Detection of Treponema phagedenis–like antibodies in serum and bulk milk from cows with and without digital dermatitis

Abstract

Digital dermatitis (DD) in dairy cows is a widespread disease linked to infection with Treponema. The traditional diagnostic method is clinical inspection, which is subjective and laborious. We explored the performance of 4 different immunogenic proteins from Treponema phagedenis in a new antibody ELISA for analysis of serum or milk. Analysis of samples from 390 cows in 25 herds showed that the ELISA could distinguish the majority of cows with DD from healthy cows. By changing the cutoff and applying parallel or serial testing, high sensitivity or specificity could be achieved. The investigation indicated that aggregated test results can be useful in the assessment of a herd’s DD status. In addition, analysis of bulk tank milk samples showed good agreement with results from individual cows. The test system could be useful in research on the epidemiology and immunology of DD.

Introduction

Digital dermatitis (DD) in cattle is a disease that causes painful lesions in the skin close to the caudal part of the claw capsule. The lesions result in lameness and decreased welfare, and the disease may have negative effects on fertility and milk production. The cause of the disease has not been clarified in full. However, it has been shown to be strongly linked to infection with bacteria of the genus Treponema.^10,16,21 Environmental factors are also known to influence the occurrence and severity of DD, and herd problems occur especially in loose-housing systems with poor maintenance of the manure gutters.^1,8 The disease is widespread and has been demonstrated also in Sweden.^7,15 According to claw trimmers and field specialists, DD seems to be increasing within the Swedish dairy population (C Bergsten, pers. comm., 2014). However, surveys on its occurrence, and on the extent of the problems it is causing, are missing. One reason for this is the lack of simple and reliable methods to diagnose DD.

The traditional diagnostic method used to detect DD involves assessment of clinical signs and requires restraint of the cow and cleaning of the claw. This work is both time-consuming and potentially hazardous for the staff involved. Above all, the assessment of what changes should be classified as DD is subjective and requires expertise in claw diseases. Tools that facilitate claw inspection have been evaluated^17,19; however, such methods still rely on subjective visual assessment. Isolation of Treponema spp. from swabs or skin biopsies can be performed, but treponemes isolated from DD are fastidious, slow-growing, anaerobic spirochetes, and only a few species have been cultured, even though ≥20 species can be present in the lesions.⁹

The presence of Treponema-specific antibodies has been shown to be associated with DD. Antibody production against Treponema phagedenis in mice and cattle has been demonstrated in experimental infections.^4,23 Antibody tests based on whole-cell preparations of Treponema from DD lesions have been described and used for research purposes in some studies.^2,5,12,20,22 These serologic results have been difficult to interpret, possibly because the tests have been based on complex material (disrupted cells), and are therefore not specific enough. Another reason for ambiguous results could be antigenic variation between isolates, which has been demonstrated.^4,11 In our previous work, we identified several immunogenic proteins^13,18 (Mushtaq M. In silico analysis of Treponema and Brachyspira genomes: assembly, annotation and phylogeny [PhD thesis]. Uppsala, Sweden: Swedish University of Agricultural Sciences, 2015, https://goo.gl/3k9dKq), and initial analysis of serum and milk from cows with and without DD has indicated that antibody tests based on some of these proteins reflect the animal’s clinical DD status.

At present, there are no established laboratory-based tests for DD. We hypothesized that tests based on the identified antigens can be used to distinguish animals with and without DD. We therefore analyzed sera and milk samples from herds with and without DD in order to explore whether the test system can be used as a tool for the detection of infected cows or herds, or demonstration of freedom from DD.

Material and methods

Study population and collection of samples

Through contact with veterinary practitioners and trained claw trimmers, farmers with dairy herds that were considered to be free from or have problems with DD were identified and contacted. Herds were considered free from DD based on the opinion of the practitioners and claw trimmers that the cows had good claw health and that DD had never been detected in the herd. The prevalence of case cows in the herds was not measured before the study, and herd status was solely based on information from the farms’ veterinary practitioner or claw trimmer. A total of 25 farmers agreed to participate in the study. The herds were located in different parts of the country (9 different counties: Halland, Kronoberg, Kalmar, Jönköping, Skaraborg, Södermanland, Uppsala, Gävleborg, and Västerbotten).

In herds with DD (n = 15), assessment of individual claw health status was performed at the time of regular claw trimming, and cows with DD were identified by the claw trimmer. The claw trimming was performed on all cows in the herd and included restraint of each cow in a restraining chute. The assessment was done as part of routine inspection by the claw trimmer, and the stage of lesions was thus not documented. Different remarks, including the identification of DD, or absence of findings (i.e., “healthy”), were based on descriptions used in the Nordic Claw Health Recording Program (https://goo.gl/C3qdyp). The program describes DD as an infection of the digital and/or interdigital skin, with erosion and bleeding, painful ulcerations (stage “M2”³). Within 0–70 d of claw trimming, individual blood and milk samples were collected from cows with DD and from 10 clinically healthy cows in the same herd. All herds were sampled once, with the exception of one large herd where sampling was performed on 3 occasions. In herds where DD had never been demonstrated (n = 10), blood and milk samples were collected from 10 randomly selected cows. At the same time, a bulk tank milk sample was collected from all herds except one. The serum and milk samples were collected in evacuated tubes without additives and plastic tubes with or without bronopol, respectively, and sent by mail to the laboratory where they were centrifuged at 1,000 × g for 10 min. Sera and skimmed milk were stored at −20°C until used. Sampling was performed between December 2011 and February 2013. The total number of cows included in the study was 390.

Information about cow identity, parity, breed, and claw health recorded at claw trimming was retrieved from the Swedish Official Milk Recording Scheme (SOMRS). One of the herds was not affiliated with SOMRS, and claw health records were missing from one of the affiliated herds (both herds were free from DD). In 2012, the average herd size and milk yield of herds in SOMRS was 72 cows and 9,261 kg of milk per cow (Cattle statistics 2014, https://goo.gl/KnPKYo). For herds that participated in the study, the average number of cows was 232 (range: 105–671) for herds with DD, and 72 (range: 38–149) for herds without DD. Information about the herds’ average milk yield was not available for our study. All herds with DD were kept in free-stall stable systems. Among herds without DD, half were kept in free-stall and half were kept in tie-stall. Three categories of DD health status were used: cows with DD (DD), cows free from DD in herds with DD (DD-F), and cows in herds free from DD (F). The distribution of cows of different parity and breed is given by DD category in Table 1. The table also includes information about DD-related lesions, based on information from the claw health database of SOMRS. Claw health observations refer to the most recent regular claw trimming, recorded before sampling. There is no specific code for clinical DD in the claw health database. The average number of days between registered claw trimming of individual cows and day of sampling was 50 (range: 1–245) for cows from herds with DD and 110 (range: 17–346) for cows from herds without DD.

Table 1.

Parity, breed, origin, and claw lesions related to digital dermatitis by 3 categories of dairy cows: cows with clinical digital dermatitis (DD), healthy cows from herds with clinical digital dermatitis (DD-F), and healthy cows from herds where digital dermatitis has never been detected (F).

	Clinical status at sampling
	DD	DD-F	F
Parity
1	64	69	28
2	42	52	19
≥3	34	21	40
Unknown	5	4	12
Breed
Swedish Holstein	64	50	38
Swedish Red	64	75	22
Other breeds, crossbreeds	14	19	27
Unknown	3	2	12
Origin
Present herd	100	109	85
Other herd	42	35	2
Unknown	3	2	12
Dermatitis (“eczema”)
No	16	133	68
Yes	118	4	6
Unknown	11	9	25
Verrucous dermatitis
No	134	137	74
Yes	0	0	0
Unknown	11	9	25
Interdigital hyperplasia
No	118	133	72
Yes	16	4	2
Unknown	11	9	25
Total	145	146	99

Numbers of lesions were based on routine registrations of claw health from the most recent claw trimming (https://goo.gl/KnPKYo).

Sampling of animals for this study was approved by the regional Animal Ethics Committee, Uppsala, Sweden. All owners of herds included in the study also gave their written consent to the sampling of their animals and retrieval of cow-level information from SOMRS.

ELISA

The antibody test system was set up as 4 different versions of an ELISA with 4 different immunogenic proteins from T. phagedenis–like strain V1 as antigens.^13,18 The proteins were purified as described elsewhere.¹⁸ The ELISA versions were optimized using serial dilutions of antigens, conjugate, and serum samples from infected and uninfected cows according to standard methodology.¹⁴

For analysis of sera, the recombinant proteins were diluted in 0.05 M carbonate–bicarbonate (pH 9.6) to final concentrations of 600 ng/mL (Ttm), 40 ng/mL (PrrA), 60 ng/mL (VpsA), and 100 ng/mL (VpsB); 100 μL per well was added to C96 Polysorb microtiter plates (Nalge Nunc, Rochester, NY). The plates were incubated at 4°C for at least 16 h and washed 3 times with phosphate-buffered saline (pH 7.4) with 0.05% Tween 20 (PBST; MilliporeSigma, St. Louis, MO). To prevent nonspecific binding, 300 μL of PBST was added to each well, and the plates were incubated at room temperature for 30–60 min. Serum samples were diluted 1:100 in PBST, and 100 μL was added to duplicate wells. Plates were incubated for 1 h at 37°C. After washing with PBST, 100 μL of peroxidase-conjugated rabbit anti-bovine IgG antibody (conjugate; MilliporeSigma A8917) diluted 1:20,000 in PBST was added. Plates were incubated for 1 h at 37°C and then washed. TMB (3,3’,5,5’-tetramethylbenzidine) substrate (100 μL/well) was added. After 10 min at room temperature, the reaction was stopped by adding 50 μL of 10% sulfuric acid. Optical density (OD) at 450 nm was measured, and the values were adjusted by subtracting the mean of a blank sample. Negative and positive control samples were included on each plate. The negative control samples were from a clinically healthy cow in a herd in which no signs of DD had been recorded in the previous years. The positive control samples were from cows in a herd investigated for problems with DD. These cows had clinical signs of DD, and T. phagedenis had been isolated from one cow. All samples were analyzed in duplicate, and the OD values were correlated to the positive control with a mean OD of 1.0. For each duplicate, the mean OD was then calculated.

For milk sample analysis, the following modifications were made: the recombinant proteins were diluted to final concentrations of 1.2 µg/mL (Ttm), 40 ng/mL (PrrA), 60 ng/mL (VpsA), and 200 ng/mL (VpsB). The conjugate was diluted 1:10,000 for Ttm, PrrA, and VpsB, and 1:20,000 for VpsA. Skim milk was diluted 1:2 in PBST before analysis.

Statistical analysis

Statistical analysis was performed using Stata v.13.1 (StataCorp, College Station, TX). The difference in antibody levels between the different categories of cows was tested for each antigen using the Dunn test for nonparametric pairwise multiple comparisons (Bonferroni adjustment) following a Kruskal–Wallis test (Stata command ‘dunntest’), and then visually assessed by plotting separate test result distributions for each antigen. ELISA OD values based on serum and milk from the same cow were compared using 2-way scatter plots and calculation of the correlation coefficient. The relative sensitivity (Se) and specificity (Sp) of the test were calculated for a wide range of cutoffs (by increments of 0.04 OD) in the ELISA, based on comparisons with clinical claw status, and visualized using 2-graph receiver operating characteristic (TG-ROC) curves⁶ (Stata command ‘roctg’).

In order to explore test performance given different purposes of testing, 3 different definitions of animals with DD versus healthy animals were used. In the first comparison, cows with clinical DD were considered cases of disease, and cows from herds without DD were considered healthy. The estimates of Se and Sp relative to this reference test reflect the highest diagnostic Se and Sp that the test can be expected to have. In addition, one comparison was set up to distinguish cows from herds with DD from cows from herds without DD, regardless of their individual clinical status. Finally, definitions were also based on clinical status regardless of herd status, to identify presence or absence of clinical disease in individual animals.

Combinations of dichotomized test results based on different antigens were investigated to enable benefits from serial or parallel testing. In parallel testing, a positive result in either test is enough to consider the animal test-positive. In serial testing, on the other hand, all tests need to be positive for the animal to be considered test-positive. Different cutoffs were tried in the different test versions to identify combinations with a relatively high overall test accuracy and potential usefulness. The number of cows that tested positive to 0, 1, 2, or 3 antigens was summarized by clinical claw status and herd, and these results were compared to results from bulk milk analysis.

Results

The distribution of test results from analysis of sera and individual milk samples using the ELISA, based on different antigens, is given in Figure 1. The difference in antibody levels between the categories DD and F was significant for all antigens; difference between DD and DD-F was significant for all antigens except Ttm. Difference between categories F and DD-F was only significant for VpsB. Difference among groups based on analysis using the Ttm antigen was considered small, and ELISA results based on Ttm were therefore excluded from further analysis. The correlation between the results from analysis of serum compared to milk was 0.78 for PrrA, 0.73 for VpsA, and 0.71 for VpsB (p < 0.001).

Figure 1.

Test results from analysis of serum and milk using an ELISA based on 3 different Treponema phagedenis antigens: A. PrrA, B. VpsA, and C. VpsB. The graphs show the distribution of test results (optical density, y-axis) for cows with clinical digital dermatitis (DD), healthy cows from herds with clinical digital dermatitis (DD-F), and healthy cows from herds where digital dermatitis has never been detected (F). For each category, results from the analysis of sera are shown to the left (gray boxes) and results from the analysis of individual milk to the right (white boxes).

The relative Se and Sp were different depending on the antigen used and whether serum or milk was tested. For analysis of serum, comparison of test results from cows with clinical DD and healthy controls in herds without DD showed that the Se and Sp at the cutoff where these parameters had the same value was 59% for PrrA, 71% for VpsA, and 76% for VpsB. For analysis of milk, the corresponding values were 78%, 72%, and 72%, respectively. It was thus not possible to achieve both high Se and high Sp in any version of the test, and the trend was that the overall test accuracy (i.e., considering both Se and Sp) was better for higher cutoffs compared to lower. In general, lower cutoffs were needed when the test results were based on the analysis of milk instead of serum. When paired samples were compared, whether analysis of serum or milk produced the highest test performance depended on the antigen and on the reference group used. The estimated relative Se and Sp at different cutoffs in the ELISA version using VspA as antigen is shown in Figure 2. Three definitions of animals with DD versus healthy animals were used, and the results shown are from analysis of serum and based on the definition in which cows with clinical DD were considered cases, and cows from herds without DD were considered healthy. The cutoff where Se ≈ Sp was the same or slightly higher for the other definitions (OD = 0.08 and 0.12, respectively), and the estimated Se or Sp at this cutoff was ~70%, compared to 64% and 66% for the other definitions. Similar trends in Se and Sp pattern across reference group definitions were seen for test versions using PrrA or VspB.

Figure 2.

Sensitivity (Se) and specificity (Sp) for different cutoffs in the ELISA based on VpsA antigen. The graph shows the relative test performance of the ELISA using cows with clinical digital dermatitis (DD) as cases of digital dermatitis, and healthy cows from herds without DD, as animals free from DD. Dashed lines indicate the 95% confidence interval of Se and Sp. Test results are based on analysis of serum from 390 cows in 25 herds with and without DD (https://goo.gl/KnPKYo).

Combinations of serum test results from test versions based on PrrA, VpsA, and VpsB, and a cutoff that ensures 95% Sp for each test version, are summarized in Table 2. The relative Se and Sp of parallel testing based on these combinations was 60.7% (95% confidence interval [CI]: 52.2–68.7) and 87.9% (95% CI: 79.8–93.6). Serial testing resulted in a relative Se and Sp of 11.7% (95% CI: 7.0–18.1) and 100.0% (95% CI: 96.3–100.0). On a herd level, this combination of test results showed that most animals (87.9%) in herds in which clinical DD was absent were test-negative in all test versions. Four of these herds had 1 or 2 animals that were positive in 1 of 3 tests, and 2 herds had 1 and 2 animals, respectively, that tested positive in 2 of 3 tests. Only 1 of 13 cows, in herds without DD that were test-positive in 1 or more test versions, was born in a different holding than the present (information about origin was missing for 3 of 13 cows). No animals in herds without DD tested positive in all 3 test versions (Table 2). In herds with clinical DD, the proportions of animals that tested positive in one or more test versions were more common. However, 2 herds with clinical DD had no animals that tested positive in more than 1 test version. In total, 4 of 15 herds with DD had proportions of test-negative cows that were similar to herds without DD.

Table 2.

Number of cows (n) that tested negative (–) or positive (+) when serum samples were analyzed using an ELISA based on 3 different Treponema phagedenis antigen (PrrA, VpsA, VpsB). The number of cows for each possible combination is presented separately for each category of clinical claw status (based on the clinical status of the cow and the herd): cows with clinical digital dermatitis (DD), healthy cows from herds with clinical digital dermatitis (DD-F), and healthy cows from herds where digital dermatitis has never been detected (F).

Test result (PrrA/VpsA/VpsB)	DD		DD-F		F
	n	%	n	%	n	%
+/+/+	17	11.7	6	4.1	0	0.0
+/+/−	6	4.1	4	2.7	1	1.0
+/−/+	9	6.2	4	2.7	0	0.0
−/+/+	14	9.7	4	2.7	2	2.0
+/−/−	17	11.7	7	4.8	4	4.0
−/+/−	12	8.3	10	6.8	2	2.0
−/−/+	13	9.0	6	4.1	3	3.0
−/−/−	57	39.3	105	71.9	87	87.9
Total	145		146		99

Bulk milk samples from 24 of 25 herds were available for analysis. All free herds had ELISA OD values <0.100 in all test versions (PrrA, VpsA, and VpsB). Among herds with clinical evidence of DD, 3 herds had equally low bulk milk OD values. These herds all had a large proportion of cows that tested negative to all antigens, and a few (0 or 1) cows that tested positive in more than 1 test version. The remaining 12 case herds had bulk milk OD >0.100 in at least 1 test version (parallel testing). On a herd level, based on the limited number of herds included in our study, this corresponds to a bulk milk test performance with a combined Se of 80% (95% CI: 51.9–95.7) and a Sp of 100% (95% CI: 66.4–100.0).

Discussion

Our study confirms that cows with DD seem to develop measurable amounts of antibodies against 3 of the proteins tested as antigen in the ELISA. However, there are some cows that lack detectable levels of such antibodies, and there is also a small number of cows in herds where DD has never been demonstrated that test positive for one or more antigens. At the individual level, the current version of the test may not be used in such a way that both the diagnostic Se and Sp remain high. However, by applying a high test cutoff, a high Sp can be achieved. In this way, the test may be used to confirm presence of disease. Results also show that testing based on the 3 different test versions may be used to further minimize the proportion of false results.

One reason for some of the seemingly incorrect test results may be the imperfect reference test. The clinical status of cows and herds in our study relies on subjective assessment by practitioners and claw trimmers. Not all DD lesions are easy to identify, and it is possible that the assessment was influenced by differing and perhaps insufficient diagnostic competence, differences in case definition, or suboptimal circumstances at the time and place of claw trimming. Moreover, the study was cross-sectional, and we cannot exclude that some of the apparently healthy control cows had a history of DD and had persisting antibodies from a previous immune response. Such cows could also have been present in herds where clinical DD was absent. There are also several possible explanations for why the test system would result in false-negative results. One is that the antigens used in the test are slightly different from the target protein of the bacteria. Other studies indicate that bacteria such as T. phagedenis may change their protein expression and thereby escape specific antibodies present in the host (Mushtaq M. In silico analysis of Treponema and Brachyspira genomes). It has also been shown that cows with DD are often infected with several species of Treponema.^10,16 In other words, although the most prevalent species in DD lesions is T. phagedenis, not all cows with DD may have antibodies against the proteins tested in our study. Cross-reactivity between all DD-related Treponema spp. is not possible to investigate because only a few species are cultivable. It is likely that the inclusion of additional Treponema proteins in the test system, preferably from other species, will help to improve test sensitivity. In addition, it is possible that sampling of some of the cows in the study was performed too soon, or too late, in relation to clinical disease. There is still very limited information on the dynamics of the antibody response related to DD, and more longitudinal studies are needed to clarify this. Also, the possibility of cross-reactivity to antibodies produced against related proteins cannot be ruled out. Finally, further refinement of the ELISA is ongoing and may remove some of the present inaccuracies.

Based on summaries of individual test results by herd, herds where DD had never been detected had a relatively high proportion of cows that did not test positive to any antigen. In contrast, herds with clinical problems with DD had a lower proportion of cows that tested negative to all 3 antigens, and in general, more cows tested positive to 1, 2, or even 3 antigens. This difference in patterns implies that aggregated test results from several individuals can be useful in the assessment of a herd’s DD status.

For test results based on analysis of bulk milk, within-herd prevalence can be expected to have an influence. Among the herds in the study that had clinical problems with DD, 3 had low levels of detectable antibodies in bulk milk, and these herds also had a high proportion of completely test-negative cows. On the whole, the results from analysis of bulk milk and the summary of individual results by herd showed relatively good agreement. In other words, bulk milk testing may be an efficient method to gain information about DD herd status. To confirm this, bulk milk test performance needs to be further investigated by analysis of samples from a larger number of herds.

There are several ways a serologic test could contribute to the control and prevention of DD. In order to design and evaluate control strategies, it is important to be able to measure the prevalence, incidence, trends in occurrence, and potential signs of disease spread within and between groups or herds. This is made possible by sampling of blood or milk, which is relatively simple and cost-efficient, and by performing repeated and objective testing of large numbers of animals. Such testing also facilitates other epidemiologic studies of the disease, including assessment of risk factors and production losses as a result of DD. It further enables studies of how the antibody response in infected animals changes over time—how quickly antibody levels rise after clinical disease, how long antibodies persist, and to what extent the presence of high antibody levels may protect against disease recurrence. This information would play a crucial part in the prospective development of a vaccine against the disease.

In the search for herds to be included in our study, it was evident that few herds are free from claw disorders. In particular, it was difficult to find herds that had never had an observed case of DD. Still, these herds do exist and we expect that there are many herds that have a very low prevalence. With the right tools, in countries or regions with a low prevalence, herds could be protected from future problems with DD. For example, purchase of animals from test-positive herds could be avoided. On an individual level, being test-positive could in some circumstances be seen as an additional argument for culling a cow. In herds that experience problems with DD, test results may be used as one of several sources of information considered in the grouping or removal of cows.

Our study shows that the ELISA described can, to some extent, distinguish cows with DD from healthy cows. A combination of results from test versions based on different antigens may be used to improve test performance. By aggregation of individual test results on a herd level, or testing of bulk tank milk samples, information on the DD status of herds can be gained. Presence of inaccurate test results shows that the test needs to be further improved in order to be considered a reliable detection tool, especially on an individual level. However, at this stage, the test could be useful in research into the epidemiology and immunology of the disease.