which reflected the neurological morbidity and severity of disease in human CNS-TB [
39,
40]. Second, H37Rv-infected mice showed an increased severity of histopathological lesions than CDC1551-infected mice, demonstrated by the greater extent of pyogranulomas and liquefactive necrosis in H37Rv i.c.vent. mice, extending from the cerebral cortex to the hippocampus which were not observed in CDC1551-infected mice, but similar to human CNS-TB histology [
28]. This is consistent with previous findings, where H37Rv is more virulent than CDC1551 in animal models of pulmonary TB both in rabbits [
41] and in mice [
42]. Although the mycobacterial load in the brain of H37Rv-infected mice had a trend to increase compared to CDC1551-infected mice, this did not reach statistical significance. A repeat experiment with a lower dose of infection, and a longer experiment with more timepoints may help to further characterize this CNS-TB model.
Previous murine CNS-TB models have employed direct injection into the brain parenchyma to induce CNS infection [
10,
18,
21], which resulted either in granulomas being restricted to the injection site with no widespread inflammation or the absence of granulomas. Thus, to better mimic the rupture of the Rich foci in human CNS-TB, with the subsequent release of
M.tb into the CSF to induce TBM [
1], we inoculated
M.tb into the third ventricle for meningeal infection. To prevent surgery-related loss of mice due to excessive bleeding or hemorrhage, we injected
M.tb at an angle into the third ventricle to avoid puncturing the superior sagittal sinus. In addition to the direct CNS inoculation of
M.tb via the i.c.vent. route, we also explored the i.v. route to mimic the natural course of hematogenous spread from the lung to the brain in human CNS-TB [
43]. However, we found the i.v. route of infection to be less suited for our murine CNS-TB model, as the mice exhibited a widespread disseminated infection resembling miliary TB, with granulomas observed in multiple organs of the lungs, spleen, heart, and kidneys, but not typical brain lesions. Dissemination of
M.tb to the heart of H37Rv i.v. mice may explain the early and uniform lethality with mortality of these mice by day 30 p.i.
Different mouse strains have different susceptibilities to
M.tb infection, which may explain the varying degree of disease and brain histopathology in existing murine CNS-TB models. To investigate whether the C3HeB/FeJ mice, which are hypersusceptible to pulmonary TB infection [
19,
20], or the
Nos2−/− mice, which have altered innate immune response, are more susceptible to CNS-TB infection, we evaluated the C3HeB/FeJ and
Nos2−/− mouse strains for our murine CNS-TB model.
M.tb-infected
Nos2−/− mice exhibited worse neurobehavioral score than C3HeB/FeJ mice and developed neurological symptoms such as myoclonic jerks and limb weakness that resembled seizures and hemiparesis, respectively, in human CNS-TB patients [
1]. In addition, infected
Nos2−/− mice demonstrated greater inflammatory cell infiltrates, higher expression of adhesion molecules and chemokines in the brain than C3HeB/FeJ mice. Although there was trend to lower mycobacterial load in the C3HeB/FeJ mice, these infected mice expressed similar level of adhesion molecules and chemokines in the brain to saline controls, indicating that the CNS response to infection in the C3HeB/FeJ mice was minimal. These findings show that
Nos2−/− mice is a better CNS-TB model than C3HeB/FeJ mice, and underscores the role of
Nos2-induced NO production in inhibiting
M.tb growth in mice [
44]. The presence of abundant neutrophilic infiltrates in the brain of
Nos2−/− mice may be due to the inability of
Nos2−/− macrophages to contain the
M.tb infection, as NOS2 upregulation by murine macrophages has been shown to be implicated in
M.tb killing [
22], which may result in the activation and recruitment of more leukocytes to the site of infection [
28]. This may explain the greater extent of brain granulomatous inflammation seen in
Nos2−/− mice compared to C3HeB/FeJ mice. Nevertheless, the role of NOS2 in
M.tb killing remains controversial as human microglia do not express NOS2 [
24,
25] unlike murine microglia [
26]. Thus
Nos2−/− mice may mimic the lack of NOS2 response in human and recapitulate human CNS-TB disease. Future studies investigating
M.tb killing and cytokine and chemokine production by bone marrow-derived macrophages and neutrophils from
Nos2−/− mice are needed to gain insight into the function and mechanism of NOS2 gene in CNS-TB pathogenesis.
Our study has limitations including the small sample size comparing C3HeB/FeJ and
Nos2−/− mouse strains in establishing the murine CNS-TB model and the use of only male mice for the study. Nevertheless, our pilot study is useful for formal sample size calculation for future studies. Findings of the CDC1551-infected
Nos2−/− mice by the i.c.vent route were successfully reproduced in a separate experiment evaluating the
M.tb strain and route of infection for the CNS-TB model. Male mice were used as there is a male predominance in TB [
45], and literature has shown males to be more susceptible to mycobacterial infection [
46,
47]. Future studies characterising the responses in both genders of mice would be required for application of research findings [
48].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.