Files in this item



application/pdfFAZAL-DISSERTATION-2017.pdf (8MB)
(no description provided)PDF


application/vnd.openxmlformats-officedocument.wordprocessingml.documentFazal, ZeeshanSupplementaryMaterial.docx (902kB)
(no description provided)Microsoft Word 2007


application/vnd.openxmlformats-officedocument.spreadsheetml.sheetFazal, Zeeshan_SupplementaryTables.xlsx (69kB)
(no description provided)Microsoft Excel 2007


Title:Gene expression patterns of encapsulated microbial cells
Author(s):Fazal, Zeeshan
Director of Research:Rodriguez-Zas, Sandra
Doctoral Committee Chair(s):Rodriguez-Zas, Sandra
Doctoral Committee Member(s):Jakobsson, Eric; Villamil, Maria; Brown, Chester
Department / Program:Animal Sciences
Discipline:Animal Sciences
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Yeast genomic responses
Microbial stress studies
Functional analysis
Microbial cell encapsulation
Abstract:To design hybrid cellular/synthetic devices such as sensors and vaccines, understanding of how the metabolic state of living cells changes upon physical confinement within three-dimensional matrices is vital. We analyze the gene expression patterns of stationary phase Saccharomyces cerevisiae (S. cerevisiae) cells encapsulated within three distinct nanostructured silica matrices and relate those patterns to known naturally occurring metabolic states. It was found that the cells for all three-encapsulated methods enter quiescent states characteristic of response to stress, albeit to different degrees and with differences in detail. By the measure of enrichment of stress-related Gene Ontology categories, we find that the AqS+g encapsulation more amenable to the cells than CDA and SD encapsulation. We hypothesize that this differential response in the AqS+g encapsulation is related to four properties of the encapsulating gel: 1) oxygen permeability, 2) relative softness of the material, 3) development of a protective sheath around individual cells, and 4) the presence of glycerol in the gel, which has been previously noted to serve as a protectant for encapsulated cells and can serve as the sole carbon source for S. cerevisiae under aerobic conditions. This work represents a combination of experiment and analysis aimed at the design and development of 3D encapsulation procedures to induce, and perhaps control, well-defined physiological behaviors. We also report on the temporal pattern of yeast gene expression patterns during encapsulation in silica matrices via a cell-directed assembly process, and upon release. Three broad classes of patterns are seen. A major shift in expression patterns is seen upon encapsulation, relative to the beginning stationary state, similar to previously reported stress response. Significant continuing shifts are seen by sampling at different intervals during a one week encapsulation. Upon release from encapsulation and reincubation in growth medium, the cells are in a state significantly different from the state prior to encapsulation and similar to the state during encapsulation. Implications are drawn for the use of encapsulated micro-organism as sensors and effectors, and for the persister state of such organisms. Ordinarily Gene Ontology (GO) enrichment analysis is subject to an arbitrary threshold for defining significance of enriched classes. In this paper, we consider replacing an arbitrary threshold with F-measure optimization to define the p-value that divides “significant enrichment” from “non-significant”. It is found that evaluation of false negatives (essential for computing recall and thus F-measure) requires a heuristic (but reasonable) assumption. We apply F-measure optimization to two sets of genes from different organisms and use Benjamini-Hochberg and random resampling to evaluate the number of false positives. It is found that the uncorrected p- value that produces optimum F-measure varies widely from one data set to another. It is also found that all three methods of FDR calculation diverge from each other within a range of uncorrected p-values that provide F-measure optimum p-values. This study includes in Appendix II a pipeline for using resampling and F-measure optimization to create lists of enriched GO classes that provide for variable weights of precision and recall.
Issue Date:2017-08-16
Rights Information:Copyright 2017 Zeeshan Fazal
Date Available in IDEALS:2018-03-13
Date Deposited:2017-12

This item appears in the following Collection(s)

Item Statistics