Old World monkeys, including the rhesus (Macaca mulatta) and cynomolgus (Macaca fascicularis) macaques, and New World monkeys (Callithrix jacchus, the common marmoset), are used as experimental models of human TB disease. They all develop the full range of Mtb disease observed in humans, including children and immunosuppressed patients, ranging from solid lesions to caseation, calcification, and cavitation (30, 56–59, 81). Each NHP model has benefits for certain types of studies; for instance, rhesus macaques have been shown to be more susceptible to Mtb infection compared with cynomolgus macaques who can show active as well as latent TB infection (58, 60). Depending on the strain of Mtb used, different rates of disease progression, as well as varying degrees of cavitation, are observed in the common marmoset (59). Two different outcomes of infection from Mtb have been seen in the cynomolgus macaque, depending on the challenge dose. It has been shown that when high doses of Mtb are administered, severe disease develops (acute TB) (55); conversely, a low-dose can result in asymptomatic infection which is similar to latent TB in humans (56, 61, 62).

Varying the route of infection does not result in different outcomes in terms of clinical parameters and behavioural changes between the macaque subspecies (63); however, lesion distribution does vary in the lung, with diffuse lesions more commonly associated with the aerosol route as compared to the intra-bronchial route (63, 64).

A range of pulmonary granulomas are observed in active TB in all NHP models. These range from early lesions consisting of small aggregates of immune cells, mainly macrophages with variable numbers of lymphocytes and neutrophils, and lacking clearly defined boundaries (65); the classical, organised, ‘solid’ structured granulomas which comprise abundant, epithelioid macrophages, neutrophils and some MNGCs in the absence of necrosis; granulomas that have a central core of necrosis but are not caseous and are surrounded by a rim of epithelioid macrophages and a peripheral rim of lymphocytes; and, large, well-defined, caseous granulomas (57, 62, 65). Cavitary lesions are rarely observed in the Old World NHP models during active disease (62); however, they have been noted in the common marmoset between 58 and 70-days post-challenge (dpc) when infected with the CDC1551 strain (59). TB pneumonia is also observed during active TB in all NHP models, and it is usually surrounding multicentric, often coalescing, large, caseous granulomas (59, 62). A summary of the main histopathological features in NHPs is represented in Table 1.

The majority of more advanced lesions mentioned above are also observed during latent TB infection (LTBI); however, granulomas without necrotic centres have not been noted. Large caseous granulomas are observed although the caseous core has often been replaced by mineralisation and/or collagenous material (62) defined as ‘fibro-calcified’ lesions by some groups (62). Also observed are lesions that are termed ‘sclerotic or fibrosing’ granulomas; lack visible mineralisation and comprise compacted, sclerotic material (62).

Depending on the strain of Mtb used and the species of NHPs, microscopic granulomas can be seen as early as 2–5 weeks post-challenge (wpc) (57, 65, 66). In rhesus macaques, early granulomas often lack clearly defined boundaries, but by 3 wpc, more ‘classical’ granulomas are seen (66). Central necrosis is also seen this timepoint when infected with H37Rv (65, 66). In cynomolgus macaques, however, the earliest manifestations of granulomas within the lung are present at 4–5 wpc, comprising multifocal to coalescing, caseous lesions (57). These granulomas are more like the ‘classical’ granuloma. A centre core of necrotic cells and amorphous material that is surrounded by a peripheral rim of epithelioid macrophages. In this zone, MNGCs may be observed where a number of macrophages have joined together to create a Langhan’s type giant cell (67, 68). In smaller, unorganised granulomas, MNGCs tend to be scattered throughout the lesion rather than in an outer rim (68). MNGC’s tend to be more abundant in the later, more advanced granulomas compared to the earlier manifestations (68). Beyond this there is rim of lymphocytes that consists of a mixture of B and T cells. It has been observed that as granulomas increase in size, B cell populations become more contained as ‘follicle-like’ structure in the lymphocyte rim, whereas in the smaller granulomas they tend to be scattered throughout this layer and the granuloma (15, 68–70). This is in contrast to findings from Fuller et al. (71) who state that B cells are only observed in follicle-like structures in caseous granulomas in cynomolgus macaques. Depending on the stage and extent of healing and containment of the granuloma a thin layer of fibrosis can develop around the external rim of lymphocytes (82). AFB are seen in most granulomas, but they are more abundant in larger granulomas with caseous necrosis (83).

The cellular composition of granulomas in NHPs used to distinguish between the type and severity of disease have been evaluated in various ways including semi-quantitative methods, for example, scoring systems (57, 65) as well as quantitative methods such as using image analysis to count the percentage of cells stained within a granuloma (31, 68). Scoring systems and lesion classifications have been used in NHP models in recent years and there are two that lead in the field; Lin et al. (57) whose system is based on 3 characteristics; the type of granulomas (caseous, solid, suppurative or mixed); the cellular composition (presence or absence of lymphocytic cuff, mineralisation, fibrosis, MNGCs and epithelioid macrophages) and the distribution pattern (focal, multifocal, coalescing and invasive). Lin et al. (57) have applied this to both active and latent TB experiments. The second is Rayner et al. (63, 65, 66) who published a more detailed description of the granulomas seen in active, early-stage TB infection which uses different stages of development to categorise the granulomas (stages 1–6). Briefly, stage 1 granulomas are small diffuse foci of cells including macrophages and lymphocytes and a few disperse neutrophils. They do not have clearly defined boundaries and they infiltrate alveolar walls and extend into the alveoli (Figure 2A). Stage 2 granulomas are unorganised but have a more defined boundary and they are larger than stage 1 but have the same mix of inflammatory cells (Figure 2B). Stage 3 granulomas are similar to stage 2, but they have a focal necrosis which is characterised by nuclear pyknosis and karyorrhexis with the loss of cellular architecture. Stage 4 lesions are circumscribed and primarily consist of macrophages mixed with neutrophils and other leucocytes and a few peripheral lymphocytes. Stage 5 granulomas are organised with a necrotic foci and degenerate neutrophils (Figure 2C). Finally, stage 6 is the classical, well demarcated granuloma that consists of central, caseous necrosis and has a variable rim of lymphocytes, mainly B cells, around the periphery (63, 65, 66) (Figures 2D–F). The use of this scoring system has been of great value evaluating vaccine candidates at the preclinical stage (32, 83).

The mouse (Mus musculus) was one model utilised by Robert Koch to demonstrate that Mtb induced comparable lesions in mice to those observed in humans during natural infection (33, 34). The broad utilisation of mice in TB immunology research is attributed to the abundance of immunological tools and reagents, as well as the availability of inbred strains, which facilitate classical genetics and adoptive transfer experiments (13).

In general, mice tend to develop an acute rather than chronic infection with Mtb, and the granulomas formed in their lungs lack the structured and organised appearance observed in human granulomas (9) (Figure 3A). It should be noted that in most of the conventional strains of mouse models, granulomas do not develop the necrotic caseous centre, which is the primary characteristic of human TB (9, 13, 20, 35). Infected macrophages tend to show an enlarged cytoplasm with high lipidic content (foamy macrophages) that are interspersed with lymphocytes and neutrophils (Figure 3B). The presence of AFB is abundant within the macrophage cytoplasm and the necrotic cells (Figures 3C,D). Additionally, the lack of MNGCs in mice makes the tuberculous granulomas of mice histopathologically different from those observed in humans (17, 20, 36) (Figure 3). Table 1 summarises the main histopathological features in mice model observed during experimental pulmonary TB.

Numerous mouse models of infection have been developed, using mostly the intranasal infection route with the aim of mimicking pulmonary TB in humans as close to a natural infection route as possible (20, 24, 28, 36–41).

Rhoades et al. (20) infected C57BL/6 mice via the aerosol route with varying doses of three different Mtb strains and studied the development of a granuloma for a period of 1 year in the lung. Although the different conditions failed to induce a caseating necrosis, the mice developed a chronic form of pulmonary TB in all cases. Rhoades et al. (20) described five histopathological stages of granuloma development; category 1: small, isolated lesions diffusely distributed throughout infected lungs, composed of a few adjacent alveoli with thickened septae, consisting primarily of mononuclear phagocytes, alveolar macrophages and occasional lymphocytes; category 2: scattered, discrete foci of alveolitis filled with mononuclear phagocytes and a few epithelioid and foamy macrophages, with perivascular and peribronchiolar lymphocytes; category 3: moderate lesions characterised by sheets of epithelioid and foamy macrophages that fill the alveoli, with mild interstitial fibrosis and tight associations of lymphocytes. In this case, while evidence of irreversible cell degeneration and death was noticed, no central necrosis was detected. In category 4 enlarged, coalescing granulomatous lesions were detected, consisting mainly of the presence of macrophages, epithelioid macrophages, and large foamy macrophages. Small foci of necrosis and polymorphonuclear cells were associated with cleft of cholesterol and fibrosis of alveolar septae was more advanced. Finally, lesions in category five are defined as extensive chronic, interstitial fibrosing granulomas, characterised by thickened alveolar septae demarcated areas filled with dead/dying epithelioid and foamy macrophages (20). With high doses of Mtb, single AFB were detected by ZN stain in several alveolar macrophages and polymorphonuclear cells of category 2 granulomas after 20 dpc. Multiple AFB were more apparent after 60 dpc in epithelioid cells and foamy macrophages and no differences between categories 3 and 4 were found (20).

Cardona et al. (36) also used the same mouse model, although with different conditions, and they proposed a new histopathological approach in the development of pulmonary lesions: primary, secondary, and tertiary granulomas. Primary granulomas (category 3 in Rhoades et al. (20) classification) consisted of an initially poorly structured lesion with neutrophils, lymphocytes and infected macrophages surrounded by a thick mantle of lymphocytes, in which periphery foamy macrophages filled the alveolar spaces. By day 220 post-challenge, macrophages are the predominant cellular element at the core of granulomas. In secondary granulomas, small numbers of infected macrophages are surrounded by lymphocytes. On day 60 post-challenge, peripheral lymphocytes formed a thick mantle which is surrounded by foamy macrophages. Tertiary granulomas consist of several neighbouring granulomas coalescing and remain confined within the alveolar spaces and did not elicit any significant fibrous reaction. Tertiary granulomas could be considered as categories 2, 4 and 5 from Rhoades et al. (20) classification because they may represent primary and/or secondary granulomas linked with foamy macrophages (36). Although Cardona et al. (36) remarked on the absence of MNGCs and the scarcity of epithelioid cells, both authors agree that foamy macrophages are a fundamental cell in mouse tuberculous granulomas (20, 36). Indeed, the continued enlargement of granulomas is heavily reliant on the presence of this particular cell type, as its proliferation plays a major role in the merging of lesions and the subsequent formation of large-sized granulomas (36).

Tsai et al. (17) evaluated significant differences as well as common characteristics in lung granulomas from humans and C57BL/6 mice with TB. A common feature observed in both human and mouse granulomas was the presence of B lymphocyte aggregates (17). However, some differences were observed between the two species. While in mouse lungs B cell aggregates were surrounded by macrophages, in human tissues this was not observed. In the latter, a characteristic association of B cells and T cells was observed, with two different patterns of expression (17).

Other groups have used the Balb/C mouse model for the study of pulmonary tuberculosis (28, 39–41). This model showed two histopathological phases of the disease. The first one or acute phase was characterised by an infiltrate of inflammatory cells in the alveoli, blood vessel and bronchiole with the formation of granulomas. The second phase was chronic pneumonia characterised by focal necrosis (40). Using the same mouse strain, Irwin et al. (39) demonstrated similar lesions to category two described by Rhoades et al. (20) at 3 wpc which became multifocal to coalescing with the progression of the disease. No evidence of necrosis or fibrosis was described with this model (28, 39–42). Pan et al. (43) demonstrated that C3HeB/FeJ mice (commonly referred to as the Kramnik mouse model) displayed lung pathology with a central caseous necrosis and encapsulation, similar to what is observed in human tissues (43). Using the same model, it had been observed that these lesions were also hypoxic (pimidazole+ by IHC), another key feature of human TB lesion, as previously mentioned (27, 28). C3HeB/FeJ mice develop three different pulmonary lesions following aerosol infection; type I is the one that most resembles classical human TB granulomas. They become evident after 35–45 dpc and are characterised by acellular caseum in the centre of the lesion, karyorrhectic debris and a distinct band of darkly stained intact neutrophils. This band is surrounded by a rim of foamy macrophages and a fibrotic capsule (24, 28, 39, 41). In humans, the central core is composed primarily of macrophages and foamy macrophages, however, in this mouse model it appears to be composed mainly of neutrophils and, to a lesser extent, foamy macrophages (24). SYBR Gold acid-fast stain, revealed large number of intracellular bacilli in the foamy macrophages and intact neutrophils and extracellular bacilli within the necrotic caseum (24). The type II lesions in infected C3HeB/FeJ mice presented fulminant granulocytic pneumonia and type III were similar to lesions found in Balb/c mice following aerosol infection (24). Occasionally, this model demonstrated the development of central cavities, like the ones observed in human tuberculosis (28, 41). Recently, a ‘humanised’ mouse model generated by transplanted human foetal liver derived haematopoietic stem cells have been proposed (HIS-NSG mice) (42). TB granulomas in these HIS-NSG mice comprised solid non-necrotic granulomas, tuberculoid pneumonia, and caseous necrotic granulomas, having the ability to model human-like pulmonary granuloma initiation and formation (42).

In contrast to mice, the use of rats (Rattus spp.) for the study of human tuberculosis is less common. However, this species could also represent a good model due to the well documented immunology and the number of monoclonal antibodies which are commercially available (84). Studies before 2006 described lung granulomas as histologically similar to classic murine models, with the absence of central necrosis (21, 84, 85). In the Lewis rat strain, Sugawara et al. (85) described granulomas of various sizes from 3 wpc, observing foamy macrophages and MNGCs with the progression of the disease (84, 85). Sugawara et al. (84) described a model of TB in F344/N-rnu nude rat strain which developed granulomas with central necrosis and encapsulated by thick collagen fibres, mimicking those observed in humans. However, unlike humans, the central necrosis was not caseous (84).

The American cotton rat (Sigmodon hispidus) is a useful model in many human respiratory diseases such as respiratory syncytial virus (RSV) or influenza virus (86). In 2007, Elwood et al. also demonstrated its value as a model for pulmonary TB (21). Cotton rats developed a granulomatous inflammation, which was visible macroscopically, demonstrating multiple white nodules in the lung parenchyma. At 8 wpc, localised and discrete granulomas surrounded by healthy lung tissue and composed of epithelioid macrophages in the centre and lymphocytes at the periphery were observed. Caseous necrosis was frequently observed which became calcified with disease progression. Multiple AFB were noted within the cytoplasm of epithelioid cells and intracellular and extracellular in necrotic granulomas (21). Later, Singhal et al. proposed another model in Wistar rats endotracheally infected with different strains and doses of Mtb, which revealed some similarities with human TB (44). Histopathological analysis of lung from rats infected with high dose of W4 Mtb strain revealed granulomas with a central area of epithelioid macrophages and a cuff of lymphocytes, together with small peribronchial aggregates from 28 dpc onwards. At 60 dpc, well organised granulomas with many lymphocytes, foamy macrophages and occasional MNGCs were observed, although necrosis and fibrosis were seldom seen (44).

Guinea pigs have been used extensively in TB research; they share many features of disease with humans, making this species a popular choice as an animal model of TB model of human disease. The histopathological features of the TB lesions in the guinea pig, represented in Table 1 and Figure 4, have been well documented. Small lesions comprising mainly macrophages, with some neutrophils and lymphocytes, arise after a few days post aerosol challenge with Mtb. These lesions are often located close to large airways (45) and it has been suggested that these initial lesions are more likely to originate in the interstitium rather than alveolar spaces (46). This occurs through a process whereby bacilli escape from macrophages adhered to the alveolar epithelial surface and subsequently infiltrate into the surrounding, pulmonary tissue. Furthermore, neutrophils are thought to facilitate early lesion growth via degranulation and release of hydrolytic enzymes. At these early time points, the pulmonary lymphatics are also affected, with lymphangitis observed as early as 5–15 dpc and involving the connective tissue supporting delicate, peribronchial and perivascular, lymphatic vessels (47). These lymphangio-centric lesions continue to develop into granulomas in parallel with those located in the parenchyma. Macrophages, both epithelioid and foamy, and lymphocytes, continue to infiltrate lesions, resulting in larger, circumscribed ‘solid’ granulomas by days 15–20 post-challenge. As the granuloma enlarges, it compresses adjacent blood vessels, causing tissue hypoxia (48, 49). Although collagen deposition may form around the periphery of the granuloma as part of its progressive, structural development, fibrosis does not play as prominent a role as in other animal models such as NHPs or cattle with bTB; furthermore, pre-existing collagen in broncho-vascular connective tissue can be incorporated into the periphery of lesions that develop in these areas.

In addition, around days 15–20 post-challenge, signs of necrosis become evident; alongside this are increased numbers of granulocytes. Around this time point, the number of acid-fast bacilli observed in lesions also plateaus, coincident with the development of adaptive immunity (50).

This process rapidly progresses to prominent, central necrosis by around day 30 post-challenge (Figure 4A). It has been proposed that neutrophils are likely to be the key cells that initiate necrosis formation (46), as opposed to macrophage ‘softening’ and degradation. Dystrophic mineralisation, mainly calcification, of the necrotic core, a key feature in disease progression in this species, follows shortly afterwards (Figure 4B). Advanced staining techniques have identified numerous clumps of over 15 AFB within the necrotic core and its periphery, the latter of which is often observed as highly eosinophilic rim (51). Moreover, special stains have identified accumulation of extracellular ferric iron in areas of lesion necrosis, and iron accumulation in macrophages at the periphery of granulomas (52); this is commensurate with the requirement of Mtb for host iron to facilitate growth and promote increased virulence.

A prominent feature of pulmonary TB in guinea pigs is the formation of secondary lesions which may arise from a number of scenarios and reflect fundamental differences in host environments (53). Orme and Basaraba (46), propose that those located in a subpleural location may arise from the passive transportation of bacilli from primary lesions through pulmonary lymphatics towards the pleura where they lodge and develop lesions. Haematogenous spread can also result in secondary lung lesions; this has been demonstrated to occur within 2–3 wpc (53). As an adaptive immune response is likely to have established by the time secondary lesions are developing, their structure differs from their primary counterparts. In general, they are small, predominantly lymphocytic, lack the discrete granuloma morphology as well as neutrophils, necrosis and calcification; the number of intra-lesional bacilli are also reduced (53). This microscopic appearance is akin to primary lesions that form in the lungs of guinea pigs vaccinated with BCG prior to infection with Mtb (53); this results in partial protection from the adaptive immune response stimulated by the vaccine, and a reduction in the severity of the inflammatory response. Infection invariably leads to progressive disease, with the time taken to reach humane endpoints dependent on variables such as the mycobacterial strain and dose (87).

Many of the features described for guinea pigs are observed in TB infection in people. Both species develop large, pulmonary granulomas with central, caseating necrosis and surrounded by epithelioid macrophages, occasional MNGCs and a rim of lymphocytes and chronic, fibrous encapsulation (Table 1). Furthermore, prominent lymphadenopathy and lymphangitis are also shared features, as well as disease dissemination. Despite the paucity of other key morphological features noted in human disease, such as post-primary, cavitating lesions that facilitate rapid spread but that are observed uncommonly in the guinea pig, this species is considered the gold standard for the preclinical evaluation of both potential new vaccines and therapeutics (88).

Evaluation of differences in the type and severity of disease has traditionally seen the employment of semi-quantitative methods, such as subjective scoring systems (72, 73) which estimate extent of disease and numeration of particular features such as number of foci of necrosis or calcification. Quantitative analysis, determining percentage of lung affected by disease, such as consolidation, can also be estimated using proprietary software packages. Lesion classification systems have been devised to grade granuloma type in the lung of the guinea pig model (45). Our group has recently developed a new scoring system to differentiate morphological characteristics in a temporal pattern and which have been observed in both parenchyma and broncho-vascular, connective tissue (74). Briefly, stage I lesions incorporate small, poorly demarcated collections of inflammatory cells, primarily macrophages, with some lymphocytes and granulocytes. Larger, circumscribed and well demarcated, non-necrotic lesions comprising macrophages (mainly epithelioid), scattered lymphocytes and variable numbers of neutrophils, are considered stage II lesions; and these progress to stage III lesions when central necrosis is visible, often with a degree of caseation. The final stage, stage IV, contains granulomas with variable but often extensive, central, dystrophic calcification and caseating necrosis (Figure 4).

Reports on immunostaining of granulomas have enabled further characterisation of granulomas; Turner et al. (45) describe the lymphocytic population in the non-necrotic, solid granulomas, comprising of significant numbers of CD4+ cells with fewer CD8+ cells; interestingly, in contrast to mice, these cells were distributed randomly within the lesions without aggregating. By 30 dpc, numbers of these cell numbers decrease prominently and an increase in B cells and granulocytes is seen, with a concomitant worsening of pathology (75). More recently, our group have used a number of cell markers (CD3, CD4, Iba1, myeloperoxidase, Mac387) to describe presence and frequency of inflammatory and immune cells in granulomas up to 27 wpc. We have identified macrophages (Iba1+) as the predominant cell marker visible in all granuloma stages. By contrast, MAC387+ (calprotectin) staining in macrophages and granulocytes was noted in scattered granulomas, with an increase in staining correlating with lesion progression up to stage III and noted early in between weeks 3–4 post-challenge. Similar changes were reported for myeloperoxidase (MPO) in granulocytes, with particularly strong staining noted in stage IV granulomas between 8 and 27 wpc. Percentage of CD3+ staining was lowest in all granulomas at the early time point of 3 wpc, whilst maximum at 8 wpc, and similar expression of staining in all granuloma types noted at the later time points, up to 27 wpc. A progressive increase in B-cell marker quantification was observed from stage I to IV. Higher expression of this marker was detected at 8 wpc in comparison with weeks 4 and 24–27 wpc in stage I and II granulomas. As more markers become available for use in the guinea pig model, further characterisation is likely to occur.

Some consider the rabbit model good as a model of immunopathogenesis and rabbits can be experimentally infected with Mtb or M. bovis and they show similar characteristics as human TB disease, for example, caseous necrosis, liquefaction and cavitary disease (79, 80, 89, 90). Kaplan et al. also have developed a model of mycobacterial meningitis that closely resembles TB meningitis in people (90). The different strains of mycobacteria cause different manifestations of disease, for example, when infected with Mtb strain CDC1551, rabbits show similarities to latent TB infection when compared to humans as they can clear the bacilli and granulomas disappear (89). However, an infection with M. bovis creates a progressive TB disease that can lead to death (80, 91).

In the 1920’s, Lurie and their team bred rabbits that had different susceptibilities to Mtb strains, and these were used to look at the pathogenesis of TB in rabbits (92, 93). Unfortunately, these inbred rabbits no longer exist, and the outbred strains of rabbits tend to be resistant to Mtb. However, in 2004, Dorman et al. evaluated a new inbred strain of New Zealand White rabbits that appeared to be more susceptible to Mtb infection when compared to the outbred strains (79).

After experimental challenge, lesions can be seen as early as 2 wpc with increased cellularity within the alveolar spaces (89). The strain of Mtb used can affect the outcome of disease. CDC1551 strain has a different outcome when compared to HN878 strain and Erdman strain. Multiple granulomas are observed with strain HN878 at 4 wpc which consist of scattered lymphocytes, neutrophils, and macrophages. These lymphoid aggregates are often perivascular and peribronchial. MNGCs are frequently observed in macrophage abundant areas (89). However, by 4 wpc with strain CDC1551, small, scattered aggregates of cells are observed (89).Inbred rabbits have large granulomas with frequent caseous necrosis and obvious AFB by 5 wpc with Erdman strain, whereas in outbred rabbits the granulomas tend to be smaller with less caseous necrosis and few viable AFB (79). By 8 wpc with HN878 strain, granulomas are more abundant and central necrosis is present with abundant AFB (89), and in rabbits with CDC1551 infection the granulomas have increased in size, started to coalesce, and have become organised structures similar to the classical granulomas in humans; with a central core of necrosis, macrophage layer and an outer rim of lymphocytes (89, 94). At 12–16 wpc in HN878 infection, central necrosis increases, liquefaction can be seen in some granulomas, and cavities can also form. Some granulomas, however, have been observed to reduce in size an become mineralised. It has been noted that the mineralised granulomas have only a few AFB (94). At 12–16 wpc in CDC1551 infection, it has been observed that granulomas are organised and some have no AFB present. It is also not uncommon for granulomas to have been reabsorbed to a point where only increased cellularity in the parenchyma is seen (89).

Infection with Mtb Erdman strain produces coalescing granulomas with some lesions developing cavities at 16–18 wpc. It has been observed that some rabbits have ‘healed’ granulomas at this time point whilst others have a mixture of small and large granulomas (80). Rabbits challenged with M. bovis tend to have a chronic infection with TB pneumonia, fibrotic granulomas and cavitary disease (26, 91). These histological features are summarised in Table 1.

Over the past few decades, the zebrafish (Danio rerio) has emerged as a widely utilised alternative vertebrate animal model in the study of mycobacterial disease (122). This species can be infected with Mycobacterium marinum (M. marinum), a genetically close relative of MTBC and a natural pathogen affecting poikilothermic species, in which it induces a systemic TB-like disease (106). Fish mycobacteriosis, resembling human TB, can manifest as either an acute infection or a chronic progressive disease characterised by the containment of mycobacteria in well-organised granulomas structurally similar to those caused by Mtb in humans (123–125). Remarkably, despite lacking lungs, zebrafish exhibit high genomic homology with humans, and possess innate and adaptive immunity systems that are comparable to mammals (122, 126, 127). Furthermore, this species displays disease phases observed in humans, including latency and reactivation (125). These shared features make zebrafish an exceptional model for understanding the disease progression and host responses involved (122, 126, 127).

In the natural environment, M. marinum spreads through water and primarily enters the host through the oral route, whereas in experimental conditions, zebrafish are commonly infected through intraperitoneal or intramuscular inoculation (124, 128, 129). The initial mycobacterial dose plays a critical role in determining the outcome of infection (125). Parenteral high-doses (>500 CFUs) lead to a more progressive and active disease with higher bacterial burden and mortality rates, while lower doses (~5–90 CFUs) can result in a chronic process with latent disease characterised by a stable numbers of granulomas (124, 128, 129). Following a successful M. marinum challenge, both larval and adult zebrafish models develop granulomas consistent with those reported in human TB, which can be found in various organs such as pancreas, gonads, kidney, liver, and occasionally the brain (106, 129). In embryos and larvae, early granuloma formation initiated by the innate immune response, can be visualised using in vivo real-time imaging within a few days post-challenge due to their transparency. These early lesions primarily consist of aggregated macrophages, few recruited neutrophils, epithelioid cells and MNGCs surrounding mycobacteria (123, 125, 130, 131). In adult zebrafish, initial TB-like granulomas characterised by cellular and bacterial aggregation have been reported at 2 wpc using both high and low infection doses. In advanced stages of the disease (16–20 wpc), mature granulomas are observed, featuring a caseous necrotic core surrounded by infected macrophages, epithelioid cells, neutrophils, infiltrating T and B lymphocytes, and a fibrous capsule (124, 129, 132–135). However, unlike mammalian tuberculous granulomas, zebrafish lesions contain fewer lymphocytes, and the presence of mineralisation has not been reported (124, 129, 131). Table 1 summarises the main histopathological features in zebrafish model observed during experimental pulmonary TB.

Various evaluation methods have been employed to study the TB-like lesions in zebrafish, enabling a comprehensive analysis of disease progression and the involvement of innate and adaptive immune responses. The transparency exhibited during the embryo and larval stages, offers a distinctive feature for investigating the initial stages of mycobacterial pathogenesis in vivo (131). Several authors have employed advanced in vivo real-time imaging techniques, including the utilisation of diverse immunofluorescence techniques to mark leukocytes and macrophages in situ (130, 136, 137). These techniques have demonstrated the ability of embryonic macrophages to phagocytose M. marinum within 1 h post-challenge, followed by the development of early granulomatous lesions 3–4 dpc (136).

Histopathological characterisation of the lesions has also been described using conventional histopathology and scoring systems. Risalde et al. (135) and López et al. (138) classified the granulomas into four histopathological stages based on their extent and cellular composition. Type I involved an infiltrate of epithelioid macrophages surrounding scattered mycobacteria, without necrotic areas or fibrous capsules. Type II represented partially encapsulated granulomas with a cluster of epithelioid macrophages and initial signs of necrosis. Type III consisted of encapsulated and well-organised granulomas containing regions of partial and complete necrosis, along with the presence of AFB. Type IV depicted encapsulated granuloma with an extensive necrotic core with a high mycobacteria concentration. Additionally, ZN staining revealed the presence of AFBs in both low and high infection dose experiments, primarily associated to necrotic areas (129, 132, 135, 138). In this context, Risalde et al. (135) and López et al. (138) developed a semi-quantitative assessment method to evaluate the presence of AFBs within the granulomas, categorised as: absent (0), very scarce (1–9), scarce (10–29), moderate (30–36, 54–71, 81–83), intense (36–53, 72–75, 84–100), very intense (>100).

Although the zebrafish has its limitations as a TB animal model, its distinctive characteristics have made significant contributions to our understanding of TB and mycobacterial disease in general. By providing valuable insights into host-pathogen interactions, immune responses, and the complex dynamics of TB granuloma formation, the zebrafish model serves as a crucial bridge between in vitro cell cultures and mammalian models, facilitating the translation of research findings to human clinical trials.

Many other animal species can be infected by MTBC bacteria. Natural TB in wildlife species such as badgers and wild boar (Sus scrofa) have been extensively studied due to its epidemiological importance in bTB, representing major reservoirs of infection for domestic livestock (139).

The histopathological features of granulomas in wild boar and domestic swine are similar to those observed in cattle and the differentiation of granulomas by developmental stages (stages I-IV) has been used in this species (140). Interestingly, wild boar can also show small lesions with heavy mineralisation and thick encapsulation, with very few immune cells and AFBs (95). The scoring system used for pulmonary lesion characterisation has been very valuable in this species to study the effect of different therapies against MTBC infection (141). Minipigs (Sus domesticus) have also been proposed as an excellent experimental model for the study of human TB, mimicking some histopathological characteristics observed in humans, such as granuloma encapsulation, and other parameters related with the immune response (142).

The Eurasian badger (Meles meles) is the main reservoir for bTB in the British Isles (143), and have been targeted for experimental vaccination to reduce the prevalence of bTB in these territories. The pulmonary lesions in naturally or experimentally infected badgers are typically solid, composed mainly of epithelioid macrophages with fewer lymphocytes, no apparent MNGCs and limited areas of necrosis and mineralisation (144). BCG has been licensed as a vaccine for TB in badgers after extensive experimental work, having used a version of the four developmental stages of the granulomas described by Wangoo et al. (5) in 2005 as a base for a histopathological scoring system principles (145, 146).

Following the 3Rs recommendations (replacement, reduction, and refinement), other less-conventional animal models have been used to study MTBC infection. For example, wax moth (Galleria mellonella) larvae, a well characterised model of infection for enterobacteria (147) has been recently used to study Mtb infection (148). It has been shown that phagocytic haemocytes can show internalisation of AFBs and cell aggregation towards a rudimentary ‘granuloma-like’ structure. Even though the model does not show many features of the typical pathology after mycobacterial infection, it can serve as a valuable tool to study the innate responses and identifying MTBC virulence genes (149). Moreover, novel in vitro models like the lung-on-chips are being developed to study the MTBC pathogenesis and will serve in the future to screen potential therapies against the disease reducing the number of animals used (150).

M. orygis has also been isolated in both humans and animals in S Asia and India. There are ongoing studies to evaluate whether this MTBC subspecies is a proxy for zoonotic TB. Currently, there is no consensus as to whether M. orygis should be identified as zoonotic TB along with M. bovis which is presently the only zoonotic MTBC subspecies defined by the WHO (151).