The cell wall and capsule of Mtb contain a large number of glycolipids, lipoproteins, and glycoproteins such as cord factor, phthiocerol dimycocerosates, phosphatidylethanolamine, diacyl trehaloses, lipoarabinomannan, phosphatidyl-myoinositol mannosides, and heparin-binding adhesin, etc. (18, 19) They can activate immune responses and serve as candidate antigens or adjuvants for TB vaccines.

Mtb can secrete numerous proteins, some of which can inhibit or induce the host immune response by promoting immune escape or activating immune signaling pathways, respectively. Most of the candidate proteins for existing TB vaccines are based on those found as secreted antigens during logarithmic growth of Mtb, such as Ag85A/B, ESAT-6, CFP10, TB10.4, MPT64, and PPE18 (20). The secretory antigens are ideal candidate antigens for the recombinant protein subunit vaccine because of their strong immunogenicity and ease of heterologous expression and amplification.

The antigens modulated under the DosR regulon are the main proteins involved in the dormant survival process of Mtb. A total of 48 structural proteins are known to be involved in aerobic respiration and carbon monoxide inhibition; representative genes include HspX, Rv2623, Rv2660c, etc. (21) Members of the durable hypoxia response (EHR) regulon are structural genes induced after exposure to hypoxia. EHR proteins are presumed to be involved in the adaptation and survival of bacteria during a long-term bacteriostatic process (22). Members of the DosR and EHR regulons are considered promising antigens to be incorporated into protein subunit vaccines for treating LTBI (23).

Resuscitation promotion factors (Rpfs) are involved in the resuscitation and reactivation of dormant Mtb infection and induce specific humoral and cellular immune responses in individuals with LTBI (24). There are 5 Rpf-like proteins (RpfA, RpfB, RpfC, RpfD, and RpfE) with partially overlapping functions in Mtb. Rpfs, especially RpfB, can trigger a memory T-cell response and has been hypothesized to be an essential antigenic target controlling bacterial activation. Rpfs can be used as candidate antigens for protein subunit vaccines against LTBI infection (25).

BCG strains have structures similar to Mtb, but 16 genomic region of difference (RD) antigens are deficient in BCG compared to Mtb (26). The RD1 gene products contain a variety of potential virulence factors, such as ESAT-6 and CFP10 (26). They play multiple roles in Mtb progression and pathogenicity, and are considered suitable candidates for use in treatment and diagnosis (27). The poor protective effect of BCG may be related to the loss of a large number of genes encoding protective antigens. Therefore, RD antigens should be emphasized in constructing recombinant protein subunit vaccines.

CAF01 comprises cationic surfactant lipid-based liposomes dimethyldioctadecylammonium (DDA) and glycolipid immunomodulator trehalose-6,6-dibehenate (TDB). DDA is a potent adjuvant capable of eliciting cellular and humoral immune responses (46). TDB is a synthetic analog of mycobacterial cord factor that is located in the cell wall of mycobacteria and has intrinsic immunostimulatory properties that activate Mincle (47). TDB incorporated with DDA creates a stable liposome by forming hydrogen bonds between the liposome membrane and the surrounding water. CAF01 has been shown to generate a Th1/Th17 polarization response via Mincle-dependent IL-1 production and subsequent MyD88 signaling (48).

AS01 is a liposome-based adjuvant that consists of the 3-O-desacyl-4′-monophosphoryl lipid A (MPL) and the saponin QS-21 (Quillaja saponaria extract), co-prepared in the presence of cholesterol (73). MPL acts as a TLR4 agonist, stimulates NF-κB transcriptional activity, and induces a Th1 response. QS-21 can enhance the antigen presentation ability of antigen-presenting cells (APCs) and activate/differentiate T cells towards Th1 immune responses. DMT is a combination of the MPL, DDA, and TDB that provides more potent and longer-lasting protective efficacy, including antigen-specific CD4⁺ Th1 response, IFN-γ⁺ CD8⁺ CTL response, and limited humoral response (42).

GLA-SE is a mixture of the TLR4 agonist glucopyranosyl lipid A (GLA) and squalene emulsion (SE) (56). GLA is a synthetic lipopolysaccharide (LPS) derivative that maintains vigorous immunostimulatory activity and has low toxicity (57). SE is able to increase the secretion of proinflammatory cytokines such as IL-6, IL-12, and TNF (58). Both GLA and SE alone can promote IgG2 response, while the combination of GLA-SE can induce considerable Th1 response (56).

IC31 comprises the synthetic, positively charged antimicrobial peptide KLKL5KLK and oligodeoxynucleotide 1a (ODN1a) (64). ODN1a is an immune stimulatory molecule that promotes Th1-biased immune responses through the TLR9/MyD88 pathway. KLK can act as an immune stimulator that aids transfer into cells in the absence of cell membrane permeability, allowing more efficient functioning of intracellular TLRs (65). KLK induces a Th2 immune response when used alone and a stronger Th1 and Th2 immunity when combined with ODNla (66, 67).

Dextran/CpG is a novel adjuvant developed with diethyl aminoethyl (DEAE)-dextran and CpG ODN (83). CpG is a TLR9 agonist with the ability to promote Th1 immune responses (secretion of IFN-γ, TNF-α, and IL-12 cytokines), opsonizing antibodies (IgG2a), and potent CD8⁺ T cell responses (84). Dextran interacts with DC-SIGN family receptors, mannose receptors, and langerin, all triggering innate immunity that promotes inflammation. Furthermore, Dextran/CpG adjuvant enhances activation of lymph node-resident APCs, thus enhancing T-cell priming (29, 83).

Advax is a novel cationic adjuvant based on the Delta inulin isoform and has a diameter of about 1-2 μm (30). Advax-based adjuvants have been shown to promote protective immunity against several pathogens in various animal species (31, 32). The potent chemotactic effect induced by Advax enables leukocyte recruitment to the site of inoculation and elicits a broad range of immune responses, including humoral response, Th1, Th2, and Th17 T-cell responses (31).

BC02 consists of BCG-derived unmethylated CpG DNA fragments and aluminum salts (Al(OH)₃) (51). CpG tends to induce Th1-type immune responses, while alum skews the response to promote the Th2 response to secrete IL-4 and IL-5 cytokines and produce IgG1 and IgE-type antibodies (52). BC02 induces robust Th1 and Th2 responses with acceptable safety (51).

DDA/poly(I:C) is composed of cationic liposome vector DDA and polyriboinosinic acid–polyribocytidylic acid, poly(I:C). Poly(I:C) mimics viral dsRNA and is a promising immune stimulator candidate for vaccines against intracellular pathogens. Poly(I:C) signaling primarily depends on TLR3 and melanoma differentiation-associated gene-5 (MDA-5) (36). Moreover, poly(I:C) induces strong Th1-skewed immune responses, with enhanced IFN-γ, IL-6, IL-12p70 as well as high antigen-specific IgG antibody (37, 38).

With the development of nanotechnology and the increasing understanding of immune responses to metals, different types of inorganic nanoadjuvants have been developed, including manganese (88), iron (89), silicon (90), magnesium (91), and gold-based adjuvants (92), etc. The commonly used polymers are poly-lactic-co-glycolic acid (PLGA), which can be constructed into nano- or larger particles to improve immune response efficiency (93). Compared with the traditional adjuvants, the novel inorganic nanoadjuvants can better activate both humoral and cellular immunity, induce a more balanced Th1/Th2 immune response and improve the safety and effectiveness of vaccines (94). Inorganic nanoadjuvants have been used in vaccines for various diseases, such as coronavirus (95), cancer (91, 96), and pertussis (97). Nanoadjuvants for TB vaccines are also being developed to enhance the immune response and extend the duration of protection (98, 99).

Mice provide the most widely used models due to the advantages of relatively low price, short experiment cycle, mature immunological evaluation indicators, abundant commercial reagents and genetically modified inbred strains (100). The mouse strains most commonly used in evaluating the immune efficacy of TB vaccines are BALB/c and C57BL/6, which are sensitive to TB vaccines immunization routes (101). However, the immune response induced by TB vaccine was different in BALB/c and C57BL/6 mice. M Carmen Garcia-Pelayo et al. found that although BCG present equally protective in BALB/c and C57BL/6 mice, it was display more enhanced Th1 and Th17 response in BALB/c mice than C57BL/6 mice (102). In another study, ChAdOx1.PPE15 as a booster vaccine for BCG improved the efficacy of BCG in C57BL/6 mice, but not in BALB/c mice (103). The susceptibility to TB and the protective responses to the vaccines vary according to the route of infection and immunization. Subcutaneous immunization is the most classic immunization method for TB vaccine, but mucosal immunization has received extensive attention in the pathogenic bacteria infected by mucosal route. A multrivalent chimpanzee adenovirus vectored vaccine developed by Sam Afkhami et al. showed strong protection against both replicating and dormant Mtb through mucosal immunization (104). Previous research by Claudio Counoupas et al. have shown that intratracheal instillation of CysVac2/Advax protected mice more effectively than the intramuscular vaccine (35).

However, despite a high genetic similarity between mice and humans, significant differences in clinical immune responses between mice and humans have stalled clinical trials of many novel vaccines that had previously shown considerable efficacy in murine models. To overcome this problem, humanized mouse models have been extensively studied in recent years. Humanized mice have a reshaped immune system, making the immune responses more like those of humans. They have been widely used in studies of epitopes and epitope-based TB protein subunit vaccine development (105). Although the use of humanized murine models has enabled many advances in TB vaccine research, deficiencies in the models such as the inability to establish LTBI and granulomas (100), abnormal T-cell responses, and the inability to control bacterial load have limited their use.

NHPs can better represent the human immune responses for assessment of the safety and efficacy of TB vaccines and adjuvants due to the close genetic and pathophysiological similarities between NHPs and humans. Rhesus macaques (RM) and cynomolgus macaques (CM) are the most commonly used primate models for TB vaccine research. It is well known that there are differences between macaque species in their ability to control disease progression, with RM showing higher rates of progression and higher levels of bacterial burden compared to CM (106). RM are often used in vaccine evaluation studies because the results of infection are more uniform than CM, while RM are often used in drug evaluation studies because they are better able to control the disease (107). NHPs provide essential insights into host-pathogen interactions during TB infection by simulating the pathogenesis of TB in humans, including the occurrence of LTBI and granuloma formation (108). NHPs can be used to evaluate the immune effect of different vaccine administration pathways and immunization strategies (75, 109).

The use of the NHP models has brought some breakthroughs in TB vaccine development in recent years. First of all, the preclinical evaluation of novel vaccines by the NHP model has facilitated the transformation of vaccines to prevent and therapy Mtb infection (110–112). The ultra-low dose aerosol-infected NHP model better simulates the course of human infection with TB and can accurately evaluate the vaccine immune efficacy (113). Moreover, using NHP makes it possible to study the interactions of cells within lung granulomas, which cannot be done in human samples. Laura Hunter et al. used infection in RM and CM models to determine the basic composition of granulomas induced after infection with the Mtb Erdman strain, as well as the spatial distribution of immune cells in granulomas in RM and CM and changes over time (114). This informs research into TB vaccines and treatments, and may provide novel immunotherapy strategies against TB. Furthermore, the development of body scanning technology, particularly the combination of PET and CT scans, has made it possible to quantitatively evaluate the protective efficacy of TB vaccines in NHP models (115–117). This strategy allows vaccine evaluation in less time and at a lower cost. However, the high cost of the animals and experimental facilities, as well as the limited quantity available, have hindered their widespread application.

Guinea pigs are also a commonly used animal model in the study of TB. Guinea pigs are more susceptible to Mtb than mice and can form classical granulomas similar to humans (118). Therefore, they are suited to studies of the pathogenesis of TB and the assessment of vaccines and drugs (119). Guinea pigs have also been used to study the response of Ag-specific T cells to mycobacterium lipids and lipopeptide-rich Ag preparations (120). Diabetes can fuel TB epidemics, and T2D co-infection with TB has been modeled in guinea pigs in recent years and used to test novel therapy approaches (121–123). However, guinea pigs are more expensive, lack test reagents, and are more difficult to genetically manipulate than mice. Adjuvant subunit vaccines tend to be less protective in guinea pigs than in mice, resulting in few successful trials of adjuvants in guinea pig models (60, 124). The cause of the limited protective immunity provided by adjuvants in guinea pig models awaits clarification, and more tools and reagents are needed for guinea pig models.

The immunity to Mtb infection in neonates is markedly distinct from that in adults. Innate and adaptive immune responses in infants cannot be inferred from adult human or animal models (125). Due to their high similarity to humans in terms of anatomy, genetics, and immune response, pigs are widely used in numerous studies (126, 127). The isolated and sterile state of the porcine fetus during pregnancy is conducive to the study of the interaction between the immature immune system and microorganisms and to determine the changes in the immune structure and function during fetus development (128). Surprisingly, pigs can undergo pathological changes to Mtb infection including caseous necrosis, liquefaction, and cavitation and mimic the immune response of vaccination BCG in humans (129). Mimicking the human neonatal immune system in pigs could improve our understanding of the infant immune response to TB. More neonatal and early-life animal models are needed to advance the development of anti-TB vaccines and drugs for neonates.

Other animal models, such as rabbits, rats, and zebrafish, have also been used in Mtb vaccine evaluation. Depending on the characteristics of each model, they have been used in different ways. Rabbits are usually infected with Mtb by aerosol route (130), and susceptibility to Mtb varies among different populations (131). Most rabbits available today are highly resistant to infection with Mtb (As Lurie’s - susceptible breed have become extinct), but highly susceptible to infection with the closely related Mycobacterium bovis. They can form granulomas, liquefaction, and cavities similar to the events found in humans, making them suitable for the study of processes leading to transmission of Mtb as well as for vaccine and drug research. In addition, the rabbit model has been used in studies of cavitary, spinal, joint, cutaneous, and meningeal TB (132, 133). However, due to the high cost, lack of immune reagents, and the inconvenience of genetic manipulation, the utility of the rabbit model is limited.

The types of rats commonly used in TB studies are American cotton rats, Wistar rats and diabetic rat strains. Several studies have found that rats exhibit delayed hypersensitivity to Mtb infection (134). Mtb infected rats can form well-organized granulomas, including epithelioid cells, multinucleated giant cells and foam macrophages, etc., which provide a common research object for the study of host control of Mtb and the establishment of latent infection (135). Rats are suitable for TB-related gene and protein research and have the advantages of low cost and simple blood collection, befitting vaccine and drug research (135). Yet, pathological changes in human lungs, such as caseous necrosis, fibrosis, calcification, and cavitation, are not formed in rats.

Recently, zebrafish have attracted increased attention as an animal model for TB. Zebrafish infected with M. marinum can form a typical granulomatous structure, which provides an excellent model for scientists to further study the mechanism of granulomatous formation (136, 137). Moreover, zebrafish have the advantages of visual monitoring, convenient genetic manipulation, fast reproduction, and low cost, they are now widely used for bacterial virulence studies and large-scale vaccine and drug screening. The immune responses during granuloma formation and necrosis can be well monitored, making zebrafish one of the best choices for studying latent TB infection (138, 139). Nevertheless, anatomical and physiological differences between zebrafish and humans impede the application of zebrafish models for vaccine development.

TB is characteristically caused by respiratory infection when the smallest aerosol droplets containing only 1 or 2 colonies reach the alveolar spaces (1). Hence, the high-dose challenge that has been typically used in animal models might have contributed to discrepant results between pre-clinical and clinical trials. To better simulate the natural human infection process, ultra-low dose infection models have been developed. Infection of conventional mice with 1-3 CFU Mtb produced granulomas with well-defined boundaries similar to human granulomas (140). In addition, the ultra-low dose aerosol-infected NHP model more closely mimicked the process of human natural TB infection. It is being used as a precise and sensitive system to assess the effectiveness of TB vaccines (113).