Applicazione della spettrometria di massa MALDI-TOF in microbiologia clinica

Abstract Mass spectrometry (MS), developed in the 70’s, thanks to the Matrix-Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) technology has been important applications in different fields: biotechnological (for the characterization and quality control of recombinant proteins and other macromolecules), medical-clinical (for laboratory diagnosis of diseases and for the development of new targeted therapeutic treatments), food and environmental. In the last few years, MALDI-TOF has become a powerful tool in the clinical microbiology setting and has revolutionised the work-flow also in our laboratory, enabling the rapid identification of bacteria and yeasts for clinical diagnosis, replacing the conventional phenotypic identification based on biochemical assays. Nowadays, two general MALDI-TOF MS methods have been proposed to characterize microorganisms: (1) mass spectra comparison with fingerprints database and (2) matching of biomarker masses to a proteome database. Recently, an other application of MALDI-TOF MS was proposed for the detection of β-lactamases, especially carbapenemases. The first aim of this study was to assess the usefullness of a new implemented database for MALDI-TOF MS by using the approach of mass spectra comparison for the identification of bacteria and yeasts for which the previous identification by the commercially available database failed due to the absence, the presence of mistakes or incomplete reference spectra. In particular, this was obtained for bacteria belonging to spirochaetes of the genera Borrelia and Leptospira, for filamentous fungi (dermatophytes) and for protozoa (Trichomonas vaginalis). The second aim of this study was to evaluate specific biomarkers for the identification and the differentiation of intestinal parasites for which a reference database is not available and its creation would be difficult due to the complexity of the biological material (faeces) analysed for their detection or the culture media (Robinson’s media) used for their cultivation. The last aim of this study was the evaluation of the applicability of MALDI-TOF MS for the study of the resistance to carbapenem by Gram-negative bacteria. In particular, a carbapenem hydrolysis assay by MALDI-TOF MS was developed in order to detect the carbapenemase production in Enterobacteriaceae. The main results reported in this study are of some interest for the application of MALDI-TOF MS in a laboratory of clinical microbiology. The effectiveness of a new database for the identification by MALDI-TOF MS based on the approach of mass spectra comparison was firstly proved for the genus Borrelia. The commercial database of the mass spectrometer currently used in our laboratory (MALDI Biotyper, Bruker Daltonics, Germany) for microbial identification, included only 3 reference spectra belonging to the species B. burgdorferi senso strictu, B. garinii, and B. spielmanii. The database was supplemented with the missing spectra belonging to Borrelia species circulating in Europe (B. afzelii) and in the world (such as B. hermsii and B. japonica), overcoming the lack of reference profile for these species in the commercial MALDI-TOF MS database. Moreover, the supplementation of the database with spectra deriving from other reference strains of the Borrelia species already present in the original database provided the evidence of the usefulness of MALDI-TOF MS for a more reliable identification of these bacteria. The result obtained on a strain of B. lusitaniae not included in the original database and identified at genus level only after its supplementation as well as those obtained at species level for the other Borrelia spp. clinical isolates, showed that MALDI-TOF MS could be used as a rapid, cheap and reliable alternative to the conventional and routine identification methods in clinical microbiological laboratories to identify strains belonging to this genus. Similarly, for the genus Leptospira, after the “ex-novo” creation of the “home-made” database with 20 reference spectra belonging to 20 reference strains, a correct identification at species level was obtained for the same strains when a second blind independent experiment was performed. The dendrogram obtained reflects the phylogenetic tree based on the 16SrRNA sequencing reported in the literature showing Leptospira species clustered according to their pathogenicity (pathogenic and saprophytic strains clearly separated into different clusters) and confirms the comparability of the results obtained by mass spectrometry and by molecular typing methods. L. biflexa and L. fainei were located in a branch while L. interrogans, L. kirschneri, L. noguchii and L. borgpetersenii clusterised in a separate one. The latter was further divided into two groups respectively containing, L. borgpetersenii on one side and L. interrogans, L. kirschneri, and L. noguchii on the other side. Moreover, the dendrogram built with the MSP-Spectra of the strains belonging to the genera Borrelia and Leptospira, analysed in this study and to the genus Brachyspira (supplemented in a previous study) showed 3 main separated clusters, one for each genus, excluding the possible mis-identification among these 3 different genera of spirochaetes. It was also interesting to observe little differences among the profiles of the analysed Leptospira strains within a given species (namely L. interrogans and L. borgpetersenii, for which more than one strain was analysed), probably due to the presence of differences among the protein patterns of distinct serovars, as subsequently confirmed by using ClinProTools statistical software that showed serovar-specific peaks. As a matter of fact, after the creation of a statistical algorithm, it was possible to discriminate at serovar level both the L. interrogans and L. borgpetersenii strains tested. However, we cannot definitely conclude that we identified universal serovar-specific peaks, since we used a selected panel of serovars in this study. Our results demonstrated that MALDI-TOF MS followed by the analysis with statistical software could be used as a first-line assay for the rapid, cheap and reliable identification of Leptospira strains also at the serovar level, limited to those analysed in this study. This features could significantly improve the approaches currently used to monitor the occurence of epidemiological outbreaks and pathogen surveillance. As already demonstrated for Borrelia and Leptospira, the supplementation of the mass spectrometer database with reference spectra of filamentous fungi (dermatophytes) was important not only for the identification of the dermatophytes species more frequently isolated in our area but also for the identification of those species of dermatophytes whose frequency was increased due to migration flows from endemic area (Microsporum audouinii, Trichophyton violaceum and, T. sudanense). Furthermore, after the database implementation, the mis-identification of T. interdigitale and T. mentagrophytes as T. tonsurans observed with the original database was avoided. The dendrogram obtained from the spectra of the 24 reference dermatophytes strains supplemented in the new in-house database revealed a clusterization that reflects the phylogenetic taxonomic tree of the dermatophytes reported in literature. However, the reference spectra of T. interdigitale and T. mentagrophytes segregated in a single cluster with low discrimination, suggesting high similarity between these two species. For this reason, besides MALDI-TOF MS, sequencing of their Internal Transcribed Spacer (ITS) region failed to distinguish between T. interdigitale and T. mentagrophytes. Noteworthy, the dendrogram created with the 12 proteic profiles already included in the Bruker Daltonics database and those deriving from 24 dermatophytes additional strains reproduced the reference phylogenetic taxonomic tree for all genera except Trichophyton. In particular, the reference spectra of all the strains (both original and additional) belonging to the species Epidermophyton floccosum, Trichopthyton rubrum, T. tonsurans, Microsporum canis, and M. gypseum cluster closely each other and in agreement with the phylogenetic taxonomic tree; on the contrary, the reference spectra already present in the Bruker Daltonics database of the strains belonging to the species, T. interdigitale and T. mentagrophytes, segregated separately from those implemented in this study. This latter result could explain the persisting mis-identification also after the implementation of the database. The usefullness of MALDI-TOF MS was also investigated for microorganisms other than bacteria and fungi for which the method was initially developed. The identification protocol was applied to Trichomonas vaginalis, by changing in the standard parameters used for bacterial and fungal identification. The use of new parameters setting for the creation of the reference spectra was necessary to overcame the overlapping of the peaks observed among the proteic profile obtained from the 2 media used in this study for the culture of this protozoan and those obtained from T. vaginalis reference strain. As a matter of fact, metabolite, pigments and/or nutrients contained in the media may interfer with the crystallisation process of the proteins and lead to low identificative score values. In particular, the peaks ranging from 3 to 6 kDa, including those referring to media components, were excluded from the analysis of clinical isolates. Due to the low sensitivity of MALDI-TOF MS technology (detection limit of 105 trophozoites/ml) the new method did not allow the identification of T. vaginalis directly from clinical samples. MALDI-TOF MS based on the detection of specific biomarkers was applied in order to identify and to differentiate Entamoeba histolytica (pathogenic) and E. dispar (non-pathogenic), the two genetically distinct but morphologically identical amoebae. The development of this application allowed to solve the difficulty in the creation of a dedicated database due to the complexity of the faecal material and of the media used for the cultivation (LYI-S-2 axenic medium, Robinson’s medium without serum, and Robinson’s medium with or without Escherichia coli), and to identify 5 specific peaks able to differentiate E. histolytica from E. dispar. In particular, two discriminating peaks for E. histolytica and 3 peaks for E. dispar were found. These 5 peaks did not show any interference with those found for the different culture media used for amoeba cultivation. For this reason these 5 discriminating peaks may be considered potential good markers to differentiate E. histolytica from E. dispar. The molecular weights of the 2 discriminating peaks of E. histolytica were found to correspond only to 2 specific proteins of E. histolytica present in GenBank: Amoebapore A and an unknown putative protein of E. histolytica HM-1:IMSS-A strain. This result confirmed the specifity of the peaks of E. histolytica identified by MALDI-TOF MS analysis. On the contrary, the E. dispar specific peaks did not have any correspondence with proteins deposited in GenBank. In our hands, the 5 discriminating peaks allowed us to differentiate 12 (7 E. dispar and 5 E. histolytica) of the 13 Entamoeba spp. clinical isolates grown in Robinson’s medium; the results were in agreement with those obtained by a Real-time PCR able to differentiate E. histolytica from E. dispar. For one clinical isolate identified as E. histolytica by Real-time PCR and by sequencing MALDI-TOF MS failed to discriminate the two species. This result allowed us to hypothesize the presence of amino acid/posttranslational differences in the proteins detected by MALDI-TOF MS for this strain. Further studies would be needed to verify this. However, when using MALDI-TOF MS as first-line diagnostic assay the identification of a similar strain, if any, in the future should be confirmed by Real-time PCR. The analysis performed at different incubation time on a faecal sample positive for E. histolytica, showed that specific peaks were detected even after 12h of incubation in Robinson’s medium without serum, that is the first step before the transfer of an aliquot in the complete medium when the culture for protozoa is performed. This result underlines the reliability of the discriminating peaks used in this study: they were detected also in the presence of fecal material. Finallly, we evaluated the applicability of MALDI-TOF MS as a rapid phenotypic assay for the detection of carbapenemase-producing bacterial strains, by developing a modified protocol based on the hydrolyis of the meropenem used as reference carbapenem both on reference and clinical strains. The meropenem hydrolysis assay by MALDI-TOF MS was performed on 3 reference strains and on 1219 clinical isolates (1185 Enterobacteriaceae and 34 non-Enterobacteriaceae). All results obtained, except one, were concordant with those obtained by the conventional phenotypic assays (Modified Hodge Test, MHT and disk diffusion inhibition Test, Sinergy Test, ST) routinely used in the majority of the diagnostic microbiology laboratories. The discordant result was referred to a Klebsiella pneumoniae strain, nonsusceptible to carbapenems and carbapenemase non-producer as assessed by MHT and ST, but positive to Meropenem Hydrolysis Assay (MHA). The spectra obtained by MHA showed the presence of both the peaks of intact meropenem and those of the hydrolysed drug. For this strain the genotypic analysis confirmed the MHA result by revealing the presence of the blaKPC gene. This is consistent with a low-expression of carbapenemase and it would have been missed by using only MHT and ST. This hypothesis should be confirmed by sequencing of the blaKPC gene and by the analysis of the deduced amino acid sequence. The meropenem hydrolysis assay by MALDI-TOF MS revealed to be an indirect phenotypic assay able to distinguish, as MHT commonly used as first line assay in clinical diagnostic laboratory, a carbapenemase-producing from a carbapenemase-non-producing strain both for diagnostic purpose and epidemiological surveillance. Compared to conventional methods used in the routine diagnostic laboratories which require trained personnel for the interpretation of the result and long execution times, the MHA by MALDI-TOF MS showed several advantages. The simple and easy preparation of samples, as well as the immediate acquisition of the data makes this technique a fast, cheap and reliable method for the detection of carbapenemase-producing strains. In conclusion, compared MALDI-TOF MS, independently from the approach used for the identification of microorganisms, confer in most cases a significant gain of both technician working time (preanalytical procedure to prepare samples) and turnaround time (automated analytical procedure to obtain results). The time effectiveness gained with MALDI-TOF identification compared with conventional identification approaches is even more accentuated when several isolates are analysed in parallel. Finally, the flexibility of the system, semplicity in the sample preparation, and the automatic acquisition of the data makes MALDI-TOF MS a rapid, accurate, and cost-effective (about 0.50/1.00 Euros per sample) method for microbial characterization and identification. All these results demonstrated that MALDI-TOF MS is becoming an important tool both in clinical and in experimental microbiology, thanks to the high efficiency in the microbial identification, due to the availability of both new commercial and in-house libraries, and to the possibility to successfully detect the bacterial resistance to antibiotic.