Serological evidence for historical and present-day exposure of North American bison to Mycoplasma bovis

Background Mycoplasma bovis causes mastitis, otitis, pneumonia and arthritis in cattle and is a major contributor to bovine respiratory disease complex. Around the year 2000, it emerged as a significant threat to the health of North American bison. Whether healthy bison are carriers of M. bovis and when they were first exposed is not known. To investigate these questions we used a commercially available ELISA that detects antibodies to M. bovis to test 3295 sera collected from 1984 through 2019 from bison in the United States and Canada. Results We identified moderately to strongly seropositive bison from as long ago as the late 1980s. Average seroprevalence over the past 36 years is similar in the United States and Canada, but country-specific differences are evident when data are sorted by the era of collection. Seroprevalence in the United States during the pre-disease era (1999 and prior) was significantly higher than in Canada, but was significantly lower than in Canada during the years 2000–2019. Considering individual countries, seroprevalence in the United States since the year 2000 dropped significantly as compared to the years 1985–1999. In Canada the trend is reversed, with seroprevalence increasing significantly since the year 2000. ELISA scores for sera collected from free-ranging bison do not differ significantly from scores for sera from more intensively managed animals, regardless of the era in which they were collected. However, seroprevalence among intensively raised Canadian bison has nearly doubled since the year 2000 and average ELISA scores rose significantly. Conclusions Our data provide the first evidence that North American bison were exposed to M. bovis many years prior to the emergence of M. bovis-related disease. Patterns of exposure inferred from these results differ in the United States and Canada, depending on the era under consideration. Our data further suggest that M. bovis may colonize healthy bison at a level sufficient to trigger antibody responses but without causing overt disease. These findings provide novel insights as to the history of M. bovis in bison and will be of value in formulating strategies to minimize the impact of mycoplasmosis on bison health and production.


Background
The bacterial pathogen Mycoplasma bovis was first isolated roughly 60 years ago, from a dairy cattle herd in the United States experiencing a severe outbreak of mastitis [1]. Since that time it has spread gradually throughout the world and today has a significant negative impact on cattle health and production in nearly all countries where beef or dairy cattle are intensively raised as a food source [2,3]. Clinical manifestations of disease are often chronic and include bronchopneumonia, mastitis, arthritis, otitis media, conjunctivitis and, more rarely, reproductive disorders [4]. Beginning around the year 2000, M. bovis additionally emerged as a health threat to North American bison. The first outbreak to be recognized occurred in the western United States in 1999, when~20% of a group of yearlings on pasture died from pneumonia (D. Hunter, personal communication). The episode was attributed to M. bovis based on the presence of lung lesions resembling those of bovine mycoplasmosis from which M. bovis was the only microbial pathogen consistently identified. The next apparent outbreak occurred in 2001, when M. bovis was identified as the cause of severe pneumonia and arthritis in a Canadian bison herd [5]. In the years following, anecdotal evidence continued to accumulate further implicating M. bovis as a primary cause of disease in bison of all ages among herds in both the United States and Canada [5]. Subsequent studies firmly established its role as a primary etiologic agent of pneumonia, polyarthritis, necrotic pharyngitis, pleuritis, dystocia, and abortion in bison, with herd-level fatalities as high as~30% [6][7][8][9][10][11].
An understanding of why M. bovis has emerged only relatively recently as a pathogen in bison would be of benefit in formulating control measures and strategies to minimize its impact on bison health and production. One hypothesis that has been proposed to account for this phenomenon theorizes that M. bovis has been circulating among North American bison for much of their history, or at least since the time it was first recognized in cattle, and that the appearance of mycoplasmosis in bison was triggered by intensified production and related changes in management and handling or by the rise of new genotypes with heightened virulence attributes [12]. An alternative theory posits that the first incursion of M. bovis into North American bison was more recent and was immediately followed by the appearance of M. bovis-related disease. No data currently exist that support or refute either of these possibilities. However, banked bison sera collected over the past 36 years provide a resource that can be used to investigate the historical and current prevalence of exposure to M. bovis in bison. The goal of our study was to determine whether there is serological evidence to suggest that M. bovis was circulating in bison well in advance of the time it was first recognized as a pathogen of bison (roughly the year 2000) and, if so, to compare historical and present-day rates of seroprevalence.

Results
Sera parsed by country of origin and/or era of collection Collated ELISA results are provided in Table 1, with sera grouped by year of collection and individual location of origin. A total of 227 sera tested positively for antibodies to M. bovis (6.9% of the 3295 evaluated), with 126 weakly positive (1+; 3.8% of all sera, 55.5% of all positives) and 101 moderately (2+ or 3+) or strongly positive (4+ or 5+; 3.1% of all sera, 44.5% of all positives). The proportion of positive sera for samples acquired on a single occasion from a single source ranges from 0 to 66.7%. Seropositive bison were identified in 16/23 (69.6%) American herds, ranches or locations represented in this study and 16/18 (88.9%) Canadian sources. Table 2 summarizes ELISA scores with results delineated according to the country in which samples were collected. Seroprevalence among individual animals in the United States and Canada over the past 36 years is similar, 7.1% (112/1587) and 6.7% (115/1708), respectively, and there is no statistically significant difference in average ELISA scores for sera from these two countries (0.134 and 0.128, respectively; P = 0.79). The proportion of seropositive animals in each country for which antibody levels were scored as 2+ or higher is also similar, 42.0% of all positives (47/112) for the United States and 47.0% of all positives (54/115) for Canada, and average ELISA scores for these groups are not significantly different (3.128 and 2.926, respectively, P = 0.37). Figure 1 is a graphical summary of the percentage of sera testing positively for each of the years represented in this study. Seropositive bison were identified throughout the entirety of the timeframe covered, beginning in the 1980s, when both weakly positive as well as moderately to strongly positive animals were found in both the United States and Canada ( To compare the seroprevalence of M. bovis before and after the time that related disease spread throughout North American bison, results from samples obtained in the year 1999 and prior were compared with those from sera acquired in the year 2000 or later (Table 3). Considering the 1504 sera collected between 1984 and 1999, 108 (7.2%) are positive while 119/1791 samples collected between 2000 and 2019 (6.6%) tested positively. There is no statistically significant difference between average ELISA scores obtained for sera representing these two  A comparison of the data obtained for each era that additionally considers the country of origin reveals that bison in the United States sampled in the 1980s and 1990s were more likely to have antibodies to M. bovis than bison from Canada sampled during the same timeframe (Table 3), with 9.4% positive (70/743) and 5.0% positive (38/761), respectively. The average ELISA score for all sera collected during the pre-disease era in the United States is double the score for all Canadian samples from that era (0.168 and 0.082, respectively; P = 0.002), although the average scores for only those sera testing positively from each country do not differ significantly (1.786 for the United States and 1.658 for Canada; P = 0.57). In contrast, 5.0% of sera collected in the United States during the year 2000 and afterwards tested positively (42/844), as compared to 8.1% of Canadian sera (77/947; Table 3). The average ELISA score for Canadian sera obtained during this era, 0.160, is significantly higher than the average for samples collected in the United States during the same timeframe, 0.103 (P = 0.05). As seen in the pre-disease era, the average ELISA score for positive sera from the United States does not differ significantly from the average for positive Canadian sera (2.071 for the United States and 2.013 for Canada; P = 0.82).
Examination of era-specific data for each country reveals that in both the United States and Canada average ELISA scores for sera collected in 1999 or prior differ significantly in comparison to scores for sera collected afterwards in the same country. However, the direction of change is different in each country. As noted above, average seroprevalence in the United States is 9.4% for samples collected between 1985 and 1999 but only 5.0% for samples collected afterwards. This decrease is accompanied by a significant decrease in average ELISA scores, 0.168 and 0.103 respectively (P = 0.03). Average ELISA scores for sera collected in Canada, where average seroprevalence is 5.0% during the pre-disease era and 8.1% in the years following, increased from 0.082 to 0.165, respectively (P = 0.003).

Herds sampled on multiple occasions
Two Canadian herds, Can-O and Can-P, and five from the United States, US-G, US-N, US-O, US-R and US-W, were sampled on more than one occasion, although only Can-O, Can-P and US-W were sampled during both the pre-disease era and afterwards (Table 1). Seroprevalence in Can-O and Can-P prior to the year 2000 is only slightly less than in the years during which samples were collected afterwards (Table 5). However, in each case the timeframe covered is relatively short, only 3-6 years. In both herds average ELISA scores rose in the years after 1999 but neither increase is statistically significant (P ≥ 0.43). Herd US-W was sampled more extensively, in 16 different years over a 22-year period ( Table 1). Although seroprevalence in this herd during the pre-disease era is more than double the prevalence found during the years afterwards (Table 5) there is no significant difference in average ELISA scores for each time period (0.082 and 0.068, respectively; P = 0.83).

Discussion
The data obtained in this study provide novel insights as to the history of M. bovis in bison, about which little was known prior to this time. Because it appears the bacterium may have circulated among bison well in advance of its first recognition as a disease problem, a "host jumping" event occurring around that time is unlikely to account for the emergence of disease. Our data, nonetheless, are consistent with the theory that recently evolved, novel genotypes of M. bovis may underlie or may have contributed significantly to the appearance of related disease. The abundance of insertion sequences, integrative conjugative elements and other mobile genetic elements found in M. bovis leads to considerable genome plasticity as a result of rearrangements, duplications and gene disruptions that can affect the biology of the bacterium through alterations in gene content and expression [13,14]. These elements may also play a role in horizontal gene transfer, both within and between different Mycoplasma species [15] [20,21]. Comprehensive genome-wide analyses, including comparison with data available for cattle isolates, are currently underway. It is well-known that M. bovis can be carried in the upper respiratory tract of healthy cattle and, on some occasions, even the lower respiratory tract [4]. Carriers sometimes seroconvert while others fail to do so and there is generally no correlation between antibody titers and the development of disease on an individual animal basis [4]. Little information is available with respect to carriage of M. bovis in healthy bison. In a prior study [9], we isolated the bacterium from the nasal cavities of 7/8 healthy bison sampled. Data reported here provide additional evidence to suggest that healthy bison may be carriers of M. bovis. We cannot discount the possibility that some bison appearing healthy at the time sera were collected may have suffered a prior, unobserved episode of mycoplasmosis. Furthermore, bison have evolved to mask signs of weakness or disease that might attract the attention of predators [22]. However, it seems unlikely that the 227 seropositive bison, or even the 101 that tested moderately to strongly seropositive, might all have experienced active disease without any outward signs being observed by herd managers or other caretakers. Accordingly, our results suggest that at least some bison carriers of M. bovis may seroconvert. It must be noted that the presence of antibodies reactive with M. bovis in healthy bison serves as only an indirect indicator of carriage or prior infection. Despite exhaustive inquiries, we were unable to acquire archived tissues or other materials collected prior to 2014 that are suitable to attempt either bacterial isolation or the detection of M. bovisspecific nucleic acids. A culture-based study currently underway will more definitively establish the present-day incidence and distribution of M. bovis carriage in healthy North American bison. As detailed in the Methods section, the ELISA used here has not been fully validated as a diagnostic test for bison. Whether the cutoff value defined for cattle sera is appropriate for bison sera is presently under investigation. Furthermore, the kit manufacturer makes no claim regarding specificity such that cross-reactivity with antibodies elicited by other species of Mycoplasma that infect bison cannot be ruled out (A. Ginter, personal communication). Therefore, our results must be interpreted with caution, particularly for samples testing only weakly (1+) positive. Based on preliminary data, we have thus far found high concordance between results for sera testing ≥2+ with the BIO K260 ELISA and results obtained from Western blots prepared with whole-cell lysates of M. bovis (K.B. Register, unpublished observations). Moreover, six of the seven healthy carriers identified in the earlier study mentioned above [9] tested either 1+ (n = 4) or 2+ (n = 2) positive when their sera were evaluated with the BIO K260 ELISA, and the single, culture-negative bison tested negatively (K.B. Register, unpublished observations). These data, while limited, provide some degree of confidence in the major conclusions of the study reported here, especially considering that sera testing moderately to strongly positive were found in both the United States and Canada during the entire continuum of years represented. Additionally, we noted that 57/101 (56.4%) sera characterized here as moderately to strongly positive have an ELISA score of 3+ or higher. In a prior assessment of various ELISA methods [23], a custom assay utilizing capture antigen prepared from bison isolates performed optimally in comparison to commercially available kits developed for use with cattle sera. The BIO K260 kit used here was among those evaluated but we found the difference in performance as compared to the custom ELISA to be nominal. It was not feasible to use the custom assay for the study reported here, due to the large number of sera tested and the time and labor required to produce a sufficient quantity of antigen. However, data reported here identify a subset of samples for which additional testing with the custom ELISA may more fully reveal the performance characteristics of the BIO K260 ELISA when used with bison sera, which could be of value in ongoing efforts to maximize sensitivity and specificity.
An additional caveat of this study, arising from the decision to include every serum available to us, is that the number of bison sampled on a single occasion from a single herd or site varies widely, from 2 to 294 (Table 1), without any knowledge of the proportion of animals represented by the sera collected. Consequently, data may be skewed, leading to erroneous conclusions, if most positive results arise from a relatively small number of herds with high rates of exposure to M. bovis that were disproportionately sampled. In counterpoint, seropositive bison were identified in 78% of the herds represented (32/41, with 16 each in the United States and Canada; see Table 1), of which 68.3% (28/41, with 14 each in the United States and Canada) include bison testing moderately to strongly positive. While this study lacks the rigor of one designed prospectively, with random, proportional sampling, the information gleaned from the samples available nonetheless provides novel insights worthy of further investigation.

Conclusions
A major conclusion from this study is that North American bison were likely exposed to M. bovis many years prior to the emergence of M. bovis-related disease. Seroprevalence averaged over the past 36 years is similar in the United States as compared to Canada, but geographic differences are evident when data from each country are further parsed by the era of collection (predisease versus post-disease, using the year 2000 as the dividing point). Seroprevalence in the United States during the pre-disease era was significantly higher than in Canada, but was significantly lower than in Canada during the years 2000-2019. In the United States, seroprevalence since the year 2000 has dropped significantly in comparison to seroprevalence during the years 1985-1999, while in Canada seroprevalence has increased significantly since the year 2000, as compared to the 16 years prior. For both eras, the level of antibody detected in positive samples from the United States is no different than the level in positive samples from Canada. Our data further suggest that M. bovis may colonize healthy bison in a manner sufficient to trigger antibody responses but without causing overt disease. These findings provide novel insights as to the history of M. bovis in bison and will be of value in formulating strategies to better control the spread and minimize the impact of M. bovis on bison health and production.

Methods
The BIO K260 M. bovis ELISA (Bio-X Diagnostics, Rochefort, Belgium), designed for use with cattle sera, was used to detect the presence of serum antibodies reactive with M. bovis. It was chosen in favor of an alternative product available from the same manufacturer (BIO K302 M. bovis ELISA) after consultation with the company's General Manager and Research & Development Department about the goals of our investigation (A. Ginter, personal communication). Although not fully validated as a diagnostic assay for bison, a prior study demonstrated results to be highly concordant with those from an in-house ELISA developed specifically for bison sera and in agreement with the exposure history of the animals from which the sera were obtained [23]. The sole alternative ELISA commercially available at the time our study commenced utilizes an anti-bovine IgG that reacts poorly with bison IgG [23]. In contrast, the BIO K260 and BIO K302 detect serum antibodies using protein G, which binds with high affinity to IgG of both cattle and bison [24]. All sera were tested on two different occasions, using independently made dilutions, as described in the manufacturer's instructions. Test values were used to assign each sample a numerical score, from 0 (negative) to 5+, using the following scoring metric provided with the ELISA kit: 0 = a value of ≤37, 1+ = a value > 37 and ≤ 60, 2+ = a value > 60 and ≤ 83, 3+ = a value of > 83 and ≤ 106, 4+ = a value of > 106 and ≤ 129 and 5+ = a value > 129. Sera were recorded as positive only if a result of 1+ or higher was obtained on both occasions tested. Positive sera for which replicate tests yielded discrepant scores (e.g., 1+ versus 2+) were classified based on the average numerical value of both tests. In all such cases the values obtained in each test were similar but were at the boundaries of the cutoff values used to define different levels of positivity. Student's two-tailed t test was used to evaluate the statistical significance of differences in ELISA scores obtained for the groups of sera indicated, with a P value ≤0.05 considered to be statistically significant.
A total of 3295 serum samples from bison in the United States (n = 1587) or Canada (n = 1708), collected from 1984 through 2019, was available for testing. Some bison were bled twice, in different years, such that the sera collectively represent 3173 bison, with 1548 from the United States and 1625 from Canada. All bison appeared healthy at the time blood was collected. Sera were stored frozen until use, at either − 20°C or − 80°C, depending on the institution from which they were obtained. Figure 1 indicates the number of sera collected during each of the years represented in this study. Eight sera in the group with a collection date of 1984-1987 were obtained in 1985; information available for the remaining 58 does not specify the exact year of collection. In some cases, the collection date was recorded as a range of several months spanning the end of 1 year and extending into the next. For those sera, the year assigned as the collection date for purposes of this study is the year in which the majority of the time period falls.
The sera collectively represent a minimum of 23 farms, herds or locations in the United States, mostly in Plains or Western states, and 18 in Canada, mostly from Alberta or The Northwest Territories. For locations Can-A and Can-D (an abattoir and auction market, respectively) no information is available regarding the ranch of origin for individual bison. In each of these instances related sera were considered to have come from a single origin. At 23 of the 36 sites of origin for which information is available regarding rearing practices bison were wide-ranging and minimally managed, with roundups or handling occurring no more than once yearly (1462 samples from 1340 bison). At the remaining 13 sites bison were being more intensively raised at the time of sample collection, with movement restricted by fencing and more frequent handling (1169 samples each representing a single bison). In some instances bison in both categories may have had close contact with cattle.
Abbreviations ELISA: Enzyme-linked immunosorbent assay; MLST: Multilocus sequence typing; MM: Minimally managed; RM: Ranch-managed; ST: Sequence type no role in the design of the study and collection, analysis, and interpretation of data or in writing the manuscript.

Availability of data and materials
The datasets used and analyzed for this study are available from the corresponding author upon request.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.