Whole cell matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) for identification of Leptospira spp. in Thailand and Lao PDR

Leptospirosis is a zoonosis with a worldwide distribution, caused by pathogenic spirochetes of the genus Leptospira. The classification and identification of leptospires can be conducted by both genotyping and serotyping which are time-consuming and established in few reference laboratories. This study used matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) as rapid and accurate tool for the identification of leptospires. The whole cell protein spectra of 116 Leptospira isolates including 15 references Leptospira spp. (pathogenic, n = 8; intermediate, n = 2; non-pathogenic, n = 5) and 101 Leptospira spp. clinical isolates was created as an in-house MALDI-TOF MS database. Ninety-seven clinical isolates from Thailand and Laos was validated with these protein spectra and revealed 98.9% correct identification when compared with 16S rRNA gene sequences method. Moreover, MALDI-TOF MS could identify spiked leptospires whole cell in urine. Biomarkers for differentiation of leptospires phylogeny and specific protein spectra for most found Leptospira spp. in this area (L. interrogans, L. kirschneri, L. borgpetersenii) based on MALDI-MS algorithm were demonstrated.


Introduction
Leptospirosis is a major public health problem in the developing countries. More than 1 million cases occur worldwide per year, including about 58,900 deaths [1]. In Thailand, prior to 1996, reported cases numbered approximately 200 cases per year, mainly in the central and southern regions. A dramatic increase occurred from 358 cases in 1996 to 14,285 cases in 2000, followed by a continual decline to 2,868 cases in 2005. Reported cases has stabilized at around 2,800-5,500 cases per year during the last 10 years [2]. Although the trend over time is consistent with a sustained outbreak, one of the problems associated with the interpretation of reported cases is that reporting can be done on the basis of a clinical diagnosis alone. Accuracy of clinical diagnosis varied from 0% to 50% between the provinces and was highest during the rainy season [3]. The current laboratory tests available for the diagnosis of leptospirosis are serological testing and molecular-based diagnosis method [4][5][6][7]. Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) is a novel method that is increasingly used for identification and classification of microorganisms from different genera, species and strains including Leptospira spp. [8][9][10]. The method detects the mass-to-charge ratio of biomolecules from whole cell bacteria and provides spectra within minutes. The protein mass spectra obtained can be added in the MALDI-TOF MS database and used as reference for identification of unknown microorganisms. However, none of known Leptospira mass spectra database are commercially available. Here, we employed MALDI-TOF MS to reveal the whole cell protein mass spectra of Leptospira spp. including pathogenic, intermediate and saprophytic groups. A large Leptospira spp. mass spectra database was established for identification of Leptospira spp. from clinical isolates in Thailand and Laos.

Bacterial strains
Fifteen references Leptospira spp. including pathogenic, intermediate and non-pathogenic groups ( Table 1); 101 of Leptospira spp. isolates and one isolate of Turneriella parva, a spirochete closely related to Leptospira, were included in this study (Table 2). Leptospira were cultured in Ellinghausen-McCullough-Johnson-Harris liquid medium containing Leptospira Medium Base EMJH (Difco, Becton Dickinson, USA), Leptospira Enrichment EMJH (Difco) with 3% normal rabbit serum (Invitrogen, USA) at 30˚C. All Leptospira from human clinical samples were isolated and cultured as described [11]. Serovar identification was performed using cross-agglutination absorption test (CAAT) at WHO/FAO/OIE Collaborating Center for Reference and Research on Leptospirosis, Brisbane, Queensland, Australia. All clinical samples were identified at specie level based on sequencing of 16S rRNA gene as described [12,13].

Ethics statement
The studies were approved by the National Ethics Committee for Health Research, Ministry of Public Health of Laos (No 25/NECHR), the Oxford Tropical Ethics Committee, UK and the Ethical Committee of Faculty of Tropical Medicine, Mahidol University, Thailand (Approval no. MUTM-EXMPT 2015-004). Urine sample from healthy donor was provided by the project investigator (Approval no. MUTM2018-054-01).

Sample preparation for MALDI-TOF Mass spectrometry
Leptospira was cultured for a week to obtain 10 8 colony forming unit (CFU) per ml as previously described [14]. Three milliliters of culture were centrifuged at 13,000 g for 2 min at room temperature and cell pellet was washed with 70% ethanol and centrifuged at 13,000 g for 2 min. The pellet was mixed with 50 μl sinapinic acid matrix solution (Bruker Daltonics, Germany) containing 10 mg of sinapinic acid in 1 ml of 2.5% trifluoroacetic acid (Sigma-Aldrich, USA) and 50% acetonitrile (Sigma-Aldrich, USA). After being well mixed by pipetting and centrifuged at 13,000 g for 1 min, then 2 μl of cell-matrix suspension were added to the ground steel MALDI target plate and dried at room temperature. Each isolates were spotted 24 dots onto the MALDI target plate to test technical replication. Mass spectra were collected using a Ultraflex II MAL-DI-TOF/TOF mass spectrometer (Bruker Daltonic, USA) operated with FlexControl software in linear positive mode, i.e. using a mass range of 2,000 to 20,000 Daltons. The instrument was externally calibrated with E. coli strain DH5α ribosomal proteins as recommended protocol.  The 97 Leptospira clinical isolates were blindly identified using Biotyper based-on MSP pattern matching. MSP of each isolates were compared to 101 reference mass spectra in the database and calculated a score value between 0 and 3 reflecting the similarity between isolate and reference spectrum. The results were recorded as identification score and displayed 10-best matching in reference pattern in ranking table. Identification scores � 2.0 were regarded as accepted for a reliable identification at species level and score � 1.7 and � 2.0 for identification at genus level. Scores < 1.7 were considered unreliable as recommended by manufactural. The reproducibility of MSP was performed. Three reference isolates including L. interrogans, L. borgpetersenii, and L. biflexa were cultured 3 times, MALDI-TOF MS analysed and interpreted using identification score. The MSP reproducibility was also evaluated on different media and different sub-culture times using low passage L. wolffii from 3 human isolates (strain H8, H9, H16) which had been kept in liquid nitrogen for 14 years. Those isolates were grown in 2 different media (EMJH with 3% normal rabbit serum, EMJH without normal rabbit serum) and 10 sub-culture times. L. wolffii from each media and culture time were subjected to MALDI-TOF MS after protein extraction. The identification scores were analyzed using ANOVA and student t-tests.
In order to test the detection limit of MALDI-TOF MS method, L. interrogans MD08 were cultured to reach 10 8 CFU/ml. After protein extraction, the mixture was diluted to 10-folded serial dilutions to 10 CFU/ml and applied to MALDI-TOF MS.

MALDI-TOF MS analysis
The statistical analysis was performed using ClinProTools software version 2.2 (Bruker Daltonics, Germany). Twenty spectra of individual isolates were analyzed and visualized for spectral comparison and automatically displayed in a three-dimensional plot based on the Principal Component Analysis (PCA) to discriminate among the analyzed strains.
To discover discriminating peaks or biomarkers specifically for analysis of leptospires phylogeny (n = 51), Leptospira species (n = 40), and leptospires serotype (n = 43), four technical replicate spectra of each isolate for training set, as listed in Table 3, were used to generate mathematical models with 3 different algorithms (Genetic Algorithm (GA), Supervised Neural Network (SNN), Quick Classifier (QC)) for classification. The generated classification models were compared based on the calculated Recognition Capacity (RC) and Cross Validation (CV) which indicated the accuracy and reliability of the model. A model with the highest RC and CV was chosen for further peak statistical evaluation. Intensity of peaks presented in the model were evaluated by statistical tests. Peak intensity with p-values < 0.05 indicated statistically significant difference. Interesting informative peaks were assessed the capability to discrimination based on Area under the Receiver Operating Characteristic (ROC) curve (AUC) values.

Detection of spiked Leptospira spp. in urine
In order to mimic the detection of leptospires in urine, 4 Leptospira spp. including 2 from pathogenic group (L. interrogans and L. borgpetersenii), one from the intermediate group (L. wolffii) and one from the non-pathogenic group (L. biflexa) were cultured to 10 8 CFU/ml. Three milliliters of culture were centrifuged at 3,000 g for 10 min, pellet was added with 3 ml urine from healthy donor. After centrifugation at 8,000 g for 10 min., pellet underwent protein extraction. The protein spectra from spiked Leptospira in urine were analyzed and compared to the protein spectra acquired from Leptospira culture. Urine without spiked Leptospira was used as negative control.

Identification of protein mass fingerprinting of Leptospires groups
Fifteen references Leptospira spp. (included 8 pathogenic, 2 intermediate and 5 non-pathogenic spp.) generated by MALDI-TOF MS revealed protein profile at mass to charge ratio (m/ z) ranged from 2-20 kDa. These protein profiles showed specificity representing individual species (Fig 1). The discrimination of Leptospira spectra was demonstrated by differentiating 3 different groups in the MSP dendrogram which related to their phylogeny (pathogenic, intermediate and non-pathogenic groups) (Fig 2). This result indicated that phylogeny of Leptospira could be clustered based on their protein profile using MALDI-TOF MS.
The reproducibility of MSP from MALDI-TOF MS demonstrated by identification score of 3 Leptospira spp. (L. interrogans, L. borgpetersenii and L. biflexa) were higher than 2.0 after triplicate tested. The identification score of low passage L. wolffii strain H8, H9 and H16 revealed no statistically significant differences in two different growth media (p-value >0.05). According to sub-culture time, identification score of less than 6 sub-culture times (1-5 times) and that of more than 6 sub-culture times (6-10 times) have no statistically differences (p-value > 0.05). These results indicated the high reproducibility of Leptospira protein spectra generated by MALDI-TOF MS. The detection limit of MALDI-TOF MS method was demonstrated in 10-fold serial dilution of L. interrogans (MD08) protein spectra at 10 6 CFU/ml (Fig 3).
The new protein spectra database of Leptospira spp. was created based on 116 Leptospira isolates of 15 references and 101 Leptospira isolates (including 100 pathogenic and 6 non-pathogenic group) as listed in Table 2. To validate this database, 97 Leptospira human isolates from Thailand and Laos were blindly analyzed based on profile matching to this Leptospira protein mass spectra database. The identification score of these isolates are shown in Table 4. The clinical isolates from Laos and Thailand were identified as L. interrogans (n = 84), L. borgpetersenii (n = 4), L. weilii (n = 7) and L. kirschneri (n = 1) with scores of 2.0-2.6. One isolates from Laos (No. UI14721) has score equal to 1.903 for L. interrogans best match and the second match (with score 1.702) was identified as the same species. The identification score evaluation of MALDI-TOF MS database demonstrated that 96 out of 97 (98.9%) correct identification at specie level.

Biomarker identification for Leptospira spp and serovars
Protein mass spectra in mass range of 3,000 to 15,000 Da were analyzed by mathematical algorithms to define the protein peaks that could represent as biomarkers for leptospires. Discriminating peaks for pathogenic and intermediate leptospires were discovered by GA models. Eight peaks specifically presented in for pathogenic leptospires (  Ten specific peaks to discriminate among L. interrogans, L. borgpetersenii, and L. kirschneri were reported by the GA model with 99.2% and 96.1% of RC and CV, respectively. Each species of Leptospira showed a unique combination of peaks (Table 7). Four peaks at m/z of 3605.24, 5529.89, 8747.26, and 11260.81 were common peaks, which were observed in three Leptospira species (L. interrogans, L. borgpetersenii, and L. kirschneri), but with differences in their relative intensities. Species-specific peaks were also detected. The presence of peaks at 6422.61 Da for L. borgpetersenii and peaks at 8084.81 Da for L. kirschneri was a unique characteristic of those two Leptospira species. Three peaks (6422.61 Da, 6924.67 Da, and 8084.81 Da)  were not detected in L. interrogans. In addition, among the absent 3 peaks, lack of 6924.67 Da was specific to L. interrogans. Using the GA model, L. interrogans serovar-specific peaks were identified with 98.1% and 82.0% of RC and CV, respectively. The peak list was reported in Table 8. Regarding to the GA classification model which demonstrated as the best capacity for serovar discrimination, 19 peaks were picked to discriminate the 8 serovars of L. interrogans including Autumnalis, Bataviae, Conicola, Grippotyphosa, Medanensis, Pomona, Pyrogenes and Sejroe. Each serovar was characterized by the unique combination of peaks in both presence or absence and relative intensity level (Table 8) Table 8. Among the serovars with absence of 2 peaks, the serovar-specific pattern could be characterized by specific lacking of the peaks to all serovars but Autumnalis and Pomona. Particularly, the serovar Bataviae, Canicola, and Pyrogenes were characterized by the absence of the peak at 8619.54 Da, 10293.62 Da, and 5638.80 Da, respectively

Spiked Leptospira spp. in urine
The protein spectra of 4 spiked Leptospira spp. (L. interrogans, L. borgpetersenii, L. wolffii, L. biflexa) demonstrated profile matching in both spiked urine and media (Fig 4). The small protein peak at m/z among 2,000-4,000 for example at m/z 3,442 were detected in protein profile of urine alone and in leptospires spiked urine. These peaks might be other ionized proteins in donor urine because they could not be detected in media. The identification score was high in both spiked leptospires in urine (range of 1.96-2.15) and in media (range of 2.0-2.5) while that of urine alone was unreliable (below 0.5).

Discussion
Whole cell MALDI-TOF MS of Leptospira was used to generate unique protein spectra or protein fingerprinting of 15 Leptospira spp. and clustered them according to their phylogeny. This study created the largest number of Leptospira protein database (116 protein spectra) that has been reported. The large protein database has strengthened the accuracy of Leptospira spp. MALDI-TOF MS identification with 98.9% correct identification of 96 Leptospira clinical isolates from different times and geographical areas of Thailand and Laos. One isolate with identification score below 2 possibly occurred because the protein spectra had low intensity. It was also possible that detection of proteins with relatively high molecular weight are more difficult to detect than protein in the low molecular weight range since this study used sinapinic acid matrix which ionized intact protein at high molecular weight [15]. MALDI-TOF MS method is simple, rapid and reliable for species identification but the method required bacterial culture which is of low sensitivity, laborious and time consuming. New culture agar called Leptospira Vanaporn Wuthiekanun (LVW) for Leptospira could reduce the isolation time but they still required a week for culture [14,16]. Urine is a non-invasive, in comparison with serum or blood, which might be of interest for diagnosis of leptospirosis. Our results using spiked 10 8 cfu/ml leptospires in urine are suggesting that MALDI-TOF MS method may be able to identify Leptospira in urine. However, there are limitations. We lacked urine from leptospirosis patients and further investigation on detection limit of Leptospira in urine is required. The lowest amount of Leptospira which could generate the protein spectra by MALDI-TOF MS was 10 6 cfu/ml of culture media which is comparable to two reports [8,9]. However, the number of Leptospira in mammal urine is varied. A meta-analysis on the quantity of Leptospira in urine of infected animals and human in Ecuador has reported that rats excreted amount of Leptospira in urine with median of 5.7 x 10 6 cells/ml while large mammals excreted and shed 5.1 × 10 8 to 1.3 × 10 9 cells Leptospira per day [17]. Moreover, protein spectra of human and animal urine need identification in order to clearly define diseaseassociated biomarker. The standardize method for detection of Leptospira in human or animal urine needed to be optimized and validated.
The major benefit from large Leptospira protein spectra database was that the database can be used as training set for machine learning and making a decision based on the computational mathematic analysis which demonstrated good classification in phylogeny. The specific peak at m/z of 6786.51 and 7006.70 were consistently found in intermediate and pathogenic Leptorspira respectively. These 2 specific peaks could be used as biomarker to differentiate between those 2 groups. Moreover, three Leptospira spp. (L. interrogans, L. kirschneri, and L. borgpetersenii) which are most found species from rodent in Thailand, Laos and Cambodia [18] could present the differentiating peaks among each other. There were few studies [8][9][10] reporting the difficulty to differentiate between L. interrogans and L. kirschneri using MALDI--TOF MS. Our data revealed that two peak at m/z of 6924.67 and 8084.81 were presented in L. kirschneri but absent in L. interrogans.
Serotyping is useful in one health epidemiological studies of Leptospirosis. The attempt of serotype differentiation by MALDI-TOF MS was examined and indicated up to 19 peaks among 8 serotypes (serovars) of L. interrogans. However, the low number of individual serovars used as training set here may reduce the result accuracy. In addition, serovar diversity of Leptospira is based upon variation of lipopolysaccharide (LPS) at the bacterial outer membrane [19]. LPS structure consists of a lipid and a polysaccharide of O-antigen, outer core and inner core. Those components may not be ionized using MALDI-TOF MS settings adapted for bacterial protein detection. Besides, Leptospira isolates in Thailand has been reported as unpredicted serovars by microscopic agglutination test (MAT) [20].
In conclusion, our whole cell MALDI-TOF MS method can cluster Leptospira spp. according to their phylogeny and identify the large scale cultured leptospiral strains at the species level based on their protein profile with highly reproducibility. The biomarkers based on protein profile could be used for identification the common pathogenic species L. interrogans, L. kirschneri, and L. borgpetersenii of human isolates in endemic areas like Thailand and Laos PDR. Therefore, whole cell MALDI-TOF MS method is a simple, faster and reliable tool for confirmation of leptospiral species.