Compilation of a MALDI-TOF mass spectral database for the rapid screening and characterisation of bacteria implicated in human infectious diseases

Abstract

A database of MALDI-TOF mass spectrometry (MS) profiles has been developed with the aim of establishing a high throughput system for the characterisation of microbes. Several parameters likely to affect the reproducibility of the mass spectrum of a taxon were exhaustively studied. These included such criteria as sample preparation, growth phase, culture conditions, sample storage, mass range of ions, reproducibility between instruments and the methodology prior to database entry. Replicates of 12 spectra per sample were analysed using a 96-well target plate containing central wells for peptide standards to correct against mass drift during analysis. The quality of the data was assessed statistically prior to database addition using root mean squared values of <3.0 as the criterion for rejection. Cluster analysis using a nearest neighbour algorithm also enabled subsets of data to be compared. This was achieved using the bespoke MicrobeLynx™ software. Columbia blood agar was used to standardise all procedures for the database, since it permitted the culture of most human pathogens and also produced spectra with a broad range of mass ions. In some instances, alternative media such as CLED were used in specific studies with greater success. Following standardisation of the procedure, a database was developed comprising ca. 3500 spectra with multiple strain entries for most species. The results to date show unequivocally that as the number of strains per species increased, so too did the success of species matching. The technique demonstrated unique mass spectral profiles for each genus/species, with the variation in mass ions among strains/species being dependent on the intra-specific diversity. The success of identification against the database for wild-type strains ranged between 33 and 100%; the lower percentage results being generally associated with poor representation of some species within the database. These findings provide a new dimension for the rapid and high throughput characterisation of human pathogens with potentially broad applications across the field of microbiology.

Introduction

It has long been recognised that the study of infectious diseases could not proceed without clear circumscription of the microbe. Hence from its inception to the present time considerable effort has been devoted to continually improving available methods for microbial characterisation. Classification and identification of bacteria have traditionally utilised characters based upon morphological and physiological properties. Consequently, all editions of Bergey’s Manual have endorsed these properties so strongly that even up to the present time new taxa may not be proposed without the description of these features (Olsen and Shah, 2003). The arrival of chemotaxonomic analysis in the 1970–1980s which included peptidoglycan analysis (Schleifer et al., 1990) and analysis of lipids such as respiratory quinones, polar lipids and long-chain cellular fatty acids (LCFA) (Collins et al., 1982) together with biochemical analyses such as multilocus enzyme electrophoresis (Williams and Shah, 1981) and SDS-polypeptide analysis (Kersters and De Lay, 1980), did much to extend the range of phenotypic characters of bacterial cells. Thus, over the next 20 years microbiology began a major overhaul of established classification systems and accepted nomenclature began changing. The arrival of nucleic acid analyses paved the way for modern approaches which began initially with the determination of mol% G+C of DNAs, DNA–DNA and DNA–RNA hybridisation (De Ley et al., 1970), rRNA cataloguing and eventually 16S rDNA sequence analysis (see e.g. Stackebrandt et al., 1988). The arrival of PCR techniques made the analysis of 16S rDNA analysis easily achievable and perhaps at no other time in the history of microbiology has the study of systematics reached such magnitude. The result has been a transformation from a classification system based upon phenotypic tests to one based upon genealogy. The current Bergey’s Manual (in preparation) will adopt this system and will include even uncultured species.

Microbial taxonomists are apprehensive about using a single criterion for the description of taxonomic units and polyphasic approaches have been advocated (Vandamme et al., 2002). The sole reliance on 16S rDNA may also be misleading since many species cannot be delineated by sequence comparison of a single gene. For example, most members of the genus Bacillus sensu stricto cannot be resolved using 16S rDNA sequencing. Bacillus anthracis, Bacillus thuringiensis and Bacillus cereus manifest very different diseases but still share 100% sequence homology in the 16S rDNA gene (Ash et al., 1991). Because of such difficulties, systematic committees who advise on the taxonomy of various groups of micro-organisms have recommended polyphasic approaches. Nevertheless, 16S rDNA sequencing provides the current structure of the microbial kingdom and represents the only public database available that can be interrogated electronically.

Despite the immense work undertaken on the chemical analysis of bacterial cells, apart from the LCFA database (Sherlock Microbial Identification System, Newark, USA) such data resources are severely lacking in Microbiology. Respiratory quinones, for example, have been systematically studied in microbes but the results are scattered and no database exists. LCFA remain an important diagnostic tool in microbiology because the methodology has been developed to make it achievable as a high throughput system while a comprehensive database was developed in tandem.

Mass spectrometry (MS) has traditionally been utilised for chemical analysis, though the advent of gentler ionisation processes has led to the emergence of more techniques for the higher molecular mass determination of biological molecules. Plasma desorption (PD), fast atom bombardment (FAB), laser desorption (LD) and electrospray ionisation (ESI) have been utilised to analyse components of living organisms (Cotter, 1992). However, the discovery of matrix-assisted laser desorption ionisation (MALDI) perhaps represents the pinnacle of these studies as it permits the analysis of biological molecules with no theoretical upper mass limit (Karas and Hillenkamp, 1988, Barshick et al., 1999). In essence, a matrix material is mixed with the biological sample and upon irradiation with a laser, molecules of the sample are ionised and desorbed to form a plume of gaseous ions. Nowadays the most common method used to detect this plume of ions is a time of flight (TOF) analyser, in which ions are separated and detected according to their molecular mass and charge. The resulting output of such analyses is a mass spectral profile determining molecular masses of ions in the original plume.

A variety of studies have shown that MALDI-TOF-MS may be used for the rapid analysis of biological components of bacterial cells (Dai et al., 1999, Nilsson, 1999, Fenselau, 1997, Chong et al., 1997, Liang et al., 1996). Furthermore, these methods have been simplified to utilise the studies of intact bacterial cells, significantly reducing preparation time (Holland et al., 1996, Claydon et al., 1996, Krishnamurthy and Ross, 1996, Welham et al., 1998, Lynn et al., 1999, Arnold and Reilly, 1998, Haag et al., 1998, Domin et al., 1999, Wang et al., 1998). The intact bacteria are applied to a target, mixed with matrix and analysed in the mass spectrometer. The molecules detected are generally surface components, which hitherto have not been systematically studied. Since many of the interesting properties associated with microbial physiology (e.g. electron transport, signal transduction, etc.), virulence and pathogenicity (toxin assemble, haemagglutinins, ligans, binding receptors, etc.) are associated with the surface of the cells, MALDI-TOF-MS offers the possibility for the large scale comparative analysis of such molecules and provides for the first time a means of gaining insight into the diversity of such components among micro-organisms. The aim of the present study is to develop the technique as a high throughput system to use in the rapid screening and characterisation of human pathogenic bacteria. We envisage that if successful, the method would support existing methodologies, for example, in the characterisation of new diversity, monitoring the emergence and re-emergence of pathogens and following changes in microbes in response to environmental pressure. To achieve this it is first necessary to assemble a database using type and reference cultures. This study outlines the parameters investigated to attain this, assembly of the database and pilot studies aimed at monitoring its success.