Loading...
Rapid methods for bacterial identification and predicting antibiotic resistance by MALDI-TOF MS
Saichek, Nicholas R.
Saichek, Nicholas R.
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2017
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Since the 1990’s, there has been an exponential rate of growth of publications regarding mass spectrometry as a means for bacterial identification. The use of matrix assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) as a high throughput, cost-effective alternative to conventional methods in clinical diagnostics has been the basis for a large portion of these. Current techniques for routine bacterial identification involve culture-, biochemical-, and molecular-based methods such as, Gram staining, catalase and oxidase activity tests, and PCR assays. While all of these are established methods with proven success of positive identification each has its respective limitation. Antimicrobial susceptibility testing is of particular interest to improve clinical outcomes and limit the spread of bloodstream infections therefore requiring a need for a rapid and accurate means for testing. In order to address this, our ongoing work with metal oxide laser ionization mass spectrometry (MOLI MS) and phage-based mass spectrometry offers an approach to accomplish simultaneous identification and antibiotic resistance profiling. The five main studies in this paper address the primary entities required for building a functional bacterial identification platform for MALDI MS. The second chapter focuses on the investigation of six metal oxide catalysts for effective cleavage and laser ionization of bacterial cell wall phospholipid extracts. Following evaluation of the six catalysts, CeO2 was found to be the most stable over time and reproducibly generated fatty acid profiles. A suite of ten bacteria, provided cross validation results of analysis of 100% correct correlation for negative-ion data. Using analysis of variance–principal component analysis (ANOVA–PCA), four sample sets collected with stored catalyst at 0, 8, 24, and 504 h showed no effect based on long-term catalyst degradation. Supervised learning by a fuzzy rule-building expert system (FuRES) that was validated with training and prediction set partitions independent of CeO2 age and unsupervised data analysis using a dendrogram of Euclidean distance confirmed that the CeO2 catalyst age had no effect on the fatty acid mass spectral profiles. CeO2-facilitated fatty acid profiling, using MOLI MS, was compared to protein profiling using the commercial Bruker Biotyper platform. Four datasets, Enterobacteriaceae, Acinetobacter, Listeria, fungi were processed by principal component analysis (PCA) and validated using leave-one-spectrum-out cross-validation (LOSOCV) to test the method’s efficacy. These tests revealed 100% correct classification. In comparison, protein profile data from the same bacteria yielded 32%, 54%, 67%, and 33% mean species-level accuracy using two MALDI-TOF MS platforms, respectively. In addition, several pathogens were misidentified by protein profiling as non-pathogens and vice versa. These results suggest novel CeO2-catalyzed lipid fragmentation readily produced (i) taxonomically tractable fatty acid profiles by MOLI MS, (ii) highly accurate bacterial classification and (iii) consistent strain-level ID for bacteria that were routinely misidentified by protein-based methods. To evaluate identification and antibiotic resistance determination capabilities of MOLI MS, fifty Staphylococcus isolates were evaluated. This genus was chosen specifically due to the increased occurrence of S. aureus infections, specifically the acquired susceptibility to -lactam antibiotics. Leave-one-spectrum-out cross-validation indicated 100% correct assignment at the species and strain level. Preliminary analysis differentiating MRSA from MSSA demonstrated the feasibility of simultaneous determination of strain identification and antibiotic resistance. As an alternative approach, phage amplification detected by MALDI-TOF MS was investigated for rapid and simultaneous Burkholderia pseudomallei identification and ceftazidime resistance determination. B. pseudomallei ceftazidime susceptible and resistant ΔpurM mutant strains Bp82 and Bp82.3 were infected with broadly targeting B. pseudomallei phage ϕX216 and production of the m/z 37.6 kDa phage capsid protein observed by MALDI-TOF MS over the course of 3 h infections. This allowed for reproducible phage-based bacterial ID within 2 h of the onset of infection. MALDI-TOF MS-measured time to detection correlated with in silico modeling, which predicted an approximate 2 h detection time. Ceftazidime susceptible strain Bp82, while detectable in the absence of the drug, owing to the reliance of phage amplification on a viable host, was not detectable when 10 µg/mL ceftazidime was added at the onset of infection. In contrast, resistant strain Bp82.3 was detected in the same 2 h timeframe both with and without the addition of ceftazidime. The final chapter offers a modified phage-based approach applied for the detection of pathogenic enterococci. An extensive host range study of vancomycin-resistant and -sensitive strains revealed promising phage candidates which were probed for unique peptide fragments by MALDI MS. Antibiotic resistance determination was confirmed by successful phage amplification.
Associated Publications
Rights
Copyright of the original work is retained by the author.