Vaccination of calves with Mycobacterium bovis Bacillus Calmette–Guerin reduces the frequency and severity of lesions of bovine tuberculosis under a natural transmission setting in Ethiopia

Summary Bovine tuberculosis (bTB) is highly prevalent in intensive dairy farms of the urban “milk‐sheds” in Ethiopia, and vaccination could be a cost‐effective disease control strategy. In the present study, the efficacy of Bacillus Calmette–Guerin (BCG) to protect against bTB was assessed in Holstein–Friesian calves in a natural transmission setting. Twenty‐three 2‐week‐old calves were subcutaneously vaccinated with BCG Danish SSI strain 1331, and matched 26 calves were injected with placebo. Six weeks later, calves were introduced into a herd of M. bovis‐infected animals (reactors) and kept in contact with them for 1 year. In vitro and in vivo immunological tests were performed to assess immune responses post‐vaccination and during exposure. Successful vaccine uptake was confirmed by tuberculin skin test and IFN‐γ responses in vaccinated calves. The kinetics of IFN‐γ responses to early secretory antigen target 6 and culture filtrate protein 10 (ESAT6 and CFP10, respectively) and tuberculin skin test responses post‐exposure suggested that the animals were infected early after being placed in contact with the infected herd as immunological signs of infection were measurable between 2 and 4 months post‐initial exposure. Protection was determined by comparing gross and microscopic pathology and bacteriological burden between vaccinated and control calves. BCG vaccination reduced the proportions of tissues with visible pathology in vaccinates compared to control calves by 49% (p < .001) with 56%, 43%, 72%, and 38% reductions in the proportion of lesioned tisues in head, thoracic, abdominal lymph nodes, and lungs, respectively (p‐values .029–.0001). In addition, the lesions were less severe grossly and microscopically in vaccinated calves than in non‐vaccinated calves (p < .05). The reduction in the overall incidence rates of bTB was 23%, 28%, and 33% on the basis of the absence of gross pathology, M. bovis culture positivity, and histopathology, respectively, in vaccinated animals. In conclusion, BCG vaccination reduced the frequency and severity of the pathology of bTB significantly, which is likely to reduce onwards transmission of the disease.

animals. In conclusion, BCG vaccination reduced the frequency and severity of the pathology of bTB significantly, which is likely to reduce onwards transmission of the disease.

K E Y W O R D S
BCG vaccination, bovine tuberculosis, natural transmission 1 | INTRODUCTION Ethiopia, with over 90 million people, is a typical example of the demographics in most developing countries in Africa and Asia with the human population increasing by 3.2% per year (CSA, 2011), leading to increased demand for food production. Hence, cereal crop production has been prioritized at the cost of grazing land for livestock (Tschopp et al., 2010), leading to overstocking and overgrazing of that land, thereby compromising the development in the Ethiopian livestock sub-sector. For this reason, intensification of livestock production is considered to be the best option. Thus, the number of intensive dairy farms in and around urban centres is increasing.
These emerging dairy farms hold cattle breeds optimized for increased milk production, as the milk production potential of indigenous zebu breeds is by far lower than that of either Holstein-Friesian or crosses between these exotic breeds and indigenous zebu breeds. Moreover, as the extensive cattle husbandry management of the Ethiopian farmers cannot satisfy the milk demand of the growing population, the government is encouraging the establishment of intensive dairy farms by the private sectors. However, both the increase in the number of exotic breeds and the intensification of dairy farming are associated with increased prevalence of cattle diseases such as bovine tuberculosis (bTB) (Ameni et al., 2007;Cosivi et al., 1998;Firdessa et al., 2012). BTB is predominantly caused by M. bovis and is characterized by the development of granulomatous lesions in the respiratory tract and also in other tissues of the animal. Globally, this disease impacts in three major ways, namely as zoonotic TB in humans; direct economic losses due to reduced livestock productivity; and indirect economic losses due to livestock trade restrictions.
The increased number of more susceptible exotic breeds together with the increased intensification of production demands prioritization of improved bTB control strategies focusing primarily on intensive dairy farms. In developed countries, the control of bTB is based on a test-and-slaughter strategy, which would be too costly to be applied on a national level in Ethiopia or in any other developing country. Hence, there is a need for exploring alternative control strategies such as routine testing and surveillance, pre-movement testing, movement restriction of infected herds, and vaccination, all of which could be combined with better bio-security and farm hygiene. In the present study, we have evaluated the performance of BCG in protecting cattle against bTB in a natural transmission setting by exposing vaccinated and non-vaccinated calves to a cattle herd known to be bTB infected, which complements our previous study (Ameni, Vordermeier, Aseffa, Young, & Hewinson, 2010).

| Study setting and sources of experimental calves
The experiment was conducted at Sebeta Agro Industry PLC, a private farm located in Sebeta about 20 km south-west of Addis Ababa, Ethiopia. At the start of this challenge experiment, its dairy herd consisted of 72 Holstein-Friesian cattle or crosses; thereof, with the zebu breed, all animals were positive in the single intradermal comparative tuberculin test (SICTT), as described further below. However, during the course of experiment the number of reactor animals decreased gradually, and to maintain a reactor to sentinel ratio > 1, 15 skin test reactor animals from Holeta Agricultural Research Centre were introduced 6 months into the challenge experiment.
Dairy farms located around Addis Ababa were randomly selected and tested for bTB using the SICTT. Thereafter, the experimental calves were recruited from seven bTB negative farms located around Addis Ababa. A total of 49 Holstein-Friesian calves were recruited for the experiment. The calves were allocated to experimental and control groups randomly using a lottery system. The allocation was carried out in batches of 10 calves, as all calves could not be recruited at once. Twenty-three calves were allocated into a vaccinated group while the remaining 26 calves were used as controls. All the calves were tested negative for bTB when tested with the Bovigam IFN-c test prior to their recruitment into the experiment.
The experiment was approved both by Ethics Committee of the Armauer Hansen Research Institute and by the Ethical Review Board at the Animal and Plant Health Agency.

| Vaccination schedule of the neonates
All calves of the vaccinated group were vaccinated within 2 weeks of birth by subcutaneous injection with 1-4 9 10 6 CFU BCG Danish SSI 1331(Staten's Serum Institute [SSI], Copenhagen, Denmark), which was supplied as freeze-dried preparation and reconstituted in Sauton's medium as per the manufacturer's instruction. Until calves were introduced into the reactor herd, at 2 months of age, they were kept isolated in a communal calf pen. During this time, the calves were fed with milk, from PPD-negative cows, concentrate, and grass on the basis of their age. At around 6 weeks post-vaccination (when they AMENI ET AL. | 97 were about 2 months old), the calves were moved to the bTB-positive herd and kept in contact with reactor animals for about 1 year. at 5lg/ml and saline (both 25ll) were used as positive and negative controls, respectively. Cultures were incubated at 37°C in a humid 5% CO 2 atmosphere for 48 hr, and supernatants were harvested and frozen. IFN-c in the supernatants were measured by an enzyme-linked immunosorbent assay using Bovigam test kit (Prionics, Schlieren, Switzerland) in accordance with the manufacturer's instructions. VLs. When gross lesions suggestive of bTB were found in any of the tissues, the animal was classified as "VL." Any animal in which TBlike lesion(s) were not found was classified as "NVL" (none visible lesions). The severity of gross lesions was scored by a semi-quantitative scoring procedure as previously described by Vordermeier et al. (2002). Briefly, lesions in the lobes of the lungs were scored separately as follows: 0, no visible lesions; 1, no gross lesions but lesions apparent on slicing of the lobe; 2, fewer than five gross lesions; 3, more than five gross lesions; and 4, gross coalescing lesions. The scores of the individual lobes were added up to calculate the lung score. Similarly, the severity of gross lesions in individual LN was scored as follows: 0, no gross lesion; 1, a small lesion at one focus (just starting); 2, small lesions at more than one focus; and 3, extensive necrosis. Individual LN scores were added up to calculate the total LN score for each LN/ tissue category. Finally, LN and lung pathology scores were added together to give the total pathology score per animal.

| Histopathological examination and grading of granuloma
For histopathology examination, tissue samples from organs displaying TB-like gross lesions were collected and immersed in fixative (10% neutral buffered formalin) for 7 days before being processed and embedded in paraffin wax. Four-micron sections were cut and routinely stained with haematoxylin and eosin (H&E) and Ziehl-Neelsen (ZN) for the detection of TB granuloma and acid-fast bacilli (AFB), respectively. Slides were examined by light microscopy to determine the distribution of granuloma development stages as defined by Wangoo et al. (2005). Thus, these four stages of granulomas were quantified and analysed as previously described (

| Statistical analysis
To estimate the incidence of bTB per group, the number of calves that developed TB was divided by the total number of calves in the group. The efficacy of the vaccine was estimated using the formula described by Orenstein et al. (1985), which considers the incidence rates of the disease in question in the vaccinated and in unvaccinated calves; i.e., the efficacy of a vaccine is the percentage reduction in the incidence rate of a disease among vaccinated calves as compared to the incidence rate in unvaccinated calves. The formula used for calculating vaccine efficacy (VE) in this study was VE = (ARU-ARV/ ARU) 9100%, where ARU is attack (incidence) rate in the unvaccinated group and ARV is the attack (incidence) rate in the vaccinated group. Chi-squared (v 2 ) test was used to compare the percentages of gross pathology in different tissues of vaccinated and control calves.
In addition, the incidence rates of bTB and efficacies of BCG in vaccinated and control calves were compared using v 2 test. Comparisons of severity of gross lesion between tissues of vaccinated and control calves were made using non-parametric t test with Mann-Whitney The calves were tested with SICTT at the fourth (a), eighth (b) and 12th (c) month post-exposure to infected herd. The means of change in skin thickness following SICTT were significantly greater (unpaired t test with Welch's correction) in control calves with VL than in control calves with NVL at eighth (p = .002) and 12th (p = .002) month post-exposure to infected herd. However, although the mean of change in skin thickness after SICTT was greater in vaccinated calves with VL than in vaccinated calves with NVL at all the months tested, the differences were not significant at any of the months The means of IFN-c response estimated by optical density (OD) measured at 450 nm were monitored for 12 months after the calves were exposed to the infected herd. The means of the IFN-c responses are compared (unpaired t test with Welch's correction) in control calves with VL and in control calves with NVL at the second (a; p = .011), fourth (b; p = .15), sixth (c; p = .0002), eighth (d; p = .001), 10 th (e; p = .0001) and 12 th (f; p = .001) month post-exposure to the infected herd. However, although vaccinated calves with VL demonstrated relatively stronger IFN-c responses than vaccinated calves with NVL at the different months of exposure, the difference between these two groups was not significant at any of the months 3.2 | Immune responses after vaccinated and control calves were exposed to infected herd The kinetics of skin test (
These figures equate to 23%, 33%, or 28% protection, respectively, when these three parameters are taken into account. While only the difference in the frequency of microscopically lesioned calves is statistically signficant (p = 0.031), the data demonstrated a consistent degree of protection against bTB disease following BCG vaccination.

| Distribution, frequency, and severity of typical bTB lesions in vaccinated and control calves
Further assessment of BCG protection against bTB was carried out by determining the reduction in the severity and within animal dissemination of disease. The severity of pathology was scored according to Vordermeier et al. (2002) in post-mortem tissues of BCGvaccinated and non-vaccinated calves and was compared as presented in Figure 4. The pathology was more severe (p = 0.04; Mann-Whitney test) in the thoracic lymph nodes of control calves (median AE SD, 1.5 AE 3.72) than in the thoracic lymph nodes of vaccinated calves (0.0 AE 1.13). While we observed similar reductions in the severity of gross pathology in vaccinated calves also in head lymph nodes and the lungs compared to unvaccinated controls, these reductions did not reach statistical significance ( Figure 4).
However, the total pathology scores that integrate the pathology scores of the different tisses were significantly lower in vaccinated as compared to control calves (Figure 4, p = 0.03; control calves scores = 5.0 AE 13.12; vaccinated calves = 3.0 AE 2.15).
The protective effect of BCG vaccination on reducing gross pathology was also confirmed when the frequencies of visible lesions found in the examined tissues were compared between control and BCG-vaccinated calves (

| Assessment of microscopic bTB lesions in vaccinated and control calves
Tissues presenting with TB-like lesion upon gross pathological examination were further subjected to microscopic histopathological analysis. The distribution, frequency, and severity of microscopic lesions were assessed, and the results are depicted in Figure 5. Microscopic lesions were scored from Stage I to Stage IV according to the classification system by Wangoo et al. (2005).
More than 500 individual granulomata were scored. The majority of granulomata were observed in the thoracic LNs (  F I G U R E 5 Frequency of occurrence of the different stages (I-IV) granulomas within the head, neck (a), and thoracic lymph nodes (b) of vaccinated and control calves. The bars represent the percentage of granulomas within each developmental stage. A significant difference was observed in the frequencies of lesions in the head and neck lymph nodes of the vaccinated and control groups (panel a, p < 0.05), but not in the frequencies of the thoracic lymph nodes (b) between the vaccinated and control groups and controls was observed in the distribution of granuloma developmental stages in the thoracic LNs (Figure 5b).

| DISCUSSION
In the present study, the efficacy of the BCG vaccine was evaluated in Holstein-Friesian calves in a natural transmission setting. When cellular immune responses were determined in vaccinated calves before they were introduced into the infected herds, neither IFN-c nor tuberculin skin test responses correlated with protection determined by post-mortem at the end of the in-contact period (VL versus NVL calves, data not shown). This therefore confirms that these two parameters are poor predictors of protection.
Based on the development of ESAT-6-/CFP-10-specific IFN-c responses following the exposure of calves to the infected herd, it can be hypothesized that the infection events took place during the first 2-4 months of calves being in contact with infected animals.
Interestingly, ESAT-6-/CFP-10-induced IFN-c responses developed at a slower pace in vaccinated VL animals compared to VL control calves. This could be because BCG vaccination reduced the extent of pathology or led to a slower disease progression compared to na€ ıve calves. Either hypothesis is supported by earlier data, demonstrating that the extent of in vitro IFN-c production after stimulation with ESAT-6 directly correlated with the extent of pathology (Vordermeier et al., 2002).
The efficacy of BCG in protecting against M. bovis infection under a natural transmission setting was estimated using either gross pathology, microscopic lesion of bTB, or isolation of M. bovis as interpretation criteria. In the present study, the efficacy of BCG for protection against disease was low, whereas its effect on reducing the extent of pathology was significant. The average efficacy of BCG to protect fully against bTB recorded in this study was around 30%.
This rate is considerably lower than the efficacy reported earlier by similar studies conducted in Ethiopia (Ameni et al., 2010) and in Mexico (Lopez-Valencia et al., 2010). This difference in full protection between the present and the previous Ethiopian natural transmission study (Ameni et al., 2010) could be attributed to differences in the severity of bTB in the infected reactor herds that served as sources of infection: Although similar ratios of calves to reactor cows were maintained in both experiments, the severity of disease was heavier in the herd used for the present study compared to the herd used for the previous study. For example, overt clinical signs of bTB were more prevalent in the infected herd used in the present study. Nevertheless, both the present and the earlier studies (Ameni et al., 2010) results are within the efficacy range of BCG (0% and 75%) reported from experimental studies and trials in cattle conducted by different researchers in different countries between 1959 and 2002 (reviewed by Hewinson, Vordermeier, & Buddle, 2003). Thus, the result of the present study may be reflective also of the inherent variability of BCG to impart protection at population and individual animal levels.
Nevertheless, in the present study, a significant level of protection by BCG vaccination could be demonstrated based both on reduction in the number of tissues with visible pathology and on the reduction in granulomata severity. Thus, the significant reduction in pathology following BCG vaccination could lead to a reduced onward transmission rate from vaccinated cattle to other susceptible cattle. This indirect vaccination effect would therefore very likely contribute to a reduction in the prevalence of bTB from vaccinated herds, as had been shown in an earlier trial in the UK (Doyle & Stuart, 1958). For example, BCG vaccination could reduce the incidence of bTB in vaccinated herds to a level where more conventional testand-slaughter approaches could be affordable. The reduction in the development of granulomas further confirms the finding of BCG imparting protection by reducing pathology. From the disease transmission point of view, this has great epidemiological implication, as lesions that are confined with bacilli contained in granulomas at certain anatomical sites may also prevent disease transmission to other animals (Johnson, Spencer, Hewinson, Vordermeier, & Wangoo, 2006).

| CONCLUSION
In conclusion, although the reduction in the proportion of susceptible animals through BCG vaccination (full protection, direct effect of vaccination) was lower than that recorded in the previous study, the significant reduction in the severity and distribution of visible and microscopical pathology in vaccinated calves is likely to reduce onward transmission to other animals (a so-called indirect vaccine effect). This would have a beneficial impact on disease control in vaccinated farms.