Combination of high throughput technologies for high dimensional single cell analysis on Tissue microarray of murine lung parenchyma

Abstract

In biomedical research, obtaining the information about cellular mechanisms and processes needed to test research hypotheses is critical, especially when studying human disease in animal models. Bioimage analysis and technology in this area are being developed because manual microscopic analysis is generally too laborious, time consuming, and error prone. Making some procedures related to image analysis automatic is necessary to process the ever-increasing amount of data from microscopy images. Currently, high-throughput molecular technologies for the assessment of marker expression or for functional analysis in research have allowed the identification of numerous putative mechanisms targeted for disease therapies; however, many of these techniques are based on destructive sampling, which does not allow the identification of specific single cell properties. Immunofluorescence, using antibodies to specific antigens, can produce images in which through image analysis one is able to identify the expression and spatial distribution of one or more biomarkers while maintaining morphological and histological structures on routinely formalin-fixed paraffin-embedded tissue. This allows the determination of spatial relationships between cells and tissues through the detection of reference morphological markers, specifically to indicate tissue architecture. The classical tissue analysis of immunofluorescent sections in a slide-by-slide manner, however, has several disadvantages: it is wasteful to stain all sections due to all the material used to perform the technique on large numbers of slides, it is difficult to standardize due to the need for group samples in different staining rounds and valuable collection samples are usually exhausted after only a few analyses. Simultaneous detection of multiple markers (multiplex immunofluorescence) allows them to be run on the same tissue section, giving high throughput, significantly expanding the range of quantifiable metrics that can be collected from tissue sections and also provide a level of consistency and accuracy needed to support the complexities of pulmonary fibrosis pathogenesis, a pulmonary disease case study, object of this thesis. The use of tissue microarray technology allows the simultaneous analysis of multiple tissue samples, selected as expertly identified regions of interest (2 mm of core diameter) on a single microscope slide. TMA has several advantages: analysis of a significant but scarce resource; simultaneous and concurrent analysis of different samples even from archival material; uniformity of staining condition; high-throughput data production. In general terms, the TMA technique allows for less variability during analysis of multiple samples by standardizing variables such as reagent preparation, time, and antibody incubation temperature. In addition, while simultaneous analysis of multiple samples with many markers can be achieved, the analysis of the resulting bio-images has an inherent complexity that is constantly a challenge to overcome for software and analysis algorithms used. Recently, advances in deep learning and scripting for customization of analysis steps have improved many crucial aspects of analysis steps (such as image coordinate registration, cell segmentation, threshold parameters) leading to ever better results. The purpose of this dissertation was to develop optimization procedures in the workflow to obtain and analyze bio-images. Thanks to the combination of multiplex immunofluorescence technology with TMA technology, it becomes possible to perform multiple reactions and investigate numerous markers simultaneously on a customized TMA paraffin blocks including entire experiments. This becomes fundamental to the imaging process as these technologies improve the ability to characterize different types of cell populations in normal and pathological tissues and their spatial distribution in relation to clinical outcomes. TMA and multiplexing technologies, associated with increasingly sophisticated digital image analysis software offer a high-quality way to study human diseases movable to animal models and in this way, it is possible to perform research bringing real contributions to the scientific community.