Pamela Pruski, Holly Lewis, David MacIntyre, Philip Bennett, Trevor Hansel and Zoltan Takats
Imperial College London, London, UK, SW7 2AZ
The mucosal membrane, a protective layer responsible for trapping pathogens in the human body, is an easily accessible and highly clinically relevant sample to diagnose pathogenic and cancerous associated diseases. Since DESI MS is not applicable for in-vivo analysis due to the potential risk of electrical shock and the use of organic solvent, medical swabs are a standard collection device for mucosal membranes that can be directly analysed with DESI MS. Chemical signatures identification of specific bacteria within minutes on the surface of swabs would provide a rapid diagnosis of infections associated with preterm delivery in comparison to standard microbial testing.
The mucosal membrane is a layer of epithelial tissue lining all passages in the human body open to the external environment such as parts of the digestive-, urogenital- and respiratory tracts. It acts as a barrier function between the body and the environment to trap pathogens (bacteria, viruses and fungi) which have an immense impact on human health and disease. Since access to mucosal membrane is easy, allowing non-invasive sampling and contains pathogenic information (e.g. lipids and secondary bacterial metabolites) it might be the sample of choice for rapid detection of infections, dysbiosis or inflammatory diseases. Medical swabs are standard sample collecting devices for mucosal membrane analysis and are used for microbial culturing, drug testing as well as genetic testing using human cells. Clinical swabs are sent to microbiology laboratories for microbial isolation and identification. This has a number of shortcomings as many bacteria are non-cultivable and the use of molecular assays is restricted due to cost and time implications. Therefore, the aim of this study was to develop a new MS-based technique for direct analysis of specific mucus models (nasal, vaginal, phyaryngeal, bronchial, oesophageal) to investigate complex interactions between the microflora and diseases. In the future this may lead to a rapid and reliable point of care identification method of pathogenic bacteria in order to allow proper treatment of patients (e.g. targeted antibiotics).
Desorption electrospray ionization (DESI) MS is an ambient ionization technique with gaining popularity in clinical applications, enabling the direct analysis of biomolecules on the spot and without sample preparation. It is mainly applied for the ionization of target compounds ranging from lipids, peptides, drug molecules to primary and secondary metabolites. In contrast to other ionization techniques (e.g. MALDI or SIMS), the sampling and ionization process with DESI can be performed on almost all surface systems such as glass, metal, porous and polymer surfaces, on human skin and non-uniform surface materials with various geometries. This study will focus on the improvement of mucosal membrane sampling from the surface of medical cotton swabs. Here, the oval geometry of the swabs and the soft material property of the cotton required the development of a new DESI MS set up. Additionally, a thin-film microextraction (TFME) approach was used on the swab surface in order to increase the overall sensitivity of target molecules.
LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) equipped with a home-built DESI source was used for the experiments. Standard medical swabs were used either directly or following chemical modification. Chemically modified swabs were prepared by immobilizing 5μm particle size ODS particles on the standard cotton swabs. Unmodified and modified swabs were used for the sampling of buccal, nasal and urogenital mucosa and were directly subjected to DESI analysis. DESI was operated at 15 μL/min flow rate, 5 bar gas pressure and 4.5 kV high voltage settings. Different solvent systems were tested for desorbing the analytes. Data were acquired as full profiles both in positive and negative ion mode at nominal and high resolution. Spectra were recalibrated and baseline subtracted prior to statistical analysis. Pattern-level identification was performed by creating statistical models using principal component analysis and linear discriminant analysis of data obtained from patients with established clinical diagnosis (training set) and localizing the unknown data points (test set) in the models. Bacterial marker database was constructed by analysing clinically relevant bacterial colonies using Rapid Evaporative Ionization Mass Spectrometry (REIMS). Taxonomical markers were obtained by performing ANOVA test followed by Tukey’s HSD test on the dataset on different taxonomical levels.
With the new developed DESI MS setup the generation of reproducible mass spectral data was achieved during the mucosal membrane analysis on standard medical swabs. In total, 300-1000 individual spectral features were obtained following de-isotoping and removal of adducts with a mass accuracy ≤3 ppm using high mass resolution mass spectrometer. High abundant analytes were tentatively identified by exact mass, isotope cluster distribution and MS/MS experiments. Identified peaks include primary metabolites (energy metabolites, amino acids, organic acids, etc.), simple and complex lipids and bacterial secondary metabolites. Medical swabs analysis of bacterial colonies and in-vitro grown epithelial cells also revealed characteristic metabolic profiles using DESI MS. Using PCA and PLS-DA a clear separation was observed between different mucosal models (buccal, nasal, pharyngeal and urogenital). The loading plot revealed unexpected information, e.g. urogenital mucosa was found to produce cholesterol sulphate as the most abundant lipid species. Also, a separation between vaginal mucosal membrane and vaginal epithelial cells was achieved within the first two principal components. In comparison to standard medical swabs, the chemically modified swabs analysis resulted in 3-10x improved signal to noise ratio for hydrophobic molecules such as fatty acid, lipids and eicosanoid inflammatory mediators. Differentiation between healthy and dysbiotic mucosal flora was studied in case of vaginal mucosa, with the long-term intention of diagnosing females with elevated risk for preterm delivery. Using statistical analysis a separation between both groups was observed using the DESI MS data. In addition, the healthy vaginal microbiome revealed almost exclusively Lactobacillus sp. In those subjects with documented preterm delivery a variety of pathogenic markers were identified by REIMS MS. The technique was also used for the detection of eicosanoid inflammatory mediators in atopic patients challenged by pollen allergen. Arachidonic acid, Leukotrienes and Prostaglandins were successfully detected by DESI from the mucus nasal lining fluid.