Development of a Skin Test for Bovine Tuberculosis for Differentiating Infected from Vaccinated Animals

Animals.To investigate the skin test performance of defined antigens in cattle naturally exposed to M. bovis, SICCT reactor cattle were recruited from United Kingdom farms with a confirmed history of bovine TB (convenience sampling dependent on availability). Animals were identified during routine surveillance operations where their SICCT response (PPD-B − PPD-A) was >2 mm and were housed in secure segregated biocontainment facilities at the Veterinary Laboratories Agency (VLA) Weybridge. A total of 62 reactors were used in the study, which included 21 male and 41 female cattle of seven different breeds, including dairy and beef stock. The age at the time of skin test investigations ranged from 9 months to 11 years, with a median age of 21 months. All reactors underwent detailed postmortem examination to assess for the presence of M. bovis infection in accordance with previously described procedures (23). Visible lesions consistent with bovine TB and/or culture isolation of M. bovis were confirmed for 55 of these reactors. For the specificity studies, Holstein-Friesian castrated male cattle (n = 54) were sourced from United Kingdom TB-free farms and housed at the VLA in accommodations separate from those of the reactors. A cohort of these calves (n = 20) was neonatally vaccinated with BCG to allow the evaluation of reagents for DIVA. A single dose of 1 × 10⁶ CFU of BCG Danish (Staten Serum Institute, Sweden) was administered subcutaneously to each of these calves at 4 to 6 weeks of age. Where a second skin test was performed for some cattle during the study, it was performed no earlier than 62 days later. All cattle experiments were cleared by local ethical review and animal procedures under a license granted by the British Home Office.

Antigens.Bovine and avian PPDs were supplied by the VLA Tuberculin Production Unit. Histidine-tagged recombinant proteins were expressed in Escherichia coli cells and purified by nickel affinity chromatography. MPB70, MPB83, ESAT-6, and CFP-10 were supplied by Lionex Diagnostics and Therapeutics GmbH (Germany), and Rv3615c was produced at the Agri-Food and Biosciences Institute (Belfast, Northern Ireland). Rv3615c is referred to in the current paper using the M. tuberculosis genome annotation (http://genolist.pasteur.fr/TubercuList/ ). In the M. bovis genome, it is annotated as Mb3645c (http://genolist.pasteur.fr/BoviList/ ). All synthetic peptides used in the study were synthesized by solid-phase F-moc (9H-fluoren-9-yl-methoxycarbonyl) chemistry and supplied at a purity of >90% (Pepceuticals Ltd., United Kingdom). The identity of each peptide was confirmed by mass spectrometry. For ESAT-6 and CFP-10, 16-mer peptides with 8-amino-acid overlaps of the complete protein sequences were used (21 peptides in total). For MPB83, a single 20-mer peptide that had previously been shown to encode a bovine T-cell epitope (p195-214) was used (24), and similarly, for Rv3615c, three bovine T-cell epitopes encoding 20-mer peptides were used (p65-84, p73-92, and p84-103) (18).

Skin test procedures.The disclosing SICCT field test was performed and interpreted according to European Communities Commission Regulations (regulation 1226/2002, amending annexes A and B of the consolidated Council Directive 64/432/EEC). For the evaluation of defined antigen combinations at VLA, up to eight intradermal injection sites were used on each animal, four on each side of the neck. PPD-B and PPD-A (2,500 IU) were always included in each test, and all antigens were administered in a 0.1-ml volume. Defined antigens were tested at a concentration of 10 μg whether used individually or as a component of a cocktail unless otherwise stated. Skin induration at administration sites was measured by using calipers by the same operator for all experiments. Measurements were recorded prior to and at 72 h after skin test and additionally at 96 and 120 h for some experiments. Results are expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings.

IFN-γ whole-blood ELISA.Heparinized whole-blood cultures were stimulated with either PPD-A (10 μg/ml), PPD-B (10 μg/ml), recombinant protein antigens (5 μg/ml), protein cocktail (5 μg/ml per constituent), staphylococcal enterotoxin B (1 μg/ml; Sigma-Aldrich, United Kingdom), or the no-antigen control. Cultures were set up within 8 h of blood collection and cultured at 37°C in 5% CO₂. Antigen-stimulated IFN-γ was measured in whole-blood culture supernatants at 24 h using the commercially available Bovigam enzyme-linked immunosorbent assay (ELISA) (Prionics AG, Switzerland). Results are expressed as the background-corrected optical density measured at 450 nm (OD₄₅₀).

Statistical analysis.Differences in response magnitudes between different antigen combinations were compared by repeated-measures analysis of variance (ANOVA), with the application of Bonferroni's multiple-comparison postanalysis test. Differences between antigen responder frequencies were analyzed by using the Fisher exact test. These statistical analyses and receiver operator curve (ROC) analysis were performed by using the Prism 5 software program (Graphpad Inc.).

A defined protein cocktail of MPB70, MPB83, ESAT-6, and CFP-10 induces skin test responses in naturally infected cattle.To assess the potential of defined antigens as diagnostic skin test reagents, we initially formulated a cocktail comprised of the purified recombinant M. bovis/M. tuberculosis protein antigens ESAT-6, CFP-10, MPB70, and MPB83. MPB70 and MPB83 were included in our cocktail since they had a previous demonstrated skin test potential for DIVA in a guinea pig model of M. bovis infection (24). In 37 SICCT-positive cattle, the protein cocktail induced a response in 29 of 37 (78%) of these cattle at 72 h posttest (Fig. 1). Skin reactions were additionally measured at 96 and 120 h since it was previously reported that the development of bovine skin test responses to ESAT-6 continued to increase after 72 h (16). In comparison with the 72-h reactions, the SICCT responses were significantly reduced at 96 and 120 h posttest (Fig. 1). The magnitudes of protein cocktail-induced reactions were comparable at each time point. The measurement of antigen-induced IFN-γ in whole-blood cultures set up on the day of the skin test demonstrated that ESAT-6 and CFP-10 were the most strongly recognized constituent antigens in the protein cocktail in vitro (Fig. 2). The protein cocktail was next tested at titrated dose concentrations of 10, 5, and 1 μg for each cocktail component. The respective responses in a cohort of 19 of the previously tested reactors are shown in Fig. 3. Skin test responses were comparable between the 10- and 5-μg doses but were significantly reduced at the 1-μg dose, demonstrating a dose response. For all further experiments, a skin test concentration of 10 μg per defined antigen component was used.

• Open in new tab
• Download powerpoint

FIG. 1.

Protein cocktail-induced skin test responses in cattle. Skin test responses were measured at 72, 96, and 120 h in cattle naturally exposed to M. bovis (n = 37). (A) Comparative SICCT PPD responses (PPD-B − PPD-A). (B) Responses induced by a protein cocktail of 10 μg each of ESAT-6, CFP-10, MPB70, and MPB83. The increase in skin induration for each animal is represented by open circles, and the horizontal line provides the mean (± standard error of the mean [SEM]), with results expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings. The statistical difference between responses was determined by using ANOVA (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

• Open in new tab
• Download powerpoint

FIG. 2.

In vitro recognition of protein antigens. The in vitro capacity of antigens to induce IFN-γ in blood from cattle naturally exposed to M. bovis (n = 37) was determined for ESAT-6, CFP-10, MPB70, and MPB83 either individually (5 μg/ml) or as a combined protein cocktail (5 μg/ml per constituent). Responses induced by PPD-B (10 μg/ml), staphylococcal enterotoxin B (SEB) (1 μg/ml), or the no-antigen control were also determined. The antigen-induced IFN-γ response for each animal is represented by open circles, and the horizontal line provides the mean (±SEM), with results expressed as the background-corrected OD₄₅₀.

• Open in new tab
• Download powerpoint

FIG. 3.

Skin test protein cocktail dose titration. A skin test dose titration was determined for the protein cocktail comprised of ESAT-6, CFP-10, MPB70, and MPB83 in cattle that were naturally exposed to M. bovis (n = 19). The protein cocktail was administered at concentrations of 10, 5, and 1 μg for each antigen component. Protein cocktail and SICCT (PPD-B − PPD-A) responses measured at 72 h for each animal are represented by open circles, and the horizontal line provides the mean (±SEM), with results expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings. The statistical difference between responses induced by the protein cocktails was determined by using ANOVA (*, P < 0.05; ***, P < 0.001).

The specificity of this protein cocktail was first determined for naïve cattle (n = 20). The cocktail induced no measureable response in any of these animals when reactions were read at 72 h (Fig. 4). While all SICCT responses were also negative, measurable PPD-A-biased responses did indicate some sensitization of these animals by environmental mycobacteria. We additionally measured reactions at 96 and 120 h. While responses remained negative at 96 h, at 120 h two animals provided positive reactions of 2 and 12 mm to the protein cocktail and 5 mm to the SICCT test (data not shown). For all further experiments, reactions were recorded only at 72 h posttest.

• Open in new tab
• Download powerpoint

FIG. 4.

Skin test specificity of a defined protein cocktail in naïve and BCG-vaccinated cattle. Skin test responses were measured for the protein cocktail comprised of ESAT-6, CFP-10, MPB70, and MPB83 (10 μg for each component) and for the SICCT test (PPD-B − PPD-A) in naïve TB-free (n = 19) and BCG-vaccinated (n = 20) cattle. Responses measured at 72 h for each animal are represented by open circles, and the horizontal line provides the mean (±SEM), with results expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings.

The potential of the protein cocktail for DIVA was evaluated by using 3-month-old calves that had been neonatally vaccinated with BCG at 4 to 6 weeks of age (n = 20). For these calves, only one animal failed to induce an SICCT PPD-B-biased skin test response, confirming the strong tuberculin skin test sensitization of BCG vaccination (Fig. 4). In contrast, the defined protein cocktail did not induce a response in any of these vaccinates. Protein cocktail and SICCT responses measured in either the naturally infected or BCG-vaccinated cattle were used to generate ROC curves to provide estimates of skin test sensitivity and specificity for DIVA. Table 1 summarizes the sensitivity and specificity estimates for SICCT cutoffs of >2 mm and >4 mm since these represent operational “severe” and “standard” test cutoffs used in United Kingdom surveillance operations (www.defra.gov.uk/foodfarm/farmanimal/diseases/atoz/tb/control/tuberculin.htm ). Also shown are the performance figures for SICCT and protein cocktail test cutoffs that provide 100% specificity for the respective tests for DIVA.

Optimization of defined skin test reagents, including the use of Rv3615c and synthetic peptides. In vitro assessment of the capacity of the constituent protein cocktail antigens to induce IFN-γ suggested the immunodominance of ESAT-6 and CFP-10 (Fig. 2). Therefore, we investigated the additive benefit of the inclusion of MPB70 and MPB83 to a cocktail already containing ESAT-6 and CFP-10. In a group of 12 previously untested naturally infected cattle, all of which induced positive SICCT responses (data not shown), 6/12 cattle induced a response to the cocktail containing all four antigens (1- to 6-mm responses), while only three of these cattle responded to the cocktail containing only ESAT-6 and CFP-10 (2-, 3-, and 6-mm responses). Although the difference in responder frequency was not significant, the result did indicate an additive benefit of including MPB70 and/or MPB83 to the cocktail.

To determine if the inclusion of both MPB83 and MPB70 was required for optimal skin responses, protein cocktails formulated with and without MPB70 were tested in a further 13 naturally infected cattle. Additionally, we used this experiment as an opportunity to evaluate the following two novel defined antigen formulations: first, a protein cocktail containing Rv3615c, a recently identified antigen for DIVA which can identify ESAT-6- or CFP-10-unresponsive M. bovis-infected cattle (18), and second, a synthetic peptide cocktail comprising overlapping peptides of ESAT-6 and CFP-10. The inclusion of MPB70 demonstrated no improvement in the responder frequency or magnitude of skin reactions in these cattle compared with those responses induced by the ESAT6-, CFP-10-, and MPB83-containing cocktail (Fig. 5). In contrast, the addition of Rv3615c increased the responder frequency from 10/13 to 13/13 and also resulted in significantly stronger responses (Fig. 5). The ESAT-6-CFP-10 peptide cocktail induced responses in 11/13 cattle, thus demonstrating their potential as skin test antigens.

• Open in new tab
• Download powerpoint

FIG. 5.

Optimization of skin test response-inducing defined antigen combinations. Skin test responses induced by the SICCT test (PPD-B − PPD-A) and a selection of protein- and peptide-based antigen combinations were determined for cattle naturally exposed to M. bovis (n = 13). Protein combinations of ESAT-6 (E6), CFP-10 (C10), MPB70, MPB83 (83), and/or Rv3615c were all tested at a concentration of 10 μg per constituent protein. The ESAT-6 and CFP-10 (E6/C10) peptide cocktail comprised 21 peptides at a concentration of 10 μg per peptide. Responses measured at 72 h for each animal are represented by open circles, and the horizontal line provides the mean (±SEM), with results expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings. The statistical difference between responses induced by the indicated cocktails was determined by using ANOVA (*, P < 0.05).

We next investigated the skin test contribution of the individual antigens present in the most promising protein cocktail comprised of ESAT-6, CFP-10, MPB83, and Rv3615c. We also extended our evaluation of synthetic peptide combinations by testing a cocktail of peptides derived from ESAT-6, CFP-10, MPB83, and Rv3615c. These reagents were tested in 12 reactor cattle that had previously demonstrated skin test responses to the protein cocktail in the previous two experiments. All nine animals that responded to any single antigen recognized CFP-10 (cutoff for positivity of >1 mm), while seven, three, and two cattle responded to ESAT-6, MPB83, and Rv3615c, respectively (Fig. 6). However, the combined protein cocktail provided maximal responder frequency and reaction size, including the three animals that failed to respond to any of the individual proteins. Encouragingly, the peptide cocktail induced responses in 10/12 of these cattle, as shown in Fig. 6.

• Open in new tab
• Download powerpoint

FIG. 6.

Skin test response-inducing capacity of individual protein antigens. Skin test responses induced by the ESAT-6, CFP-10, MPB83, and Rv3615c proteins were determined either individually (10 μg) or in combination (10 μg per protein) for cattle naturally exposed to M. bovis (n = 12). Additionally, responses to a cocktail of 25 peptides (10 μg per peptide) derived from ESAT-6, CFP-10, MPB83, and Rv3615c were measured in these cattle, along with SICCT skin test responses. Reactions measured at 72 h for each animal are represented by open circles, and the horizontal line provides the mean (± SEM), with results expressed as the difference in skin thicknesses (mm) between the pre- and post-skin test readings.

Finally, the specificity of the most potent protein combination of ESAT-6, CFP10, MPB83, and Rv3615c was evaluated for 14 new naïve cattle and 19 of the BCG neonatally vaccinated calves previously tested. The protein cocktail did not induce responses in any of these naïve or vaccinated cattle. While 16/19 of the BCG vaccinates still demonstrated a PPD-B-biased IFN-γ response at this time (PPD-B − PPD-A, >0.1 OD₄₅₀ [data not shown]), it should be noted that only 3/19 of the BCG vaccinates still elicited a PPD-B-biased skin reaction with this test due to the extended period of nearly 12 months following neonatal vaccination at this time (data not shown).