Performance of MALDI biotyper compared with Vitek™ 2 compact system for fast identification and discrimination of Staphylococcus species isolated from bovine mastitis

Abstract This study was designed to evaluate the ability of MALDI Biotyper (MBT) compared with Vitek™ 2 compact system for accurate identification of Staphylococcus aureus (S. aureus) and coagulase‐negative staphylococci (CNS) strains and discriminate methicillin‐sensitive S. aureus (MSSA) from methicillin‐resistant S. aureus (MRSA). Throughout Al‐Qassim region, Saudi Arabia, a total of 198 isolates of S. aureus (132 MSSA and 66 MRSA) and 44 CNS were collected from five dairy farms where the prevalence of staphylococcal mastitis was reported. The results produced by Vitek™ 2 compact system demonstrated that 123/132 MSSA isolates (93.18%), 61/66 MRSA (92.42%), and 37/44 CNS species (84.09%) were correctly identified. However; 130/132 MSSA (98.48%), 64/66 MRSA (96.96%), and 44/44 CNS (100%) were correctly identified by MBT with score ≥2. 00. The principal component analysis (PCA) dendrogram generated by MBT illustrated that the tested isolates were classified into two groups of Staphylococcus species at the distance level of 600. S. aureus isolates were found to be closely related with higher peak intensities in the mass of 3,993 Da, 4,121 Da and 5,845 Da were detected in MRSA, whereas, that were lost in MSSA. Conclusion: This study verified that MBT is an alternative powerful tool for precise identification and discrimination of Staphylococcus species.

characters are not accurate and laborious. In general, the morphological identification of bacteria is frequently shadowed by biochemical analysis of virulence genes as coagulase and thermonuclease (Wang et al., 2013). However, phenotypic and genotypic analyses are considered the most truthful methods for identification of S. aureus mastitis; these methods are laborious, expensive, and are not up till now been usually utilized in discriminating methicillin-resistant S. aureus (MRSA) from methicillin-sensitive S. aureus (MSSA). Although, MRSA was recognized ( Koivula et al., 2007;Wolters et al., 2010) and discriminated from MSSA due to spectral variations ( Böhme et al., 2012;Du et al., 2002), there is little evidence that has been presented regarding the use of matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF-MS) protein fingerprinting as a typing technique for grouping the different isolates of S. aureus at the strain level. Furthermore, one of the most common restrictions associated with the precise diagnosis of CNS is the absence of an exact, fast, and suitable technique that can discriminate the microbial species including this group (Cantekin et al., 2015;Koivula et al., 2007;Wang et al., 2013).
To facilitate the investigation of eruptions, the contagious isolates must be typed; on the other hand, nearly all molecular techniques are luxurious or labor-intensive. For this reason, there is still a need for supplementary quick, low-cost, and a precise technique for detection of various microorganisms of infectious diseases. In general, innovative techniques such as MALDI-TOF-MS for accurate and fast characterization of different microbes is considered an important step toward suitable prevention and control of communicable diseases in medical and veterinary diagnosis and currently are of major concern (El Behiry et al., 2014;Hays & Van Leeuwen, 2012;Sauer et al., 2008).
Formerly, the MALDI mass spectral imaging technique was developed as a tool for intact protein imaging of the tissue surface (Stoeckli et al., 2001;Stoeckli et al., 2002). Recently, a powerful mass spectral (MS) technique has developed the fields of proteomics, lipidomics, and metabolomics. This MS revolution has fortified researches to value the individual characterization and quantitation of biomolecules in living systems. In general, proteomics are defined as the identification and quantification of all the expressed proteins of a biological sample, with the aim of understanding their functions, their interactions, and their contribution to biological processes. Correspondingly, metabolomics has the goal to list and enumerate the entire range of metabolites. The developing field of lipidomics concentrates on the whole changes of lipid molecules .
Although the MS was recognized at the beginning of 1900s, its application was restricted to the chemistry sciences. Therefore, the progress of MALDI-TOF-MS in 1980s extended the scope of MS to different biological molecules such as proteins (Singhal et al., 2015).
In an attempt to identify a cheaper and accurate method of identification, we employed protein fingerprinting using MALDI-TOF-MS as a diagnostic method for typing and fast detection of Staphylococcal species. MALDI-TOF technology allows precise detection of numerous types of microorganisms in limited time, from 2 days to a few minutes, with a slight amount of bacteria needed for an accurate investigation (10 4 to 10 6 CFU). Mass spectral Identification of microbes by MALDI-TOF-MS can be used for a large variety of bacteria, yeasts, and molds (Liu et al., 2007;Pignone et al., 2006). Limited numbers of cells are enough for rapid and exact identification.

MALDI Biotyper (MBT) is a diagnostic method in which chemical
substances are ionized into charged molecules and the mass charge ratio (m/z) is considered. Spectra ranging from 2,000 to 20,000 m/z were examined using the MBT system's automated control and the identified values of ≥2.000 specified species-level identification, values ranging from 1.700 to 1.999 pointed to the identification of genus level, and scores of 1.700 showed no faithful identification (Chen et al., 2014;Cheng et al., 2015;Deng et al., 2014). Our current study aims to evaluate the performance of MBT compared with an automated colorimetric method using the Vitek ™ 2 compact system in the identification of S. aureus and CNS species and discrimination of MSSA from MRSA strains isolated from milk of dairy cows suffering from mastitis to deliver a novel opportunity for fast, real identification, and control of serious staphylococcal mastitis.

| Isolation of genomic DNA
DNA was extracted from all S. aureus isolates using Nexttec TM cleanColumn (nexttec GmbH Biotechnologie, Leverkusen, Germany).
The extraction method was performed according to the manufacturer's guidelines to obtain a purified DNA.

| PCR amplification
PCR amplifications were carried out with a triple of primers specific for the nuc gene, which encodes to specific regions of the thermonuclease of S. aureus, mecA gene, which is specific to methicillin resistance and 16S rRNA, a genus-specific gene for staphylococcal | 1063 species. The primers were shown in Table 1 and designed according to the sequences published in Biomers, Ulm, Germany.

| Colorimetric identification of Staphylcoccus species by Vitek ™ 2 compact system
All preliminary identified isolates of S. aureus (MSSA and MRSA) and CNS were tested by Vitek ™ 2 compact system (BioMe′rieux, France) according to the procedures provided by the company.
The results were finally interpreted and tabulated automatically by the ID-GPC library. The confidence value was expressed by seven various categories of results: excellent, very good, good, acceptable identification (only one result is delivered), low discrimination (if more than one result is recorded, the software recommends supplementary investigations to be achieved), or the data are inconclusive or unidentified.

| Bacterial test standard (BTS) preparation
A quantity of 50 μl of standard solvent was added to the in vitro diagnostic product (IVD) BTS pellet and dissolved by pipetting up and down for 20 times at room temperature. The IVD BTS solution was then mixed for 5 min by pipetting up and down for 20 times and centrifuged at 13, 000 rpm for 2 min. 2 μl aliquots of the supernatant were pipetted into a separate screw cap, micro tubes and the rest was stored at −18°C.

| Sample preparation
For preparation of the samples for MBT, a single fresh colony of overnight culture, incubated for 18-24 hr at 37°C was used for each isolate. Ethanol/formic acid extraction technique was performed as described by Bruker Daltonik, Germany. In brief, 1-2 colonies were transferred into a sterilized Eppendorf tube and mixed carefully in 300 μl of deionized water. A quantity of 900 μl of 100% ethanol was then added, the contents of the tube were judiciously mixed, and the tube was then centrifuged at 13, 000 rpm in a desktop centrifuge (centrifuge 5430, Eppendorf, USA) for 2 min.
The liquid that lay above the sediment was discarded and the sediment was dried in air for at least 5 min. From 10 to 80 μl of formic acid (70%) was then supplemented according to the dried sediment size and mixed vigorously. An equivalent volume of acetonitrile, like formic acid, (70%) was then added and the mixture was centrifuged again at 13, 000 rpm for 2 min and then 1 μl of the liquid, floating on the surface above the precipitate was placed onto 96-well target plate and permitted to dry at 25°C. Consequently, each sample was inoculated with 1 μl of matrix solution, which formed a saturated solution of α-cyano-4 hydroxy-cinnamic acid (HCCA) in 50% acetonitrile and 2.5% trifluoroacetic acid (final concentration: 10 mg HCCA/ml) purchased from Sigma Aldrich (# 19182) and dried in air at 25°C. The target plate was successively presented into the MBT device for direct measurement and data interpretation via FlexControl software with Compass Flex Series version 1.3 software and a 60-Hz nitrogen laser (337 nm wavelengths). All tested samples were blinded and run in duplicates for obtaining accurate identification.

| Data analysis
In the range from 0 to 3, the score value of unknown spectrum was detected by matching the unknown spectrum with a recognized spectrum from the MBT taxonomy. The precision of the species identification was determined as described by the manufacturer procedures of Bruker Daltonics, Germany. The extremely accurate identification was carried out by MBT, when the log score was in the range from 2.30 to 3.00; however, species and probable genus identification were detected in the range from 2.000 to 2.299 and 1.700 to 1.999, respectively. In addition, the log score range from 0.000 to 1.699 meant that the identification was not reliable. The different spectra produced by compass software were evaluated in a m/z range between 2, 000 Da and 20, 000 Da.
The acceptance criteria were based on 50 laser shots per spot.
To discriminate between individual strains, an equivalent number of spectra per strain was required: 25 spectra were selected and mathematical testing of the datasets was done on the basis of Principal Component Analysis (PCA) and the results were demonstrated in a three-dimensional (3d) score plot, automatically created by the software. According to the MBT database, containing 132 main spectra of staphylococcus species and subspecies, the PCA dendrogram setting was utilized to group species. Moreover, a dendrogram was created from the minimal spanning tree (MSP) dataset. The MSP dendrogram was created according to the assessment of the principal spectra of analyzing the strains. Firstly, each main spectrum of the MBT library was compared with each of the other main spectra resulting in a matrix of cross-wise identification scores. This matrix was utilized to estimate the distance values for each pair of main spectra. According to these values, a dendrogram was produced by means of MBT software.
T A B L E 1 Oligonucleotide sequences and product length of S. aureus gene-specific primers

| Species identification of S. aureus and CNS isolates by MBT
In this study, the tested isolates were analyzed by MBT and the resulting spectra were compared with the spectra in the library  Concerning the calculation of the PCA algorithm, every peak could get loading values which originate from the calculation of PCs. In this study, each peak was specified with three loading values originated from three PCs (PC1, PC2, and PC3) calculation. Briefly, the impacts of PC1, PC2, and PC3 to the generation of the profile in a percentage conspiracy of the difference elucidated, were about 52%, 13%, and 10% for S. aureus and 40%, 29%, and 18%, correspondingly for CNS (Fig. 3).
An effective gel view over the viewing representation of data derived from 198 well-characterized strains of S. aureus and 44 CNS is revealed in Figure 4. All single spectra are demonstrated on a density scale and as can be seen from this figure, the intensity distribu-

| DISCUSSION
Most of the pathogens present in the environment are difficult to control as they acquire latest means of antibiotic resistance from other organisms through horizontal transfer (Barlow, 2009).  198 S. aureus (132 MSSA and 66 MRSA) isolates and 44 CNS isolates from apparently healthy cows and cows that showed signs of mastitis were preliminarily identified and differentiated by tube coagulase test and then tested by the Vitek ™ 2 compact system. According to the obtained results, 37/44 isolates of CNS (84.09%) and 184/198 isolates of S. aureus (92.92%) were correctly identified. Similar findings agreed with that recorded by previous study (Spanu et al., 2003), which assessed the capability of the VITEK 2 system for fast detection of 130 S. aureus and 275 CNS isolated from blood cultures and illustrated that 90.5% of CNS and 99.2% of S. aureus strains were correctly identified.
Moreover, another study recorded the performance of Vitek 2 System on 11 S. aureus isolates and stated that the percentage of identification was 100% on genus and species levels (Da Silva Paim et al., 2014).
In our study, the Vitek ™ 2 compact system established equivalent rates of correctness in the detection of mastitis caused by staphylococcus species. Therefore, microorganisms with low detection levels included a very restricted number of strains to provide a valuable indication for improvement by Vitek ™ 2 compact system. The fact that staphylococcal isolates which were not identified properly could be described by their sluggish metabolism, leading to indistinct results in the reaction wells. Similar findings agreed with that recorded previously (Da Silva Paim et al., 2014;Ligozzi et al., 2002;Pfaller & Herwaldt, 1988), who evaluated the other automated identification systems used for identification of various staphylococcus species.
Although, several gram-positive microorganisms were correctly identified by Vitek 2 Compact system, some restrictions of this technique lead to failure of detection of uncommon bacteria. Subsequently, supplementary tools may be compulsory to precise detection of certain strains at the species level (Carbonnelle et al., 2007).
MALDI-TOF-MS therefore, represents one of the major, fast and correct tool for identification and typing of various microorganisms isolated from clinical and subclinical samples (Carbonnelle et al., 2007;El Behiry et al., 2014;Singhal et al., 2015). In our study, detection and typing of S. aureus and CNS strains were carried out by proteomic analysis and the frequency of precise identification at the genus and species levels was 194/198 (97.97%) and 44/44 (100%), respectively, with the score level ranging from 2,000 to 3,000 Da. Parallel results were recorded previously (Loonen et al., 2012) Despite the capital cost and maintenance of MALDI-TOF-MS, it is considered as one of the potential drawbacks Neville et al., 2011;Pavlovic et al., 2013). This technique represents a powerful new tool that has the potential to replace conventional identification tools for a majority of routine isolates encountered in conventional identification methods, about €4.6-8.23/sample (Bizzini et al., 2010;Seng et al., 2009). Furthermore, another study indicated that the reagents compulsory for phenotypic identification using modern automated instruments cost about $10 per isolate, whereas, the required reagents for MALDI-TOF MS do not exceed $0.50 per sample (Cherkaoui et al., 2010).
Furthermore, the proteomic identification was fast and took about 30 min per isolate from target plate to obtain the final results, and nearly 2 hr to investigate a full 96-spot target plate. Therefore, when we compared the MALDI-TOF-MS with the colorimetric method in this study, it was noticed that the utilization of the MALDI-TOF-MS for a routine identification of various Staphylococcus strains is reliable which is characterized mainly by simple procedures, large amount of sampling, high accuracy, more sensitivity, and reproducibility.
Through this study, we verified that identification of S. aureus and CNS strains isolated from mastitic cows by MBT combined with compass software was faster, economic, and more precise than The Vitek 2 compact system and also was successfully discriminated MSSA from MRSA according to the peak intensities. Consequently, the opportunity of solving the problem of clinical and subclinical bovine mastitis was possible by precise and early diagnostic method.