Management of Mycobacterium avium subsp. paratuberculosis in dairy farms: Selection and evaluation of different DNA extraction methods from bovine and buffaloes milk and colostrum for the establishment of a safe colostrum farm bank

Abstract The aim of this study was to develop and validate different innovative DNA extraction methods to detect Mycobacterium avium subsp. paratuberculosis (MAP) DNA from bovine and buffalo colostrum. Paratuberculosis is a chronic inflammatory infection of domestic and wild animals, especially ruminants, caused by MAP. The primary route of disease transmission is feces, but MAP can also be excreted in milk and colostrum. In 2015, the Italian Ministry of Health has issued a voluntary control plan of MAP in order to allow risk‐based certification of bovine and buffaloes farms. In addition to the annual diagnostic screening and to the clinical surveillance of animals the plan includes the adoption of biosecurity and management measures to progressively mitigate the incidence of MAP. To achieve this goal it is crucial to ensure the accuracy of the methods used to detect the presence of MAP in bovine and buffaloes milk and colostrum, in order to: (1) support a "safe colostrum farm‐bank" set‐up and thus prevent the main within‐farm MAP transmission route and (2) to allow the MAP‐free certification of milk products for export purposes. To achieve these goals, seven different DNA extraction protocols were identified from bibliography, out of which three methods were finally selected after the adoption of an evaluation procedure aimed at assessing the efficiency of extraction of DNA, the purity of DNA and the adaptability of the DNA amplification: NucleoSpin® Food Kit (Macherey‐Nagel), NucleoSpin® Food Kit (Macherey‐Nagel) combined with the magnetic beads, and QIAamp Cador Pathogen Mini kit (QIAGEN). In particular, the NucleoSpin® Food Kit (Macherey‐Nagel) and the QIAamp Cador Pathogen Mini kit (QIAGEN) were tested on bovine and buffalo colostrum, showing a LOD between 4 × 104 (2.6 × 106 cfu/ml) and 4.08 (26.7 cfu/ml) IS900 target copies and a LOD between 5.3 × 105 (4.1 × 106 cfu/ml) and 53 (4.1 × 103 cfu/ml) IS900 target copies, respectively.


| INTRODUC TI ON
Mycobacterium avium subsp. paratuberculosis (MAP) is the etiological agent of paratuberculosis (Johne's Disease), a contagious and chronic gastrointestinal disease affecting domestic and wild ruminants. The disease has a primarily oral-fecal transmission route and the main source of infection is represented by feces and milk of the affected animals. In addition, there is the possibility of a bacterial streaming in other body districts, such as womb, testicular parenchyma, and breast representing another source of dissemination (Ayele, Svastova, Roubal, Bartos, & Pavlik, 2004;Sweeney, Whitlock, & Rosenberger, 1992;Whittington & Windsor, 2009). After the infection, the disease can remain in a subclinical phase for several years, and only few animals show typical symptoms of paratuberculosis (Klinkenberg & Koets, 2015).
Moreover, many reports refer to the presence of MAP DNA in powdered infant formula (Hruska, Bartos, Kralik, & Pavlik, 2005, Donaghy, Johnston, & Rowe, 2010, Hruska, Slana, Kralik, & Pavlik,2011. The serological ELISA test on bovine and buffalo serum is usually used as a screening test, because it is a rapid and low-cost assay, but its sensitivity is very low (15%) during the subclinical stage (Chui, King, & Sim, 2010). The cultural isolation is considered the gold standard in the detection of MAP, but it is time consuming especially in the case of samples with a low number of viable bacteria, for which long incubation times are needed to obtain bacterial growth. However, the sensitivity of this test in the animals affected by subclinical infection is reduced (23%-29%), while the specificity is up to 100%. (Nielsen & Toft, 2008). On the other hand, PCR represents a rapid test to detect MAP in feces, milk, and clinical samples (Fang et al., 2002;Grant et al., 1998;Millar et al., 1996;O'Mahony & Hill, 2002;Pillai & Jayarao, 2002).
The most common molecular target for this assay is the IS900, an insertion element present in 15-18 copies in MAP genome (OIE, 2014;Ricchi et al., 2016). However, the detection of IS900-like sequences in non-MAP isolates represents a possible cause of false-positive results (Green, 1993;Laurin, McKenna, Chaffer, & Keefe, 2015;Taddei et al., 2008;Tasara, Hoelzle, & Stephan, 2005). To overcome the problem, single-copy F57 target has been used to confirm the IS900 assay from different matrix, including milk (Meadus, Gill, Duff, Badoni, & Saucier, 2008;Ricchi et al., 2009;Schonenbrucher, Abdulmawjood, Failing, & Bulte, 2008;Stephan, Schumacher, Tasara, & Grant, 2007;Tasara et al., 2005). Since the global objective is to progressively mitigate the risk of incidence of paratuberculosis infections in cattle and buffaloes and possibly reduce the exposure of humans to MAP through the food chain, it is crucial to ensure the accuracy of the methods used to detect its presence in bovine and buffaloes milk and colostrum. In this scenario, the availability of accurate methods and protocols aimed at establishing a safe colostrum farm-bank, representing an important step to prevent one of the main whitin-farm MAP transmission route.
In Italy, the Ministry of Health set off in 2015 a voluntary control plan of MAP in order to allow a risk-based certification of both bovine and buffaloes farms and products also for export purposes (SANCO, UE, European Commission, 2000; Veterinary certificate for import of milk and milk products into India, IN-LO1; Health Certificate for milk, milk products intended to be exported from Italy to people's Republic of China, RPC-LO1). In addition to the annual diagnostic screening and to the clinical surveillance of animals the plan includes the adoption of biosecurity and management measures to progressively mitigate the incidence of MAP. The aim of this study was to set up and validate different innovative DNA extraction methods to detect MAP DNA from bovine and buffalo milk and colostrum. In the first stage of the work, seven different DNA extraction methods have been tested on artificially contaminated pasteurized commercial milk, selecting the three best protocols according to the yield and the bibliographic research.
Then, experiments on artificially contaminated and negative bovine and buffalo milk and colostrum samples have been conducted aimed at the performance evaluation of the selected extraction protocols through the interrater agreement assessment; finally, the protocols (26.7 cfu/ml) IS900 target copies and a LOD between 5.3 × 10 5 (4.1 × 10 6 cfu/ml) and 53 (4.1 × 10 3 cfu/ml) IS900 target copies, respectively.

K E Y W O R D S
Colostrum, DNA extraction, evaluation, MAP, Real-time PCR were tested on buffalo colostrum field samples and interrater agreement was again assessed.

| Materials
The protocol was validated using the Mycobacterium avium subsp.
A three-step evaluation framework was applied.

| Artificial contamination of milk
Two liters of commercial pasteurized bovine milk were purchased from a local supermarket, then aliquotated in 50 ml test tubes and stored at −20°C until analysis.
The aliquots were thawed at room temperature and half of them were artificially contaminated with 1.15 × 10 5 cfu/ml MAP strain (ATCC 19689), grown on Herrold's egg yolk medium for 3 months at 37°C.
The remaining aliquots were contaminated with 3.6 × 10 11 TOPO-TA-IS900 recombinant plasmid, corresponding to 3.1 × 10 10 cfu/ml. The two groups of aliquots have been subjected to two different analyses for the DNA extraction kit selection and the LOD verification.
All the methods were tested on pasteurized commercial bovine milk purchased from a local supermarket and contaminated with MAP strain (ATCC 19689). Then, the kits were selected on the basis of the DNA yield and purity evaluated through photometric reading and the Real time PCR results. (See Table 1).
In addition, further parameters for the kit selection were con-

| DNA extraction protocols
For the three selected protocols, 50 ml of commercial pasteurized milk was centrifugated at 2,500g for 20 min. The supernatant was discarded and the pellet was used for the DNA extraction.
• Protocol (D): NucleoSpin ® Food kit 740,945.50,Dürer,Germany) The pellet was resuspended in 550 µl of Buffer CF. The sample was added to a 2 ml tube containing 10 µl of Proteinase K and incubated at 65°C for 30 min. The sample was centrifugated at 10,000g for 10 min, and then the DNA extraction was proceeded with kit instruction.
Finally, the sample was eluted with 100 µl of elution buffer. The pellet was resuspended in 50 µl of magnetic beads (Adiapure) and incubated in continuous stirring at room temperature for 30 min.

TA B L E 1 Criteria for the DNA extraction kit selection
The pellet was recovery putting the sample on magnetic support for 20 min. After, the liquid phase was discarded and the sample was resuspended with 550 µl of CF Buffer and 10 µl of proteinase K in a 2 ml tube. Then, the solution was incubated at 65°C for 30 min and the DNA extraction was proceeded with kit instruction. Finally, the sample was eluted with 100 µl of elution buffer.

Hilden, Germany)
The pellet was added to a 2 ml tube containing 100 µl of VXL buffer and 10 µl of proteinase K and the sample was incubated at 25°C for 15 min. The DNA extraction was proceeded with kit instruction.
Finally, the sample was eluted with 100 µl of elution buffer.

| Real Time IS900
The master mix was composed by:

| Limit of detection (LOD) for applicability of selected molecular protocols.
For each considered matrix two standard curves were analyzed. The first was drawn up using the ATCC 19689 MAP strain, while the second using the recombinant TOPO-TA plasmid containing IS900 target concentration. Both the obtained standard curves showed values between 5.3 × 10 10 target copies (equivalent to 4.1 × 10 11 cfu/ml), and 5.3 IS900 target copies (equivalent to 4.1 × 10 1 cfu/ml), with corresponding Ct values between 3.6 × 10 11 target copies (equivalent 3.1 × 10 10 cfu/ml) and 3.6 IS900 target copies (equivalent to 2.4 × 10 1 cfu/ml). In addition, LOD were verified for all three selected DNA extraction protocols using three different operators.

| Step 2 -Performance evaluation of the selected protocols on bulk tank milk and individual colostrum
Bovine and buffalo bulk tank milk and individual colostrum were taken from two farms periodically tested for MAP and always nega- Milk and colostrum samples were thawed at room temperature and artificially contaminated at increasing concentrations with TOPO-TA-IS900 recombinant plasmid.

| Panels constitution.
Two different panels of milk and colostrum samples were constituted as follows: • 10 negative (non contaminated) samples.
The milk and colostrum panels were tested in triplicate with each of the three different selected extraction protocols and using three different operators, for a total of 120 samples/protocol.

| Evaluation schedule
Through the evaluation process the following statistic parameters were calculated: sensitivity (Se) by determining the LOD, specificity (Sp), accuracy, positive predictive value (VPP), negative predictive value (VPN), and the statistic K value (Cohen concordance), based on the OIE manual (OIE, 2008(OIE, , 2013.The sensitivity was calculated according to the formula: The specificity was calculated according to the formula: The accuracy was calculated according to the formula: The positive predictive value was calculated according to the formula: The negative predictive value was calculated according to the formula: where "TP" represents true positive, "FP" represents false positive, "TN" represents true negative and the "FN" represents false negative according to test results of the standard assay of comparison.
The statistic K value was calculated considering that K > 0.75 indicates an optimum concordance level between multiple operators not attributable to the impact of the case. For the evaluation was used the following formula: where "P 0 " represents the sum of observed concordance rate and "P e " represents the sum of the products of the marginal level of percentages.
The values of K test were established on the basis of procedures reported on the OIE manual, in which the optimal K value is between 0.81 and 1 (OIE, 2008; Cap.1, 27-43).

| Step 3 -"On-field evaluation" of the selected protocols on individual buffalo colostrum
Two liters of buffalo individual colostrum were collected from two naturally infected but asymptomatic animals, 1 L each. The animals were classified infected as they resulted positive to ELISA test on blood and to bacteriological examination and to Real-Time PCR on feces.
Two liters of buffalo individual colostrum were collected from two clinically healthy noninfected animals (ELISA negative and negative to bacteriological examination and to Real-Time PCR on feces) from a farm MAP free since 2014.
All samples were aliquotated in 50-ml test tubes and stored at −20°C until analysis.
Milk and colostrum samples were thawed at room temperature.
Before each experiment, a bacteriological analysis was performed on each sample in order to evaluate possible interferences with the results due to the presence of other pathogenic or commensal microorganisms (Enterobacteriaceae, Streptococcaceae, Listeria Monocytogenes, Pseaudomonas spp., sulfite-reducing clostridia).
A panel of buffalo colostrum samples was constituted as follows: • 10 negative samples from clinically healthy and tested negative animals (blood and feces) • 10 positive samples from naturally infected and tested positive animals (blood and feces) The panel was tested in triplicate with each of the three different selected extraction protocols and using three different operators, for a total of 60 samples/operator.

| Step 1 -Selection of DNA extraction protocols on pasteurized milk
The photometrical reading reported a yield of 21.5 ng/µl and a DNA purity of 1.75 for the Cador Pathogen kit, a yield of 21.5 ng/µl and a DNA purity of 1.75 for Macherey-Nagel kit and a yield of 18.15 ng/ µl and a DNA purity of 1.67 for Macherey-Nagel kit combined with magnetic beads.

| Step 2 -Performance evaluation of the selected protocols on artificially contaminated bovine and buffalo bulk tank milk and individual colostrum
The best yield for the bovine bulk tank milk was of 15 ng/µl for Macherey-Nagel kit.
In both bovine and buffalo samples part of magnetic beads remained trapped in the milk fat component, reducing the total extraction yield. The presence/absence of somatic cells in the samples was also analyzed, observing that their presence can hinder the beads recovery and consequently the extraction yield. However, the Macherey-Nagel kit presents a pretreatment buffer that effectively reduces the cellular and fat component.
Analogous results were obtained for bovine and buffaloes colostrum, related to the extraction yield both for the Macherey-Nagel kit and for the Macherey-Nagel kit combined with magnetic beads (22.45 ng/µl). In addition, the LOD obtained from bovine colostrum resulted equivalent for both DNA extraction kits, with a range values from 4 × 10 4 (2.6 × 10 6 cfu/ml) and 4.08 (26.7 cfu/ml) IS900 target copies.
The QIAamp Cador Pathogen kit for the bovine colostrum showed a lower yield extraction equal to a 10 ng/µl and a LOD with a range value from 4 × 10 3 and 10 IS900 target copy.
Regarding the extraction from buffalo colostrum, the Macherey-Nagel kit showed an evident great yield value (16.4 ng/µl) compared with Macherey-Nagel kit combined with the magnetic beads (10 ng/µl).
For the QIAamp Cador Pathogen kit were reported only the results on the buffalo and bovine colostrum.
This result is due to the high presence of fat that traps a large portion of magnetic beads.
To recover a significant quantity of magnetic beads, the centrifugation phase has been eliminated, as the beads remain concentrated at the bottom of the tube, together with the colostrum fat particles, making it impossible to obtain a good magnetic separation.
The statistical results of selected DNA extraction methods from bovine and buffalo colostrum were reported (see Table n.5).
The Macherey-Nagel kit without magnetic beads and QIAamp Cador Pathogen kit showed perfect statistic values, evidencing 100% sensitivity and specificity with a K value equal to 1, underlining an optimal concordance between operators.
The Macherey-Nagel kit with magnetic beads gave a 100% specificity and only 97.7% sensitivity due to the misclassification by two operators of two positive samples as negative at a 5.3 × 10 5 TOPO-TA-IS900 target copies (4.1 × 10 6 cfu/ml) concentration in the buffaloes colostrum panel.

| Step 3 -Results for individual buffalo colostrum (field samples)
For the on field evaluation of the selected protocols on individual buffalo colostrum, were considered 10 positive samples with the same level of MAP contamination (1.8 × 10 5 ) and 10 negative samples for each operator. In the Table 6 was reported the panel constitution and the average Ct detected on the individual buffalo colostrum using three operators (see Table 6).
A 100% sensitivity and specificity resulted for the three selected protocols used. All positive and negative samples of the panel were correctly classified with a complete accordance (100%) between operators and between the selected protocols.

| D ISCUSS I ON
The bibliographic review conducted on commercial and home-made kits routinely used for the DNA extraction from complex matrices, rich in fats and proteins, allowed a preliminary selection of kits and protocols with the best theoretical potential for a further development and for their application to MAP DNA extraction from milk and colostrum.
Specifically, the DNA extraction protocols subjected to evaluation were those routinely employed to detect GMO in different foods, such as chocolate products, cocoa, nougat products, breakfast cereals, muesli, nut/chocolate spread, jam and fruit concentrates, pollen, lecithin, spices, bread, raw processed products and cosmetics (plant and animal ingredients).
The evaluation process (step 1) allowed to verify the actual suitability of three protocols for MAP DNA extraction from the matrix "milk," confirming their potential validity of application on bovine and buffalo milk and colostrum taken in the field.
The results obtained from the evaluation process on artificially contaminated bulk tank milk and individual colostrum revealed (step TA B L E 3 Sperimental data obtained from bovine and buffalo colostrum on contaminated field samples Nucleo Spin Food kit (Macherey Nagel) combined with magnetic beads Bovine milk 10 ng/μl 1.7 × 10 7 −7.7 1.75 × 10 9 −7.7 Nucleo Spin Food kit (Macherey Nagel) combined with magnetic beads buffalo milk 47.6 ng/μl 3.86 × 10 6 −9.7 2.54 × 10 9 −5 2), for two protocols out of three, 100% sensitivity and specificity for MAP DNA extraction at increasing concentration of TOPO-TA-IS900 recombinant plasmid, along with a perfect repeatability between operators.
Both LOD detected with the two selected protocols were within the range of bacterial count observed in dairy cattle with subclinical symptoms (Stabel et al., 2014), specifically between 1.24 × 10 4 and 1.4 cfu/ml. As a consequence, this protocol showed an overall 100% specificity and a lower sensitivity (97.7%), with an interoperators agreement classified as "good" (K statistic value = 0.966). The probability that a negative sample is actually negative. f Statistical coefficient representing the degree of accuracy and reliability in a statistical classification. Recently, agreements have been signed between Italy and some third countries, for example China and India, to ensure the export of dairy products certified free from several diseases, such as tuberculosis, brucellosis, anthrax, listeriosis, and paratuberculosis.
In this context, the use of the validated MAP DNA extraction protocols opens up realistic prospects for the specific certification of PDO dairy products, such as milk itself, Parmesan cheese, Pecorino Romano cheese, and buffalo's Mozzarella, also for export purposes.

ACK N OWLED G M ENTS
The work was supported by the research project IZSLT 07/14 of Istituto Zooprofilattico del Lazio e della Toscana "M. Aleandri" financed by the Italian Ministry of Health. We thank Luca Buttazzoni (Research Centre for the meats production and genetic improvement-CREA) for providing the zootechnical farm of Ministry of Agriculture, the zootechnical farmer Cesare Riccioni for the assistance during the sample collection and Giancarlo Micarelli of public veterinary service to making available the veterinary staff.

CO N FLI C T O F I NTE R E S T S
None declared.

AUTH O R CO NTR I B UTI O N S
Fabrizio Gamberale, Gabriele Pietrella, and Antonella Cersini were involved in data curation and visualization.