Analysis of proteomes released from in vitro cultured eight Clostridium difficile PCR ribotypes revealed specific expression in PCR ribotypes 027 and 176 confirming their genetic relatedness and clinical importance at the proteomic level

Eight well characterized clinical isolates of C. difficile were selected from the Czech National C. difficile strain collection (Table 1) [13].

Seven C. difficile isolates were cultured from diarrheal glutamate dehydrogenase (GDH) and toxin A/B positive stool samples of hospitalized patients with CDI. One C. difficile isolate (RT 010, non-toxigenic) was cultured from diarrheal GDH positive and toxin A/B negative stool sample of patient with Candida-acquired diarrhea. All isolates were sensitive to metronidazole (MTZ).

Clostridium difficile isolates were recovered from the frozen stocks by inoculating on the Schaedler Anaerobe Agar CM0437 (Oxoid) and cultured for 48 h at 37 °C under anaerobic conditions. Toxin production of all strains in the study was confirmed using commercial immunochromatographic assay (Vidia, Czech Republic) for the detection of free toxins A and B in the stool samples when the bacterial suspension was investigated as a stool sample. The bacterial mass was resuspended in Thioglycolate medium USP (Oxoid) and the number of bacteria (CFU) was analyzed via optical density (OD) analysis at 595 nm (Multiskan Spectrum plate reader, Thermo Fisher Scientific), considering that OD 1 in 1 mL of Thioglycolate medium corresponds to 2.4 × 10⁶ CFU. Later, 9 mL of Thioglycolate medium was inoculated in triplicate for each representative strain to OD 1.99 and cultivated for 5 days at 37 °C under anaerobic conditions. OD was also measured at the end of the cultivation and reached comparable values among the cultures (Additional file 1: Table S1). Capillary electrophoresis ribotyping of C. difficile isolates was performed, using primers described elsewhere [17], before resuspension in Thioglycolate medium USP and after 5 days of culture before proteomic analysis.

Following the pelleting of bacterial cells by centrifugation (18,000g/20 min) to remove all bacterial cells from the proteomes released from in vitro cultures, the pH of the supernatants was adjusted to 3.5 with 3 M sulfuric acid (Sigma-Aldrich). After an overnight precipitation at 4 °C, the pellets were recovered by centrifugation (18,000g/20 min). Because of the potential of interfering substances in the supernatant, the sample preparation workflow was applied and based on FASP (Filter aided sample preparation—FASP) [18]. Pellets were resuspended in 100 mM ammonium bicarbonate (Sigma-Aldrich), and proteins were quantified by bicinchoninic acid assay, (QuantiPro™ BCA Assay Kit, Sigma-Aldrich), [19]. Data are available in the Additional file 1: Table S1. Resuspended pellets were transferred onto Amicon^® Ultra—10 kDa filters (Millipore) and washed twice by 100 mM ammonium bicarbonate. Subsequently, the samples were denatured by 8 M guanidinium chloride (Sigma-Aldrich), reduced with 100 mM Tris (2-carboxyethyl) phosphine hydrochloride (TCEP, Sigma-Aldrich) and alkylated with 300 mM iodoacetamide (Sigma-Aldrich). Finally, the samples were digested with 2 μg of sequencing grade trypsin (Promega) overnight at 37 °C. Empore™ SPE Cartridges, C18, standard density, bed I.D. 4 mm (Sigma-Aldrich) were used to desalt peptide mixtures before drying to completion in a speed-vac. Before the mass spectrometry analysis, the samples were resuspended in 30 μL of 2% acetonitrile (ACN)/0.1% trifluoroacteic acid. The samples were furthered analyzed by LC–MS/MS techniques involving targeted mass spectrometry and LFQ. Subcellular localization of the proteins released from in vitro cultured panel was evaluated by bioinformatic tools. Detailed description of procedure is described in the Additional file 1.

To assess the applicability of the LFQ approach, we examined the similarity of individual proteomes using hierarchical clustering and principal component analysis. Unsupervised cluster analysis of protein expression profiles was performed using Euclidean distances. Statistical procedures were performed using the computational platform Perseus [9]. Both hierarchical clustering (Fig. 1c) and PCA (Fig. 1d) generated eight distinctive groups encompassing each biological triplicate of analyzed RT representatives and showed the applicability of this workflow. For example, RTs 027 and 176 nearly co-cluster, on the contrary proteomes from RT 078 created a distinctive group.

Using the targeted mass spectrometry, TcdA was detected as highly produced in RTs 027 and 176. Lower quantities of TcdA were observed in RTs 005 and 012. The protein was not detected in RTs 010, 078, and 001. TcdB was detected exclusively in RTs 027 and 176. Employing heavy labeled peptides and targeted mass spectrometry for TcdA and TcdB, the analysis confirmed these results (Additional file 1: Table S5, Figure S2) and approved the LFQ approach relevancy. The expression differences of selected proteins connected with pathogenicity are depicted in Table 2.

CdtA and CdtB/(binary toxin) were observed in the secreted fractions of RTs 027, 078 and 176, and in high levels in RTs 027 and 176.

The expression of Sigma-54 dependent regulatory protein (C9YK92), responsible for creating swift and precise responses to environmental change [24], was observed among RTs 027, 176, and 005. Furthermore, the high expression of nucleic acid zinc binding protein (C9YLG4) and the cell surface, putative penicillin binding protein, which belongs to the cell wall binding repeat 2 family (C9YLL2), were observed exclusively in RTs 027 and 176.

The nitroreductase C9YRI2, an enzyme putatively involved in the activation of the MTZ, was expressed at similar levels in RTs 001, 005, 010, 012, and 014, but was not detected in RTs 027, 176, and 078. In contrast, Abc-type Fe3+ transport system periplasmic component-like protein C9YQW5 was found in high levels only in RTs 027, 176, and 078.

A higher expression of flagellar proteins FlgE, G, K, L, Fli C, D, K, and flagellar basal body protein C9YI80 were observed among RTs 027, 176, 005, and 014 with the exception of FlgK and FliD quantifiable also in RT 078. Moreover, FlgC levels were increased only in RTs 027, 176, and 014, whilst FlgM and FliL were found to be increased in RTs 027, 176, and 005 (Table 2). Interestingly, FliE was expressed only in RTs 027 and 176, and glycosyltransferase C9YI34, which is involved in the post-translational modification of flagellum, was expressed in RTs 027, 176, 005, and 014.

The Pro–Pro endopeptidase (PPEP-1) expression was observed in all RTs without relevant differences (Table 2). In addition, the search for substrates of PPEP-1 was also performed. Since PPEP-1 gene is probably not present in R20291 genome/proteome, the reference strain 630 [25, 26] was employed. The substrates CD2831/Q183R6 and CD3246/Q17ZZ0 were detected only in the replicates of RT 078 isolate (Table 2).

HisG (C9YLE5) a HisZ (C9YLE4) were not detected in non-toxigenic RT 010 (Table 2). In other RTs, the expression of these proteins was observed in high levels but without distinct differences.