Natural Transmission of Plasmodium berghei Exacerbates Chronic Tuberculosis in an Experimental Co-Infection Model

Human populations are rarely exposed to one pathogen alone. Particularly in high incidence regions such as sub-Saharan Africa, concurrent infections with more than one pathogen represent a widely underappreciated public health problem. Two of the world’s most notorious killers, malaria and tuberculosis, are co-endemic in impoverished populations in the tropics. However, interactions between both infections in a co-infected individual have not been studied in detail. Both pathogens have a major impact on the lung as the prime target organ for aerogenic Mycobacterium tuberculosis and the site for one of the main complications in severe malaria, malaria-associated acute respiratory distress syndrome (MA-ARDS). In order to study the ramifications caused by both infections within the same host we established an experimental mouse model of co-infection between Mycobacterium tuberculosis and Plasmodium berghei NK65, a recently described model for MA-ARDS. Our study provides evidence that malaria-induced immune responses impair host resistance to Mycobacterium tuberculosis. Using the natural routes of infection, we observed that co-infection exacerbated chronic tuberculosis while rendering mice less refractory to Plasmodium. Co-infected animals presented with enhanced inflammatory immune responses as reflected by exacerbated leukocyte infiltrates, tissue pathology and hypercytokinemia accompanied by altered T-cell responses. Our results - demonstrating striking changes in the immune regulation by co-infection with Plasmodium and Mycobacterium - are highly relevant for the medical management of both infections in humans.


Introduction
Today, 3.3 billion people live in malaria endemic areas with sub-Saharan Africa being one of the most affected regions in the world (WHO). Plasmodium falciparum causes the most serious form of the disease with the highest rates of complications and mortality. In sub-Saharan Africa, malaria is co-endemic with tuberculosis in most regions. Despite this obvious epidemiological overlap, very little is known about interactions between the malaria parasite and the tubercle bacillus.
Tuberculosis is the most prevalent bacterial infectious disease in humans. The causative agent, Mycobacterium tuberculosis, is transmitted from infected people by aerosols and establishes infection in the lung, from where it can spread to any organ through blood or the lymphatic system. After entering the lung, M. tuberculosis is phagocytosed by alveolar macrophages, which provide a niche for its survival and replication and initiate a local inflammatory reaction, while the pathogen is transported to the draining lymph node to induce an antigen-specific T cell response [1,2]. IFN-c, the hallmark Th1 cytokine, is central in protection against tuberculosis by activation of macrophages to generate microbicidal effectors and pro-inflammatory cytokines, such as tumor necrosis factor alpha (TNF-a) which contributes to macrophage activation, granuloma formation and control of mycobacterial infection [3,4]. This initial immune response leads to protective immunity in more than 90% of the infected people, but usually fails to achieve sterile eradication of the pathogen leading to latent infection without clinical signs of tuberculosis. Reactivation can occur after years or decades leading to active tuberculosis. The risk of reactivation increases with conditions that modulate the immune status of the host such as disease, drug treatment, age, malnutrition or stress.
Together with HIV and tuberculosis, malaria comprises the triad of main infectious threats to humankind. Malaria is a vectorborne disease caused by the protozoan parasite Plasmodium and is naturally transmitted by the bite of a female Anopheles mosquito. Each year, approximately 350-500 million cases of malaria lead to the death of 1-3 million people, predominately young children and pregnant women in sub-Saharan Africa [5].
Severe malaria is often complicated by malaria-associated acute respiratory distress syndrome (MA-ARDS), characterized by pulmonary inflammation, edema and hemorrhages [6]. MA-ARDS is more common in adults than in children, with a higher prevalence in pregnant women and non-immune individuals [7]. As tuberculosis is primarily a disease of the lung, MA-ARDS may affect the course of tuberculosis in co-infected patients. However, no data on the clinical outcome of tuberculosis in people with MA-ARDS are available to date. The rodent malaria parasite Plasmodium berghei NK65 (PbNK65) has recently been identified as an experimental model for MA-ARDS [7]. Co-infection with M. tuberculosis and PbNK65 therefore is a highly relevant model to study the effects of malaria-induced pulmonary pathology on the outcome of tuberculosis.
Of note, experimental Plasmodium-Mycobacterium co-infection studies have been exclusively carried out by infecting mice with parasitized erythrocytes, giving rise to blood-stage malaria infection while excluding the clinically-silent liver-stage phase. However, the liver stage is an obligatory step during infection and relevant for anti-plasmodial immunity [8][9][10][11]. Here we report on a novel experimental model system to study malariatuberculosis co-infection in mice after challenge with both pathogens via their natural routes. Using this model, we observed exacerbated lung pathology, hypercytokinemia and dysregulated T cell responses together with significantly increased mycobacterial loads in co-infected animals, demonstrating that immunity to M. tuberculosis was severely compromised after naturally transmitted malaria infection.

Co-infected Mice are More Refractory to Plasmodium Sporozoite Infection but Exacerbate Tuberculosis
To investigate whether co-infection with PbNK65 affects the outcome of M. tuberculosis infection, C57BL/6 mice were infected via the aerosol route with 100 CFU M. tuberculosis H37Rv per lung, and 40 days later, when M. tuberculosis infection had reached the chronic phase, mice were challenged with PbNK65 sporozoites by mosquito bite. Approximately 10 sporozoite-infected mosquitoes were feeding on each mouse, which resulted in a 100% infection rate as confirmed by the presence of blood-stage parasites 4-5 days later (Table 1).
Twelve days post co-infection the mycobacterial loads in lungs and spleens were significantly increased in animals co-infected with PbNK65 ( Fig. 1 A), indicating that malaria caused reactivation of chronic tuberculosis. While co-infection promoted M. tuberculosis infection, it led at the same time to better control of PbNK65 sporozoite infection as reflected by significantly lower parasite levels in peripheral blood ( Fig. 1 B) and less body weight loss ( Fig. 1 C) as compared to animals infected with PbNK65 alone.

Severe Inflammation in Lungs of Mice Co-infected with M. tuberculosis and PbNK65
Histopathology of the lungs revealed cellular infiltration as well as hemorrhage (Fig. 2 B) in PbNK65 infected animals as described before [12]. Hence, when M. tuberculosis infected mice were coinfected with PbNK65 they presented with more extensive pulmonary leukocyte infiltrations and increased total lung weight than after M. tuberculosis infection alone ( Fig.2 A-E). Highnumbers of leukocytes  Control mice were infected with M. tuberculosis or PbNK65 alone, respectively. A) Mice were sacrificed 12 days after co-infection and serial dilutions of lung and spleen lysates were plated for CFU determination. B) Parasitemia in peripheral blood was monitored by daily Giemsastained blood smears. Note, that co-infected animals had significantly lower parasite numbers than mice infected with PbNK65 alone. Results are shown as means 6 SD (n = 10). C) Weight loss was reduced when mice were pre-infected with M. tuberculosis before PbNK65 challenge. Results are shown as means 6 SD (n = 10). Results from one representative experiment out of two independent ones are shown. Statistical analysis was performed using the Student's t test (A) or ANOVA (B and C) (*p,0.05; **p,0.01; *** p,0.001). doi:10.1371/journal.pone.0048110.g001 were observed marginating along vessel walls and infiltrating the M. tuberculosis infected lung tissue upon PbNK65 co-infection, with polymorphonuclear neutrophils and monocytes being the most abundant cell types ( Fig. 2 C lower panel, arrows and arrowheads). Alveolar reduction was much more pronounced in the presence of PbNK65 while tissue necrosis was comparable in co-infected and M. tuberculosis infected lungs (Fig. 2 D). Frequently, macrophages with abundant malaria pigment (hemozoin) were observed in PbNK65 infected lungs (Fig. 2 B lower panel, asterisks), which were however less frequent in co-infected animals reflecting the lower parasitemia levels in those animals. FACS analysis of lung leukocytes revealed an increase in CD11b + cells in co-infected compared to M. tuberculosis infected mice ( Histological analysis of the liver revealed some periportal inflammation and tissue necrosis in PbNK65 infected animals which however, was reduced in animals co-infected with M. tuberculosis ( Fig. 3 A and B). Reduced liver pathology together with a reduction in peripheral blood parasitemia ( Fig. 1 B) indicates an increased resistance to PbNK65 infection in Mtb infected animals.
Taken together, PbNK65 induced leukocyte infiltration led to exacerbated inflammation and tissue pathology in M. tuberculosis infected lungs.

Impact of Plasmodium-Mycobacterium Co-infection on T Cell Responses in Lung, Spleen and Liver
The histopathological alterations observed in co-infected mice indicate that PbNK65 modulates inflammatory immune responses to M. tuberculosis. To assess whether T cell responses were affected by malaria co-infection, lung, spleen and liver cells from co-and singleinfected animals were analyzed by FACS. Frequencies of CD4 and CD8 T cells were significantly altered upon co-infection in all three organs when compared to mice infected with M. tuberculosis alone ( Fig. 4 A). While the frequencies of CD4 T cells were, except from the spleen, significantly decreased upon co-infection with PbNK65, frequencies of CD8 T cells were significantly increased, resulting in a reversed CD4/CD8 T cell ratio.
Functional analysis of CD4 and CD8 T cells recruited to the respective organs by intracellular cytokine staining upon ex vivo restimulation revealed a shift in the balance of T cell derived TNFa/IL-10 towards IL-10 in co-infected animals ( Fig. 4 B and Figure  S1 B). Particularly in spleen and liver, the frequencies of TNFa producing CD4 and CD8 T cells were dramatically decreased while the percentage of IL-10 producers was significantly elevated compared to those of M. tuberculosis infected animals ( Fig. 4 B middle and lower panel). In the lung, CD8 but not CD4 T cells were subject to Plasmodiuminduced immunomodulation, as similar percentages of pulmonary CD4 T cells but significantly reduced frequencies of CD8 T cells produced IL-2, TNF-a or IFN-c in coinfected compared to M. tuberculosis infected mice ( Fig. 4 B upper panel). In addition, IFN-c producing CD8 T cell frequencies were increased in spleens from co-infected mice ( Fig. 4 B middle panel) and frequencies of hepatic IL-2 producing CD4 and CD8 T cells were reduced upon co-infection in comparison to those infected with M. tuberculosis only ( Fig. 4 B lower panel). Of note, overall cytokine profiles of CD4 and CD8 T cells from co-infected animals were comparable to those from animals solely infected with PbNK65, indicating that PbNK65 overwrites T-cell responses induced in mice only infected with M. tuberculosis. Alterations in leukocyte recruitment and T cell cytokine production in malaria-tuberculosis co-infected mice prompted us to investigate cytokine protein levels in tissue and serum of those animals. We found significantly elevated protein concentrations of IL-10, monocyte chemoattractant protein-1 (MCP-1), IFN-c, TNF-a and IL-6 in lungs, spleens, livers as well as sera of malaria-tuberculosis co-infected mice (Fig. 5 A-D). Highly elevated levels of cytokines primarily produced by macrophages and dendritic cells (IL-6, MCP-1, TNF-a, IL-10) in M. tuberculosis-PbNK65 co-infected animals indicate that immune regulation is significantly impaired when malaria is concurrent with tuberculosis in mice.

Discussion
Using PbNK65, a rodent Plasmodium strain that has been described as a new model for MA-ARDS in C57BL/6 mice [12], we have established a novel experimental model system to study the effect of malaria associated lung pathology on tuberculosis in the mouse. The advantage of PbNK65 over the infection of mice with PbANKA, another rodent malaria parasite known to cause pulmonary complications such as acute lung injury and ARDS [13], is the fact that PbNK65 infected C57BL/6 mice survive significantly longer compared to PbANKA infected mice.
PbANKA is the prime model for cerebral malaria [14]. Susceptible C57BL/6 mice infected with PbANKA die very rapidly from experimental cerebral malaria usually within 6-8 days ( [15] and our own observations). This time window is too narrow to detect any potential effects of malaria infection on the containment or replication of M. tuberculosis. Therefore, we found the PbNK65 infection model more appropriate to study the effect of malaria induced lung pathology on the outcome of M. tuberculosis infection. Importantly, in order to experimentally reproduce aspects of both diseases more precisely as it occurs during naturally acquired human infections, mice were challenged with both pathogens via their natural routes of infection, i.e. aerosol and mosquito bite, respectively.
In our model, PbNK65 co-infection of mice chronically infected with M. tuberculosis exacerbated tuberculosis. Upon co-infection with naturally transmitted PbNK65, mice failed to control M. tuberculosis replication and presented with more severe lung pathology as characterized by massive influx of inflammatory leukocytes including neutrophils. Although neutrophils are the most commonly infected phagocytes in patients with active tuberculosis [16], their role is controversially discussed. There is however, good evidence that neutrophils play only a minor role restricting or eliminating M. tuberculosis but rather contribute to pathology and tissue destruction [17][18][19][20]. While they seem to contribute to very early defense against mycobacteria [21][22][23][24][25], massive presence of neutrophils in M. tuberculosis infection have been associated with poor disease outcome in both, humans and mice [26][27][28][29][30]. Increased neutrophil influx together with increased total leukocyte recruitment therefore indicates loss of control of inflammation in co-infected mice.
Highly elevated cytokine levels in tissues and sera of co-infected mice further indicate impaired immune regulation. Both, M. tuberculosis and PbNK65 induce Th1 immune responses, which in concert may cause an overwhelming pro-inflammatory cytokine storm difficult to be counter-regulated similar to what has been observed in sepsis [31][32][33][34][35]. IL-10 levels are also significantly enhanced in co-infected mice like in sepsis where it is associated with a compensatory anti-inflammatory response. IL-10 is a negative regulator of Th1 responses and of central importance in immunity to malaria, where it ameliorates immunopathology at the expense of parasite elimination [36][37][38][39]. However, high levels of IL-10 can result in profound immunosuppression. It has been shown that IL-10 deactivates macrophage effector function against intracellular pathogens [40][41][42][43]. In fact, IL-10 antagonizes proinflammatory responses essential for protective immunity to M. tuberculosis [44][45][46] such as phagosome maturation [47] and IFN-c induced production of reactive nitrogen intermediates [42], which mediate the killing of M. tuberculosis [48,49]. Importantly, both macrophage and T cell derived IL-10 suppress macrophage function in mycobacterial infection [50,51]. IL-10 not only counteracts IFN-c mediated macrophage activation but also inhibits macrophage programmed cell death [52][53][54], which is an important mechanism to eliminate intracellular mycobacteria [48,55,56]. Macrophage deactivation by IL-10 has been shown to exacerbate disease [50] and to be associated with reactivation of tuberculosis in humans and mice [57][58][59]. In line with this, despite high levels of both IFN-c and TNF-a, co-infected mice were unable to control M. tuberculosis replication. Overproduction of IL-10 associated with concurrent PbNK65 infection is therefore likely to exacerbate the outcome of tuberculosis by interfering with macrophage associated bacterial clearance.
Apart from its role in macrophage activation, TNF-a is a central mediator of granuloma formation and plays a major role in the control of persistent M. tuberculosis infection [3]. However, tight regulation of TNF-a production is required to protect the host from its detrimental activities. Both, loss and overproduction of TNF-a have fatal effects on the outcome of tuberculosis, be it due to a loss of granuloma structure and excessive pathology or due to impaired macrophage activation [60,61]. TNF-a directly effects immune cell recruitment by upregulation of endothelial adhesion molecules [62] and induction of chemokine production which further recruit leukocytes to the site of infection [63,64]. Hence, overproduction of TNF-a in co-infected mice most likely contributes to exacerbated inflammation and immunopathology.
CD4 and CD8 T cell responses are significantly altered in coinfected mice when compared to those infected with M. tuberculosis alone. Of note, the profile of the T cell response in co-infected animals resembles that of animals infected with PbNK65 alone, indicating that PbNK65 infection overwrites the M. tuberculosis biased T cell responses. CD8 T cells play a major role in cellular immunity to malaria as reflected by increased frequencies of CD8 T cells in co-infected compared to M. tuberculosis infected animals. Moreover, particularly in spleen and liver, the frequencies of TNFa producing CD4 and CD8 T cells were dramatically decreased while the percentage of IL-10 producers was significantly elevated, shifting the balance of T cell derived TNF-a/IL-10 towards IL-10. Hence, Plasmodium co-infection modulated established immune responses triggered by M. tuberculosis infection with a detrimental impact on the outcome of tuberculosis.
Higher susceptibility to tuberculosis upon concurrent Plasmodium infection has been described before [65,66]. Those studies were carried out exclusively using parasitized erythrocytes for infection, and only one study used virulent M. tuberculosis, while the other one studied co-infection between Plasmodium and BCG, the tuberculosis vaccine strain. More importantly, neither P. yoelii nor P. chabaudi used in those studies cause MA-ARDS in mice as observed upon PbNK65 infection. This explains the more pronounced differences in pathology and bacterial burden observed herein when compared to other studies. We believe that our co-infection model is more relevant to the human situation as it not only mimics the natural course of malaria (by mosquito bite transmission) and tuberculosis infections, but also takes MA-ARDS into account. Of note, our study is the only one considering the fact that natural infection by mosquito bite is more efficient than simulated infection via needle infections (e.g. iv injections of malarial sporozoites) [67].
Malaria-tuberculosis co-infected mice were more resistant to PbNK65 infection as reflected by reduced parasitemia levels, liver pathology and body weight loss. Non-specific protection to Plasmodium infection by mycobacteria has been reported before  [68][69][70]. Presumably, mycobacteria-elicited pro-inflammatory responses protect mice against subsequent Plasmodium infection. One possible mechanism could be IFN-c and TNF-a mediated activation of macrophages, which contribute to parasite clearance particularly in blood and spleen of infected animals. Chronically M. tuberculosis infected mice may already carry pre-activated macrophages at the time of Plasmodium infection, which most likely accelerate clearance of malaria parasites. This is in line with the observation that the development of blood-stage parasitemia following PbNK65 sporozoite transmission was delayed in animals pre-infected with M. tuberculosis and, upon appearance of parasites in the periphery, blood parasite levels were significantly reduced compared to those in mice infected with PbNK65 alone.
Together our data suggest an overly aggressive innate response in co-infected animals, with severely increased and uncontrolled recruitment of pro-inflammatory leukocytes to the lung and hypercytokinemia. The fact that Th1 T cell responses are rather decreased in co-infected animals, with an enhanced ratio of IL-10 to IFN-c/TNF-a producing T cells, suggests that the cytokine storm is rather caused by uncontrolled cytokine secretion from innate immune cells such as macrophages.
In summary, our data demonstrate that immune responses to tuberculosis and malaria mutually influence each other during coinfection. A recent clinical study in Guinea-Bissau seems to support our data as it reported improved clinical outcome and reduced mortality among severely ill tuberculosis patients after malaria prevention had been carried out [71]. In conclusion, our data indicate that timely diagnosis of malaria-tuberculosis coinfection and MA-ARDS is critical for the clinical management of co-infected individuals and the control of exacerbated lung pathology in tuberculosis patients.

Plasmodium Life Cycle
Anopheles stephensi mosquitoes were raised at 28uC, 75% humidity under a 12-h light/12-h dark cycle and maintained on a 10% sucrose/PABA solution during adult stages. 4-5-day-old female mosquitoes were blood-fed on anaesthetized NMRI mice that had been infected with Plasmodium berghei NK65. Rodents were assayed for high levels of parasitemia and the abundance of gametocytestage parasites capable of exflagellation. After the infective blood meal, mosquitoes were maintained at 21uC, 80% humidity. On day 10 post feeding, mosquitoes were dissected in RPMI 1640 medium, and isolated midguts were examined for the infection rate. Infectious salivary gland sporozoite populations were separated as described previously [72,73].

Bacterial Strains and Culture
M. tuberculosis H37Rv was grown in Middlebrook 7H9 broth (BD Biosciences) supplemented with 0.5% glycerol and asparagine (1 g/l), and OADC (Oleic acid, Albumin, Dextrose, Catalase) enrichment medium (BD). Bacterial cultures were harvested and aliquots were frozen at -80uC until later use. Viable cell counts in thawed aliquots were determined by plating serial dilutions of cultures onto Middlebrook 7H11 agar plates followed by incubation at 37uC.

Mice, Infection and Colony Forming Units (CFUs)
For all experiments female C57BL/6 mice aged between 6-8 weeks were used, which were obtained from Charles River Laboratories. Mice were maintained under specific barrier conditions in BSL 3 facilities.
For infection of experimental animals, Mycobacterium tuberculosis stocks were diluted in sterile distilled water/1% v/v Tween-80/ 1% w/v albumin at a concentration providing an uptake of 100 viable bacilli per lung. Infection was performed via the respiratory route by using an aerosol chamber (Glas-Col, Terre-Haute, IN, USA). Animals were exposed for 60 min to an aerosol generated by nebulising the prepared M. tuberculosis suspension. The inoculum size was quantified 24 h after infection by determining bacterial loads in the lungs of infected mice.
Bacterial loads in lung, spleen and liver were evaluated at different time points after aerosol infection by mechanical disruption of the organs in 0,05% v/v Tween 20 in PBS containing a proteinase inhibitor cocktail (Roche Diagnostics) prepared according to the manufacturer's instructions. Tenfold serial dilutions of organ homogenates in sterile water/1% v/v Tween 80/1% w/v albumin were plated onto Middlebrook 7H11 agar plates and incubated at 37uC. Colonies were enumerated after 3-4 weeks.

Natural Malaria Parasite Infection Experiments
Naïve mice or animals pre-infected for 40 days with M tuberculosis were infected with Plasmodium berghei NK65 by mosquito bite. Mice were exposed for 15 min to highly infected mosquitoes that contained an average of 30,000 wild-type salivary gland sporozoites each. Successful blood feeding was confirmed by mosquito dissection after the co-infection experiment. Parasitemia (occurrence of intraerythrocytic stages) was monitored by daily Giemsa stained blood smears.

Cell Isolation and Purification from Lungs, Spleens and Livers
Lungs were removed, sectioned and incubated in PBS containing liberase (1,5 mg/ml; Worthington Biochemical) and DNAse I (250 mg/ml; Worthington Biochemical) at 37uC for 90 min. Digested lung tissue was broken into single cell suspension by subsequent passage through a 100 mm pore size nylon cell strainer. Spleens and livers were also passed through a 100 mm pore size nylon cell strainer to obtain single cell suspensions. Liver infiltrating lymphocytes were isolated using a Percoll Gradient. Remaining erythrocytes were lysed and cells were resuspended in RPMI 1640 medium supplemented with 2 mM glutamine, 1% Hepes (v/v), b-mercaptoethanol (50 mM), and 10% v/v heatinactivated fetal calf serum (complete RPMI 1640 medium).

Multiplex Cytokine Assays
The concentrations of IL-6, IL-10, MCP-1, IFN-c, and TNFa in lung, spleen and liver homogenates and serum were determined by cytometric bead array (Mouse Inflammation Kit, BD Biosciences) according to the manufacturer's protocol.

Histology
Organs from infected mice were fixed with 4% w/v PFA for 24 h and embedded in paraffin. Sections (3 mm) were rehydrated by running through xylenes, alcohols of decreasing concentrations and finally water. Sections were stained with hematoxylin and eosin (H&E) and analyzed with an Axioplan 2 light microscope and Volocity software (Perkin Elmer).
Histological sections of infected lungs were scored in a blind manner using the following pathological scores (scores are given in parentheses after each category):

Statistical Analysis
Statistical analysis was performed using ANOVA compared with Tukeỳs or Bonferroni post-testing. The unpaired Student's t test was employed to compare bacterial titres. Figure S1 Flow cytometric analysis. C57BL/6 mice were infected by aerosol with M. tuberculosis H37Rv (100 CFU/lung) and challenged with PbNK65 infectious sporozoites by mosquito bite 40 days later. Control mice were infected with M. tuberculosis or PbNK65 alone, respectively. A) 12 days upon PbNK65 infection, lung leukocytes were analyzed for surface expression of CD11b and GR-1. Representative cell-frequency dot plots of electronically gated CD90.2 negative lung cells stained with anti-CD11b and anti-GR1 are shown, in which the numbers represent the frequencies as percentages. B) T cells in lung, spleen and liver were analyzed for the production of IL-2, TNF-a, IL-10 and IFNc. Representative dot plots of electronically gated CD90.2 + CD8 + (left dot plots) or CD90.2 + CD4 + (right dot plots) spleen cells are shown. Numbers represent the frequency of the cells as percentage.