Simon Cameron (Presenter)
Imperial College London
Bio: I joined Prof. Zoltan Takat's research group in October 2015 as a Research Associate in Microbial Populations and Metabonomics in the Division of Computational and Systems Medicine at Imperial College London. I received my BSc (2011) and PhD (2015) from Aberystwyth University, Wales, UK. My research interests are in the broad fields of microbial ecology and metabolomics, with a focus on the human microbiome and metabolome and its response to disease and stress. I currently work as part of the MicrobeID team at Imperial College London, which is developing rapid evaporative ionisation mass spectrometry (REIMS) as a high-throughput, a to assign taxonomic and functional classifications to microbial isolates and to the direct-from-sample profiling of mixed microbial communities.
Authorship: Simon Cameron (1), Alvaro Perdones-Montero (1), Frances Bolt (1), Adam Burke (1), Zsolt Bodai (1), Tony Rickards (1,2), Kate Hardiman (1), A Abdolrasouli (1,2), Monica Rebec (2), Zoltan Takats (1).
(1) Imperial College London, London, UK (2) Imperial College Healthcare NHS Trust, London, UK
| Short Abstract Mass spectrometry (MS) has revolutionised the workflow of microbiology laboratories; leading to a substantial reduction in diagnosis times. Unlike commercially available MS platforms, rapid evaporative ionisation MS (REIMS) requires no sample preparation, such as the addition of a matrix, before sample analysis. We have developed an automated, high-throughput REIMS platform which is capable of analysing up to 5,000 microbial colonies in 24 hours. This platform classifies isolates with an accuracy of >99% for Gram, >95% for genus, and >93% for species. This presentation will cover the ongoing technical optimisation of the platform, including the addition of laser REIMS and optimised monopolar diathermy electrodes, the development of a reference spectral database to allow identification of unknown isolates, and phenotypic classifications including antimicrobial resistance profiles. |
Long Abstract
This presentation will encompass the on-going work within the MicrobeID team at Imperial College London to develop rapid evaporative ionisation mass spectrometry (REIMS) as a high-throughput, automated platform for the taxonomic identification and functional classification of microorganisms. Four main areas will be covered: (1) optimisation of the electrical diathermy monopolar electrodes; (2) adaptation of a surgical laser as an alternative to electrical diathermy to conduct REIMS analysis to improve classification accuracy and sample throughput; (3) development of a REIMS mass spectral reference database to allow taxonomic identification of unknown isolates; and (4) phenotypic classification of microbes, including virulence and pathogenicity factors and antimicrobial resistance profiles.
INTRODUCTION: Prior to the introduction of mass spectrometry platforms, such as matrix-assisted laser desorption ionisation time of flight (MALDI-ToF), into microbiology laboratories, the majority of microbial identifications were reliant upon extensive biochemical tests. These included sugar utilisation and fermentation, and were time consuming, variable, and required expertise to perform. The implementation of mass spectrometry platforms has allowed for rapid microbial identification; improving patient care through reduced diagnosis times, and targeted clinical interventions. MALDI-ToF based platforms are the most widely used in clinical laboratories. Identifications using this approach rely upon the characterisation of ionised proteins within the mass range of 2−20 kDa; of which over 50% are of ribosomal origin [1,2]. Although, the introduction of MALDI-ToF has reduced identification times, it still requires the addition of a matrix to aid ionisation. Furthermore, for some microbial groups, such as for yeasts, additional extraction steps are necessary for accurate species level identifications to be achieved. In the majority of investigations, MALDI-ToF is reliant upon the isolation of pure microbial cultures before analysis can be carried out. This is frequently the major factor in determining the length of time required for microbial identification. In addition, MALDI-ToF based systems can also have difficulty in the species-level classification of some groupings, such as Salmonella spp., and in sub-species analysis.
REIMS has been shown to provide accurate species-level classification of bacteria and yeasts direct from colonies; without the need for additional preparative steps [3,4]. In comparison to MALDI-ToF based platforms, REIMS utilises the lipidomic profile of bacteria and fungi for species-level classification. REIMS works by applying a radiofrequency electrical current to a microbial biomass, causing it to heat rapidly and evaporate. The resulting vapour, containing gas phase ions of metabolites and structural lipids, is then channelled to a mass spectrometer for analysis; under the instrument’s normal vacuum. This allows for a real-time mass spectral signal to be generated as detection of ions happens in less than one second since heating of the sample has occurred. The REIMS technology has recently been incorporated into a high-throughput, automated platform which includes culture plate handling, colony imaging and selection, and REIMS analysis using a monopolar probe. This platform is capable of processing up to 5,000 microbial colonies in a 24 hour period; making it highly suitable for a high-throughput clinical microbiology laboratory. We have recently shown that this high-throughput platform is capable of achieving species-level taxonomic classifications which are comparable with the handheld bipolar probe REIMS approach [5].
METHODS: We have previously reported on the analysis of 25 clinically important microbial species using both handheld bipolar REIMS and high-throughput monopolar REIMS, showing comparable classification accuracies are achievable using both. To extend the breadth of microbial taxonomy analysed using REIMS, over 45 microbial species have been analysed using the high-throughput monopolar REIMS platform. This collection of species, constitutes our preliminary REIMS mass spectral reference database, by which we are able to assign taxonomic identifications to unknown isolates. To ensure the REIMS taxonomic designation is accurate, full-length 16S rRNA gene and ITS region sequencing, MALDI-ToF mass spectrometry using the Bruker Biotyper system (Bruker Daltonics), and other species-specific biochemical and genetic tests have been completed. REIMS analysis is conducted using a Xevo G2-XS QToF instrument in negative ion detection mode. Sample heating occurs directly, such as from a microbial agar culture plate, without sample preparation or processing. The analyte containing vapour is aspirated using the instrument’s native vacuum towards the mass spectrometer. Prior to its entry, the vapour is mixed with 2-propanol containing leucine encephalin, which acts as an external lock mass compound for mass drift correction, and impacts with the REIMS collision surface, in the form of a heated kanthal ribbon to break up ion clusters. Mass spectral data underwent background subtraction, mass drift correction, total ion count normalisation, and mass binning to 0.1 Da using the Offline Model Builder (OMB) software (Waters Corporation). Classification models were then constructed using a Random Forest algorithm and used to assign classifications to each isolate.
In addition to the development of a mass spectral reference library, we are also optimising the high-throughput REIMS platform to improve throughput, ease of use, and taxonomic identification accuracy. This work currently includes optimisation of the REIMS monopolar electrode surface geometry, and the utilisation of a carbon dioxide surgical laser, rather than electrical diathermy as previously employed, to apply heat to a microbial biomass in the rapid evaporation process. Furthermore, we are also conducting a substantial amount of application development work. This includes the analysis of mixed microbial cultures, the analysis of filamentous fungi isolates directly from culture, the functional classification of antimicrobial resistant isolates, including methicillin resistant/sensitive and carbapenamase producing/non-producing isolates, assessment of virulence metabolites produced by pathogens, and direct-from-sample pathogen detection from urine and faecal diagnostic specimens.
RESULTS AND DISCUSSION: The high-throughput REIMS platforms is highly suited to the workflow and workload of clinical microbiology laboratories. It is capable of processing up to 5,000 microbial colonies in a 24 hour period, with data generated in real-time. The platform requires minimal user input, thereby reducing variability and associated costs. Assessment of the ability of the high-throughput REIMS platform to accurately and robustly assign taxonomic classification to clinically important microbial species was completed through the analysis of 15 isolates for over 40 species. Random Forest classification models were used to assign Gram stain, genus, and species level classification for all isolates. Using this approach, classification accuracies of >99% for Gram stain, >95% for genus, and >93% for species were achieved. Utilisation of this data set as a REIMS mass spectral reference database allowed for the identification of unknown microbial cultures, which were not used in the construction of the reference classification model, achieved taxonomic identifications with accuracy rates of >99% for Gram stain, and >93% for species.
Optimisation of the high-throughput REIMS platform, including the utilisation of a shortened stainless steel sharp tip electrode, and a round end electrode has increased classification accuracies. In addition, preliminary work on the use of a carbon dioxide surgical laser to act as the source of sample heating suggests that improved species-level classification accuracies can be achieved, with increased sample throughput.
For analysis of mixed cultures of bacteria and yeasts, a variety of species and concentration ratios were tested using both our high-throughput REIMS method and the previously reported handheld bipolar REIMS approach [3,4]. For all species combinations and ratios, species specific biomarkers were detected that act as a proof-of-principle that REIMS has the utility to detect clinically important microbial species from mixed cultures and/or communities.
Using exactly the same workflow as for bacterial and yeast cultures, to date 300 filamentous fungi isolates, collected as part of routine clinical microbiology diagnostics, have been analysed using REIMS, directly from their culture medium. After species level identifications were confirmed through sequencing of the ITS region, classification models were constructed using Random Forest and were able to achieved genus and species level classification accuracies in excess of 90%.
In addition to the taxonomic classification of clinically important microbes, REIMS possesses the potential to also give important functional information on isolates; such as identifying methicillin resistant or sensitive Staphylococcus aureus isolates, without exposure to the relevant antibiotic. A total of 150 methicillin sensitive S. aureus and 150 resistant S. aureus isolates have been analysed using the high-throughput REIMS platform, with preliminary data analysis suggesting that REIMS has utility in the determination of antibiotic susceptibilities.
CONCLUSIONS: REIMS employed in a high-throughput, automated platform is highly suited to clinical microbiology laboratories and is capable of achieving species-level classification accuracies which are comparable with commercially available MALDI-ToF platforms. In addition, REIMS has a utility in providing clinically important phenotypic data, beyond just taxonomic identifications, including virulence profiles and antimicrobial susceptibilities.
References & Acknowledgements:
References:
[1] Krause, E., Wenschuh, H. & Jungblut, P. R. The Dominance of Arginine-Containing Peptides in MALDI-Derived Tryptic Mass Fingerprints of Proteins. Anal. Chem. 71, 4160–4165 (1999).
[2] Ryzhov, V. & Fenselau, C. Characterization of the Protein Subset Desorbed by MALDI from Whole Bacterial Cells. Anal. Chem. 73, 746–750 (2001).
[3] Strittmatter, N. et al., Analysis of Intact Bacteria using Rapid Evaporative Ionisation Mass Spectrometry. Chem. Commun. 49, 6188–90 (2013).
[4] Strittmatter, N. et al., Characterization and Identification of Clinically Relevant Microorganisms using Rapid Evaporative Ionization Mass Spectrometry. Anal. Chem. 86, 6555–62 (2014).
[5] Bolt, F. et al., Automated High-Throughput Identification and Characterisation of Clinically Important Bacteria and Fungi using Rapid Evaporative Ionisation Mass Spectrometry (REIMS). Anal. Chem. (2016).
Acknowledgements:
This work is supported by a research grant from the UK Biotechnology and Biological Sciences Research Council (ref: BB/L020858/1). Additional financial (staff salaries at Imperial College London, provision and equipment of consumables, and other research funding) and technical support is given by the Waters Corporation. This abstract/presentation does not promote any commercially available product of the Waters Corporation.
| Description | Y/N | Source |
| Grants | yes | The Waters Corporation and BBSRC |
| Salary | no | |
| Board Member | no | |
| Stock | no | |
| Expenses | no |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | yes |