The Many Hosts of Mycobacteria 8 (MHM8): A conference report

Abstract

Mycobacteria are important causes of disease in human and animal hosts. Diseases caused by mycobacteria include leprosy, tuberculosis (TB), nontuberculous mycobacteria (NTM) infections and Buruli Ulcer. To better understand and treat mycobacterial disease, clinicians, veterinarians and scientists use a range of discipline-specific approaches to conduct basic and applied research, including conducting epidemiological surveys, patient studies, wildlife sampling, animal models, genetic studies and computational simulations. To foster the exchange of knowledge and collaboration across disciplines, the Many Hosts of Mycobacteria (MHM) conference series brings together clinical, veterinary and basic scientists who are dedicated to advancing mycobacterial disease research. Started in 2007, the MHM series recently held its 8th conference at the Albert Einstein College of Medicine (Bronx, NY). Here, we review the diseases discussed at MHM8 and summarize the presentations on research advances in leprosy, NTM and Buruli Ulcer, human and animal TB, mycobacterial disease comorbidities, mycobacterial genetics and ‘omics, and animal models. A mouse models workshop, which was held immediately after MHM8, is also summarized. In addition to being a resource for those who were unable to attend MHM8, we anticipate this review will provide a benchmark to gauge the progress of future research concerning mycobacteria and their many hosts.

1. Introduction

The 8th Many Hosts of Mycobacteria (MHM) conference was held at Albert Einstein College of Medicine (EINSTEIN) in Bronx, NY, March 4–6, 2019, for the purpose of identifying gaps in our collective knowledge of mycobacteria biology, mycobacteria-host interactions, mycobacterial disease surveillance and treatment. The Mycobacteria genus consists of ~200 bacterial species, a number of which cause significant diseases in numerous hosts (Fig. 1). The hosts of mycobacteria include humans and animals that are directly affected by the spectrum of infections caused by these pathogens, ranging from latent and asymptomatic to active and disease. Furthermore, lower eukaryotes are potential environmental reservoirs (e.g. amoeba, aquatic insects), higher eukaryotes facilitate zoonotic transmission (e.g. armadillo, deer), and inanimate materials are potential fomites or biofilm substrates (e.g. bronchoscopes, shower heads).

Fig. 1.

Depiction of the general phylogenetic relationships between mycobacterial species, alongside several of the many hosts of mycobacterial pathogens: fish, aquatic insect larvae, red squirrel, armadillo, human, African buffalo, nine-banded mongoose, brushtail possum, deer, British badger, wild boar, cow and elephant. The phylogenetic tree is based on the work of Fedrezzi et al. [225]. Watercolor images were generously provided by artist Bridget Hecox (Ostrander, OH).

Since 2007, the MHM series has brought together basic, clinical and veterinary scientists with a shared interest in mycobacterial disease research. The workshop is held approximately every other year in a location of historical significance to those in this field. Previous MHM meetings have resulted in new collaborations, as well as a book that was authored by many of the meeting attendees [1]. The 2019 location (Albert Einstein College of Medicine, Bronx, NY) is significant to the field for its long history and reputation for making seminal advances in the fields of mycobacterial genetics, pathogenesis, and immunity. Participants in this year’s conference were from eight different countries and represented universities, government agencies, academic medical centers, agricultural centers, zoos, and wildlife parks. Remarkably, and perhaps reflecting the worldwide distribution of pathogenic mycobacteria, the themes and challenges that were discussed at MHM8 were anything but country-specific.

2. Overview of the mycobacterial diseases discussed at MHM8

The mycobacterial diseases that were a focus of MHM8 included leprosy (M. leprae), Buruli Ulcer (M. ulcerans), other nontuberculous mycobacteria (NTM) infections, human tuberculosis (M. tuberculosis) (TB) and its associated comorbidities, as well as animal TB (M. bovis) in livestock, wildlife, and humans (zoonotic TB, zTB). An overview of these diseases was provided during the first MHM8 panel by Drs. Henry Boom (Case Western Reserve University), Kathryn Dupnik (Weill Cornell Medical College), Ken Olivier (National Institutes of Health) and Richard Robinson (The Ohio State University).

• Leprosy:

Leprosy is a debilitating and stigmatizing disease caused by M. leprae, which infects peripheral nerves and induces sensory-motor deficits that can progress to deformities and disability. For the last 10 years, there have been >200,000 new diagnoses of leprosy (also known as Hansen’s Disease), with 80% of all new cases arising in just three countries, India, Brazil and Indonesia, and millions more individuals with long-term sequelae of M. leprae-induced nerve damage [2]. There are microbiological and epidemiological challenges to the scientific study of M. leprae and leprosy: microbiological challenges include the inability to grow M. leprae in culture, and imperfect animal models (i.e. the nine-banded armadillo and mouse foot-pad model) which do not recapitulate the pathologic immune reactions, reversal reaction, and erythema nodosum leprosum that can develop in people with leprosy; epidemiological challenges include the inability to identify people at risk of developing disease and which are latently infected (this is particularly significant since early diagnosis and treatment of leprosy are essential for avoiding disability). An underappreciated aspect of leprosy is that it is also a zoonosis in the southeastern United States and Brazil, transmitted via contact with nine-banded armadillos [3], and that its etiological agents have been isolated from wild red squirrels from the United Kingdom [4]. Infected armadillos and red squirrels develop signs of leprosy that are similar to but not identical to those in humans.

• Nontuberculous mycobacteria (NTM) and Buruli ulcer:

The term “nontuberculous mycobacteria” is often used interchangeably with “environmental mycobacteria” or “atypical mycobacteria”, and is an umbrella term for >160 species which are primarily acquired from the environment and can cause pulmonary and non-pulmonary disease forms (e.g. cervical lymphadenitis). Like TB, NTM lung disease in the US is a chronic condition associated with substantial direct and indirect financial burdens [5], as well as morbidity and mortality [6, 7]. The incidence of NTM lung disease in the US has increased over the past decade and exceeds that of TB [8–10], with increased risk among US veterans with COPD [11,12] and postmenopausal women [10,13,14]. The majority of NTM lung disease in the US is caused by environmental exposure to M. avium complex [9], although other species contribute in a state- or community-specific manner (e.g. M. abscessus, M. porcinum, M. chimaera) [15–17]. Buruli ulcer, on the other hand, is a skin disease caused by infection with M. ulcerans. Previously referred to as Bairnsdale or Dainstree ulcer, Buruli ulcer is characterized by intravascular thrombosis and coagulation necrosis, and may be ulcerating or non-ulcerating. A major contributor to M. ulcerans pathology is mycolactone, a necrotic cytotoxin produced by the bacterium, which also suppresses the human host’s ability to mount an immune response to M. ulcerans. M. ulcerans transmission is very focal and relates to environmental conditions such as deforestation, and possibly also wild animals that harbor M. ulcerans, such as common ringtail (Pseudocheirus peregrinus) and brushtail possums (Trichosurus vulpecula) [18]. M. ulcerans strain differences and variability in properties of mycolactone may account for different clinical presentations and responses to treatment in Australia and Africa [19]. As with M. leprae, there is no screening tool to detect exposure to M. ulcerans or pre-clinical disease.

• Animal tuberculosis (animal TB):

The term “animal TB” is commonly used to describe disease due to M. bovis; however, it can also apply to infection by other Mycobacterium tuberculosis (Mtb) complex members (MTBC), such as M. pinnipedii, M. orygis, M. caprae, Dassie bacillus, M. suricattae, and M. mungi. Here, we use “animal TB” to refer to disease caused by M. bovis in non-humans and “zoonotic tuberculosis” or “zTB” to describe human disease caused by M. bovis. Whereas Mtb is the primary cause of human TB worldwide and has a single natural reservoir (i.e. humans), M. bovis has several natural reservoirs including Eurasian badgers (Meles meles), Australian brushtail possum (Trichosurus vulpecula), wild boar (Sus scrofa), African buffalo (Syncerus caffer), and white-tailed deer (Odocoileus virginianus) (Fig. 1). M. bovis is an important cause of disease in cattle and other animals, as well as in people (zTB) who come into direct contact with infected animals (e.g. farmers, veterinarians, hunters, slaughterhouse workers, pastoralists) or their products (e.g. unpasteurized milk, cheese) [20]. Whereas human TB is spread via aerosol, animal TB is spread via aerosol or ingestion of M. bovis-laden milk or saliva, or contact with bacteria left on the grasses, hay bales, water troughs, or other farm equipment shared by multiple animals [21, 22]. Outside the infected host, M. bovis is resilient and can survive for long periods of time on numerous substrates (soil, water, hay, corn) [23], in soils of varying pH under controlled or natural climate conditions [23–25], and are possibly aided by environmental amoeba in which M. bovis can survive [26]. The global burden of zTB in humans is difficult to know, as most clinical diagnostic tests do not discriminate between disease caused by Mtb and M. bovis. What is known, however, is that animal TB has significant and detrimental financial consequences for cattle farmers in the U.S., Mexico, UK, and Ethiopia [27], Sika Deer farmers in China [28], and wildlife conservationists in South Africa [29].

• Human tuberculosis (TB):

Globally, TB is the leading cause of death by an infectious disease. The majority of TB cases are caused by Mtb (the minority of TB cases being caused by the other MTBC members listed in Fig. 1). The relationship between Mtb and humans is ancient, having evolved over the past 10,000 to 70,000 years [30]. Mtb can be maintained in humans through latent or subclinical infection. Latency is hypothesized to have been important for this long-term host/pathogen relationship, allowing Mtb to infect an individual without impacting their evolutionary fitness until they become immunocompromised with age [31,32]. While latent TB is asymptomatic and not infectious, active TB is associated with “consumption” of the lung tissue and dissemination of Mtb into the airways. Extensive coughing can create Mtb-laden aerosols that enable Mtb infection of a new generation of hosts. It is speculated that the balance of activity versus latency not only ensures Mtb survival but also selects for Mtb variants that are hypo-inflammatory and, thus, more capable of evading immune recognition [30].

3. Advances in leprosy research

Recent advances in leprosy research that were made using the mouse and armadillo models, along with recent progress in the search for leprosy biomarkers, mechanisms of zoonotic transmission, and the discovery of a second leprosy-causing species (M. lepromatosis) were presented at MHM8. MHM8 speakers who were focused on these and other aspects of leprosy include Linda Adams (National Hansen’s Disease Programs), Karen Dobos (Colorado State University), Maria Pena (National Hansen’s Disease Programs), Kathryn Dupnik (Weill Cornell Medical College), Ramanuj Lahiri (National Hansen’s Disease Programs), and John Spencer (Colorado State University).

As mentioned in our Overviews section above, M. leprae—the causative agent of leprosy—cannot be cultivated on artificial laboratory media but must be grown in animal models. The two major animal models are the nine-banded armadillo in which M. leprae multiply in the internal organs, and athymic nude mice in which bacilli are cultivated in the footpads; a zebrafish model of M. leprae granulomatous infection has also been reported [33]. Because of the difficulties in initiating and maintaining the infrastructure required for these models, the National Institute of Allergies and Infectious Diseases sponsors a centralized resource for the propagation of M. leprae for production of leprosy research reagents (AAI-15006). M. leprae-infected armadillo tissues are subjected to fractionation and purification of various antigenic components from the bacterial cell wall, membrane and cytosol, as well as DNA, monoclonal and polyclonal antibody reagents recognizing various native and recombinant antigens. These leprosy reagents are available, free of charge, to the global leprosy research community through BEI Resources (www.beiresources.org) [34]. Through BEI Resources, live M. leprae derived from the athymic nude mouse foot pads are supplied directly to qualified investigators; a current catalogue of four M. leprae and two M. lepromatosis strains are available, as well as other high-quality research reagents for M. leprae, Mtb, and other mycobacteria, as well as reagents for numerous other pathogens.

In addition to providing a means to culture large numbers of M. leprae, armadillos exhibit the full spectrum of histopathological responses to M. leprae and develop extensive nerve involvement that closely recapitulates human leprosy. They can provide an abundant source of leprotic neurologic fibers and tissues, and this, along with a controlled and known infection status, make them an excellent model for leprosy neuropathy [35,36]. Armadillos have been successfully used to test the safety and efficacy of post-exposure immunization with a new leprosy vaccine [37], screen neuroprotective drugs, and evaluate M. leprae skin test antigens. A variety of non-invasive tests to assess neuropathy have been adapted to the armadillo model in recent years, including measurement of nerve conduction velocity and compound motor action potential, epidermal nerve fiber density, and ultrasound. These clinically relevant endpoints enable early and long-term monitoring of the evolution of neuropathy.

Biomarkers that could substantially aid in the clinical treatment of leprosy are needed. Active research on development of leprosy biomarkers includes biomarkers that can (i) diagnose early leprosy; (ii) classify leprosy into multibacillary or paucibacillary disease; (iii) predict the probability of reaction in newly diagnosed patients; (iv) differentiate reactions from relapse; (v) identify high-risk persons as candidates for vaccination; (vi) detect early neuropathy; and (vii) determine responsiveness for vaccine development. One recent study analyzed RNASeq performed on peripheral blood monocytes (PBMCs) of armadillos experimentally infected with M. leprae (manuscript in preparation). In these animals, 161 genes were upregulated. These genes were involved in host innate immunity and immune response regulation, especially genes associated with cell migration and receptor function. More than 100 genes were downregulated, including those in the IL-10 pathway. Another study examined predictive biomarkers for leprosy reactions [38]. Increased IgM, IgG1 and C3d-associated immune complexes with decreased complement 4 (C4) at leprosy diagnosis were associated with individuals who subsequently developed erythema nodosum leprosum (ENL), an inflammatory reaction that occurs in 30–50% of all leprosy patients and is responsible for increased nerve damage (unless treated effectively with corticosteroids). Decreased C4 at diagnosis was also found in those who developed reversal reaction (RR), a mostly cell mediated inflammatory response that can also lead to nerve damage. Subsequent RR and ENL were both associated with elevated anti-M. leprae antibody levels at diagnosis.

Although the primary mode of leprosy transmission is thought to be human-to-human, armadillos are a known zoonotic host for M. leprae in the southern United States [3] and in Brazil [39]. The M. leprae strain 3I-2-V1 is found in 20% of the armadillo population and in 64% of human endemic leprosy cases in Texas and Louisiana. A recent study conducted in a hyperendemic region in western Pará state in Brazil [39] detected M. leprae DNA in 10 of 16 armadillos tested. Based on a survey of 146 residents of this area, 65% had some contact with armadillos, such as hunting or processing the meat for cooking. Those who reported eating the meat more than once per month had significantly higher titers to the M. leprae-specific antigen, PGL-1, a known biomarker for M. leprae exposure. These data indicate an association between armadillo consumption and frequency of infection with M. leprae.

Another closely related mycobacterial species, M. lepromatosis, was recently identified as a second causative agent of leprosy [40]. Like M. leprae, M. lepromatosis cannot be cultured on laboratory media but can be propagated in mouse footpads [41]. The two species are very similar genetically [42]. M. lepromatosis has three genes not found in M. leprae, including one hypothetical gene, a gene that is predicted to encode a lipoprotein, and hemN, a heme biosynthesis gene. Conversely, M. leprae has 24 genes not found in M. lepromatosis, 23 of which are hypothetical genes and one a possible D-ribose binding protein. Growth rate and division time, and immunological, metabolic, and drug susceptibility parameters are also similar. Clinically, M. lepromatosis infection is diagnosed primarily in Mexico and co-infection with both M. leprae and M. lepromatosis can occur. M. lepromatosis infection has also been reported in the United States, Brazil, Myanmar, Singapore, and Canada. Interestingly, both M. lepromatosis and M. leprae infection have been found in red squirrels in the British Isles [4]. No M. lepromatosis, however, has been detected in armadillos in the United States.

4. Advances in nontuberculous mycobacteria and Buruli ulcer research

NTM are ubiquitous in outdoor environments (e.g. soil) and indoor environments (e.g. showerheads). This panel focused on two NTM that are present in environmental reservoirs and of growing clinical significance, Mycobacterium ulcerans and Mycobacterium abscessus. MHM8 speakers included Paul Converse and Eric Nuermberger (Johns Hopkins University), Alison Kraigsley (Center for Infectious Disease and Research, University of Minnesota), Luanne Hall-Stoodley (The Ohio State University), and Diane Ordway (Colorado State University). NTM topics covered drug discovery, biofilm development, models of infection, repurposed drugs, development of novel regimens to treat infection, repurposed vaccines/immunotherapy, and understanding how discoveries in TB vaccination might translate to vaccine development for NTM disease-.

Paul Converse gave an overview of Buruli ulcer, reviewing the discovery and identification of mycolactone, the polyketide-derived toxin produced by M. ulcerans, the causative agent of Buruli ulcer. Mycolactone is encoded by genes on a giant 174-kb virulence plasmid pMUM001 [43] and has cytotoxic, immunosuppressive, and analgesic properties [44]. Host targets of mycolactone also contribute to the downregulation of inflammatory mediators and defective host cell responses and include: Wiskott-Aldrich Syndrome Protein (WASP) [45,46] leading to apoptosis, Sec61 in multiple cell types [47–50], AT2R in neurons [51,52], and SNARE proteins involved in wound healing [53]. The toxin also appears to bind to LDL, HDL, and other lipid binding proteins [44]. Although mycolactone is still cytotoxic when associated with lipoproteins, further data on the kinetics of the association with lipid binding proteins are needed to make a conclusive determination. While infection with M. ulcerans is initially intracellular, it becomes extracellular, possibly due to mycolactone-mediated destruction of host cells and inhibition of immune cell recruitment to infected sites [54]. In addition, the resulting painless lesions likely delay treatment. A para- doxical reaction can develop in the area of the skin ulcer [55,56], resulting in temporary exacerbation of inflammation, which may correspond to the reduced production of mycolactone as the bacteria die or mycolactone synthesis is blocked. In mice treated with effective antibiotic regimens, footpad swelling, mycolactone concentrations, and bacillary load all decline [54,57–60]. Tissue remodeling associated with lesion resolution proceeds when toxin production or bacillary elimination occurs without inflammatory or paradoxical reactions. In the mouse at least, detectable toxin does not appear to persist in tissue correlating with an absence of residual scarring [54,57]. While the toxin is specific for M. ulcerans/Buruli ulcer and, thus, considered a prime target for disease diagnosis, mycolactone is a very small non-proteinaceous molecule and its cytotoxic and immunosuppressive properties have hindered the production of monoclonal antibodies [61]. Approaches such as fluorescent thin layer chromatography are often difficult to interpret making TLC a poor diagnostic platform [44]. Although liquid chromatography/mass spectrometry is rapid and highly sensitive, it is also expensive, and results can vary between instruments. While variation between instruments is a manageable problem, this technology is not typically accessible in resource poor settings where the bulk of the disease occurs. All diagnostic strategies require a stable supply of synthetic mycolactone. Phage and yeast display are promising approaches to produce recombinant antibodies to mycolactone, which could then be adapted to a simple lateral flow assay in resource poor settings [61].

Eric Nuermberger discussed mouse models for M. ulcerans and M. abscessus drug efficacy testing. Until 20 years ago, treatment of M. ulcerans disease (Buruli Ulcer) was primarily surgical (i.e. wide local excision followed by skin grafting). The curative potential of rifampin + aminoglycoside combination was demonstrated using a mouse footpad infection model [58,62]. Promising results with rifampin (RIF) + streptomycin (STR) in 21 Ghanaian patients [63] led to provisional WHO recommendations for an 8-week treatment regimen, plus surgical excision when needed. Subsequent trials affirmed the efficacy of RIF + STR and, more recently, that of an all-oral regimen of RIF + clarithromycin (CLR). Major objectives for drug and regimen development are to shorten treatment with well-tolerated, all-oral regimens. Promising results have been obtained in mice using high-dose rifamycins (e.g. RIF, rifapentine) [64], replacing STR or CLR with clofazimine [57], and incorporating the potent QcrB inhibitor, Q203 [65]. Despite its track record, questions regarding the mouse footpad infection model persist, including: what is the most relevant efficacy endpoint (i.e. resolution of swelling, bacterial clearance, relapse prevention) [66]; how well qualified is the model for drug/regimen efficacy testing; would an actual ulcer model be more useful for testing topical therapies or assuring adequate drug distribution to all lesion compartments; and would better representation of specific disease states (e.g. nodules, plaques, ulcers) enable more stratified therapeutic approaches.

In contrast, current therapy of M. abscessus lung infection leaves much room for improvement. Combinations of multiple injectable (e.g. amikacin, β-lactams, tigecycline) and oral (e.g. macrolides, clofazimine) drugs used for as long as tolerated, or up to one year past sputum culture conversion, still result in poor cure rates despite substantial toxicity and intolerance. Outcomes are strongly influenced by macrolide susceptibility and whether the infected tissue is amenable to resection. The major questions facing drug and regimen developers include how to improve safety/tolerability and sputum conversion rates, identify effective oral drugs, shorten treatment, identify the pathological niches and location(s) that M. abscessus persists in despite antibiotic therapy, and testing the utility of inhaled antibiotics. Promising agents for repurposing against M. abscessus infections include clofazimine and bedaquiline [67], new β-lactam/β-lactamase inhibitor and dual β-lactam combinations [68–72], new tetracycline derivatives (e.g. omadacycline, eravacycline) [70] and compounds with novel mechanisms of action (e. g. MmpL3 inhibitors) [73]. Unfortunately, there is a dearth of well qualified animal models of M. abscessus infections for use in drug development. Early studies show that most immunocompetent mouse strains clear M. abscessus in a few weeks after infection [74]. High-dose infection models in severe combined immunodeficiency (SCID) or athymic nude mice are costly, difficult to maintain and do not develop lung pathology characteristic of their human counterparts. An improved model would be less costly, demonstrate greater bacterial persistence and more human-like pathology (e.g. biofilm, necrotic granulomas), but at issue is how to achieve a better model of persistent infection with desirable pathology. Potential avenues may include using transient immunosuppression and/or underlying lung disease models to promote establishment of bacterial infection and pathology that may promote persistence [69].

Once considered rare, human NTM pulmonary infections are rising in North America and Europe, and are increasingly recognized in areas where TB is endemic, such as Africa, South America, and India [75]. While M. abscessus is an NTM that can cause infections in immunocompetent individuals, its emergence as a pulmonary pathogen in patients with chronic lung diseases, has been particularly troubling [75, 76]. Although biofilm formation has been primarily considered an attribute of mycobacteria growing in the environment, M. abscessus can present as biofilm-like aggregates in the lungs of patients with chronic obstructive lung disease (COPD) or cystic fibrosis (CF), leading to the hypothesis that M. abscessus pulmonary infections involve biofilm development [77,78]. Alison Kraigsley gave an overview of biofilm development, which can compound intrinsic drug resistance in many types of bacteria. Luanne Hall-Stoodley discussed M. abscessus biofilm development, which can occur with either glycopeptidolipid-producing (smooth) or glycopeptidolipid-deficient (rough) colony forming variants. M. abscessus biofilm aggregates are highly tolerant of host cell antimicrobial effectors and antibiotics used in NTM therapy, surviving low pH, high concentrations of hydrogen peroxide, and 100-fold higher concentrations of amikacin and azithromycin compared with non-aggregated (single-cell) variants [79]. Like other pathogenic mycobacteria, M. abscessus is a facultative intracellular pathogen that survives in human macrophages. M. abscessus variants are similarly taken up by human macrophages and survive in the presence of high concentrations of amikacin or azithromycin, indicating that macrophages are a likely reservoir of infection [79]. Additionally, M. abscessus aggregate development can occur on human bronchial epithelial (HBE) cells differentiated at airway liquid interface from patients with CF. HBE cells differentiate into ciliated and mucus-producing cells that recapitulate airway epithelial cell self-organization, maturation, morphology, and function, and are highly suited to studying human airway cell-pathogen interactions [80,81]. Preliminary studies indicate that, along with macrophages, biofilm-like aggregates associated with HBE may act as an additional reservoir for persistent M. abscessus infection in the lungs in patients with structural lung disease. Current efforts are underway to use primary differentiated macrophages and HBE as promising cell infection models to better understand M. abscessus pathophysiology in the context of human cells and to address the limited animal models available for studying M. abscessus infections.

Diane Ordway discussed therapeutic vaccine approaches for NTM. While effective new treatment regimens against NTMs are achievable, a notable challenge is that patients infected with NTM have a high chance of being re-infected. Thus, a therapeutic vaccine strategy that controls opportunistic NTM and mitigates subsequent re-infection is needed against NTMs. Several approaches may be exploited to develop and advance vaccines against NTM. For example, if similar virulence factors are expressed during infection, cross-reactive antigens produced by other mycobacterial pathogens may allow the development of an NTM vaccine strategy, which combines effective treatment and post-exposure vaccine to prevent re-infection [82]. One such approach tested vaccine safety and efficacy in a SCID mouse model. SCID mice were vaccinated with Bacille Calmette-Guerin (BCG), and with the new vaccines IKE-M. smegmatis Δesx-3 mutant, IKEPLUS-Mtb Δesx-3 mutant, and IKEPLUS-Mtb (developed by Bill Jacobs, Albert Einstein Medical School). Eight weeks later mice received an intravenous infection with the outbreak strain, M. abscessus OM194, and were evaluated for bacterial burden in each group on day 92 post-infection. The IKEPLUS-Mtb Δesx-3 vaccination group resulted in a 4 log₁₀ reduction in M. abscessus bacterial burden. Studies are currently underway to repeat these studies in a cystic fibrosis mouse model to evaluate bacterial burden, organ histology and immune responses using flow cytometry. Additional studies are being completed in the Jacob’s laboratory to vaccinate RAG mice with BCG, IKE-M. smegmatis Δesx-3 mutant, IKEPLUS-Mtb Δesx-3 mutant, and IKEPLUS-Mtb followed by infection with M. abscessus OM194 and evaluation of both organ bacterial burden and immune responses in each group.

In summary, MHM8 speakers reviewed two environmental NTM that increasingly cause human infections. NTM infections are understudied and much remains to be discovered about these mycobacterial pathogens. Future directions for research highlighted infection models that better recapitulate human pathology and represent specific disease states (e.g. nodules, plaques, ulcers), and the development of animal models that can be used to better evaluate therapeutic efficacy and vaccine testing.

5. Advances in human tuberculosis & immunology research

The importance of innate and adaptive immunity for mycobacterial disease resistance is well established. At MHM8, selected areas of progress and important remaining knowledge gaps in the immunology of infections with Mtb, NTM and other intracellular bacterial pathogens were presented and discussed. These included the roles of T cells (Steven Porcelli, Albert Einstein College of Medicine), B cells (John Chan, Albert Einstein College of Medicine), antibodies (Jacqueline Achkar, Albert Einstein College of Medicine), and trained immunity (Kristina de Paris, University of North Carolina at Chapel Hill) in the defense against Mtb, the comparative immunology between Mtb and NTM infections (Diane Ordway, Colorado State University), and lessons learned from the immunology of other intracellular bacteria (Karen Elkins, Food and Drug Administration). Irrespective of which topics were addressed, a major theme remained that, regardless of the host, we still don’t know all the immune responses, mechanisms, and antigens involved in the protection against Mtb and other mycobacteria. Furthermore, although the specific roles of cell-mediated and other immune components in the protection against TB ultimately need to be confirmed in humans, some panelists pointed out that the mouse remains a valuable animal model for generating knowledge about the detailed mechanisms of TB pathogenesis.

While it is well-established that CD4 T cells are critical for the control of Mtb and other mycobacteria, the detailed contribution of other T cell subsets involved in protection against infection and disease in humans is less well-known. Interferon-gamma (IFN- γ) is a key cytokine, but the degree to which it can be a correlate of protection on its own remains unclear. Tumor necrosis factor further contributes to protection, but other cytokines are less well-studied. Furthermore, despite some tangible evidence of progress in TB vaccine efficacy demonstrated in recent human phase 2 trials [83,84], the combination of T cell antigens needed for optimal protection remains to be determined.

In contrast to T cells, we don’t have all the tools needed to track B cells. Nevertheless, increasing evidence in mice, other animal models, and humans demonstrates the role of B cells and humoral immunity in regulating the immune response to Mtb (reviewed in Refs. [85–89]). B cells form prominent aggregates in the lungs of Mtb infected mice, non-human primates, and humans, some of which have been associated with disease outcome [90,91]. Studies with B cell knock-out mice and B cell transfer have shown that B cells limit inflammation and influence lung cytokine levels during acute Mtb infection, and studies with isotype knockout mice and serum transfer have established the role of immunoglobulins in the protection against Mtb.

Despite the increasing evidence, the role of antibodies in the protection against TB and other mycobacteria remains controversial and insufficiently investigated. Over the past few years, in vitro and in vivo studies of polyclonal human serum antibodies have further established the contributing role of antibodies in the protection against TB in humans [85–89]. Most, but not all, studies suggest that serum immunoglobulin G (IgG) from certain asymptomatic Mtb infected or exposed and/or BCG vaccinated asymptomatic individuals is protective, in contrast to patients with active TB [92–95]. Both glycan compositions of the IgG constant region [94] and mycobacterial surface-specificity of the IgG variable region [93,95] have been shown to influence protective efficacy but detailed knowledge of protective antigens is limited.

Recent increasing evidence further supports a role of epigenetic changes and trained innate immunity in the protection against TB. For example, priming of innate immune cells, such as hematopoietic stem cells, with BCG can lead to epigenetically modified macrophages that provide better protection against Mtb than naïve macrophages [96]. Assessing blood samples from individuals with a range of Mtb infection states, the use of mycobacterial growth inhibition assays (MGIAs) has allowed a holistic approach correlating involvement of components from all immune arms with mycobacterial growth reduction ex vivo. Recent work with MGIAs revealed that blood samples from TB patients resulted in pronounced mycobacterial growth reduction, which correlated with increased involvement from all immune components - humoral, cell-mediated, innate as well as trained innate immunity [92].

In contrast to Mtb, NTM have expanded exposure risks but are less virulent [97,98]. NTM exposure includes contact with water or fomite sources and infection outcomes range from clearance or colonization to latent infection and overt disease [99–101]. It is conceivable that prior NTM exposure or infection could prime immune responses against Mtb later in life [102]. While IFN-γ appears to be more important in the protection against Mtb than NTM, TNF-α is essential to control infection with both [103,104]. Whereas most immunocompetent individuals can clear NTM infections, treatment with TNF-α inhibitors, chronic pulmonary damage, and reduced leptin are among several conditions that have been shown to increase the risk for persistent NTM infections and disease [105–107]. The emergence of NTM infections underscores the need for a better understanding of immune responses required for regulating and killing NTMs, which in turn could enhance our understanding of mechanisms involved in controlling Mtb infection [108].

The slow intracellular growth of Mtb can make it difficult to for researchers to identify mechanisms of protection; for this reason, time was taken at MHM8 to consider what lessons may be gleaned from other intracellular bacterial pathogens that divide more rapidly. Specifically, one approach to identifying vaccine correlates of protection against TB could come from exploring and comparing functional bioassays and other methods, i.e. immune cell gene expression signatures, with other intracellular bacteria, such as Francisella tularensis [109]. F. tularensis has a much faster bacterial replication rate and disease course than Mtb and an excellent mouse model exists. Furthermore, Francisella is similar to Mtb in replicating primarily in host macrophages and in causing respiratory disease. Importantly, Th1 T cells and their products play major roles in protective immune responses to both pathogens. Using the Live Vaccine Strain (LVS) of Francisella, an in vitro tissue culture assay was recently developed to measure the ability of T cells to control bacterial growth within macrophages in a variety of tissues, including spleen and peripheral blood, [110]. Studies in mice and rats have demonstrated that the intramacrophage bacterial growth control during in vitro lymphocyte-macrophage co-cultures reflects the relative efficacy in vivo of different Francisella vaccine candidates [110,111]; studies in vaccinated Rhesus macaques and humans are ongoing and appear promising. Thus, this bioassay may be a functional correlate of vaccine-induced protection. Similar results have now been obtained in studies using Mtb vaccines in mice suggesting the value of assays/models and lessons learned from other intracellular bacteria for the field of TB [112,113].

6. Advances in animal tuberculosis research

Animal TB is a global One Health issue, particularly in livestock and wildlife; multiple susceptible host species and infection with various members of the Mtb complex increase the complexity and potential impact of this disease in different ecosystems. The theme of complexity and lack of fundamental knowledge on disease burden, transmission dynamics, spillover and spillback between and within animal hosts, and inadequate diagnostic tests was repeatedly emphasized by the panel speakers: Peter Buss (Kruger National Park, South Africa), Michele Miller (Stellenbosch University, South Africa), Tyler Thacker (National Animal Disease Center, USDA, USA), Liliana Salvador (University of Georgia, USA), Natalie Parlane (AgResearch, New Zealand) and Marian Price-Carter (Hopkirk Research Institute, New Zealand).

In wildlife, the knowledge gaps of mycobacterial diseases are even greater due to the paucity of research and variability in routes of transmission and pathogenesis between species, including ingestion of infected prey by predators (ex. lions, African wild dogs) and scavengers, indirect environmental transmission (ex. M. mungi in anal gland secretions of banded mongooses) [114] as well as aerosol in social animals (ex. M. bovis in African buffaloes, M. suricattae in meerkats) [115,116]. Key questions need to be addressed in order to facilitate TB control such as identifying risk factors, conditions for a species to become a reservoir host, role of the environment in transmission, and pathogenesis. These investigations require accurate methods, for detecting infection and disease in the different species, with their variable immune responses, sites of infection, and disease dynamics. However, there are limited, if any, tests available for performing surveillance, much less control in wildlife species.

The global burden of M. bovis is often concentrated in developing countries, with prevalence varying significantly between countries and regions of the world. However, due to effective eradication programs, the prevalence in the U.S.A. in cattle is currently around 0.006%, with only 6 new infected herds (out of 913,000 herds) detected in 2018. Epidemiological investigations are facilitated by the routine use of whole genome sequencing for all M. bovis isolates in the U.S., which have shown that several new genotypes have been introduced into the country [117]. In addition, this has allowed identification of a livestock case of Mtb, highlighting the increasing disease threat that people may present to animals.

New Zealand also has a plan to eradicate M. bovis by 2055, which requires that all livestock have permanent identification to allow movement tracing. However, the presence of a wildlife reservoir, the brushtailed possum, complicates eradication. This is being addressed with the use of whole genome sequencing to determine the source of outbreaks and is being employed as part of the eradication strategy [118]. In addition, involvement and support by the various stakeholders underpins the successful implementation of this plan [119].

Understanding networks of cattle movement is essential to investigating disease transmission. Statistical modeling permits identification of risk factors; however, roles need to be disentangled to understand the different mechanisms that contribute to disease outbreaks. For example, risk factors for bovine TB in England and Scotland include herd size and cattle movement from high-risk areas [120]. Spatial spread over long distances is largely explained by cattle movement, resulting in outbreaks outside the endemic area, emphasizing the importance of identifying infected animals and their role in the network. However, it is more difficult to discern the role of wildlife, cattle movement, and persistence of M. bovis in herds that are spatially clustered [121], and highlights the need of use of simulation modeling together with local field data collection for the design of more effective surveillance strategies.

Although mycobacterial diseases in animals are primarily due to M. bovis infection, M. avium is a significant pathogen in humans, mammals, and birds. This organism is found in a variety of environmental niches, and usually spreads by aerosols. M. avium spp. paratuberculosis (MAP) causes Johne’s disease and chronic wasting in ruminants, resulting in significant production losses in livestock as well as having a potential (albeit debatable) link with Crohns’ disease in humans. Although there have been new developments for detection using phage assays, transcriptomics, and transmission pathways (for example, persistence in free-living amoebae [122–125]), many questions remain. Key questions include whether there is a link between MAP and human disease, determining whether movement control is effective for a highly prevalent disease such as Johne’s disease, and risk associated with spreading effluent and waste water containing viable mycobacteria that survive water treatment.

7. Mycobacterial disease comorbidities

Mycobacterial diseases do not exist in isolation; rather, they are often found with comorbidities that have developed before or after an initial mycobacterial infection event. Mycobacterial disease comorbidities are health conditions that originate independent of a mycobacterial disease (e.g. TB, leprosy), but exacerbate or worsen mycobacterial disease outcome or diagnosis. Comorbidities that were a focus of MHM8 included HIV/AIDS, adipose tissue loss, anti-helminthic therapy, and environmental NTM. MHM8 speakers who focused on these and other aspects of comorbidities include Janice Endsley (University of Texas Medical Branch, Texas), Deanna Hagge (The Leprosy Mission, Nepal), Jyothi Nagajyothi (Rutgers University, New Jersey), Kristina DeParis (University of North Carolina, North Carolina) and Tyler Thacker (US Department of Agriculture, Iowa).

Whatever equilibrium may have existed between Mtb and its human hosts was altered in 1981, with the beginning of the HIV/AIDS pandemic [126]. HIV is a retrovirus that is transmitted through sexual intercourse (via semen or vaginal fluid), blood contamination (via shared needles), or from mother to child during birth or breastfeeding (via breastmilk). A long-term goal of MHM8 speaker Kristina DeParis is develop a dual pediatric vaccine for HIV and TB by taking advantage of the intrinsic adjuvant activity of mycobacteria. Live attenuated Mtb was engineered to express SIV antigens and this vaccine was tested in neonatal macaque model that mimics breastmilk transmission of HIV-1 [127]. These studies provided a wealth of information on pediatric immune responses to both SIV and mycobacteria.

HIV induces significant changes in immunological responses that can tip the balance in host control of Mtb. HIV infects and modulates leukocyte lineages which are key to maintaining Mtb latency, including T helper (T_H) cells and macrophages [128], results in HIV-positive (HIV^POS) individuals being more likely to develop active TB than HIV-negative (HIV^NEG) individuals [129–132]. Thus, whereas the transition of Mtb from latency to active disease may take decades in HIV^NEG individuals, HIV infection accelerates the loss of T_H cells and macrophage (MΦ) dysfunction [128], resulting in higher rates of active disease in individuals with comorbid TB and immunodeficiency.

HIV affects the T_H and MΦ interaction which promotes Mtb-containment and referred to as Type IV delayed-type hypersensitivity (DTH). Following Mtb-infection of the lungs, T_H cells are activated following Mtb-antigen presentation in the draining mediastinal lymph node [133]. Following activation, these T_H cells differentiate into T_H1 cells in an IL12 and IL12Rβ1-dependent manner [134–137]. Upon T_H1 cell emigration into the lung and direct recognition of Mtb-infected cells [138], T_H1 cells produce IFNγ, which activates MΦ and limits Mtb replication. The lag time between initial Mtb antigen exposure and T_H1-mediated MΦ activation underlies the “delay” in DTH (relative to other forms of hypersensitivity that are antibody-mediated). Although greater numbers of Mtb-specific T_H1 cells are not indicative of greater TB resistance [139,140], the essential role DTH plays in TB resistance is well established in both animal models and humans [99]. How HIV affects DTH and TB resistance is difficult to study in animal models, as HIV’s tropism for human cells limits the utility of wild type mice as a model system for HIV/Mtb coinfection studies. MHM8 speaker Janice Endsley demonstrated that these challenges can be mitigated by the use of humanized mice. Humanized mice are immunodeficient at birth and subsequently reconstituted with a human immune system via stem cell transplantation. The Endsley lab has demonstrated that Mtb-infected humanized mice develop granulomatous regions which resemble those found during human TB, and have human T cells that are functionally competent [141]. Using the same model system, they further demonstrated that TB progression is exacerbated by HIV/Mtb co-infection [142].

Additional comorbidities that affect mycobacterial disease outcome discussed at MHM8 were adipose tissue loss, chronic helminth infection, and co-infection with NTMs. The association of TB with malnutrition has been known for a long time; however, until the recent report of MHM8 speaker Jyothi Nagajyothi [143,144], a direct link between lipid homeostasis and Mtb-infected macrophage function had not been demonstrated. Using a transgenic “fatless” mouse model system, the Nagajyothi lab demonstrated that Mtb bacilli are present in fat tissue following aerosol infection of mice and show that loss of fat cells is associated with an increase in pulmonary Mtb burden and lung pathology. Adipose tissue contains very important populations of macrophages [145]; although future research will distinguish whether the phenotype of Mtb-infected “fatless” mice is due to the absence of adipose tissue per se, or the absence of adipose-associated macrophages, these data nevertheless support a model wherein an acute loss of adipose tissue during latent TB infection may favor development of active TB disease. Helminth infection is a common comorbidity in countries with a high leprosy burden. The majority of new leprosy cases are diagnosed in regions where soil-transmitted helminths are endemic. Gastrointestinal helminth infections skew T cells towards Th2 bias, which attenuate the efficacy of Th1 cells, DTH and mycobacterial clearance. In a study of 145 Nepalese leprosy patients, the Hagge lab demonstrated that 55% of participants harbored 1 or more soil-transmitted helminth co-infections, and that these co-infections were significantly and inversely associated with leprosy reaction [146]. Furthermore, anti-helminthic chemotherapy (i.e. deworming) may initiate immune reconstitution and lead to leprosy reaction development [146].

Complicating the problems of TB prevention and diagnosis in animals are comorbid infections or exposures to nontuberculous mycobacteria (NTM). NTM comprise 0.02–2.0% of soil bacteria depending on their environment [147], and can be found in the water supplies of urban and rural areas [148,149]. NTMs generally do not cause disease, but nevertheless elicit an adaptive immune response that is similar to that which is elicited by M. bovis: a mycobacterial-antigen specific, type IV DTH that is detected using either a Mantoux/PPD skin test or interferon-gamma release assay (IGRA). Whether a comorbid NTM infection blocks the protective effect of M. bovis BCG has been debated for some time. This probably varies by NTM species and is an important consideration in vaccine efficacy studies [150–152]. In cattle, exposure to the NTM species M. avium elicits an IGRA response that is comparable to that elicited by M. bovis and confers a modest degree of protection against subsequent M. bovis challenge, but nevertheless interferes with M. bovis diagnosis [153]. MHM8 speaker Tyler Thacker urged consideration of NTM and other common co-infections such as γ-herpesviruses [154] in the design and testing of vaccines against M. bovis infection.

8. Advances in mycobacterial genetics

The increasing availability of population genomics and genetic data has allowed an improved understanding of genotypic differentiation among organisms, disease transmission dynamics, and vaccine development. The MHM8 speakers on this panel included Marcel Behr (McGill University, Canada), Liliana Salvador (University of Georgia, Georgia), John Spencer (Colorado State University, Colorado), Lief Kiresbom (Uppsala University, Sweden), and Bill Jacobs (Albert Einstein College of Medicine, New York).

Marcel Behr focused his presentation on the genomic differences between opportunistic and professional pathogens. He started by providing insights into the emergence and evolution of the different NTM lineages. Phylogenetic relationships among Mycobacterium genus have distinguished rapidly growing species from slow ones. M. abscessus is a rapidly growing NTM for which the taxonomy is unclear, while Mtb, M. canettii and M. kansasii belong to the slow group of evolved mycobacteria [155]. Mtb and M. kansasii are phylogenetic close to one another and both cause pulmonary disease; however, Mtb is a global health problem, while M. kansasii is an opportunistic pathogen. Comparative genomic analysis, and side-by-side in vitro and in vivo studies on these two organisms, have shown that there are differences in conserved genes and genomic regions, in metabolic capacity, and in the acquisition of Mtb specific genes since they branched from the common ancestor. They also have distinct epidemic profiles, with a high difference (higher than three orders of magnitude) on the number of bacterial counts in in vivo infections in a given time period. These results suggest that M. kansasii could mirror the environmental ancestor of Mtb before its emergence as a professional pathogen, and that it could be used as a model system to understand how pathogens shift from being opportunistic (environmental) to professionals (host-specific). Brites and colleagues [156] have proposed that some MTBC ecotypes start as generalists and then subsequently adapt to multiple host species. At least nine members of the MTBC infect animals other than humans. Phylogenomic and comparative analyses have shown that the animal-adapted MTBC members are paraphyletic with some members more closely related to the human-adapted M. africanum Lineage 6 than to other animal-adapted strains. Furthermore, four main animal-adapted MTBC clades that might correspond to four main host shifts have been identified, in which two are distinct cattle lineages (M. bovis and M. orygis); the authors have hypothesized that these might be related to independent cattle domestication events. However, these animal-adapted lineages are not necessarily related to host relatedness but are driven instead by ecological factors such as contact rates and demographic aspects of the host population. This session ended by challenging the audience with what exactly is a host since there are mycobacteria that only cause disease, while others use the hosts as a reservoir (professional pathogens). For professional pathogens, why do they follow a non-random distribution of targets? Is it just opportunity, biology, or both?

Liliana Salvador discussed how whole genome sequencing can provide key insights into the management and control of M. bovis in animal species in an endemic area. M. bovis can infect a variety of species, and the role of infected wildlife in the persistence and spread of the disease to livestock is difficult to quantify and control. Phylodynamic approaches can be used to investigate important epidemiological, immunological, evolutionary and ecological processes, such as epidemic spread and spatio-temporal dynamics [157,158]. Of particular interest, host movement and bacterial migration are examples of cryptic processes that contribute to the spread of the disease. Several past studies have used whole genome sequencing data to understand M. bovis dynamics across species [118,159–161]. White-tailed deer are known to act as a reservoir of infection in Michigan [162,163], and over the years there have been cases of infected elk and cattle as well. Elk are known to be a maintenance host of this disease in other areas of the world, therefore, understanding their role in the circulation of M. bovis is fundamental – they are fewer in number, but range further than deer, so may enable long distance spread. By combining M. bovis genomic data from populations of elk, deer, and cattle with spatio-temporal locations, Salvador et al. has shown strong spatio-temporal admixture of M. bovis isolates [164]. Clustering of these isolates in elk and cattle suggest either intraspecies transmission within the two populations, or exposure to a common source. No support for cross-species transmission amongst elk and cattle was found, and the data support that interspecies transmission and maintenance is likely only due to deer. This study demonstrates how phylogenomic studies can be used to study transmission between species, providing insights into disease management and control, particularly in a multi-host system.

John Spencer presented his work on the dynamics of M. leprae using whole genome sequencing. The M. leprae genome was completely sequenced in 2001 and is one of the most highly conserved genomes in the bacterial world. Whole genome sequence analysis of hundreds of clinical strains from all over the world have been characterized into four main SNP types (SNP 1–4) and 16 subtypes that have a fairly restricted geographic range worldwide [165,166]. In 2013, genome sequences were obtained from skeletons of five medieval leprosy cases from the United Kingdom, Sweden, and Denmark [167]. The ancient M. leprae sequences were compared with modern strains (representing diverse genotypes and geographic regions), and a remarkable genomic conservation during the last millennium was detected. Interestingly, a single Brazilian SNP type 3I isolate from São Paulo state was found to branch between two medieval 3I strains, Jorgen_625 from Denmark and SK2 from the UK [168], and represents, for the first time, an ancestral 3I lineage in a modern sample, supporting the hypothesis of the European origin of the 3I lineage in the Americas. Despite several studies and available drugs, it has been difficult to eradicate this disease in humans. Wildlife reservoirs of infection have been found in different parts of the world. Two species of leprosy-causing organisms caused warty growths on the faces and extremities of red squirrels in the British Isles [4]. Armadillos have been shown to be a natural reservoir of M. leprae in the southern states of the U.S. and to have a zoonotic impact in Brazil [39]. Spencer and colleagues are currently analyzing the genetic diversity and distribution of M. leprae strain types from different regions of Brazil and relating them with existing strains around the world. Comparative and phylogenetic analysis like these provide insights into the evolution and antimicrobial resistance around the world, uncovering lineages and phylogenetic trends of particular importance [168].

ESX genes encode a Type VII secretion system that facilitates the export of ESAT6 and other mycobacterial substrates into the phagosome [169]. Following their export into the phagosome lumen, ESX substrates cause lysis of the phagosome, which in turn provides mycobacteria with access to the cytosol compartment and enhances mycobacterial survival. Depending on the species, mycobacteria have several ESX systems and paralogs of varying importance to overall virulence [170]. Lief Kiresbom demonstrated that the expression of these systems is increased under low pH conditions, which would correspond to the acidification experienced by mycobacteria during phago-lysosomal fusion. Nevertheless, all mycobacteria within a given culture or infection model do not respond uniformly to environmental stimuli such as lower pH, which allows for the emergence of culture variants; it was discussed how transcriptome analysis of such culture variants could provide interesting information about the biology of mycobacteria and a means to identify genes that regulate diversity within a culture/infection. In addition to having different effects on mycobacterial virulence, each ESX system also has unique consequences for host immunity [171]. For example, in the absence of ESX5, M. marinum is able to translocate from the phagolysosome to the cytosol but—relative to wild type or complemented strains—unable to elicit inflammasome activation, IL1β and antimicrobial peptide secretion [172]. Building on this work, the lab of Bill Jacobs generated and used an ESX5-deficient strain of Mtb (Δesx5 Mtb) in a heterologous prime-boost vaccine study, wherein mice were primed with BCG and subsequently boosted with Δesx5 Mtb. These studies are ongoing, but support the development of a TB vaccine for individuals in the developing world who—despite having already been BCG-vaccinated—are nevertheless still at risk for TB.

Bill Jacobs focused his session on strategies to enable in vitro growth M. leprae by gene addition. A vast amount of work has been done to understand the molecular basis and expression of M. leprae genes, however, M. leprae still cannot be cultured in microbiological culture media or in cell culture systems [173]. M. haemophilum, which is a close relative of M. leprae (and sometimes is misdiagnosed as such) [174], can grow in culture with hemin or ferric ammonium citrate [175]. IrtAB (iron-regulated ABC transporter) is absent in both M. haemophilum and M. leprae [176], and it is not clear how M. leprae acquires iron if it does not have mycobactin. This raises the questions if it would be possible to make M. leprae grow by giving it the same ABC (ATP-binding cassette) transporters that make M. haemophilum grow. A strategy is being tested in which an IrtAB plasmid is electroporated into M. leprae to determine if acquisition of IrtAB would enable in vitro growth of M. leprae.

9. Advances in mycobacterial ‘omics

Advancements in mycobacterial research include the incorporation of advanced methods, including the development of ‘omics approaches for understanding mycobacterial physiology, and provide fundamental knowledge that will improve the ability to advance research on interventions. At MHM8, the leading approaches to ‘omics data, including transcriptomics, metabolomics, lipidomics and proteomics were presented and discussed for mycobacteria, including Mtb and NTMs. Specific topics included the transcriptional and translational profiling (Keith Derbyshire, Wadsworth Center, New York), proteomics (Karen Dobos, Colorado State University, Colorado), metabolomics (Michael Berney, Albert Einstein College of Medicine, New York), and lipidomics (Travis Hartman, Weill Cornell Medicine, New York). The panel recognized that the value of the data from each method is enhanced when taking a systems biology approach to integrate these data. Obtaining DNA sequence is important, but what does that mean? We still need to get from sequence to function. Transcriptional and translational profiling allow us to see what proteins are encoded, however, proteomics is required to demonstrate if the proteins are made, modified, and what the relative levels of each protein are available under any given condition. Proteomics alone will not necessarily reveal protein function, but some of the proteins will be metabolically active. Metabolomic and lipidomic studies can provide information about any lipids that may be produced by protein catalyzed reactions.

Keith Derbyshire began with a discussion of mycobacterial transcriptomics. Bioinformatic pipelines employ base gene annotations on the assumption that all bacteria use the same genetic architecture, based on the genetic architecture as observed in E. coli. Instead, if we explore science with an open mind and take an empirical approach to define genes for M. smegmatis and Mtb under a variety of growth conditions it becomes clear that additional genes exist with non-traditional genetic elements. To functionally annotate genes in mycobacterial genomes ribosomal profiling (Ribo-Seq) is used in combination with RNA-Seq. RNA-seq is used to map the transcribed regions of the genome and Ribo-seq can then map the translated regions. In this approach RNA-Seq was used to determine the total RNA profile, ribosomes were isolated, and the bound 70S RNA fragments were sequenced. These sequences were then mapped back to the reference gene sequence to identify the 5’ boundaries of genes, which mark the transcriptional and translational start sites [177]. In a classic gene profile, genes are leadered and the Ribo-Seq boundary spans the 5′-UTR transcriptional start site and the Shine-Delgarno sequence and is about 20–30 nucleotides from the initiation methionine codon. Using empirical gene determination, about 70% of genes in mycobacteria have this leadered transcript architecture, however, an additional, novel gene architecture was identified and approximately 30% of the gene were actually leaderless. For the leaderless transcripts, there is no 5′-UTR or Shine-Dalgarno sequence. An ATG or GTG acts as both the transcriptional and translational start site and the ribosome binds directly to the 5′ of the RNA. As a result of this previously unidentified novel gene architecture, mycobacterial genes have been mis-annotated, and many small proteins have been ignored. The Derbyshire lab identified about 1000 leaderless genes in Mtb and about 500 of these have been mis-annotated in public databases. An additional 200 new genes were also identified many of which encode for proteins under 50 amino acids in size (sproteins). These sproteins are hard to predict, ignored by traditional annotation pipelines and are hard to detect using traditional genetic and biochemical methods. Using this empirical gene determination, a number of leadless open reading frames initiating novel genes were identified, ~130 in M. smegmatis and ~197 in Mtb, but there could be more if leadered genes are included. These novel genes are only about 50 amino acids in size, so it is hard to envision a structure or function, but many are conserved, indicating they are likely to have a functional role. As an example, two sproteins were identified that were demonstrated to have a functional role as part of the M. smegmatis ribosome. Additional novel sproteins and newly identified genes can be explored on the Wadsworth interactive genomics site (http://www.wadsworth.org/research/scientific-resources/interactivegenomics/). Transcriptional and translational profiling better informs about the coding capacity of the genome and can define novel gene starts, novel proteins, including functional small proteins and non-coding RNAs, as well as novel gene architectures, leaderless genes and functionally relevant isoforms and antisense genes. The limitations to this approach include: data are collected as a population average, and thus do not inform on the level of a single cell; genes will only be detected if they are expressed under the conditions being used; it does not provide information on post-translational modifications or protein abundance and turnover. This approach will identify potential novel genes but will not provide a functional annotation, and requires validation with additional genetic, proteomic and metabolic analyses. What is critical is that gene annotation is accurate and comprehensive and is most informative when taking an integrated systems biology approach.

Karen Dobos from Colorado State University followed with a review of proteomic methods and studies in mycobacterium. Approaches include traditional label free methods like spectral counting, which allow for quantitation of individual proteins in a sample, as the more of a protein is present in a sample, the more tandem MS spectra are collected for the peptides from this protein [178]. Other methods require labeling, including DiGE (2-D Fluorescence Difference Gel Electrophoresis), iTRAQ (Isobaric Tag for Relative and Absolute Quantitation), and other related methods. Methods are available to apply proteomics to evaluate changes in proteins levels in both in vitro and in vivo settings [179,180]. Proteomics has also been proposed to be developed for clinical applications. One recent example is the application of antibody enhance MS detection of Mtb biomarkers in blood as an improved TB diagnostic. This method combines antibody-labelled, energy-focusing nanodisks with high-throughput mass spectrometry to enhance the detection of Mtb-specific peptides in digested serum samples. MS methods not only detect peptides, but also can determine the quantity of peptides this method can also be used to monitor response to therapy [181]. For many applications, a comprehensive view of all proteins in a sample is needed and a number of new methods has been developed to address this need. DIA/SWATH-MS (Data Independent Acquisition/Sequential Window Acquisition of all THeoretical Mass Spectra) is a new method that permits quantitative monitoring of all peptides and proteins during a reaction regardless of the dynamic range and sample complexity. This method does not require technique development for specific peptides and proteins, can be retroactively mined, however, it requires knowing what specific proteins are being monitored. For DIA/SWATH-MS, non-labelled protein samples are digested with trypsin and the resulting peptides are analyzed by liquid chromatography coupled to a tandem mass spectrometer and all ionized compounds of a given sample that fall within a specified mass range are fragmented in a systematic and unbiased fashion [182]. A method that can be used for absolute quantitation of proteins is Multiple Reaction Monitoring Mass Spectrometry (MRM MS), which is a targeted mass spectrometry approach that has been used for quantification of small molecules in clinical samples, including blood serum and urine. MRM MS for Mtb in vitro was shown with a study that demonstrated strain variation in levels of Antigen 85 protein, despite a consistency in transcription [183]. Another example of a clinical application of MRM MS is the identification of exosomes in serum for patients with LTBI, stratified from classical plasma proteins, and detecting products from tissue leakage was able to identify LTBI specific serum biomarkers. Except for a few scant examples, the majority of most mycobacteria proteomics studies use bottom-up proteomic approach while relatively untapped top-down proteomic methodologies may be of value to explore in the future [184].

Michael Berney from the Albert Einstein College of Medicine presented on mycobacterial metabolism and metabolomic methods. Relative to those grown in synthetic media, intra-macrophage mycobacteria exhibit a high degree of metabolic heterogeneity [185]. Metabolomics is the most direct window into what is going on inside Mtb in either media at a given time and under a given condition, as metabolites are substrates and products of active enzymes which maintain Mtb survival [186]. For these reasons, metabolomics can be used to better understand how Mtb adapts to different stresses [187], identify vulnerable pathways that can be exploited for new drug targets [188,189], and detect mycobacteria-derived volatiles in human fluids (e.g. urine) and breath [190]. Metabolomics methodologies fall into two distinct groups (untargeted metabolomics, which look at a comprehensive analysis of all the measurable analytes in a sample including chemical unknowns; and targeted metabolomics, which measure defined groups of chemically characterized and biochemically annotated metabolites, and does not capture unknown metabolites), and have demonstrated that Mtb are incapable of scavenging intermediates of methionine and S-adenosylmethionine (SAM) biosynthesis from the host [189], and the aspartate pathway as being particular important for Mtb bacilli that survive within granulomas [188]. The methionine, SAM and aspartate pathways therefore represent new spaces for anti-TB drug development that were revealed by metabolomics. Metabolomics has also been used to predict the mode of action by which orphan compounds limit mycobacterial growth [191], identified novel ethambutol metabolites in urine [192] (which may be useful for monitoring antibiotic treatment adherence), and identify volatile organic compounds (VOCs) that consistently distinguish between healthy controls and individuals hospitalized for suspicion of pulmonary TB [193]. It may come as no surprise that TB-specific VOCs exist, as the ability of trained African giant pouched rats to detect TB via smell is well documented [194–198], and may serve as the basis for a “triage test” for case finding among smear-negative TB patients in resource-limited settings [199].

The final talk of the session was presented by Travis Hartman from Weill Cornell Medicine on lipidomics. Although mycobacterial lipids have long been acknowledged as important immunogenic components and virulence factors [200–202], new analytical methodologies [203, 204] enable us to appreciate their role in cellular physiology. Like metabolomics, transcriptomics, and proteomics - lipidomics furnishes a high-resolution snapshot that enable both qualitative and quantitative assessment of samples using dedicated bioinformatics tools. Because as much as 40% of the dry weight of Mtb is comprised of lipids, understanding the chemical perturbations associated with the lipid pool can be an informative readout of Mtb physiology and provide evidence for mechanistic explanations of phenotype [205]. These inferences are possible because there are specific lipid modifications that are well-established biomarkers of oxidative stress [206], for example. Within the lipidome of Mtb, there is considerable diversity, including lipids that are unique to Mtb. The lipidome can now be characterized at a high-resolution using LC-MS (and MS/MS) using methods previously described [203]. To capture the heterogeneity of the in vitro lipidome, Mtb is grown as a “cake” on a filter that can be rapidly transferred into a quenching solvent - an experimental methodology that is also amenable to transcriptomics and metabolomics. Manipulations of this setup, including the use of different media or drug concentrations under the filter, are performed in order to understand how conditions influence the lipidome. Additionally, an unpublished “infectomics” analysis using matched samples from a mouse model, a rabbit model, and human clinical samples demonstrated there are many TB-specific lipid markers that are sufficient to differentiate caseous versus uninfected tissue samples. This work might identify biomarkers to predict bacterial burden in tissues and describe disease progression in a high-throughput manner. The value of big data and databases were emphasized, including The Tuberculosis Database (TBDB) [207], and sharing of data through public databases was encouraged. Particularly, searchable data repositories that contain well-annotated experimental data deserve a consistent and ongoing funding source, as a shared resource that benefits groups with varied research interests.

10. Animal models of mycobacterial disease

Animal models of mycobacterial diseases have been essential to establishing our knowledge of mycobacterial disease pathogenesis and immunity and have provided directions for translational studies. The purpose of this panel discussion was to evaluate the strengths, limitations, and special considerations required when using different animal species for TB research. While the primary focus was on their suitability for Mtb translational research, models for M. bovis and M. marinum were also discussed. Species covered in detail in this session included guinea pigs, ferrets, fish, cattle, marmosets, and macaques. Current research, when relevant, was also mentioned. It was clear from the presentations that each species had a heterogenous combination of advantages and disadvantages, making it crucial to identify precisely the goal of a given research project. MHM8 speakers who focused on these and other aspects of animal models included Carly Kanipe (Iowa State University, Iowa), Don Ennis (University of Louisiana, Louisiana), Ann Rawkins (Public Health England, UK), Angelo Izzo (Colorado State University, Colorado), Fred Quinn (University of Georgia, Georgia), Frank Verreck (Biomedical Primate Research Center, The Netherlands), Sally Sharpe (Public Health England, United Kingdom), and Laura Via (National Institutes of Health, Maryland).

An overarching theme was the ability of each research model to mimic human disease both immunologically and pathologically. In natural infections, cattle have been found to be infected with Mtb but field data are scarce. In an experimental infection model of cattle aerosol infection with Mtb, cattle produced Mtb-specific immune response but no detectable disease [21,208]. Ferret studies have demonstrated that aerosol transmission of Mtb occurs but whole-body exposure improves transmission efficiency. While non-human primates could be naturally or experimentally infected with M. bovis, this was not a prominent focus. Fish were the only represented species to naturally or experimentally host M. marinum, and guinea pigs were considered solely experimental hosts to both Mtb and M. bovis. With regards to pathology of disease, in contrast to mice, all presented models developed characteristic granulomas. Of distinction however, only non-human primates develop cavitary lesions similar to humans. Non-human primates had an additional benefit of having well-defined age ranges, which could be matched to corresponding human ages. Interestingly, it was pointed out that not all non-human primates are immunologically equivalent. For example, rhesus macaques (M. mulatta) appear more prone to disease and have increased anti-inflammatory responses than do cynomolgus macaques (M. fascicularis), which have an increased monocyte TNF response. Finally, the topic of coinfections with parasites and other pathogens was discussed. It was emphasized that ferrets serve as a valuable model for natural influenza coinfection, a trait only shared with non-human primates. Research evaluating coinfection with influenza and other respiratory pathogens is unique to the ferret model, and was considered at MHM8 to be an important future research direction given the number of countries that are simultaneously affected by TB and influenza [209,210].

Surprisingly, in addition to cost, the size of the animal was another topic that garnered attention. The use of cattle, the largest of the research models, allows for tissue and repeated large quantities of blood to be collected. Unfortunately, while readily available and moderately priced, the cattle’s size proves a significant disadvantage when it comes to housing infectious animals and the use of advanced imaging techniques, such as PET scans. Therefore, granuloma dynamics studies are severely lacking. Fish, in contrast, are the smallest animal model allowing rapid gross examination, but with limited volumes of individual samples. Fortunately, their cost and housing allow for small volumes of individual samples is made up for by large numbers of animals that can be used in treatment groups. Guinea pigs, ferrets, and marmosets, while significantly larger than fish, can only tolerate small volumes of blood sampling, significantly reducing the size of a testable sample. While requiring more space than fish, ferrets and guinea pigs remain relatively easy to house and reasonably affordable. Moderate amounts of blood can be sampled from macaques and marmosets while these animals are still small enough for imaging using PET and MRI. With the exception of non-human primates and mice, available immunological reagents for the other model systems vary greatly. Cost and housing of non-human primates is however the most expensive of the models discussed.

At the conclusion of each model overview, urgent needs and goals were presented. All models continue to look for immunological markers to indicate protection and disease severity. If they can be found, it would dramatically improve vaccine efficacy testing. While species-specific immunological reagents have improved, many of the panelists felt this was a continuing limitation in their research. Using animal models, researchers have identified species-specific differences in current diagnostic products for TB in cattle and pigs (e.g. Bovigam) [211–213]; the ability to rapidly identify tuberculous cattle and pigs is economically important, since M. bovis readily transmits between bovinae/suidae and can have a significant, detrimental impact on farmers’ income [1]. In sum, the models presented possess significant advantages as well as unique challenges compared not only to the mouse model, but also to each other. This makes determination of the research question crucial in selecting the most appropriate model. All models presented during this panel discussion developed characteristic tuberculous granulomas, in contrast to mice, making them especially relevant in translatable pathologic research. Additional considerations when choosing an animal model include housing capabilities, volume and number of samples required, and species of mycobacterium under investigation.

11. Mouse models and imaging workshop

Mice are the most widely used animal model for studying host responses to TB and other mycobacterial diseases. The size, cost, genetic tractability, and availability of mouse-specific reagents all contribute to the attractiveness of using mice as a small animal model. Early studies that used mice predominantly relied on inbred laboratory strains including C57BL/6J mice. The mouse model workshop was organized by Igor Kramnik (Boston University, Massachusetts) as a follow up to a mouse model workshop he chaired as part of the MHM7 meeting in 2017. This mouse model workshop brought together researchers who are pushing the boundaries of the traditional murine models, both in terms of strains of mice being used and the tools being employed to study mycobacterial diseases. Several topics that were covered in the workshop included the studies of genetically engineered mice that better recapitulate facets of human disease, novel populations of genetically and phenotypically diverse mice, and novel tools that can be used for phenotypic and genetic analyses of mice, including new advances in technologies for the characterization of disease in vivo. Mouse model workshop speakers included Janice Endsley (The University of Texas Medical Branch, Texas), Gillian Beamer (Tufts University, Massachusetts), Sherry Kurtz (Food and Drug Administration, Maryland), Clare Smith (University of Massachusetts Medical School, Massachusetts), Andreas Kupz (James Cook University, Australia), Igor Kramnik (Boston University, Massachusetts), Eric Nuermberger (Johns Hopkins University, Maryland), Vitaly Ganusov (University of Tennessee, Tennessee), Shumin Tan (Tufts University, Massachusetts), Alvaro Ordońez (Johns Hopkins University, Maryland), and Ismaheel Lawal (University of Pretoria, South Africa).

Many aspects of the host immune response to mycobacteria were discovered using the C57BL/6 mouse model, yet these mice lack several of the characteristic pathologies found in human disease, such as caseating granulomas. Additionally, C57BL/6 mice are also an inbred mouse population with little genetic diversity. Several of the presenters at this workshop are utilizing novel mouse populations that overcome some of these limitations. Igor Kramnik discussed new findings based on the mouse model he developed, the C3HeB/FeJ mice. When these mice are given an aerogenic, low dose Mtb challenge, they develop well defined lung granulomas with central caseous necrosis that more closely resemble the granulomas seen in humans or non-human primates [214]. The type I interferon pathway is central to the development of this pathology, and he presented several new pieces of data that further develop the role of type I interferons and the stress response pathways in host immunity to Mtb infection.

Several investigators at the workshop are utilizing new, related mouse populations; the Collaborative Cross (CC) mice and the Diversity Outbred (DO) mice. The CC mice are a panel of inbred mouse lines created from a strategic breeding of 8 founder mouse strains [215]. The DO mice were created as an outbred population during the generation of the inbred CC lines [216]. Both sets of animals maintain a large pool of genetic diversity, though the CC lines are maintained as individual inbred lines, and the DO mice as genetically heterogenous outbred individuals. Clare Smith used the CC lines to interrogate host genetic factors involved in primary resistance to Mtb, as well as those related to vaccine-induced protection [217]. Additionally, she has employed transposon mutant libraries of Mtb strains to evaluate the linkage between bacterial physiology and factors required for pathogenesis within a diverse host genetic background. Gillian Beamer demonstrated that in the genetically heterogenous DO mice, there was a wide diversity of outcomes in the hosts following aerogenic Mtb infection [218]. Quantitative trait loci (QTL) mapping of animals from the primary infection studies led to several novel genetic loci candidates that may be linked to primary host resistance to Mtb. Sherry Kurtz presented work using a complimentary approach to elucidating host genetic factors necessary for vaccine-induced protection against Mtb using the DO mice. As was seen with primary infection, vaccinated Mtb-challenged DO mice exhibited a wide range of disease outcomes.

Drug development and understanding co-morbidities of mycobacterial infection were also discussed in this session. HIV co-infection with Mtb is of great public health interest, and Janice Endsley developed a mouse model to help understanding the interplay of these infections within a host. A bone marrow, liver, thymus humanized mouse model has well organized granulomatous lesions following Mtb infection, and new data demonstrated that HIV-coinfection alters the course of TB [141]. Another approach by Andreas Kupz investigated the utility of a lymphatic infection model of Mtb to study host responses, and also to mimic the situation of an HIV co-infection [219]. Additionally, mice can also be used to screen novel drug candidates and drug regimens. Eric Nuermberger used variable doses and routes of delivery of Mtb and other mycobacterial species to establish various infection models in which to test new anti-mycobacterial compounds [220]. This work addressed the efficacy of drugs in combination therapy in hopes of reducing relapse rates.

Examining infection from the perspective of the bacteria can also lead to discoveries in host biology. There have been long standing questions about the status of the Mtb bacilli in the host over time, especially during quiescent/chronic/latent infection. Vitaly Ganusov examined replication rates of Mtb in both the mouse and rabbit models, using a reporter plasmid to evaluate chronic infection and whether Mtb strains are dormant/non-replicating, slowly replicating, or replicating and dying at equal rates. Enumerating plasmid segregation in replicating bacteria and mathematical modeling of segregation rates suggests that there is a dynamic bacterial population where replication rates change over time, including during the chronic stages of infection [221]. Shumin Tan presented work that exploits various reporter strains of Mtb that can read out aspects of the host microenvironment surrounding the bacteria, such as pH, hypoxia, and oxidative stress [222]. These bacterial strains contain florescent proteins whose expression is driven by promoters that respond to the different environmental cues in vitro and in vivo and have led to important insights into the host-pathogen interactions.

The MHM8 Mouse Model workshop ended with presentations in the field of whole-body imaging. Radiography and PET/CT have been used in humans and non-human primates to visualize the progression of TB over time, though these technologies have been less widely adapted for use in small animals such as mice. Alvaro Ordonez has been advancing functional imaging technologies to visualize the disease environment in the lungs of mice. These studies employ various nuclear imaging reagents that can inform about aspects of host biology at the sites of infection, such as hypoxia and apoptosis [223]. Ismaheel Lawal also presented work using PET/CT and FDG labelling to examine lung pathology in patients during the course of TB drug treatment after infection and used the PET/CT results to predict treatment failures [224].

In summary, there have been many advances made in the field of TB research and mouse models presented at this workshop. The new strains of mice and techniques being employed to study Mtb infection are leading to important new discoveries about the pathogenesis of the organism, the immune response to infection, and mechanisms to treat the disease.

12. Concluding remarks

The 2019 MHM8 meeting that was held at the Albert Einstein College of Medicine (EINSTEIN) in Bronx, NY, comprised presentations and ideation sessions by stakeholders working to advance research into leprosy, NTM/environmental mycobacteria infection, Buruli ulcer, human TB, animal TB, mycobacterial disease comorbidities, and animal models of mycobacterial disease. A workshop was also held to promote refinement of the mouse TB model, as mice remain the most widely used animal for in vivo testing. As with previous MHM meetings, the collegial atmosphere and opportunity to develop new collaborations drew in scientists, veterinarians, physicians, and policy makers from across the globe. Finally, in addition to the scientific discussions, time was taken to acknowledge the EINSTEIN Animal Institute staff and technicians whose work is essential for scientific progress, as well as to commemorate the passing of two important figures in the field: Drs. Ian Orme and Don “Mark” Estes. Throughout his career at the Trudeau Institute (Saranac Lake, NY) and Colorado State University (Fort Collins, CO), Ian Orme pioneered the mouse and guinea pig model for TB research and made seminal discoveries regarding the roles of T cells during TB progression and vaccination. Mark Estes dedicated his career to vaccine and therapeutic drug discovery in the fields of TB, HIV, and bioterrorism research; his career spanned professorships at the University of Missouri (Columbia, MO), University of Texas (San Antonio, TX), University of Texas Medical Branch (Galveston, TX) and—most recently—the University of Georgia (Athens, GA). Drs. Orme and Estes will undoubtedly continue to impact the mycobacterial field through their literature and multiple trainees.