Serological detection of Mycobacterium Tuberculosis complex infection in multiple hosts by One Universal ELISA

Tuberculosis (TB), a contagious disease mainly caused by Mycobacterium tuberculosis (M. tb), Mycobacterium bovis (M. bovis), and Mycobacterium caprae (M. caprae), poses a major global threat to the health of humans and many species of animals. Developing an ante-mortem detection technique for different species would be of significance in improving the surveillance employing a One Health strategy. To achieve this goal, a universal indirect ELISA was established for serologically detecting Mycobacterium tuberculosis complex infection for multiple live hosts by using a fusion protein of MPB70, MPB83, ESAT6, and CFP10 common in M. tb, M. bovis, and M. caprae as the coating antigen (MMEC) and HRP-labeled fusion protein A and G as a secondary antibody. After testing the known positive and negative sera, the receiver operating characteristic curves were constructed to decide the cut-off values. Then, the diagnostic sensitivity and specificity of MMEC/AG-iELISA were determined as 100.00% (95% CI: 96.90%, 100.00%) and 100.00% (95% CI: 98.44%, 100.00%) for M. bovis infection of cattle, 100.00% (95% CI: 95.00%, 100.00%) and 100.0% (95% CI: 96.80%, 100.00%) for M. bovis infection of sheep, 90.74% (95% CI: 80.09%, 95.98%) and 98.63% (95% CI: 95.14%, 99.76%) for M. bovis infection of cervids, 100.00% (95% CI: 15.81%, 100.00%) and 98.81% (95% CI: 93.54%, 99.97%) for M. bovis infection of monkeys, 100.00% (95% CI: 86.82%, 100.00%) and 94.85% (95% CI: 91.22%, 97.03%) for M. tb infection of humans. Furthermore, this MMEC/AG-iELISA likely detects M. caprae infection in roe deer. Thus this method has a promising application in serological TB surveillance for multiple animal species thereby providing evidence for taking further action in TB control.

, which share a wide range of hosts but maintain different host preferences [1]. For example, human TB (hTB) is mainly caused by M. tb, but M. bovis of animal origin can infect humans and cause hTB either. According to the Global Tuberculosis Report in 2020 by the World Health Organization, there were about 10.0 million new TB cases and 1.4 million new deaths in 2019 [2]. Among the hTB cases, the proportions of cases caused by M. bovis infection varied in different countries and areas. For example, in Ethiopia, Tanzania, Turkey, and the west bank of Palestine, 17.0%, 26.1%, 5.3%, and 6.5% of hTB cases were reported to be caused by M. bovis respectively [3][4][5][6][7][8]. In addition, M. caprae infection accounted for 0.3% of hTB cases in Spain [9]. On the other hand, more and more publications reported M. tb infection in animals especially bovines [10][11][12]. Sheep TB cases are mainly caused by M. caprae and M. bovis transmitted from cattle under the condition of cohabitation or sharing grazing land. In Spain, sheep cohabiting with infected cattle or goats were reported to have a prevalence of 50.44% caused by the infection with M. bovis or M. caprae [13]. Additionally, wild animals such as deer and monkeys are also susceptible to these three MTBC members [14,15]. Therefore, OIE added the infection of bovids, cervids, goats, and New World camelids by M. bovis, M. tb, and M. caprae into the disease list of the Terrestrial Animal Health Code since 2019. In May 2014, the World Health Assembly passed a resolution of the post-2015 End TB Strategy with its ambitious targets to be fully realized in 2035. However, the maintenance of MTBC infection in the multi-host system is a great barrier to achieving this goal worldwide. To quickly investigate the epidemiological status of TB in many different species of live animals, it is ideal to have one common universal test for all.
Currently, commercial detection methods including reagents and kits are mainly developed for bovine TB (bTB) and hTB. Diagnosis of TB for other animals caused by M. bovis, M. caprae, and M. tb in sheep, goats, cervids, monkeys, and some other wild animals close to humans and domestic animals is usually based on the conventional methods such as the tuberculin skin or eye tests, mycobacterial acid-fast staining and culture, postmortem examination for typical tubercles and histopathological analysis, but these conventional detection methods either have low-sensitivities or low-specificities, or are unsuitable for the examination of live animals [16]. Therefore, a universal method to detect MTBC infection for multiple live hosts infected by three common MTBC members is urgently needed. Antibody detection is an ideal choice to identify the risk of TB transmission in multiple species due to its advantages of simple, easy, rapid, and high-throughput screening.
Another limitation of the antibody test with a traditional ELISA lies in the species specificity of the secondary antibodies which varies with host species. One solution addressing this is to develop an antigen sandwiched ELISA such as the commercial MPB83 sandwiched ELISA kit (INGEZIM Tuberculosis DR, Ingenasa, Spain) which has been validated in the detection of M. bovis infection in some wildlife (deer, badgers, wild boars, etc) and goats. However, as mentioned above, by targeting a single antigen, it might not display sufficient sensitivity. Alternatively, an ELISA using a universal ligand of Ig such as Staphylococcal protein A and/or Streptococcal protein G and fused multiple antigens would overcome the limitation in species and sensitivity [24,25].
Therefore, this study aimed to develop a single universal serological detection method for MTBC infection named MMEC/AG-iELISA by using a fusion protein made from four specific MTBC proteins MPB70, MPB83, ESAT6, and CFP10 (MMEC) as the coating antigen and HRP-labeled fusion protein of Staphylococcal protein A and/or Streptococcal protein G (AG-HRP) as the secondary antibody to bind Ig of various hosts species. This MMEC/AG-iELISA was optimized to detect antibodies to MTBC infection in bovines, sheep, cervids, monkeys, and humans.

Ethics statement
Animal experiments to collect sera were conducted in strict accordance with the Guide for the Care and Use of Laboratory Animals, Hubei Province, China. The protocols were approved by the Ethics Committee of Huazhong Agricultural University (agreement no. HZAUSH-2019-005).

Serum sample collection
The information of serum samples from various hosts is presented in Table 1. Human sera were collected previously by the Tuberculosis department, Wuhan Medical Treatment Center, and stored at -80˚C by this laboratory [26]. Briefly, the 24 human serum samples were collected from 24 TB patients who were firstly suggested by clinical signs and chest X-ray test, and then confirmed by M. tb isolation from sputum with mycobacterial culture and molecular typing. The hospital used Xpert to rapidly type M. tb and test drug resistance gene rpoB for the patient treatment, while we used multiplex PCR to further differentiate MTBC members for our research described as follows. In addition, the 233 human negative sera were collected from healthy volunteers who were negative in intradermal tuberculin test (induration area < 5 mm), chest X-ray test, and in vitro blood IFN-γ release test.
Bovine serum samples included the following: 120 positive sera were confirmed by the comparative intradermal tuberculin test, in vitro blood IFN-γ release test, antibody test to bTB kit (IDEXX, America), and ELISA kit to detect the antibody against M. bovis (KEQIAN, China); 242 negative sera were confirmed by the comparative intradermal tuberculin test, in vitro blood IFN-γ release test, and with the above mentioned ELISA kits; the samples of antisera against M. paratuberculosis, Bovine herpes virus 1, Brucella abortus, and bovine viral diarrhea virus (BVDV) were purchased from China Institute of Veterinary Drug Control; 17 positive sera were collected from M. bovis-infected cattle, the tissue with pathological change of the 17 cattle were collected in an abattoir. The collected samples were homogenized and decontaminated with 4% sodium hydroxide. After neutralization and centrifugation, the sediment is inoculated on 7H11 medium (Becton Dickinson, Franklin Lakes, NJ) for a minimum of 8 weeks at 37˚C with CO 2 for MTBC culture. When growth is visible, the genome of the clone was extracted by boiling. Then, typing of the MTBC members was conducted by multiplex PCR targeting seven genes 16S Rrna, Rv0577, IS1561', Rv1510, Rv1970, Rv3877/8, and Rv3120 as described before [27]. Briefly, each PCR mixture was prepared with 25 μL of PCR Master (Tsingke Biotechnology Co., Ltd., China), 20 μL of water, 2.5 μL of dimethyl sulfoxide, 1 μL of each primer at 20 μM, and 0.5 μL of DNA. PCR amplifications were performed with an initial denaturation step of 5 min at 94˚C followed by 35 cycles of 60 s at 94˚C, 60 s at 60˚C, and 90 s at 72˚C, and ending with a final elongation step for 10 min at 72˚C. PCR products were visualized by 1% agarose gel electrophoresis and sequenced. Meanwhile, the genome of MTBC in collected tissue was also directly extracted with nucleic acid extraction instruments (TIANLONG, China), and the subspecies of the MTBC were typed by multiplex PCR [27]. By mycobacterial culture and multiplex PCR typing, all 17 cattle were determined to be infected with M. bovis. However, for the tissue direct PCR typing, animals 9, 13, and 14 showed negative results (S1 Table).
Then 113 negative sheep sera were collected from healthy sheep (Chinese Hu sheep) which were determined with the cervical intradermal comparative test with bovine and avian purified protein derivative (PPD) [28]. After clipping hair, cleaning the skin, and measuring skin-fold thickness, 2500 IU/0.1 mL PPD of M. bovis abbreviated as PPD-B (CZ VeterinariaS.A., Porriño, Spain) or 2500 IU/0.1 mL PPD of M. avium abbreviated as PPD-A (CZ VeterinariaS.A., Porriño, Spain) was injected on each side of the neck at an identical position in the center of the middle third of the neck with a 1 mL sterile syringe. Seventy-two hours later, the skin-fold thickness was measured again. The reaction was considered to be negative if the increase in skin-fold thickness at the PPD-B injection site was less than or equal to the increase in the skin-fold reaction at the PPD-A injection site, and without clinical signs, such as diffuse or extensive oedema, exudation, necrosis, pain or inflammation of the lymphatic ducts in that region or the lymph nodes [28]. Then, 46 positive sheep sera to PPD-B were prepared as follows: the TB negative sheep were subcutaneously inoculated with 2500 IU/0.1 mL PPD-B and boosted with the same dose at 30 days later in the center of the middle third of the neck with a 1 mL sterile syringe, the positive sera were collected one week later, and titers were determined by PPD-ELISA as previously described [24]. Besides, one TB negative sheep was immunized with the mixture of 10 mL recombinant fusion protein MMEC (0.2 mg/mL) and 10 mL Freund's complete adjuvant (Sigma, America), and boosted with the mixture of the same amount of MMEC and Freund's incomplete adjuvant (Sigma, America) 15 days later, the positive sera were collected one week later and used as a positive control. Meanwhile, the serum sample collected from the sheep before immunization with MMEC was used as a negative control. Deer sera included: 52 positive sika deer sera and 146 negative sika deer sera determined by PPD-ELISA and the INGEZIM kit [24], one positive serum sample to M. bovis was collected from a Pere David's deer which was determined by clinical signs, pathological changes, mycobacterial culture, and PCR typing [27] with the samples from the tuberculous nodule in the lung; one positive serum to M. caprae was collected from a roe deer which was determined by clinical symptoms, mycobacterial culture, and PCR typing [27] from the animal with the abscesses in the right submaxillary face, meanwhile three positive serum samples collected from other three roe deer in the same group were determined by PPD-ELISA.
Monkey sera included: 2 positive sera determined by upper eyelid intradermal PPD-B test and PPD-ELISA, 84 negative sera collected from the animals which were determined by upper eyelid intradermal PPD-B test, PPD-ELISA, and the INGEZIM kit [24].

Expression and purification of MMEC
The recombinant plasmid encoding fusion protein MMEC was previously constructed in our lab [29]. Briefly, the genes were amplified from the genomic DNA of M. bovis and linked by a linker (GSGGGGSGGGGSGS). Further, the fusion DNA fragment was inserted into pET28a (+) plasmid, which was transformed into E. coli BL21 (DE3) (TIANGEN, China) for expression [29]. MMEC was purified by Ni SepharoseTM 6 Fast Flow resin gravity column (GE healthcare, America). The purity of MMEC was evaluated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and the concentration of MMEC was determined by Enhanced BCA Protein Assay Kit (Beyotime, China).

HRP labeled protein AG
A volume of 100 μL of 20 mg/mL HRP (Sigma, America) was mixed with 100 μL of 21 mg/mL NaIO4 (Sigma, America) and the mixture was incubated for 30 min at 4˚C, then 2 μL ethylene glycol (Sinopharm, China) was added, and the obtained mixture was incubated for 30 min at room temperature in the dark. The oxidized HRP was mixed with 1 mL of 2 mg/mL protein AG (Prime gene, China), and the mixture was incubated for 2 h at room temperature, then 20 μL of 20 mg/mL NaBH4 (Sinopharm, China) was added and incubated for 2 h at 4˚C. Afterward, the AG-HRP was dialyzed overnight in PBS and stored at -80˚C after being mixed with 50% glycerol.
The concentration of AG-HRP was determined by Enhanced BCA Protein Assay Kit. The titer of AG-HRP was determined as follows: 100 μL of MMEC (100 ng/mL) was added to the 96-well plate and incubated overnight at 4˚C, then 200 μL blocking buffer (1% fish gelatin (Sigma, America) in PBS) was added and incubated for 1 h at 37˚C. Whereafter, the plate was incubated with 100 μL of positive or negative control serum sample pre-diluted with blocking buffer at 1:10000 for 30 min at 37˚C. Then the plate was incubated with 100 μL of two-fold serially diluted AG-HRP (diluted with blocking buffer) for 30 min at 37˚C. The plate was washed six times with PBST (0.5% Tween-20 in PBS) after each step during this procedure. Afterward, the plate was incubated with 100 μL of SureBlue TMB microwell substrate (KPL, America) for 10 min in the dark, then mixed with 100 μL of 1 N HCl to terminate the reaction. Subsequently, the optical density (OD) values at the wave length of 450 nm were measured with a spectrophotometer.

Establishment of MMEC/AG-iELISA
After the MMEC coating concentrations of 200 ng/mL, 100 ng/mL, 50 ng/mL, 25 ng/mL, and 12.5 ng/mL and AG-HRP concentrations of 4 μg/mL, 0.8 μg/mL, 0.4 μg/mL, and 0.27 μg/mL were examined, the optimum coating concentration of MMEC and dilution of AG-HRP were determined by checkerboard titration as described above. Then, the optimum serum dilution was determined with pooled positive bovine antiserum (120 positive bovine sera equally pooled) and pooled negative bovine serum (242 negative bovine sera equally pooled) at 1:10, 1:25, 1:50, 1:100, 1:200, 1:400, 1:800, and 1:1600 dilutions. The ratio of OD 450nm values for the positive (P) serum to the negative (N) serum sample (P/N) was calculated. When the P/N was the highest and/or the OD value of the positive serum was slightly over 1, the reaction conditions including MMEC coating concentration, AG-HRP dilution, and serum dilution were determined to be optimal [30].

Determination of cut-off value, diagnostic sensitivity and specificity of MMEC/AG-iELISA for each species
The positive and negative sera of each species described above were diluted with blocking buffer at two or three titers around the optimal serum dilution determined in the above checkerboard titration. For each test, the sheep positive control (PC) serum sample and negative control (NC) serum sample were added in duplicates. Only when the average OD NC was smaller or equal to 0.09 , while the average OD PC was larger or equal to 1.26 which were determined by the checkerboard titration, the test is valid. Then the S/P value for each sample was calculated as follows: S/P = (OD sample -OD NC )/(OD PC -OD NC ). The cut-off value of human, bovine, sheep, and deer sera was determined by receiver operator characteristic (ROC) analysis for maximum diagnostic sensitivity and specificity. In addition, the cut-off value of monkey sera was alternatively defined as the mean S/P value of the negative sera plus 3 standard deviations (SD) with 99% confidence due to the insufficient positive sera [24].

Evaluation of analytical sensitivity and specificity of MMEC/AG-iELISA
One two-fold serially diluted positive serum of cattle with M. bovis infection was tested in parallel with both MMEC/AG-iELISA and the IDEXX kit. Similarly, two-fold serially diluted positive serum of sheep, sika deer, roe deer, and macaque monkey with M. bovis or M. caprae infection were tested in parallel with MMEC/AG-iELISA and the INGEZIM kit. In addition, cattle antisera against M. paratuberculosis, B. abortus, Bovine herpes virus 1, and BVDV were used to evaluate the analytical specificity of MMEC/AG-iELISA.

Application of MMEC/AG-iELISA for the serological diagnosis of TB
To compare MMEC/AG-iELISA and commercial kits, all the bovine positive and negative sera were tested in parallel with MMEC/AG-iELISA and the IDEXX kit; all the sheep positive and negative sera were tested in parallel with MMEC/AG-iELISA and the INGEZIM kit. In addition, 145 sheep or goat sera collected in Gansu province were tested with MMEC/AG-iELISA for evaluation of the seroprevalence of sheep or goat TB in Gansu province between 2019 and 2020, 60 of them were collected in the summer of 2019, and 60 of them were collected in the summer of 2020, while 25 of them were collected in the winter of 2020.

Statistical analyses
By testing the antibody positive and negative sera, the ROC curve was constructed for each species. Then the area under the curve (AUC) (Area ± SE), cut-off values, diagnostic sensitivity, and diagnostic specificity of MMEC/AG-iELISA with 95% CI were determined by the software built in Graphpad 8.3 (GraphPad Software, Inc, USA). Kappa value was determined by SPSS 25 (International Business Machines Corporation, America) as follows: poor (<0.20), fair (0.20-0.40), moderate (0.41-0.60), good (0.61-0.80) and very good (0.81-1.00) [31]. Chisquare test was used to determine the difference between the positive and negative proportions with SPSS 25. The difference is significant when p<0.05, and is extremely significant when p<0.01.

Development of MMEC/AG-iELISA
The recombinant fusion protein MMEC was expressed in E. coli and purified to a stocking concentration of 0.57 mg/mL, and confirmed to be a single band with a molecular mass of about 75 kDa in the gel after 10% SDS-PAGE ( Fig 1A). Meanwhile, the titer of AG-HRP was determined as 1×2 21 with a stocking concentration of 4.02 mg/mL (Fig 1B). Through the checkerboard titration, the optimum coating concentration of MMEC was 50 ng/mL, the final dilution of AG-HRP was 0.4 μg/ mL, and serum dilution of 1:100 (S2 and S3 Tables). In the practical test for the serum samples from different species, the dilutions of serum samples were finely adjusted around 1:100 and the one was selected with the highest sensitivity and specificity. In detail, 120 positive bovine sera and 242 negative bovine sera were diluted at 1:50 and 1:100; 46 positive sheep sera and 113 negative sheep sera were diluted at 1:25, 1:50, and 1:100; 54 positive deer sera and 146 negative deer sera were diluted at 1:25, 1:50, and 1:100; 2 positive monkey sera and 84 negative monkey sera were diluted at 1:25, 1:50, and 1:100; 24 positive human sera and 233 negative human sera were diluted at 1:100, 1:200, and 1:500. Then all the pre-diluted sera were tested with MMEC/AG-iELISA to further determine the parameters of this assay.
Then the cattle antisera to pathogens such as M. paratuberculosis, Bovine herpesvirus 1, Brucella abortus, and BVDV were tested by this MMEC/AG-iELISA and determined to be negative based on their S/P values of 0.114, 0.011, -0.010, and 0.089, respectively, which were less than the cut-off value of 0.30.
To compare its analytical sensitivity with the commercial IDEXX kit, one cattle positive serum was two-fold serially diluted and tested with both tests. The results indicated that the highest serum dilution for MMEC/AG-iELISA was 1.6 × 2 14 , which was the same as that for the IDEXX kit test (S5 Table).
Further, all the 120 positive and 242 negative bovine sera to bTB were tested in parallel with MMEC/AG-iELISA and the IDEXX kit. The results indicated the observed agreement is 100.00% (362/362) and Kappa value is 1, suggesting a very good agreement between these two methods (S6 Table).
Finally, 17 sera without antibody background but confirmed as M. bovis infection were detected by both tests. These 17 sera were collected from the cattle which had gross pathological change and confirmed as M. bovis-infection by mycobacterial culture and PCR typing (S1 Table), of which, 12 were determined as positive by MMEC/AG-iELISA with a sensitivity of 70.59% (95% CI: 44.04%, 89.69%), while only 9 of them were determined as positive by the IDEXX kit with a sensitivity of 52.94% (95% CI: 27.81%, 77.02%) (S1 Table). Although there is no significant difference between them (p = 0.481), the findings indicate this MMEC/AG-iELISA has a tendency of high sensitivity for serological diagnosis of bTB.
Meanwhile, all the positive and negative sheep sera were tested with a commercial double antigen sandwiched INGEZIM kit. As a result, although all the negative sera were correctly detected by the INGEZIM kit, only 14 out of 46 sheep positive sera were successfully detected with a sensitivity of 30.43% (95% CI: 17.70%, 45.80%) (S8 Table), indicating that the INGEZIM kit test was significantly less sensitive than MMEC/AG-iELISA for test of M. bovis PPD induced antibodies in sheep (p < 0.001). Then one positive sheep serum was two-fold serially diluted and tested with MMEC/AG-iELISA and the INGEZIM kit. The results indicated that with MMEC/AG-iELISA, the highest serum dilution was 1:3200, which was 16 folds higher than that with commercial INGEZIM kit (1:200) (S9 Table).
To determine the capacity of MMEC/AG-iELISA to differentiate MTBC and nontuberculous mycobacteria (NTM) infection in sheep, the comparative tests were performed by using the intradermal comparative test, MMEC/AG-iELISA and PPD-ELISA for 26 sheep that were defined as NTM infection by the intradermal comparative test because the increase of skin fold thickness reactive to avian PPD was larger than that to bovine PPD, and larger than 4 mm. As a result, both MMEC/AG-iELISA and PPD-ELISA determined these animals were negative to TB (S10 Table).
Further, both assays were used to test one roe deer serum infected with M. caprae. The results indicated that with MMEC/AG-iELISA, the titer of this serum was 1:50. However, this serum was determined to be negative by the INGEZIM kit (S13 Table). For the other three serum samples collected from other three roe deer cohabited with the M. caprae infected roe deer, all three sera were determined to be positive by MMEC/AG-iELISA, but only two of them were determined to be positive by the INGEZIM kit.

Detection of TB in monkeys with MMEC/AG-iELISA
For monkey serum test, 84 negative macaque monkey sera and two positive sera determined by the upper eyelid PPD intradermal test, PPD-ELISA, and the INGEZIM kit were used at 1:25, 1:50, and 1:100 dilution. The cut-off S/P values were determined by the mean + 3 SD of negative samples to be 0.21, 0.17 and 0.11 respectively at three dilutions, then the corresponding diagnostic specificities were 97.62% (95% CI: 91.66%, 99.71%), 98.81% (95% CI: 93.54%, 99.97%), and 97.62% (95% CI: 91.66%, 99.71%) (S14 Table). Therefore, the serum dilution 1:50 and cut-off S/P value 0.17 were selected. Then, the two positive sera were determined as positive at these conditions, and the diagnostic sensitivity was 100.00% (95% CI: 15.81%, 100.00%). Then one positive macaque monkey serum sample was two-fold serially diluted and tested in parallel with MMEC/AG-iELISA and the commercial INGEZIM kit. The results indicated that with MMEC/AG-iELISA, the highest serum dilution was 1:400, however, this serum was determined to be negative by the INGEZIM kit (S15 Table).

Discussion
Since TB is an important zoonotic disease, the maintenance of MTBC infection in animals is a great barrier for achieving this goal of the End-TB strategy. Development of a detection technique for different live species is of significance in improving surveillance efficiency and control effect of a One Health strategy. In the current study, the universal MMEC/AG-iELISA was developed and demonstrated to have the potential in serological diagnosis of TB in humans, cattle, sheep, cervids, and monkeys.
To evaluate the performance of MMEC/AG-iELISA in serological detection of M. bovis infection in cattle, two sets of samples were used. One set was the serum samples with a known antibody background. For this set of samples, a complete agreement was observed between our MMEC/AG-iELISA and the commercial IDEXX kit. The very high agreement might be attributed to the strict selection of the samples because all the positive and negative samples were commonly determined by two or more tests. Further, we tested the other set of 17 serum samples from the cattle without known antibody background but identified to be infected with M. bovis through bacterial culture and genotyping which is considered as the gold standard method. The results showed that this MMEC/AG-iELISA had a higher sensitivity 70.59% (95% CI: 44.04%, 89.69%) than its commercial counterpart 52.94% (95% CI: 27.81%, 77.02%). It might be attributed to the differential antigens used in these two methods. The current MMEC/ AG-iELISA uses the fusion protein MMEC that consists of MPB70, MPB83, ESAT6, and CFP10. Among them, ESAT6 and CFP10 are well-known as the early secreted proteins during MTBC infection. On the other hand, the IDEXX kit uses only MPB70 and MPB83 as its coating protein and could miss the detection of early infection. Previously, inconsistent diagnostic sensitivities of the IDEXX kit such as 74% in Great Britain, 69% in Ireland, 46% in the United States, and 40% in New Zealand were reported regarding the samples diagnosed with mycobacterial methods. The inconsistent results were probably affected by the infection stages of the sampled animals because antibody development is usually at the late stage of MTBC infection [32]. Of course, more samples are needed to investigate the reason causing this difference in sensitivities. Unfortunately, it is usually difficult to get sufficient live cattle diagnosed with M. bovis infection for serum antibody detection. In addition, the potential reason for no antibodies being detected in five sera by MMEC/AG-iELISA might probably be that the infection was at the very early stage when the antibodies wouldn't have been significantly developed yet [28].
Although M. tb infection in animals has been increasingly reported in different countries in recent years [11,12,33], it is difficult to collect serum samples from confirmed M. tb-infected animals due to the difficulties in mycobacterial detection and genotyping for live animals. As an alternative, we used the positive and negative hTB serum samples and confirmed that this new method had a sensitivity as high as 100.00% (95% CI: 86.82%, 100.0%) and specificity as high as 94.85% (95% CI: 91.22%, 97.03%) for diagnosis of M. tb infection. Different sensitivities for the serological tests of hTB were previously reported, which might be attributed to the patients' different disease stages and severity degrees of TB [34,35]. Theoretically, MMEC/AG-iELISA could also be used for detecting animal M. tb infection. However, the performance of MMEC/AG-iELISA for serological diagnosis of M. tb infection in animals needs to be further evaluated in the future.
The tuberculosis of sheep can be caused by either M. bovis or M. caprae [13,36], but TB in sheep and goats is usually neglected likely due to its short lifetime of sheep and goats, less intensive farming system, and not as close contact or relationship to humans as dairy cows. In addition, it is difficult to collect serum samples from live sheep and goats diagnosed as M. bovis or M. caprae infection. As an alternative, we used the antibody positive serum samples developed by immunization of PPD-B and confirmed by PPD-ELISA. To evaluate the performance of MMEC/AG-iELISA, we compared the detection results of MMEC/AG-iELISA and the only available commercial ELISA kit (INGEZIM kit). Although the two methods were in complete agreement as to the detection of the negative samples, the sensitivity of our MMEC/ AG-iELISA was significantly higher than that of the INGEZIM kit in the detection of positive samples (p<0.001). The most likely reason might be that the INGEZIM kit used only MPB83 as the capturing antigen. Furthermore, to determine whether this MMEC/AG-iELISA could be used in the detection of natural TB in sheep and goats, 145 random serum samples of sheep and goats were clinically collected from Gansu province and tested. An unexpected high seroprevalence of TB in sheep and goats (11.72%, 95% CI: 6.98%, 18.11%) was obtained between 2019 and 2020 with seasonal variation indicating the surveillance and control of TB in sheep and goats should be conducted in China. To our best knowledge, this is the first report on a serological survey on TB in sheep and goats. However, more samples from wider areas should be covered in future investigations to elucidate the epidemiological status of TB in sheep and goats in China. For the 113 sheep which were diagnosed as TB negative with the intradermal comparative test, 26 of them were considered as NTM infection as the increase in skin-fold thickness at the avian injection site was greater than 4mm, and greater than that in the bovine injection site. All the sera of the 26 sheep were tested by MMEC/AG-iELISA and PPD-ELISA, and the results were negative (S10 Table). Therefore, NTM infection wouldn't affect the test results of MMEC/AG-iELISA.
Since TB in wildlife poses a great threat to humans and domestic animals, surveillance of wildlife TB is important. Among wildlife species, cervids and monkeys are very susceptible hosts and close to both humans and domestic animals [37][38][39]. However, wildlife can be very aggressive and are difficult to capture and handle. Therefore, conventional methods such as skin test including catching animals, measuring skin-fold thickness, injecting PPD, and measuring the increase of skin-fold thickness 72 h later is usually not practical. Therefore, developing an in vitro blood/serum test with high accuracy is critical for the surveillance of wildlife TB. This current study confirmed this test has high sensitivity and specificity for sika deer in serological diagnosis of M. bovis infection which is in line with the other serology tests [40,41]. Similarly, this method was demonstrated to work well for TB testing in macaque monkeys. However, because only two positive samples were collected during the period of this study, the cut-off value for TB serological tests in the monkeys was determined as the mean S/P value of the negative sera plus 3 SD [24]. To our knowledge, wild monkeys might have a high prevalence of TB. Since we checked the monkeys raised as the experimental animals and these monkeys have experienced several rounds of quarantine and testing, the TB prevalence in these monkeys is low. Therefore, more samples, especially positive samples, should be collected and tested to further evaluate the performance of this technique in the future.
Although M. caprae-caused TB has been added to the OIE disease list, M. caprae infection is really neglected and greatly underestimated. In addition to sheep and goats, which are susceptible hosts of M. caprae [42], cervids such as red deer are natural hosts and maintenance reservoirs of M. caprae [43]. Fortunately, we have identified one case of M. caprae infection in roe deer by clinical symptoms, mycobacterial isolation, and genotyping through PCR and genome sequencing, and confirmed MMEC/AG-iELISA could be used for serological diagnosis of deer M. caprae infection.

Conclusions
In conclusion, we established a universal MMEC/AG-iELISA for serological diagnosis of MTBC infection in multiple hosts including cattle, sheep/goats, cervids, and humans mainly caused by M. bovis and M. tb with high sensitivity and specificity. Therefore, it would be a promising application in serological TB surveillance for multiple animal species thereby providing evidence for taking further action in TB control of a One health strategy.