Rapid Identification of the Foodborne Pathogen Trichinella spp. by Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry

Human trichinellosis occurs through consumption of raw or inadequately processed meat or meat products containing larvae of the parasitic nematodes of the genus Trichinella. Currently, nine species and three genotypes are recognized, of which T. spiralis, T. britovi and T. pseudospiralis have the highest public health relevance. To date, the differentiation of the larvae to the species and genotype level is based primarily on molecular methods, which can be relatively time consuming and labor intensive. Due to its rapidness and ease of use a matrix assisted laser desorption / ionization time of flight mass spectrometry (MALDI-TOF MS) reference spectra database using Trichinella strains of all known species and genotypes was created. A formicacid/acetonitrile protein extraction was carried out after pooling 10 larvae of each Trichinella species and genotype. Each sample was spotted 9 times using α-cyano 4-hydoxy cinnamic acid matrix and a MicroFlex LT mass spectrometer was used to acquire 3 spectra (m/z 2000 to 20000 Da) from each spot resulting in 27 spectra/species or genotype. Following the spectra quality assessment, Biotyper software was used to create a main spectra library (MSP) representing nine species and three genotypes of Trichinella. The evaluation of the spectra generated by MALDI-TOF MS revealed a classification which was comparable to the results obtained by molecular methods. Also, each Trichinella species utilized in this study was distinct and distinguishable with a high confidence level. Further, different conservation methods such as freezing and conservation in alcohol and the host species origin of the isolated larvae did not have a significant influence on the generated spectra. Therefore, the described MALDI-TOF MS can successfully be implemented for both genus and species level identification and represents a major step forward in the use of this technique in foodborne parasitology.

acquire 3 spectra (m/z 2000 to 20000 Da) from each spot resulting in 27 spectra/species or genotype. Following the spectra quality assessment, Biotyper software was used to create a main spectra library (MSP) representing nine species and three genotypes of Trichinella. The evaluation of the spectra generated by MALDI-TOF MS revealed a classification which was comparable to the results obtained by molecular methods. Also, each Trichinella species utilized in this study was distinct and distinguishable with a high confidence level. Further, different conservation methods such as freezing and conservation in alcohol and the host species origin of the isolated larvae did not have a significant influence on the generated spectra. Therefore, the described MALDI-TOF MS can successfully be implemented for both genus and species level identification and represents a major step forward in the use of this technique in foodborne parasitology.

Introduction
Nematodes of the genus Trichinella are one of the ten foodborne parasites with the greatest global impact [1]. Human trichinellosis has been documented in 55 countries with an average yearly incidence of approximately ten-thousand cases [2]. Infections occur through consumption of raw or inadequately processed meat or meat products containing the parasitic larvae [3]. Pork represents the most important source of human infection but meat of horses, wild boars, bears and badgers can also play a significant role [4]. Trichinellosis is mainly controlled through the examination of each animal carcass which may be infested with Trichinella and intended for human consumption [5]. Several countries have monitoring programs for assessing the prevalence of infection in wildlife and to identify the circulating etiological agents. According to the International Commission on Trichinellosis all positive samples should be forwarded to a qualified reference laboratory for the determination of the Trichinella species involved [6].
Currently, nine species, namely T. spiralis, T. nativa, T. britovi, T. pseudospiralis, T. murrelli, T. nelsoni, T. papuae, T. zimbabwensis, T. patagoniensis and three genotypes, Trichinella T6, Trichinella T8 and Trichinella T9, are recognized. [7,8]. All species and genotypes can infect mammals. T. pseudospiralis also develops in birds and T. papuae and T. zimbabwensis can infect some reptile species [2]. T. spiralis and T. britovi are the most widespread etiological agents of Trichinella infection in wild and domestic animals and cause the most human infections worldwide. To date, the differentiation of the larvae to the species and genotype level is based primarily on molecular methods. Eight species and one genotype can be distinguished by multiplex PCR analysis [9,10]. The remaining species and genotypes are identified by PCR-RFLP or PCR amplification and sequencing [7]. However, these methods are relatively time consuming.
In recent years, matrix assisted laser desorption / ionization time of flight mass spectrometry (MALDI-TOF MS) has evolved to a routine method for bacterial species identification due to its comparability to molecular methods, rapidness, minimum sample preparation and cost effectiveness per sample [11]. MALDI-TOF MS species identification is based on the protein profiling in which a unique spectrum is generated using the intact proteins extracted from an organism and compared with the spectra from the known organisms stored in the database through a pattern matching algorithm. Recently, we demonstrated that MALDI Biotyper, a commercial software, can also be used for the creation of reference spectra for species other than bacteria [12,13]. As increasing number of routine diagnostics laboratories are implementing MALDI as a method of rapid microbial species identification, development of a Trichinella spectra database would be of great benefit without any extra cost or labor training. Therefore, the aim of this study was to establish a protocol for extraction of proteins and to create Trichinella specific reference spectra database using MALDI Biotyper software for rapid species identification similar to that of the microorganisms.

Sample collection
As listed in Table 1, nine Trichinella species and three Trichinella genotypes were utilized to establish a MALDI-TOF MS spectra reference library. All Trichinella isolates were passaged in mice and provided by Dr. Edoardo Pozio from the International Trichinella Reference Center. Larvae from domestic pigs, wild boar and guinea pigs originated from our own laboratory.
Trichinella larvae were isolated from muscles using the magnetic stirrer method [14] and stored in ethanol (95%) at -18°C.
Hyostrongylus rubidus and Trichuris sp., obtained from domestic pigs, were utilized to assess the spectral specificity and were provided by Dr. Peter-Henning Clausen, FU-Berlin. Additional evaluation was also carried out using Metastrongylus sp., isolated from wild boar muscles and subsequently identified by 18S sequencing in our laboratory [15]. Two female guinea pigs (inbred strain 2A) were kept together in a 4800 cm 2 large cage with 600g Licnocel Flakes and autoclaved hay (110g), with day light and a 12h light regime. There was a second floor in the cage with houses. The atmosphere was 20-22°C at 55% humidity. Water and food ('Alleinfuttermittel für Meerschweinchen' by SNIFF) was given ad libitum. Animals were infected with 500 Trichinella larvae through non-invasive application by pipette. In total 4 guinea pigs were infected.

Animal work and ethics statement
Two female domestic pigs (Deutsche Landrasse) were kept in a 9m 2 box from 12 weeks of age. They were kept on approximately one bale of hay, had access to toys, a 12h light regime and atmospheric conditions as described for the guinea pigs. Water was freely available and the pigs were fed 1.2-2.5 kg/d pig feed from Ruckdeschel. One pig was infected with 40000 Trichinella spp. at 18 weeks through handfeeding of spiked food.
Both domestic pigs and small animals do not show any clinical signs of a Trichinella infection and did not need any treatment. Domestic pigs were electrically stunned inducing instantaneous insensibility, followed by bleeding. Guinea pigs were anesthetized using 0.4 mg

Molecular species confirmation
The Trichinella species was confirmed by molecular techniques, which were defined as the gold standard. DNA extraction and PCR was performed as described by Pozio et al. [10] and Mayer-Scholl et al [16].

Sample preparation for MALDI-TOF mass spectrometry
For the generation of the master spectra library (MSP) containing most reproducible peaks of the species /genotypes, a total of ten Trichinella larvae of same species or genotype were pooled and the protein extraction was carried out with 10 mg zirconia/silica beads (0.5 μm) and 100 μl 70% formic acid in a 2 ml Eppendorf tube. The samples were sonicated on ice for 3 min (amplitude 100%, pulse 60: UP200St, Hielscher Ultrasound Technology, Teltow, Germany). 100 μl 100% of acetonitrile were added and sonicated again. The suspension was centrifuged at 14.000 g for 3 min at room temperature and 150 μl of clear supernatant was collected for protein enrichment by vacuum centrifugation (12 min, 45°C, Eppendorf Concentrator 5301). One μl of each sample was spotted onto the target plate nine times (MSP 96 target polished steel (MicroScout Target) plate; Bruker Daltonik, Bremen, Germany), air dried and then overlaid with 1 μl of saturated α-cyano-4-hydroxy-cinnamic acid matrix solution (200 mg in 2.5% TFA/50% ACN) and dried completely.

Influence of conservation methods, Trichinella hosts and number of larvae
The influence of different conservation methods such as freezing (T. spiralis infested meat at -20°C for 7 days) and 70% alcohol on the quality of the obtained spectra was measured. The influence of selected host species on the MALDI-TOF MS spectra mouse, guinea pig, domestic pig was also appraised. To determine if the method has a comparable sensitivity to the PCR, proteins were extracted from a single T. spiralis larva (fresh and frozen) and analyzed. All examinations were performed in triplicate. Further, to test the reproducibility of the method, Trichinella isolates IS 328, 559 and 003 were analyzed on different days by different technicians.

Generation of Trichinella master spectrum library
The MALDI-TOF MS measurements were carried out using MALDI-TOF Microflex LT (Bruker Daltonics, Bremen, Germany) on a range of 2,000-20,000 m/z (mass to charge ratio) following the calibration with Bacterial Test Standard (Bruker Daltonics, Bremen, Germany). A total of 27 spectra per specimen were acquired from nine spots where each spot was measured three times. For each spectrum, 200 laser shots in 40 shot steps from different positions of the target spot (random walk movement) were automatically acquired using AutoXecute acquisition mode in Flex control 3.4 software. The quality of each spectrum was assessed with FlexAnalysis 3.4 software, the flat-liners and spectra with peak deviations exceeding 500 ppm were replaced with fresh measurements. The composite correlation index (CCI) matrix was calculated with MALDI Biotyper 3.14 software with the following settings: lower bound 3,000, the upper bound 15,000, the resolution of the mass range four, and the number of intervals for CCI four. Following which, a master spectra library (MSP) was created using the default MSP creation method of the Biotyper software. The created master spectra were then compiled as a Trichinella specific MALDI-TOF MS spectra database. Master spectrum dendrogram cluster analysis was carried out with the correlation distance measure and single linkage. The threshold distance value for a single organism and zero for related organisms was computed based on the score value for species differentiation.

Evaluation of generated spectra by score values
Score values function as a measure of the reliability of the isolate identification; these are acquired through the matching of the unknown spectrum against the reference library and vice versa and through the correlation of the intensities of the matched peaks. The following cutoff values recommended by the manufacturer were used for sample identification: 0 to 1.699 indicates 'no reliable identification'; 1.7 to 1.999 indicates 'probable genus identification'; 2.0 to 2.299 indicates 'secure genus identification and probable species identification', and 2.300 to 3.000 indicates a 'highly probable species identification'.

Results
The four Trichinella species (T. spiralis, T. britovi, T. pseudospiralis and T. nativa), which play the most important role in world-wide human infection clearly displayed different peaks, at least in terms of peak patterns and their intensities which indicates that the spectra are suitable for the creation of distinct reference spectra for each species (Fig 1). The nematodes included in the study to verify the specificity of the method displayed a different peak pattern in comparison to all Trichinella species (Fig 1, S1 and S2 Figs).
As shown in a heat map of all 33 tested Trichinella strains (Fig 2), all tested isolates of T. spiralis, T. pseudospiralis, T. nelsoni, T. papuae and T. zimbabwensis could be distinguished with high confidence. Some isolates within the T. spiralis species showed log score values between 1.7 and 1.99 (e.g. T. spiralis ISS 1604), however, interspecies comparison showed log score values lower than 1.70. T. patagoniensis and T9 could also be identified at the species level with high confidence, but congruencies on the genus level with other Trichinella species were identified (log score < 1.82).
Hyostrongylus rubidus, Metastrongylus sp. and Trichuris sp. obtained log score values were lower than 1.10 (Fig 2). Comparison of the Trichinella spectra to the entire Bruker reference library did not reveal a single clear match. The highest log score value of 1.471 was found with Lactobacillus crispatus.
The relatedness of the spectra sets was analysed by composite correlation index analysis (Fig  3). The CCI ranged from 0.56 and 0.89 within the T. spiralis population, indicating high similarity. The highest inter-species CCI value for T1 was 0.51 with T12. The T. pseudospiralis isolates showed a lower degree of intra-species relatedness (CCI: 0.56-0.68), but a higher degree of incongruence with other Trichinella species (0.12-0.35). Conversely, T. britovi showed a high variability within the species (CCI: 0.39-0.81) but the highest degree of similarity with other Trichinella species, especially T. murelli (CCI: 0.39-0.65) and T8 (CCI: 0.31-0.61).
As shown in Fig 4, the parasitic species Trichuris sp., Metastrongylus sp. and Hyostrongylus rubidus could be clearly separated from the Trichinella spp. cluster. All nine Trichinella species and three genotypes were separated into their distinct clusters, albeit with different levels of  similarity. Only isolate ISS 346 (T. murrelli) did not form a cluster with the remaining four T. murrelli isolates.
Important criteria that could have an effect on the quality of the protein spectra and are important for future potential use in a routine setting include the reproducibility of the method, the larval host species, conservation methods, the number of larvae and the complexity of the protein extraction protocol.
The isolates tested on different days and by different technicians resulted in species level pattern matching log score values as recommended by the manufacturer: ISS 328 (2.5-2.6), ISS 559 (2.2-2.4) and ISS 003 (2.3-2.4), indicating good reproducibility and possible usefulness of this method for rapid species identification. Comparable peak patterns were found in ISS 003  When generating MSPs using 10 larvae conservation methods (70% alcohol, freezing) did not have an effect on the log score values (2.3-2.6) or the peak patterns, in comparison to the log score values of native larvae (2.4-2.6)(). Only differences in peak intensities were identified (e.g. an intensity of 30000 vs 5000 in the major peaks). Also, fresh single larvae tested in triplicate did not show varying log score values (2.1-2.2) and could be identified with high confidence. In contrast, when larvae were frozen, quality protein spectra resulting in comparable log score values could only be generated using 5 larvae.
Only protein extraction protocols including sonication with beads resulted in high quality spectra (S4 Fig). As seen in S5 Fig, sonication with solvent and beads is necessary to completely disrupt the larvae to obtain sufficient proteins to correctly identify all Trichinella species.

Discussion
In this study, we created a MALDI-TOF MS reference spectra database using Trichinella reference strains of all known species and genotypes. To date, two distinct Trichinella clades have been characterized by the presence or absence of an intramuscular collagen capsule [17]. Trichinella species that encapsulate in the host muscle tissue are T. spiralis, T. nativa, T. britovi, T. murelli, T. nelsoni, T. patagoniensis and the genotypes T6, T8 and T9. The second clade represents the Trichinella species which do not induce the formation of a capsule. Here, the species infecting mammals and birds (T. pseudospiralis) is clearly partitioned from those infecting mammals and reptiles (T. papuae, T. zimbabwensis) [7,17]. As shown in the dendrogram, the evaluation of the spectra generated by MALDI-TOF MS mirrored the classification obtained by molecular methods.
As the current focus of our work was not the development of a taxonomic classification but to determine if the described method can be used in a routine setting, the discussion of the usefulness of the method will focus on the uniqueness of the created reference spectra.
The MALDI Biotyper and its integrated database were chosen to create a Trichinella-specific spectra database.
The four most prevalent species, T. spiralis, T. britovi, T. pseudospiralis and T. nativa resulted as highly probable species level matching with the correct species. Except in a few isolated cases, genus level matching was not observed among the strains belonging to the different species.
In contrast, two of the five T. britovi strains included in this study appeared to overlap with at least one strain of T. murelli and T8 at species level (>2.0). As T. murelli is the vicariant species of T. britovi in North America [7] difficulties in a reliable distinction between the two species is expected. Similar observations were also made among T. nativa and T britovi at genus level matching. However, this is not relevant as these stains matched with other strains of the same species with much higher log score values. This underscores that the database needs further improvement by addition of isolates representing several geographical regions and different hosts. Another possibility could be to define reduced log score cutoff values for species or genus level identification [18].
In routine Trichinella testing, fresh larvae are isolated from meat, identified morphologically and subjected to multiplex PCR analysis using one larva. Correct species identification was also achieved by MALDI-TOF MS when a single fresh larva was used. The larvae frozen at -20°C displayed reduced peak intensities, and therefore, 3-5 larvae are needed to obtain reliable spectra, which is comparable to the number of frozen larvae needed for the multiplex PCR.
Mixed Trichinella infections involving two separate species have been detected [19][20][21]. Through the visualization of additional bands, these infections can be detected by multiplex PCR, if more than one larva is examined. We found that samples containing two different Trichinella species are not identified reliably. In order to identify mixed infections by MALDI-TOF MS, a number of single Trichinella larvae need to be examined separately.
The mass spectrometry described here allows a fast verification of Trichinella larval isolates on genus level, which might be crucial for confirmatory testing after questionable findings during meat inspection. Furthermore, reliable determination of the four most prevalent Trichinella species worldwide is possible by means of MALDI-TOF MS analysis. As the method is easy to handle and can be performed within 1 hour, it represents a major step forward in the use of this technique in foodborne parasitology. Future studies will focus on the expansion of the Trichinella database to increase the accuracy and precision of identification and the validation of MALDI-TOF MS for typing of Trichinella field isolates. However, one should be cautious while interpreting the identification results up to the species and genotype level as this is the first attempt in a Trichinella database creation with a relatively very small number of isolates per species. It is proposed to reinforce the database with addition of strains representing various hosts and geographical regions. Furthermore, the database and protocol will be made available to large laboratories to test the performance of this technique under routine conditions.