Real time human micro-organisms biotyping based on Water-Assisted Laser Desorption/Ionization

Gram negative (E. coli D31, Pseudomonas aeruginosa); Gram positive (Staphylococcus aureus, Enterococcus faecalis) and Yeast (Candida Albicans) were cultivated as described previously (21). Briefly, bacteria were cultivated in poor broth medium. Microbial growth was assessed by the increase in A 620 nm after 8-hour incubation at 37°C.

The basic design of the instrument setup is already described in a previous study (17, 22). In these experiments, the prototype was equipped with a fibered tunable InfraRed Optical parametric oscillator (OPO) system between 2.8 μm to 3.1 μm (Radiant version 1.0.1, OPOTEK Inc., Carlsbad, USA) pumped by a 1.064 μm radiation delivered by a Q-switched 10 ns pulse width Nd:YAG laser (Quantel Laser, Les Ulis, France). A 1 meter length biocompatible laser fiber with 450 μm inner diameter (HP fiber, Infrared Fiber Systems, Silver Spring, USA) was connected to the exit of the OPO system and focused by a 20 mm focal length CaF2 lens attached at its end. A Tygon® ND 100-65 tubing (2.4 mm inner diameter, 4 mm inner diameter, Akron, USA) was used to aspirate the ablated material and was directly connected to the inlet of the mass spectrometer (Synapt G2s, Waters, Manchester) through a modified atmospheric pressure interface described elsewhere (18) (Fig. 1). The samples was irradiated over 5 sec at 1.6 J/cm². The fluence was decreased to 0.8 J/cm2 when the wavelength was tuned to allow for keeping it constant over the screened range. The spectra acquisition was performed in positive resolution mode with a scan time of 0.5 sec. For the experiments related to the evaluation of the transfer tube temperature, a flexible heating cable system (isopad T7000; Thermocoax, Suresnes, France) was uniformly winded around the aspiration tube.

Figure 1

Special care was taken when working with infrared laser beams. The laser used for this study is classified as safety class IV which requires the wear of specific laser safety goggles throughout all experiments. The samples were irradiated over 5 sec at 1.6 J/cm². Only when the wavelength was tuned the fluence was decreased to 0.8 J/cm² to allow for keeping it constant over the screened range. Three sample spots were deposited onto a glass slide to obtain significant results. The spectra acquisition were performed in positive resolution mode with a scan time of 0.5 sec.

For real time analysis, the built model was exported to a second software module called OMB Recognition. The parameters for this analysis were: a TIC threshold of 1E+4 count for irradiation detection, a normalized intensity threshold of 1E+6, and an outlier limit of 5 standard deviations. The software analyzes 1 scan per spectrum and has a 3s timeout waiting for good spectrum. The cross-validated models are exported from the OMB software and then loaded in the OMB Recognition software module before any real time piece of tissue analysis by the SpiderMass (18, 20). The real time acquired data are then directly interrogated, giving an immediate feedback using a color scale predefined from the specified classes.

m/z intervals corresponding to loadings with the largest contribution to the explained variance observed in the different groups were selected for MS/MS-based identification. For these experiments, the settings were exactly the same as described in the Instrumentation section and published (18, 20). The identifications were performed directly on the tissue by doing a full scan first to verify the presence of the targeted masses. Then after switching to MS/MS mode, they were subjected to collision-induced dissociation (CID) in the transfer cell with respectively 20 and 35 V. The resulting spectra were annotated manually, and assignments were verified by interrogating the high accuracy mass measurements of the precursor ions using the LipidsMap database.

To assess the performance of SpiderMass for biotyping, we studied two different of sampling and analytical conditions instrument (Fig. 1). Time course analyses from T0 to T24 hours of E. Coli growths in liquid suspension were performed either directly from the agarose of the petri dish or from a smear spread on a glass slide. First, 10 μl of bacteria in suspension were harvested and dropped either on the agarose or smeared onto a glass slide and analyzed just after with SpiderMass in both negative and positive ion mode (Fig. 2). In negative of positive mode at T0, no specific m/z can be detected whatever the ion mode. Then bacteria were left growing for longer in time until 24H (T24) (Fig. 2). In time course, specific m/z are detected at T4h bacteria growth in negative mode i.e. 688.490, 719.491, 747.519 and 749.522. M/z 702.514 appears at T8 bacteria growth (Supplementary Data 1) and m/z at 733.515 increased at T24h bacteria growth (Fig. 2) In positive mode, first specific m/z (734.388, 748.390, and 764.370) appears at T6 bacteria growth. At T8 bacteria growth, these m/z increased and T24, ion at 748.790 is major (Supplementary Data 1). We then

Figure 2

compared tests performed on culture medium directly, agarose (petri dish) and on smears on glass slides for bacteria collected at T24 (Fig. 3). Results established that the smears performed on glass slides of the bacteria suspension gave the best results in term of intensity i.e. 1.32E5 vs 8.33E4 in agarose and 4.44E3 directly in medium, in negative mode (Fig. 3A). Some tiny differences are registered between glass slides and agarose, in fact ions m/e of 702.518 and 733.515 have not the same level of intensity in both conditions (Fig. 3A). PCA analyses of these two conditions in time course of bacteria growth (Figs. 3Ba, 3Bb) revealed that the discriminate ions between the two conditions are the same (Figs. 3Ba, 3Bb). Comparison between direct medium, agarose and smear by PCA confirmed that the 3 methods are different in negative and positive modes (Fig. 3Bc). Specific common ions are also detected in positive mode ranged from 549.432 to 564.453 (Fig. 3C). Based on these data, smears on glass slide was then selected instead direct analyses on agarose. Reproducibility experiments were conducted on 3 different slides containing smears of bacteria collected at T24. Experiments were realized on the same slides analyzed 3 times during 3 consecutive days in negative (Fig. 4A) and positive (Fig. 4B) modes. The same m/z were found in negative and positive modes whatever the slides and the time of the analysis reflecting the robustness of the technology.

Figure 3

The next step consisted to perform biotyping on different species of Gram negative (E. coli D31, Pseudomonas aeruginosa); Gram positive (Staphylococcus aureus, Enterococcus faecalis) and Yeast (Candida Albicans) (Fig. 5). None of them have the same ions detected. C. alibicans have characteristic ions centered on the most intense at 835.41. E. Faecalis presents 4 peak ranges (centered on m/z of 281.18, 747.40, 1403.89, 1545.89, and 1373.9). For P. aeruginosa 5 peaks ranges are found (centered on m/z of 270.12, 716.41, 47.4, 1403.89, and 1603.92) and for S. aureus, a wide peak range centered on 735.39 is mostly discriminative. These results established the clear possibility to biotype human pathogens using the SpiderMass based on their metabolites and lipids profiles. In fact some of the m/z have been characterized by MS/MS using SpiderMass in MS/MS real time mode. In E. coli, ions at m/z of 688.49 corresponds to PE (16:0/16:1). The ones at m/z 688.35 to PE (16:0/cy17:0), m/z

of m/z 719.35 to PG (16:0/16:1) and m/z 733.36 to PG (16:0/cys17:0), 773.38 to PG (18:1/18:1) (Fig. 6).

Figure 4

Figure 5

Figure 6