Development of methods for the microbiome analysis of formalin fixed paraffin embedded tissue specimens

Abstract

For Synthetic Biology to reach its potential, it necessitates foundational knowledge of the organisms that can be engineered. The remarkable influence our microbiome has on our health status has made it a focus of attention for engineering possibilities aiming at its modulation. As the field of the human microbiome expands, it necessitates access to high-quality nucleic acid samples which are truly representative of the community of bacteria under study. Formalin-fixed, paraffin-embedded (FFPE) samples represent the most comprehensive collections of patient materials in hospital pathology archives. However, for this sample to become reliably accessible for microbiome studies, the effects of FFPE processing on bacteria must be considered. Any sample processing method should be based upon specific study aims, target organisms and sample types. It is only through a holistic understanding of FFPE-induced changes to the bacterial cellular structure and its DNA content, that a reliable method can be developed It is hypothesised here that with a sample-prep workflow considering the effects of FFPE on bacterial cells, their DNA content and the overall contamination introduced, a reliable and reproducible analysis of the microbiome of FFPE samples could be achieved. As such, the overall aim of this thesis was to characterise FFPE induced changes to the bacterial cell walls/membranes and their DNA content, and with this information, to propose strategies for purifying and repairing DNA suitable for microbiome analysis, while also characterising the common contaminants found in samples processed in this manner. To achieve this, an appropriate FFPE bacterial study model was first developed. With this in place, a thorough characterisation of the state of bacterial FFPE DNA was performed and strategies to reduce this damage assessed. Finally, to develop an appropriate method for bacterial DNA extraction from FFPE samples (unavailable at the time of writing), the state of the bacterial cell wall/membrane was assessed and strategies for a uniform bacterial lysis and host depletion evaluated. Chapter 2 Describes methods for creating a mock bacterial FFPE block (Protoblock) that serves as a standard for FFPE samples. The Protoblock is a cell matrix which can be populated with cell types and numbers as desired, so as to resemble those of the FFPE tissue specimens. Its accuracy for representing bacterial load and cell architecture was validated by microscopy. With this model, the performance of the human gold-standard FFPE kit for microbiome analysis of FFPE samples was evaluated and found unsuitable for microbiome research. Additionally, the Protoblock permitted the characterisation of bacterial FFPE DNA, where it was found to be highly fragmented, a poor PCR template and featured significant sequence alterations. Finally, this model also permitted the characterisation of contaminants originating from the FFPE process, the most common being Xanthomonadaceae, Pseudomonadaceae and Clostridiaceae. Chapter 3 Makes a thorough investigation of the state of bacterial FFPE DNA in terms of PCR readability, formalin crosslinking, and the presence of sequence artefacts. Here, bacterial FFPE DNA was found to be highly fragmented, with a significant inverse correlation between fragment size and PCR recovery and a log-fold reduction between the recovery of 200 bp and 500 bp fragments. It was also evident that 95-97% of DNA present in these samples was crosslinked and that the most evident sequence artefacts were those derived from oxidative damage. Two strategies to reduce this damage were investigated. (1) An optimised decrosslinking procedure (10 oC lower than current methods) significantly reduced sequence artefacts. (2) The in vitro reconstitution of the Base Excision Repair pathway targeting oxidative DNA damage, using FPG and Endo VIII DNA glycosylases. Samples treated with both strategies showed a 3X increase in fragment length and a significant reduction in sequence chimeras and SNPs, leading to a significant improvement in sequencing readability. Chapter 4 Investigates the state of the bacterial cell wall/envelope and mammalian membrane to assess the state of their permeabilisation in FFPE samples. In this chapter, mammalian and Gram-negative bacterial cells were found to be impermeable to molecules with dimensions of 3-5 nm. A host depletion strategy was devised using a combination of Saponin and DNAse (Benzonase). It was also found that FFPE bacterial cells require a lysis strategy, and the use of a mix of bacterial-lytic enzymes was found to provide a uniform cross-taxa bacterial lysis. The collection of methods developed were tested by 16S rRNA gene sequence analysis of protoblocks, murine FFPE faeces and human breast tumour samples. The collection of methods provided an overall increase in recovery of 16S PCR amplicons, a higher uniformity in bacterial lysis, and a higher bacterial to host DNA ratio in high biomass models. However, these improvements were obscured for low biomass samples, where contaminants dominated the sequencing reads. It is concluded from this work that to unlock the potential of FFPE specimens for the microbiome field, a full dedicated workflow, comprising not only sample-prep, but also QC, 16S PCR and 16S sequencing, needs to be in place. This workflow should be directed by a robust QC system. In addition, a database for known FFPE derived common contaminants is essential to inform future strategies for the biological removal of contaminants from these samples.