Potential of immunomodulatory agents as adjunct host-directed therapies for multidrug-resistant tuberculosis

Treatment of multidrug-resistant tuberculosis (MDR-TB) is extremely challenging due to the virulence of the etiologic strains of Mycobacterium tuberculosis (M. tb), the aberrant host immune responses and the diminishing treatment options with TB drugs. New treatment regimens incorporating therapeutics targeting both M. tb and host factors are urgently needed to improve the clinical management outcomes of MDR-TB. Host-directed therapies (HDT) could avert destructive tuberculous lung pathology, facilitate eradication of M. tb, improve survival and prevent long-term functional disability. In this review we (1) discuss the use of HDT for cancer and other infections, drawing parallels and the precedent they set for MDR-TB treatment, (2) highlight preclinical studies of pharmacological agents commonly used in clinical practice which have HDT potential, and (3) outline developments in cellular therapy to promote clinically beneficial immunomodulation to improve treatment outcomes in patients with pulmonary MDR-TB. The use of HDTs as adjuncts to MDR-TB therapy requires urgent evaluation.


Background
The World Health Organization (WHO) presently ranks tuberculosis (TB) as the leading cause of death due to infectious disease worldwide, with an estimated 1.5 million deaths from TB in 2014 [1]. Multidrug-resistant TB (MDR-TB) accounted for more than 3 % of newly diagnosed TB cases, and over 10 % (190,000) of all TB-related deaths [1]. Further, 12 % of individuals diagnosed with TB were also co-infected with HIV, comprising more than 20 % (390,000) of deaths due to TB in 2014 [1]. Current WHO guidelines for MDR-TB treatment recommend the use of combination TB drug therapies that target Mycobacterium tuberculosis (M. tb). These drugs, several of which have serious side effects, have to be taken for a period of 12-18 months, with the long duration resulting in poor patient compliance. Furthermore, current MDR-TB treatment regimens cure only 50 % of MDR-TB cases and impose a huge financial toll on national health services [1,2]. Other important limitations of current MDR-TB treatment include drug interactions and insufficient drug concentrations in patients' lungs [3] as well as aberrant or excess host inflammatory responses to M. tb, resulting in lung tissue destruction, compromising the body's ability to eradicate the M. tb bacilli, and leading to permanent lung tissue damage [4][5][6].
Pulmonary pathology observed in patients with all forms of TB can be attributed to intrinsic exaggerated inflammatory responses to M. tb [4]. Since HIV coinfection is the major comorbidity associated with poor clinical outcome in patients with active TB, early restoration of immune cells in HIV-TB patients following antiretroviral therapy (ART) can result in the development of immune reconstitution inflammatory syndrome (TB-IRIS) [5]. In TB-IRIS, existing M. tb-specific T cells proliferate uncontrollably, consequently perpetrating local and systemic overt immune reactions and potentially leading to death [5]. An exaggerated release of pro-inflammatory cytokines, such as tumour necrosis factor alpha (TNF-α), interleukin 6 (IL-6), IL-8, IL-18, and interferons alpha and beta (IFN-α/β), invariably perpetrate tissue damage in active TB as well as TB-IRIS [6,7]. Although TB-IRIS is more closely associated with drug-susceptible TB patients starting on ART, it has been reported in South African patients with MDR-TB and HIV co-infection [8]. An exigent need therefore exists to develop new strategies to treat and prevent harmful inflammatory responses in MDR/extensively drug resistant-TB patients to avoid long-term, if not permanent, lung damage [9][10][11][12].
Considering these limitations and intricate clinical scenarios in TB, host-directed therapies (HDTs) have emerged as a new and promising intervention avenue against all forms (drug-sensitive and drug-resistant) of the disease [13]. HDTs generally target clinically relevant biological pathways in the host to modulate and rectify pathological immune responses. In TB, HDTs may neutralise excessive inflammation in organs and decrease M. tb proliferation while facilitating tissue repair [7,10].

Therapeutic targeting of immunological and cellular processes in MDR-TB
Over the past several years, extensive preclinical and clinical studies have revealed a myriad of immunological pathways and biomarkers that either influence the outcome of M. tb infection or indicate the state of disease, respectively [6,[14][15][16][17][18]. These pathways include cytokinemediated signalling, intracellular antimicrobial processes, and establishment of long-term immunological memory as well as physiological homeostasis [13]. The use of immunomodulatory agents as HDTs for TB treatment has therefore garnered great interest as a means of improving clinical outcome of therapy, based on previous experience with other infectious diseases and cancer. Many of these agents are already licensed for non-TB indications, thus bearing US Food and Drug Administration approval, while others are in advanced clinical trials [18,19]. A summary of the various HDT agents with therapeutic potential in pulmonary MDR-TB is presented in Table 1 and Fig. 1.

Immunomodulatory small-molecule drugs
Several non-antibiotic drugs with immunomodulatory properties have already shown clinical efficacy in ameliorating infection-induced pathology. For example, interventional use of glucocorticoid receptor agonists, i.e., prednisone and dexamethasone, has mortality-reducing effects in community-acquired pneumonia, concomitant with lowering of pro-inflammatory cytokine release [20,21]. However, the clinical outcome of prednisone therapy in pulmonary TB has been inconclusive; further studies are required to ascertain a clear beneficial effect in patients [22].
Eicosanoid regulation dependent on arachidonic acid (ARA) metabolism has been associated with modulation of clinically relevant immune responses and pathology in patients with active TB [23]. Two major arms of enzymatic reactions carry out ARA catabolism: the first involving cyclooxygenases 1 and 2 (COX1 and 2) and the other, lipoxygenases (LO). Breakdown of ARA by COX enzymes progresses to prostaglandin synthesis, while LO metabolise ARA to produce leukotrienes, lipid mediators responsible for fever induction during active infection [24]. COX2 is particularly associated with regulation of pro-inflammatory cytokine production and inflammation via prostaglandin E2 (PGE2) production [25]. Leukotrienes are also involved in initiation of inflammation, angiogenesis and progression of cancer [25].
It has been shown in TB that balance between concentrations of lipoxin A4 (LXA4) and leukotrienne B4 controls subsequent TNF-α release by activated immune cells in the lung [23]. Overproduction of LXA4 may instead cause immunosuppression, highlighting that maintenance of this axis in TB is pivotal in determining patient survival or death. The use of aspirin (acetylsalicylic acid), a nonsteroidal anti-inflammatory drug (NSAID) inhibits COX2 activity, thus enhancing LXA4 synthesis, which in turn would decrease TNF-α levels and subsequently neutralise pulmonary inflammation in TB patients [23,26]. The same may be achieved with ibuprofen (another NSAID), which has been shown to improve survival of experimental mice with severe lung TB by dampening tissue pathology while reducing M. tb load [27]. Ibuprofen also has direct anti-M. tb activity [28].
Zileuton is an anti-inflammatory drug currently licensed for the treatment of asthma that exerts its effects by inhibiting 5-LO activity [29]. As opposed to NSAIDs, zileuton blocks LO-mediated leukotriene production, facilitating COX2 activation and PGE2 synthesis. In a mouse model of TB, zileuton was shown to dramatically abate IFN-α/β levels in the lung to improve lung pathology, in addition to reducing M. tb load [30]. Combination therapy comprising zileuton and PGE2 augmented interleukin 1 beta (IL-1β)-mediated control of M. tb growth. Furthermore, patients with latent TB infection who successfully contained infection and did not develop active pulmonary TB appear to maintain balanced serum levels of PGE2 and LXA4 [30].
In addition to their cholesterol-lowering properties, statins also possess anti-inflammatory properties, which enhance their efficacy in the clinical management of atherosclerosis [31]. Statins can reduce IFN-γ-mediated MHC-II upregulation, and thus CD4 T cell activationhaving clinical benefits in cardiac transplantation [32]. Statins may also downregulate TNF-α and IL-1β release via activation of the transcription factor Kruppel-like factor 2 (KLF2), as well as reduce IL-6, IL-8 and CCL5    secretion by myeloid cells and lymphocytes [31]. Prior use of statins has been shown to ameliorate inflammationinduced organ damage, thus reducing mortality among patients with sepsis and community-acquired pneumonia [33][34][35][36]. The therapeutic potential of statins has been evaluated in TB. Peripheral blood mononuclear cells derived from patients with familial hypercholesterolemia undergoing atorvastatin therapy were found to be more resistant to M. tb infection and replication [37]. It was established in a mouse model of TB that the cholesterol-lowering properties of atorvastatin potentiated this effect, where simvastatin treatment was concomitant with reduced pulmonary, hepatic and splenic M. tb burden and tissue pathology [37]. It was also shown that simvastatin potentiates autophagic cell death of M. tb-infected macrophages, thus having a dual approach in controlling TB disease. In a separate study, coadministration of simvastatin to an anti-TB chemotherapy regimen combining rifampicin, isoniazid and pyrazinamide led to improved bactericidal activity of the drugs against intracellular M. tb as well as in a mouse model of TB [38]. Clinically, patients with type 2 diabetes mellitus (T2DM) on statin therapy (as well as anti-diabetic therapy) seem to be at lower risk of developing active TB [39]. Metformin (MET) is an essential drug used in the treatment of T2DM, which acts by affecting the mitochondrial respiratory chain [40]. In a recent preclinical study, MET was evaluated as an adjunct therapeutic for drug-susceptible (DS)-TB and MDR-TB [41]. MET treatment induced superoxide anion production in macrophages infected with DS or MDR M. tb in anadenosine monophosphate-activated protein kinase (AMPK)-dependent manner [41]. Furthermore, MET therapy potentiated the activity of the first-line anti-TB drugs isoniazid and ethionamide in a mouse model of TB, commensurate with improved anti-TB CD8 T cell responses, reduced concentrations of inflammatory cytokines and lung pathology [41]. A retrospective evaluation of patients with T2DM receiving MET therapy showed that they are less prone to develop cavitary pulmonary TB. In addition, T cell non-reactivity to ESAT-6 in a T-SPOT IFN-γ assay also suggested less incidence of latent TB infection among these individuals [41]. The ability of MET to promote maintenance of memory CD8 T cells via AMPK-driven fatty acid oxidation, as seen in mouse models of solid tumours [42,43], may be an additional avenue for its potent immunomodulatory properties in TB.
Imatinib mesylate, a highly specific and successful drug that inhibits the Bcr-Abl tyrosine kinase in chronic myelogenous leukaemia has also been evaluated in TB [44]; treatment of M. marinum-infected mice with imatinib resulted in enhanced myelopoiesis in the bone marrow. Furthermore, an early influx of neutrophils into the lungs of infected mice helped reduce the bacterial burden. Imatinib in combination with rifampicin greatly reduced liver pathology in mice infected with M. marinum, while reducing lung bacterial load also in M. tb-infected animals [45].
The DNA alkylating agent cyclophosphamide (CP) is a potent anti-proliferative cancer drug and pre-conditioning agent metabolised by cytochrome P450 to produce reactive metabolites that crosslink with guanine residues [46,47]. Cell types expressing low amounts of aldehyde dehydrogenase are most susceptible to CP treatment since its enzymatic activity neutralises drug-derived cytotoxicity [48]. Regulatory T cells (Tregs) express high cellular levels of ATP-binding cassette B1, which binds CP, thus making these cells especially vulnerable to the drug [49]. Conversely, aldehyde dehydrogenase overexpression can promote resistance to CP toxicity [50]. Prior CP administration to patients participating in a clinical trial of a renal cell cancer vaccine candidate (IMA901) led to Treg decline in peripheral blood of the participants, afforded better CD8 T cells priming and improved overall survival [51]. CP treatment may remove Treg cells as well as pathological lymphocytosis in patients with MDR-TB [52][53][54].
Histone deacetylase inhibitors (HDIs) comprise several clinically approved pharmacological agents which block the activity of the chromosome-modifying enzymes histone deacetylases to promote gene transcription [55]. Histone acetylation mediated by HDIs has been clinically beneficial in cancer therapy [56], vorinostat (suberoylanilide hydroxamic acid) was licensed in 2006 for the treatment of cutaneous T cell lymphoma [57], and valproic acid (VPA, currently in use for epilepsy and bipolar disorder) is in clinical trials for solid tumours [58], while many others are in late-stage clinical trials for a plethora of cancers [59]. VPA and vorinostat can induce apoptosis and autophagy in treated cells via oxidative burst and inhibition of the mammalian target of rapamycin (mTOR), respectively [60][61][62]. The effectiveness of HDIs in infectious disease has been evaluated against HIV infection and West Nile virus infection. VPA or vorinostat treatment of CD4 T cells from individuals latently infected with HIV-1 can reactivate viral progeny, making virus-infected cells highly susceptible to ART as well as CD8 T cell-mediated immune attack [63][64][65][66]. These observations have led to investigating the clinical efficacy of HDIs in HIV-infected individuals [67,68]. In a mouse model of West Nile virus infection, co-treatment of vorinostat with an experimental antiviral drug candidate ameliorated brain inflammation and neuronal damage in addition to reduction in viral load [69]. Evaluation of phenylbutyrate (PBA), another HDI, against intracellular M. tb has been shown to potentiate pro-inflammatory signalling (upregulation of CASP1, IL12, and VDR genes, among others), antimicrobial peptide production, reduced M. tb survival, thus implying enhanced anti-TB T cell activity [70]. Clinical evaluation of PBA and vitamin D3 treatment in patients with pulmonary TB resulted in quicker sputum smear conversion and increased production of cathelicidin by peripheral blood monocyte-derived macrophages [71].

Blockade of cytokine signalling
As previously mentioned, the uncontrolled release of pro-inflammatory cytokines, such as TNF-α, IL-6 and IFN-α/β, drives destructive immunopathology and severe lung disease in TB. The use of anti-TNF-α therapy (adalimumab) in a patient with severe pulmonary TB proved to be life-saving [72], and set the stage for anti-cytokine therapy in clinical management of TB. IL-6, a major player in exacerbating disease severity in patients with TB, is another potential candidate for clinical blockade [6,73], further confirmed by reduction of IL-6 serum levels in TB patients post-treatment [74]. Preclinical evaluation of IL-6 receptor (IL-6R) blockade in a mouse model of TB improved specific anti-TB T cell responses in conjunction with less severe pathology and reduced M. tb burden in the lungs [75]. Monoclonal antibodies that inhibit the IL-6/IL-6R pathway are currently endorsed for treatment of immune-mediated diseases such as idiopathic juvenile arthritis and Castleman's disease in humans (anti-IL-6, siltuximab; anti-IL-6R, tocilizumab) [76,77], and are currently under investigation for cancer treatment [78]. Blockade of the IL-6/IL-6R pathway may, therefore, be considered for clinical management of severe pulmonary MDR-TB. Preclinical evaluation of bevacizumab (anti-vascular endothelial growth factor (VEGF) monoclonal antibody) in a rabbit model of TB showed that inhibition of VEGF-A was commensurate with modified angiogenesis in the granuloma, which improved penetration of a small molecule dye (representative of an anti-TB drug) as well as oxygenation [79]. Oehlers et al. [80] showed that VEGF promotes granuloma formation in M. marinum-infected zebrafish, while pharmacological inhibition of VEGF activity with pazopanib reduced neovascularisation as well as the mycobacterial burden in the animals. IL-4 is another cytokine associated with poor prognosis in clinical TB that quantitatively declines in response to anti-TB treatment [81][82][83][84]. A clinical trial investigating the adjunctive potential of IL-4 blockade (anti-human IL-4 monoclonal antibody, pascolizumab) in patients with DS pulmonary TB undergoing anti-TB treatment is currently underway (NCT01638520).

Vitamin D3
Vitamin D3 (VD3) is important for host resistance to TB; pulmonary TB patients with VD3 deficiency are not able to mount adequate control of primary M. tb infection in the lung [85]. Further, VD3 has been implicated in immunological control of TB in humans involving orchestration of IFN-γ, IL-32 and IL-15 signalling [86,87]. Recent evaluation of the therapeutic potential of VD3 in combination with PBA has shown improved antimycobacterial effects and faster conversion to AFBnegative sputum by the participating TB patients [71]. In vitro evaluation of M. tb-infected macrophages treated with VD3 resulted in upregulation of a collage of anti-inflammatory (IL-10, ARG1) and pro-inflammatory (IL1B, TNF) genes over a 72-hour exposure period [70] this profile was relatively unchanged with PBA cotreatment. However, other clinical studies have not found a correlation between VD3 therapy and improved treatment outcome in TB patients [88,89]. The therapeutic potential of VD3 in MDR-TB patients therefore warrants further clinical evaluation in larger cohorts of patients in endemic countries.

Immune checkpoint inhibition
One of the most remarkable immunomodulatory therapies of recent times is currently licensed for melanoma treatmentantibody-based inhibition of the immune checkpoints programmed cell death 1 (PD-1) and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4). This strategy activates and mobilises tumour-infiltrating T lymphocytes as well as those in peripheral blood of patients with metastatic melanoma [90]. Although immunologically purposed to regulate T cell responses for avoiding immune hyper reactivity, PD-1 and CTLA-4 expression on tumourspecific T cells abrogates anti-tumour effector functions and renders the T cells irresponsive [91]. Tregs have abundant expression of PD-1 on their surface; interaction of this molecule with its ligands (PD-L1/2) on target cells leads to expansion and dampening of antigen-specific effector T cell responses [92]. Importantly, a myriad of cytokines, including IFN-γ, TNF-α, type 1 interferons, GM-CSF, IL-2, IL-7 and IL-15, are actively involved in enhancing the expression of PD-1 on T cells and its ligands on antigen-presenting cells and tumour cells at sites of disease [93]. CTLA-4 binds to CD80 and/or CD86 to control the activation state of effector T cells in the aftermath of a successful immune response, and initiates tolerance by abrogating IL-2 production and downstream activation of signal transducer and activator of transcription 5 (STAT5) [94].
Results from clinical trials assessing monoclonal antibodies targeting PD-1 (pembrolizumab/nivolumab) and CTLA-4 (ipilimumab) in cancers other than melanoma, i.e., lung cancer and ovarian cancer, have been very promising [95,96]. Anti-PD-1 therapy (nivolumab) has been fruitful also in treating patients with chronic hepatitis C virus infection, resulting in a dramatic reduction in viral load [97]. There are several ongoing clinical studies evaluating the efficacy of anti-PD-L1 therapy (avelumab, MEDI4736) in cancer (e.g., but not limited to, NCT02572843, NCT02584829, NCT02588131, NCT02088112). Several clinical studies have evaluated the role of the PD-1/PD-L1 pathway in the pathogenesis of TB. PD-L1, the main ligand for PD-1, is markedly expressed on human macrophages upon infection with virulent M. tb [98]. In vitro blockade of PD-L1 with monoclonal antibodies induced IFN-γ-triggered killing of the infected cells by autologous peripheral blood T cells from patients with pulmonary TB [98,99]. Clinically, PD-1 expression on T cells declines over a course of standard anti-TB therapyconcordant with improved in vitro responsiveness of Th1 cells to antigenic stimulation [99,100]. In addition, expression of PD-1 on the surface of neutrophils, Tregs and natural killer T cells indicates that destructive inflammation in TB patients may further impair the M. tb-specific immune response [53,54,101,102]. Similar observations have been reported for CTLA-4 expression on Tregs and effector T cells isolated from peripheral blood of patients with active TB [53,103,104].
Lymphocyte-activation gene 3 (LAG3) is another immune checkpoint generally expressed by Tregs to modulate CD4+ T cell responses, and binds to MHC-II molecules with high affinity to induce T cell exhaustion several days after activation [105]. LAG3 was recently shown to be highly expressed in the lungs of macaques with active pulmonary TB and not in latent TB infection, suggesting a link between low LAG3 expression and successful containment of M. tb infection [106]. Combined blockade of PD-1 and LAG3 is able to rescue specific CD4+ and CD8+ T cell activity directed against Plasmodium falciparum in an experimental mouse model of malaria [95,107]. Several clinical trials are currently registered for testing the efficacy of anti-LAG3 monoclonal antibodies (BMS-986016, GSK2831781, LAG525) in various cancers (NCT02061761, NCT01968109, NCT02195349, NCT02460224).
The fourth clinically relevant immune checkpoint molecule is T cell immunoglobulin and mucin domain 3 (TIM3). Immunologically, TIM3 binds to galectin 9 (Gal9) on the surface of antigen-presenting cells (as well as tumour cells) to induce apoptosis of activated T cells [108]. Inhibition of TIM3 (MBG453) in cancer therapy affords improvement of anti-tumour T cell responses, marked by IFN-γ secretion and CD8 T cell effector functions [108]. Pertinent to TB, in vitro TIM3 blockade in co-culture experiments with M. tb-infected macrophages from TB patients with or without HIV co-infection fostered bacterial killing and enhanced IL-1β secretion by infected cells as well as IFN-γ release by T cells [109,110]. T cell-mediated reduction of bacterial burden in infected macrophages from HIV-TB patients was further improved in combination with anti-PD-1 blockade [110]. Along with PD-1 and CTLA-4, treatments targeting LAG3 and TIM3 may improve targeted immune responses in patients with MDR-TB for better clinical outcomes.

Cellular therapy
In addition to chemical compounds, monoclonal antibodies and nutraceuticals, cell-based therapies have also been very promising in cancer and infectious disease [111,112]. Adoptive transfer of tumour-infiltrating T lymphocytes has saved many lives in the last 20 years, with objective response rates reaching 56 % in patients with metastatic melanoma [113]. Transfer of donorderived cytomegalovirus-and Epstein-Barr virus-specific CD8 T cells following stem cell transplantation is actively pursued in the clinic [111]. Pertaining to TB, re-infusion of bone marrow-derived mesenchymal stromal cells isolated from patients with MDR/extensively drug resistant-TB was shown to be safe [114], and a phase 2 trial is underway in Durban, South Africa, to evaluate their efficacy in improving treatment outcomes and effects on immune responses. Mesenchymal stromal cell treatment also re-programmed anti-TB T cell responses, with focus on specific M. tb antigens. Infusion of M. tbspecific T cells has yet to be clinically evaluated in TB patients, although in vitro evidence for feasibility, comprising subsets of CD8 T cells, NK T cells and TCR γδ T cells, has been established [115][116][117][118][119].

Conclusion
The poor treatment outcomes of MDR-TB treatment using current WHO recommended TB drug treatment regimens now warrants newer approaches to therapy. The use of immunomodulatory agents as adjunct HDT for improving management outcomes warrants urgent evaluation in well designed, controlled randomized clinical trials. These studies will also provide opportunities to study the M. tb-specific immune responses in the lung and peripheral blood (B cells, T cells, macrophages, antibodies) as markers of inflammation, disease activity, cure or relapse. Well-designed, multicentre randomised clinical trials involving aptly defined patient cohorts are now required.

Authors' contributions
All authors contributed equally to this review. All authors read and approved the final manuscript.
Authors' information AZ is Professor of Infectious Diseases and International Health at the University College London and Consultant Infectious Diseases Physician at the UCL Hospitals NHS Foundation Trust. MR is a postdoctoral scientist in the Therapeutic Immunology Unit (TIM) at the Department of Laboratory Medicine (LABMED), Karolinska Institutet, Sweden. ED is the chief of neurosurgery at Karolinska University Hospital (Solna campus), and adjunct professor at Karolinska Institutet, Sweden. MM is Professor of Clinical Immunology and Head of TIM LABMED, Karolinska Institutet, Sweden.

Competing interests
The authors declare that they have no competing interests.
Author details 1 Division of Infection and Immunity, University College London, and NIHR Biomedical Research Centre, UCL Hospitals NHS Foundation Trust, London,