Mycobacterium bovis (M. bovis) is the primary agent of Bovine tuberculosis (bTB), a zoonotic disease affecting cattle and wild animals. It represents not only a threat to human health, but it also negatively affects the global cattle industry due to the negative impact on animal production and detrimental effects on animal health (1). This chronic disease is distinguished by the progressive development of the characteristic granulomas, mainly in the lungs and lymph nodes, which limits the spread of mycobacteria into the host (1, 2). The disease can disseminate within a herd before the development of obvious clinical symptoms or can stay latent for years and spread only when the animal is immunocompromised due to additional stresses or old age (1).

Control of bTB relies on early diagnosis, removal of infected animals, and tracing and containment of contact cases. Identification of infected animals is carried out with slaughterhouse surveillance (to identify animals with characteristic TB-lesions) or through ante-mortem assays, such as the intradermal tuberculin test (IDT) and the interferon-gamma (IFN-γ) release (IGRA) blood assay (1, 3, 4).

IDT measures dermal swelling primarily due to cell-mediated immune response (CMI) 72 h after intradermal injection of purified protein derivative (PPD) in the skin of tested animals (Shiller et al., 2010). The IGRA assay is instead a laboratory-based test which measures release of IFN-γ from whole blood (predominantly T lymphocytes) stimulated for 18–24 h with PPDs (5).

Nevertheless, both tests present some disadvantages: they are unable to differentiate M. bovis infection from bTB disease and are incapable of identifying infected animals in the first stage of infection (3, 5). Therefore, several efforts have been made to improve bTB diagnosis and understanding (6) and one of the main areas of bTB research is the discovery of new biomarkers able to early detect M. bovis infected animals.

Cytokines are small proteins that play a crucial role in orchestrating immune responses. They have been widely studied as biomarkers for several diseases, including tuberculosis (7, 8). Their quantification in biological fluids, such as blood, is also characterized by lack of invasiveness and relatively low cost (8).

An early and accurate diagnosis of subclinical infection can indeed improve bTB control strategies (7). Another important achievement would be the identification of biomarkers able to differentiate early infection from a late stage of the disease (6, 7). In this way, the available resources can be focused toward removing animals that pose transmission risks to preserve those with higher economic, and/or genetic value (7). Previous studies reported that bTB infection in cattle could be enhanced by a combination of IFN-γ and the related chemokine IFN-γ-induced protein 10 (IP-10) (7). Other cytokines are involved in immune response against this pathogen, and their potential use as biomarkers of M. bovis requires further investigation (7, 9).

In our work, we evaluated the ability of 16 key immune cytokines to identify M. bovis infected cattle. In addition, we investigated whether these cytokines were able to differentiate infected animals presenting or not TB-like lesions at post-mortem examinations, thereby exploring their value not only for diagnosis but also for the immunological staging of bTB.

Whole blood samples from healthy (N = 19), infected (N = 17), and affected (N = 19) cattle were stimulated with a specific M. bovis antigen (PPD-B), alongside the nil control antigen (PBS) and a control of lymphocyte reactivity (PWM). After 18–24 h, the release of 16 key immune cytokines was determined through singleplex and multiplex ELISA.

Statistically significantly higher levels of IFN-γ, IL-1α, IL-4, IL-10, IL-17, and IP-10 were detected in PWM-stimulated samples compared to PBS-samples (Table 1). This indicated that leukocyte reactivity was not altered by inadequate preservation of the sample or immunosuppression of the cattle due to other pathological events or treatments (e.g., corticosteroids).

Differences between PPD-B-stimulated samples and PBS-samples were also identified. In all groups, higher levels of IL-6 and MIP-1α, were observed in the PPD-B compared to the PBS conditions were observed for IL-6 and MIP-1α, suggesting an inflammatory response to the PPD-B antigen, regardless of previous animals’ exposure to M. bovis (Table 1). On the contrary, the infected and affected groups, but not healthy animals, presented statistically significantly higher levels of IFN-γ, IL-1α, and IP-10 in the PPD-B condition compared to the baseline control (PBS) (Table 1). Moreover, the infected and affected groups, but not healthy animals, presented lower levels of VEGF-A in the PPD-B condition compared to PBS, with statistical significance (p< 0.05). In the affected group only, we observed a higher release of TNF in response to stimulation with PPD-B compared to the PBS control, with statistical significance (p = 0.0004).

Then, M. bovis specific cytokine responses were analyzed. For each cytokine, the differences between the levels in the PPD-B and PBS conditions were quantified. These response differences were then compared among the three groups (healthy, infected, and affected).

T-cell cytokines were first analysed: IFN-γ (hallmark of Th1 response), IL-4 (hallmark of Th2 response), and IL-17 (mainly released by Th17). Affected and infected animals released higher levels of M. bovis-specific IFN-γ (818 ± 99; 409 ± 105, respectively) compared to healthy cattle (5 ± 99) with p < 0.0001 and p< 0.0195, respectively (Figure 1). In addition, differences in M. bovis-specific IFN-γ levels were observed between the affected and infected groups (p = 0.0178) (Figure 1). Affected animals, but not infected cattle, released higher levels of M. bovis specific IL-4 and IL-17 compared to healthy cattle, with p < 0.0007 and p< 0.0003, respectively (Figure 1). In addition, affected animals released higher levels of M. bovis specific IL-4 and IL-17 compared to infected subjects, with p < 0.0195 and p< 0.0236, respectively (Figure 1).

Release of M. bovis specific pro-inflammatory cytokines (IL-1α, IL-6, TNF) in healthy, infected, and affected cattle was then analysed. No significant differences were observed among groups for IL−1α or IL−6. In contrast, affected cattle showed higher levels of M. bovis specific TNF compared with healthy animals (Figure 2).

Subsequently, the release of M. bovis specific anti-inflammatory cytokines (IL-10, IL-36Ra) in the three groups was investigated. For both the anti-inflammatory IL-10 and the receptor antagonist IL-36Ra, no differences between groups were observed (Supplementary Figure 1).

IP-10, MIP-1α, and MCP-1 are chemokines that trigger leukocyte recruitment into the inflammatory site. We observed that affected and infected animals released higher levels of M. bovis-specific IP-10 (1567 ± 135; 1414 ± 148, respectively) compared to healthy cattle (227 ± 135), both with p< 0.0001 (Figure 3). No differences in M. bovis-specific IP-10 levels were instead observed between the affected and infected groups (Figure 4). For both MIP-1α and MCP-1, no differences between groups were detected (Figure 3).

IL-23, IL-27, and IL-35 are members of the IL-12 family, and their M. bovis specific levels in healthy, infected, and affected cattle were determined and compared. For all these cytokines, no differences among groups were observed (Supplementary Figure 2).

The levels of M. bovis specific VEGF-A (vascular endothelial growth factor) and THBS-1 (thrombospondin-1, an adhesive glycoprotein that mediates cell-to-cell and cell-to-matrix interactions) were also analyzed. Affected and infected animals released lower levels of M. bovis-specific VEGF-A (-68 ± 16; -88 ± 17, respectively) compared to healthy cattle (-37 ± 16), although without significance (p = 0.3769; p = 0.0816) (Supplementary Figure 3). For THBS-1, no statistically significant differences between healthy, infected, and affected cattle were observed (Supplementary Figure 3).

The correlation analysis didn’t highlight either high or significant interdependence among the tested variables, as shown in the Supplementary Table 1.

Finally, canonical discriminant analyses (CDA) were used to generate predictive cytokine profiles by groups, to identify potential diagnostic biomarkers of M. bovis infection. A CDA was performed with the 16 cytokines monitored in the study (Figure 4A). The first canonical (Can1) differentiates between healthy and M. bovis exposed animals, whereas the second canonical (Can2) could only modestly allow a separation between infected and affected cattle (Figure 4A). Table 2A reports the factor loading (FL) for each cytokine in canonical variables, but only Can 1 showed a positive and high correlation with Δ_IP-10 (FL = 0.82), Δ_IFN-γ (FL = 0.75), Δ_IL-17 (FL = 0.58), and Δ_IL-4 (FL = 0.52).

To identify the most discriminating combination of cytokines, another CDA was performed using the 5 cytokines that presented statistically significant differences among groups (IP-10, IFN-γ, IL-17, IL-4, TNF). The resulting scatter plot of multivariate outcomes is shown in Figure 4. Using these 5 cytokines, Can1 could still differentiate healthy from M. bovis exposed animals, except for three animals (Figure 4B). The second canonical (Can2) could only modestly allow a separation between infected and affected cattle (Figure 4B). Table 2B reports the factor loading (FL) for each cytokine in canonical variables, showing a positive and high correlation with Δ_IP-10 (FL = 0.89), Δ_IFN-γ (FL = 0.76), Δ_IL-17 (FL = 0.58), and Δ_IL-4 (FL = 0.55) and Δ_TNF (FL = 0.50) in Can 1, while in Can 2 only Δ_IL-17 had a high FL (0.54).

Bovine tuberculosis is a zoonotic disease threatening the cattle industry worldwide. To date, several studies have been conducted to improve its diagnosis and understanding, including the discovery of new biomarkers (7). M. bovis infection triggers the development of a cell-mediated immune response, which precedes humoral responses; accordingly, IFN-γ remains one of the most widely used biomarkers for detecting M. bovis infection in cattle and other species. Several additional cytokines are involved in immune response against this pathogen and their potential use as biomarkers of M. bovis infection in cattle requires further investigation (7, 9). In this study, we evaluated the utility of 16 key immune cytokines as diagnostic biomarkers for M. bovis infection in cattle.

IFN-γ is a cytokine mainly released by NK and activated T cells in the framework of a Th1 response (17). It is routinely used in the diagnosis of M. bovis infection in cattle, and in our work, it was indeed able to discriminate healthy from infected animals. Differences were also observed between cattle presenting (‘affected group’) or lacking (‘infected group’) visible TB-lesions, suggesting that the magnitude of the antigen-specific IFN-γ response might increase with disease progression and lesion development during M. bovis infection.

IL-4 is regarded as a signature cytokine of a Th2 response (18). Previous studies reported that this cytokine was able to distinguish M. bovis infected cattle from healthy subjects, although IL-4 response was delayed compared to the IFN-γ response (19). It was speculated that the IL-4 response reflected a switch from Th1- to Th2-dominated responses in the later stage of M. bovis infection (7). Accordingly, we observed that affected cattle released higher PPD-B-specific IL-4 compared to those belonging to the infected and control groups. Our data suggest that this cytokine might be useful for differentiating stages of bTB infection in cattle.

IL-17 is a cytokine mainly released by Th17 and γδ-T cells and it is a hallmark of a Th17 response (20). Previous studies reported that M. bovis infected cattle presented higher PPD-B-specific release of IL-17 compared to healthy subjects (21, 22). In addition, Blanco and collaborators observed that IL-17 expression was associated with the presence of TB-lesions in infected cattle (23). Accordingly, we observed that cattle, belonging to the affected group (with visible TB-lesions), released higher PPD-B-specific IL-17 compared to those belonging to the infected and control groups, suggesting that this cytokine might be useful in cattle to distinguish bTB infection stage.

IL-1α, IL-6, and TNF are key pro-inflammatory cytokines released during the early stages of infection, which enhance inflammation and trigger the release of chemokines, which in turn enhance the recruitment of leukocytes to the infection focus. We observed no differences among groups in terms of PPD-B specific release of IL-1α and IL-6, whereas affected cattle presented higher levels of TNF compared to healthy subjects. Previous studies in humans described that antigen-specific release of TNF by CD4⁺ T cells was able to distinguish patients with active and latent infection (24) and we recently observed that the frequency of TNF⁺ producing CD4⁺ T cells enabled discrimination between infected/exposed and non-infected Mediterranean buffaloes (25). In agreement, we previously reported that Mediterranean buffaloes with visible TB lesions at the slaughterhouse released higher levels of TNF in response to PPD-B stimulation compared to healthy subjects and those reactive only to the IFN-γ assay (14). Overall, these data suggest that TNF might be a valuable biomarker for distinguishing bTB infection stages across different host species, including cattle.

Two anti-inflammatory cytokines were then evaluated: IL-10 and IL-36Ra (26, 27). Previous studies reported that IL-10 promotes M. bovis survival within macrophages (28) and plays an important role in granuloma formation (29). Nevertheless, no differences between groups were observed in terms of PPD-B specific IL-10 release and this might be due to the diverse cytokine pattern in cells in peripheral blood (evaluated in our study) compared with those present within granulomatous lesions, where IL-10 production is typically more pronounced. No differences were also observed for IL-36Ra in all tested cattle, according to what we previously observed in Mediterranean buffaloes (14).

Chemokines are low-molecular-weight mediators that promote cell recruitment to infected tissues (30–32). We observed no differences between groups in terms of PPD-B specific release of MIP-1α and MCP-1, whereas both infected and affected groups presented higher levels of IP-10 compared to healthy subjects. In agreement, several studies in cattle and buffaloes showed the utility of this chemokine in identifying M. bovis infected animals (3, 14, 33, 34). Studies in humans reported indeed that IP-10 can be a biomarker for tuberculosis in both adults and children. However, IP−10 does not reliably distinguish active TB from latent TB (35, 36). In agreement, our data suggest that IP-10 is not informative for differentiating stages of bTB in cattle, but it might improve early detection of the infection. Future studies with a larger number of animals will be essential to confirm these findings and to assess whether IP-10 can improve early diagnosis of M. bovis.

IL-23, IL-27, and IL-35 are members of the IL-12 family (37) and little is known about their role during M. bovis infection. We observed no differences among healthy, infected, and affected cattle. Our data disagree with those of Sharma and co-authors, where researchers described that PBMC from M. bovis infected cattle presented higher expression of IL-23 in response to PPD-B stimulation compared to those of healthy controls (38). This difference might be due to post-transcriptional mechanisms that inhibit IL-23 protein release from PBMC, despite elevated mRNA expression, in response to PPD-B stimulation.

VEGF-A stimulates angiogenesis and plays a key role in the maintenance of the vascular and lymphatic systems (39) and its role during M. bovis infection in cattle is largely unexplored. Previous studies in humans and mice reported that VEGF-A is implicated in granuloma formation and likely enhances tuberculosis spread in the host (40, 41). Higher serum levels of this growth factor were observed in human patients with symptoms of tuberculosis compared to healthy controls (40). Surprisingly, in our study we observed that cattle from both the infected and the affected groups, but not healthy controls, released lower levels of VEGF-A in response to PPD-B stimulation compared to nil control (PBS). Similar results were previously reported in Mediterranean Buffaloes (14). We might speculate that this downregulation is due to higher PPD-B-specific release of IFN-γ in the infected/affected groups compared to controls. Supporting this hypothesis, previous studies reported that IFN-γ can suppress VEGF-A release by monocytes (42).

THBS-1 is an adhesive glycoprotein which mediates cell-to-cell and cell-to-matrix interactions (43). In cattle, THBS-1 was reported to be down-regulated in cattle experimentally infected with M. bovis (44) and another study described that healthy and naturally infected bTB cattle presented a differential transcription of the THBS-1 gene (45). On the contrary, in our study, we observed no differences between healthy and M. bovis exposed cattle; this might be due to post-transcriptional mechanisms which nullify the difference between groups.

Finally, we aimed to identify the cytokine combinations that better differentiated the three groups. Two canonical analyses were performed: the first one included all the sixteen cytokines analyzed in the study, and the second one included only the five cytokines (IFN-γ, IP-10, IL-4, IL-10, TNF) with statistical differences among groups. In both analyses, a clear differentiation was observed between healthy and M. bovis exposed animals, whereas the separation between infected and affected cattle remained modest. Our data suggest that the quantitative determination of IFN-γ and the related chemokine IP-10 could be useful in the diagnosis of M. bovis infection in cattle, in agreement with previous studies (7). These preliminary observations should be validated to establish whether the parallel measurement of these two cytokines can improve the diagnosis of M. bovis infection in cattle in terms of sensitivity and specificity, and whether IP-10 could allow an earlier identification of infected cattle. In addition, our data suggested that IL-4, IL-10, and TNF might be useful to distinguish cattle in diverse stages of TB infection. Future studies on a larger set of samples will be essential to establish whether the combined measurement of these three cytokines can allow the identification of cattle with limited transmission risks, which might allow temporary retention of animals with high economic and/or genetic value under appropriate management and biosecurity conditions.

In parallel to the discovery of new biomarkers, another main area of bTB research is the identification of new antigens to improve IGRA specificity. PPD-B provides a wide variety of antigens that can be presented to lymphocytes, reflecting the range of antigens to which the host is exposed during infection, but shares epitopes with other mycobacteria, thus it shows limited specificity (46). Several studies investigated new antigens to improve IGRA specificity, such as ESAT6 and CFP10, which are potent T-cell–stimulating proteins secreted by M. tuberculosis and M. bovis (47–49). More recently, more complex antigen formulations have been evaluated in both IGRA and IDT, such as DST-F (ESAT-6/CFP-10 plus Rv3615c) and MDT, which comprises ESAT-6/CFP-10, Rv3615c, Rv3020c, Rv1789, Rv3478, and Rv3810 (50, 51). Other studies investigated the ability of the protein complex P22 to improve bTB diagnosis (52, 53). P22 is composed of several immunodominant antigens recognized by T cells, including MPB70, MPB83, ESAT-6, and CFP-10, and may provide greater specificity than PPD-B (53).

Future studies should combine both strategies (discovery for new biomarkers and identification of more specific antigens) to improve bTB diagnosis. The selected cytokines (IFN-γ, IP-10, IL-4, IL-10, TNF) should be quantified in vitro not only in samples stimulated with PPD-B, but also in response to stimulation with other antigens such as ESAT-6, ESAT10, or P22.

Overall, the data generated in our study provide a foundation to improve both the diagnosis and the immunological staging of M. bovis infection in cattle.