Browsing by Author "You, Lingchong"

An essential property of microbial communities is the ability to survive a disturbance. This is readily observed in bacteria, which have developed the ability to survive every antibiotic treatment at an alarming rate, considering the timescale at which new antibiotics are developed. Thus, there is a critical need to use antibiotics more effectively, extend the shelf life of existing antibiotics and minimize their side effects. This requires understanding the mechanisms underlying bacterial drug responses. Past studies have focused on survival in the presence of antibiotics by individual cells, such as genetic mutants. Also important, however, is the fact that a population of bacterial cells can collectively survive antibiotic treatments lethal to individual cells. This tolerance can arise by diverse mechanisms, including resistance-conferring enzyme production, titration-mediated bistable growth inhibition, swarming and interpopulation interactions. These strategies can enable rapid population recovery after antibiotic treatment and provide a time window during which otherwise susceptible bacteria can acquire inheritable genetic resistance.

To further explore bacterial antibiotic responses, I focused on bacteria producing β-lactamase, an enzyme that has drastically limited the use of our most commonly prescribed antibiotics: β-lactams. Through the characterization of clinical isolates and a computational model, my Ph.D. thesis has three implications:

First, survival can be achieved through resistance, the ability to absorb effects of a disturbance without a significant change, or resilience, the ability to recover after being perturbed by a disturbance. Current practices for determining the antibiotic sensitivity of bacteria do not characterize a population as resistant and/or resilient, they only report whether the bacteria can survive the antibiotic exposure. As resistance and resilience often depend on different attributes, distinguishing between these two modes of survival could inform treatment strategies. These concepts have long been applied to the analysis of ecological systems, though their interpretations are often subject to debate. This framework readily lends itself to the dissection of the bacterial response to antibiotic treatment, where both terms can be unambiguously defined.

Second, the ability to tolerate the antibiotic treatment in the short term corresponds to resistance, which primarily depends on traits associated with individual cells. In contrast, the ability to recover after being perturbed by an antibiotic corresponds to resilience, which primarily depends on traits associated with the population.

And finally, understanding the temporal dynamics of an antibiotic response could guide the design of a dosing protocol to optimize treatment efficiency for any antibiotic-pathogen combination. Ultimately, optimized dosing protocols could allow reintroduction of a repertoire of first-line antibiotics with improved treatment outcomes and preserve last-resort antibiotics.

Although antibiotics have proven to be one of the great achievements of modern medicine, their efficacy has dramatically decreased over the past several decades. This is due, in part, to the rapid pace of natural bacterial evolution, but also to the overuse and misuse of antibiotics in general. This often selects for drug-resistant pathogens, and allows them to flourish in the face of antibiotic treatment. In addition to the emergence of genetic resistance, bacteria often utilize a number of population-level behaviors to survive antibiotic treatment. This is referred to as collective antibiotic tolerance (CAT). Taken together, antibiotic resistance and tolerance have led to the re-emergence of infectious diseases throughout the world. In general, there are two strategies to combat this risk: develop novel antibiotics, and/or use existing drugs more effectively, so as to minimize the chance of resistance emergence. Novel drug development is a time- and resource-intensive process, and pharmaceutical companies are not financially incentivized to develop these types of drugs. Therefore, it is of increasing importance to understand the population dynamics underlying various bacterial survival mechanisms, and exploit this knowledge to design better antibiotic treatment protocols.

My dissertation research focuses on a prevalent phenomenon called the postantibiotic effect (PAE), which refers to the transient suppression of bacterial growth following antibiotic treatment. Although PAE has been empirically observed in a wide variety of antibiotics and microbial species, heretofore there has not been a definitive mechanistic explanation for this pervasive observation.

In this work, I use a combination of high-throughput microfluidic experiments and computational modeling to examine the relationship between dosing parameters and the degree of bacterial inhibition, quantified by population recovery time. I found that recovery time is a function of total antibiotic, regardless of how the dose profile. Moreover, a minimal model of transport and binding kinetics was sufficient to recapture this trend, suggesting a unifying explanation for historical observations of PAE in a variety of contexts. I validated this modeling using both in silico and in vitro perturbation studies.

Moreover, I showed that efflux inhibition, a common strategy in antibiotic treatment, is effective in certain dynamic-dependent situations. This work puts forth a possible mechanism for PAE, which could serve as a clinical aid in selecting effective antibiotic/adjuvant combinations, as well as in designing periodic antibiotic treatments.

Programmed death is often associated with a bacterial stress response. This behavior appears paradoxical, as it offers no benefit to the individual. This paradox can be explained if the death is `altruistic': the sacrifice of some cells can benefit the survivors through release of `public goods'. However, the conditions where bacterial programmed death becomes advantageous have not been unambiguously demonstrated experimentally. Here, I determined such conditions by engineering tunable, stress-induced altruistic death in the bacterium Escherichia coli. Using a mathematical model, we predicted the existence of an optimal programmed death rate that maximizes population growth under stress. I further predicted that altruistic death could generate the `Eagle effect', a counter-intuitive phenomenon where bacteria appear to grow better when treated with higher antibiotic concentrations. In support of these modeling insights, I experimentally demonstrated both the optimality in programmed death rate and the Eagle effect using our engineered system. These findings fill a critical conceptual gap in the analysis of the evolution of bacterial programmed death, and have implications for a design of antibiotic treatment.

From biosynthesis to bioremediation, microbes have been engineered to address a variety of biotechnological applications. A promising direction in these endeavors is harnessing the power of designer microbial consortia that consist of multiple populations with well-defined interactions. Consortia can accomplish tasks that are difficult or potentially impossible to achieve using monocultures. Despite their potential, the rules underlying microbial community maintenance and function are not well defined, though rapid progress is being made. This limited understanding is in part due to the greater challenges associated with increased complexity when dealing with multi-population interactions. For example, although metabolic pathways are often engineered in single microbial populations, the introduction of heterologous circuits into the host can create a substantial metabolic burden that limits the overall productivity of the system. This limitation could be overcome by metabolic division of labor (DOL), whereby distinct populations perform different steps in a metabolic pathway, reducing the burden each population will experience. While conceptually appealing, the conditions when DOL is advantageous have not been rigorously established. In my dissertation, I analyzed 24 common architectures of metabolic pathways in which DOL can be implemented. My analysis revealed general criteria defining the conditions that favor DOL, accounting for the burden or benefit of the pathway activity on the host populations as well as the transport and turnover of enzymes and intermediate metabolites. These criteria can help guide engineering of metabolic pathways and has implications for understanding evolution of natural communities. I next investigated utilizing horizontal gene transfer (HGT) as a tool to engineer microbial consortia. Due to the role of HGT in spreading and maintaining diverse functional traits such as metabolic functions, virulence factors, and antibiotic resistance, suppressing plasmid transfer in microbial communities has profound implications for consortia engineering. However, existing tools for inhibiting HGT are limited in their modes of delivery, efficacy, and scalability. I demonstrated a generalizable denial-of-spread (DoS) strategy that can target and eliminate specific conjugative plasmids from communities. My strategy exploits retrotransfer, whereby an engineered DoS plasmid is introduced into host cells containing a target plasmid via the target’s own conjugative machinery. Within the same host, DoS eliminates the target plasmid through a combination of transfer competition and plasmid incompatibility, after which DoS can be removed via induced suicide. DoS’s design is highly tunable and scalable to various conjugative plasmids, different plasmid curing mechanisms, or environmental contexts. Together, my findings contribute to a greater understanding of consortia stability and establish a potential new tool for precision engineering of said consortia.

Bacteria communicate, coordinate, and cooperate as a population and this `social' behavior is key to their proliferation. Quorum sensing (QS) is the cell-cell communication mechanism by which bacteria sense their population density and modulate their target gene expression accordingly. While QS is ubiquitous among bacteria, there is tremendous diversity in terms of the sensory elements used and the biochemical and transport properties of signaling molecules. Further, the targets of QS include a wide range of cooperative actions, such as the secretion of enzymes for nutrient foraging, virulence toxins, and biofilm-forming compounds. Here I investigate what role QS and cooperation play, as universal social characteristics, in promoting bacterial proliferation.

Engineered biological circuits offer the potential to test our understanding of natural systems under well-defined contexts, by focusing on the key characteristics and components of interest. In my doctoral work, I have taken advantage of this methodology to study bacterial social behavior. Combining mathematical modeling with quantitative experiments using gene circuits, my research has (1) elucidated the `core' components of cell-cell communication across bacteria, (2) explained how communication and cooperation advantage bacterial growth, and (3) opened up the important application of this research in generating novel antibacterial therapies.

Bistable switches are commonly observed in the regulation of critical processes such as cell cycles and differentiation. The switches possess two fundamental properties: memory and bimodality. Once switched ON, the switches can remember their ON state despite a drastic drop in stimulus levels. Furthermore, at intermediate stimulus levels with cellular noise, the switches can cause a population to exhibit bimodal distribution of cell states. Till date, experimental studies have focused primarily on cellular mechanisms that generate bistable switches and their impact on cellular dynamics.

Here, I study emergent bistability due to bacterial interactions with either synthetic gene circuits or antibiotics. A synthetic gene circuit is often engineered by considering the host cell as an invariable "chassis". Circuit activation, however, may modulate host physiology, which in turn can drastically impact circuit behavior. I illustrate this point by a simple circuit consisting of mutant T7 RNA polymerase (T7 RNAP*) that activates its own expression in bacterium Escherichia coli. Although activation by the T7 RNAP* is noncooperative, the circuit caused bistable gene expression. This counterintuitive observation can be explained by growth retardation caused by circuit activation, which resulted in nonlinear dilution of T7 RNAP* in individual bacteria. Predictions made by models accounting for such effects were verified by further experimental measurements. The results reveal a novel mechanism of generating bistability and underscore the need to account for host physiology modulation when engineering gene circuits.

In the context of antibiotic treatment, I investigate bistability as the underlying mechanism of inoculum effect. The inoculum effect refers to the decreasing efficacy of an antibiotic with increasing bacterial density. Despite its implication for the design of antibiotic treatment strategies, its mechanism remains poorly understood. Here I show that, for antibiotics that target the core replication machinery, the inoculum effect can be explained by bistable bacterial growth. My results suggest that a critical requirement for this bistability is sufficiently fast turnover of the core machinery induced by the antibiotic via the heat shock response. I further show that antibiotics that exhibit the inoculum effect can cause a "band-pass" response of bacterial growth on the frequency of antibiotic treatment, whereby the treatment efficacy drastically diminishes at intermediate frequencies. The results have implications on optimal design of antibiotic treatment.

The human microbiome is a dynamic, integrated ecosystem that interacts with the host to influence cancer development and progression, as well as affect response to anti-cancer therapies, suggesting opportunities for diagnostic and therapeutic approaches. Many microbe-microbe and host-microbe interactions relevant to cancer are expected to take place at the tumor site. However, obtaining and sequencing biological samples for the interrogation of these interactions is costly, while the exponential growth of sequencing data for such samples poses analytical and interpretive challenges. Thus, there is a growing need for comprehensive resources, reference databases, and analytical tools for understanding host-microbe interactions relevant to human cancers and other diseases. Herein, I demonstrate that the creation of such resources does not necessitate massive investments into new research programs, and can instead be accomplished by utilizing preexisting, public information. In two trans-kingdom, pan-cancer analyses of sequencing data from The Cancer Genome Atlas (TCGA), it is shown that both bacteria and fungi are involved human tumors samples, and that these signatures are predictive of patient outcomes. In doing so, novel methods for mitigating contamination and false-positive signals in such datasets are described. Lastly, a widely applicable analytical tool and reference database for microbe set enrichment analysis is proposed, which can be used to interpret large microbiome datasets.

Advancements in the biotechnology and pharmaceutical fields have led to the development of an expanding number of applications for certain recombinant proteins of interest. As such, the demand for efficient and cost effective protein production systems is growing. A great deal of research, cost and time goes into improving and optimizing the production of commercially valuable proteins of interest. Many current methods involve growing a culture of cells to its maximum capacity, all of which are producing a certain protein of interest, and then killing off the entire culture to extract the protein. By doing so, regrowth of the entire cell population is required, taking additional time and resources. Controlled lysis could allow for a more continuous protein release through the killing of only a portion of the population and allowing recovery in the exponential growth phase. This study acts as a proof of concept for the implementation of programmable suicidal lysis circuits into bacteria, Escherichia coli, being cultured for protein production for the sustained production and release of said proteins.

To test the viability of suicidal lysis as a mechanism for sustained protein release a robust oscillator circuit, ePop, was used. The ePop circuit controls the synthesis of E gene, producing a protein that incites cellular lysis by attacking the cell wall. By culturing cells for long term growth and extracting small volumes of the culture at various time points for protein quantification, the protein release capabilities of ePop were observed. Protein quantities in the lysates and supernatants of the extractions were determined using SDS-PAGE Coomassi Staining and a Pierce BCA Protein Assay. Also, western blotting was performed on supernatant samples to show the effective release of a specific protein of interest, GFP. The focus was on the presence of protein in the supernatant which is correlated to the release during the lysis cycle of the bacterial population oscillations.

Protein release via the ePop circuit was shown to be effective and robust. The oscillator circuit released measurable quantities of protein in the supernatant of the culture extractions as predicted. The green fluorescent protein of interest used as a pilot protein was effectively released into the supernatant and shown through a western blot with a GFP specific antibody. Population oscillator circuits through cellular lysis were shown to be a viable method for protein release and could be applied to protein production processes as well as other technologies.

From gene expression to protein abundance, biological measurements have traditionally been taken as population averages. Studies over the past decade, however, have established the biological relevance of population heterogeneity, even for cells with identical genetic background. The advancements of our understanding, albeit transformative, have been almost exclusively focused on the generation and the regulation of population heterogeneity and its implications in cell-fate decisions. What’s less appreciated is the potential of using population heterogeneity measurements as an information source of cellular physiology and network dynamics. While gene expression from a signaling network in a single cell is stochastic and unpredictable, the distribution of the gene expression in a sufficiently large cell population is uniquely determined by the signaling network and is deterministic. This distribution represents a quantitative fingerprint for the cell population under a specific environmental condition. I established and characterized a computational platform using stochastic modeling of the Myc/Rb/E2F network that supports the analysis of distributional data—how it changes as a function of perturbation and how it can be used to infer cellular or external variables of interest. I then demonstrated that a viral-mediated gene expression probe can be effectively and efficiently employed to generate heterogeneity fingerprints of cell populations and differentiate different cell lines as well as characterize drug activities.

Control of E2F transcription factor activity, regulated by the action of the retinoblastoma tumor suppressor, is critical for determining cell cycle entry and cell proliferation. However, an understanding of the precise determinants of this control, including the role of other cell cycle regulatory activities, has not been clearly defined.

Recognizing that the contributions of individual regulatory components could be masked by heterogeneity in populations of cells, we made use of an integrated system to follow E2F transcriptional dynamics at the single cell level and in real time. We measured and characterized E2F temporal dynamics in the first cell cycle where cells enter the cell cycle after a period of quiescence. Quantitative analyses revealed that crossing a threshold of amplitude of E2F transcriptional activity serves as the critical determinant of cell-cycle commitment and division.

By using a developed ordinary differential equation model for Rb/E2F network, we performed simulations and predicted that Myc and cyclin D/E activities have distinct roles in modulating E2F transcriptional dynamics. Myc is critical in modulating the amplitude whereas cyclin D/E activities have little effect on the amplitude but do contribute to the modulation of duration of E2F transcriptional activation. These predictions were validated through the analysis of E2F dynamics in single cells under the conditions that cyclin D/E or Myc activities are perturbed by small molecule inhibitors or RNA interference.

In an ongoing study, we also measured E2F dynamics in cycling cells. We provide preliminary results showing robust oscillatory E2F expression at the single-cell level that aligns with the progression of continuous cell division. The temporal characteristics of the dynamics trajectories deserve further quantitative investigations.

Taken together, our results establish a strict relationship between E2F dynamics and cell fate decision at the single-cell level, providing a refined model for understanding the control logic of cell cycle entry.

Non-genetic reprogramming, including but not limited to metabolomic and epigenetic, play an equally significant role in cancer development compared to genetic mutations. In most scenarios, non-genetic alterations of tumor cells are associated with their tumor microenvironment, which is highly related to tumor progress and efficiency of the treatment. Nevertheless, how cancer cells adapt their microenvironment by metabolomic or epigenetic reprogramming remains largely unknown. This dissertation started with exploring two scenarios in colorectal cancer (CRC) studies: the metabolic reprogramming of CRC liver metastasis and the epigenetic remodeling of CRC patient-derived models of cancer. In the study of CRC liver metastasis (Chapter 2), we found that metastatic CRC cells promote their fructose metabolism in the liver by upregulating ALDOB. Knocking down ALDOB or restricting the dietary fructose can suppress CRC liver metastasis. In addition, we examined the potential therapeutic approach for liver metastasis with a KHK inhibitor. For the patient-derived models of cancer (PDMC) project (Chapter 3), we developed six matched PT-PDMC sets and performed ATAC-seq and mRNA to study the chromatin accessibilities of CRC cells. We found two-axis chromatin remodeling separating PDMC from the original patient sample (axis #1) as well as the different cancer models (axis #2). PDOX is more similar to PDX than organoids suggesting the chromatin remodeling of CRC cells is under the pressure of tumor microenvironment in PDMC. We also identified the two transcript factors, KLF14 and EGR2, which respond to the xenografts’ environment by footprinting analysis. These two TFs and their downstream gene, EPHA4, altered CRC tumor growth and drug sensitivities. Therefore, chromatin remodeling of different PDMC may interfere with their ability to predict therapeutic outcomes. In the last part of the dissertation (Chapter 4), I developed a novel system that can label and manipulate the tumor niche in situ. This method provides tools for studying the non-genetic alterations of CRC cells when they interact with the tumor microenvironment. Taken together, this dissertation presents a comprehensive understanding of the non-genetic reprogramming of colorectal cancer and its tumor microenvironment. It advances both the knowledge of non-genetic reprogramming in colorectal cancer and technologies to study the tumor microenvironment.

Microbial communities are an integral and indispensable part of biogeochemical processes, health of plants and animals, and human activities [1-3]. The community structure (the members and their relative abundance) is a key variable that determines the dynamics, stability, functions, and evolution of a microbial community [4]. However, our capability of interpreting, predicting, and controlling microbial community structure is still lacking. This is manifested in the challenges we face in maintaining the stable colonization of beneficial strains [5], reducing the colonization of pathogenic strains [6], and controlling the community structure for engineering purposes [7].

To solve this problem, one approach is to develop methods and tools specific to each system. However, due to the diversity and complexity of microbial communities, these system-specific approaches can be labor intensive and ad hoc. The opposite approach is to distill general principles that can be widely applicable to any microbial communities without characterizing low-level details, such as at the metabolic level. If there exist such general rules, we can then build a toolbox to solve classes of problems related to microbial community structure in an efficient manner.

In this dissertation, I lay out the reasoning and steps to distill general principles that underlie microbial community structure. Using this approach, I then study two types of factors that have significant impacts on microbial community structures. One is a study on mutualism, where I established a general mathematical criterion that underlies the coexistence of mutualistic systems and developed a machine learning method to apply the criterion to diverse experimental systems for interpretation and prediction of system outcomes. The other study is on spatial partitioning, where I established that spatial partitioning increases biodiversity for communities dominated by negative interactions and decreases biodiversity for those dominated by positive interactions. When strong positive and negative interactions are present, biodiversity peaks at an intermediate partitioning level. This general principle can be used to interpret, predict, and control microbial community structure in a robust manner.

Bacteria are found throughout the environment, from the air to the soil, but more importantly, they reside within the human body. Crucial to their survival in each of these environments is the constant interplay between these organisms and their surroundings. Inadvertently, the ways in which these stimuli are processed can have a profound impact on human health. With potentially negative or positive consequences, it becomes critical to understand how microorganisms encode and decode signals.

Understanding bacterial signal processing is crucial to tackling the treatment of infectious diseases, especially with the rise of antibiotic resistant organisms. Antibiotic resistance has become a global health issue as bacteria have developed or acquired genes that confer resistance to all antibiotics currently in use today. This has serious implications for the future treatment of infectious diseases, potentially limiting options to those from a pre-antibiotic era. However, as with other external factors, antibiotics are just another signal that bacteria need to decode and encode a response to. As such, it is of utmost importance to better understand how bacteria process stimuli.

In my dissertation, I analyzed the ways in which bacteria both encode and decode information. In particular, I focused on how information is processed from signals with a temporal domain. To start, I developed a computational framework to understand how organisms decode signals, specifically oscillatory signals. With this model, I examined the capability of an incoherent feedforward loop motif to exhibit temporal adaptation, in which a network becomes desensitized to sustained stimuli. I discovered that this property is crucial for networks to distinguish signals of varying temporal dynamics.

In terms of information encoding, I utilized the complexity of this process to predict bacterial characteristics of interest. The fundamental premise behind this work is to increase the information content of phenotypes for the prediction of bacterial characteristics. Specifically, I used the temporal domain of growth for the prediction of genetic identity and traits of interest. I demonstrated that temporal growth dynamics under standardized conditions can differentiate among hundreds of strains, even strains of the same species. While growth dynamics could, with high accuracy, differentiate between unique strains, it was insufficient to quantify how genetically different these strains were. This absence highlighted the challenges in using genomics to infer phenotypes and vice versa. Bypassing this complexity, I showed that growth dynamics alone could robustly predict antibiotic responses. Together, my findings demonstrate the ability to develop applications that take advantage of the complexity of bacterial information encoding.

This work highlights the importance of understanding how bacteria decode signals with temporal dynamics. Additionally, I demonstrated one application for utilizing bacterial signal encoding, the prediction of bacterial characteristics.

Bacteria are found throughout the environment, from the air to the soil, but more importantly, they reside within the human body. Crucial to their survival in each of these environments is the constant interplay between these organisms and their surroundings. Inadvertently, the ways in which these stimuli are processed can have a profound impact on human health. With potentially negative or positive consequences, it becomes critical to understand how microorganisms encode and decode signals.

Understanding bacterial signal processing is crucial to tackling the treatment of infectious diseases, especially with the rise of antibiotic resistant organisms. Antibiotic resistance has become a global health issue as bacteria have developed or acquired genes that confer resistance to all antibiotics currently in use today. This has serious implications for the future treatment of infectious diseases, potentially limiting options to those from a pre-antibiotic era. However, as with other external factors, antibiotics are just another signal that bacteria need to decode and encode a response to. As such, it is of utmost importance to better understand how bacteria process stimuli.

In my dissertation, I analyzed the ways in which bacteria both encode and decode information. In particular, I focused on how information is processed from signals with a temporal domain. To start, I developed a computational framework to understand how organisms decode signals, specifically oscillatory signals. With this model, I examined the capability of an incoherent feedforward loop motif to exhibit temporal adaptation, in which a network becomes desensitized to sustained stimuli. I discovered that this property is crucial for networks to distinguish signals of varying temporal dynamics.

In terms of information encoding, I utilized the complexity of this process to predict bacterial characteristics of interest. The fundamental premise behind this work is to increase the information content of phenotypes for the prediction of bacterial characteristics. Specifically, I used the temporal domain of growth for the prediction of genetic identity and traits of interest. I demonstrated that temporal growth dynamics under standardized conditions can differentiate among hundreds of strains, even strains of the same species. While growth dynamics could, with high accuracy, differentiate between unique strains, it was insufficient to quantify how genetically different these strains were. This absence highlighted the challenges in using genomics to infer phenotypes and vice versa. Bypassing this complexity, I showed that growth dynamics alone could robustly predict antibiotic responses. Together, my findings demonstrate the ability to develop applications that take advantage of the complexity of bacterial information encoding.

This work highlights the importance of understanding how bacteria decode signals with temporal dynamics. Additionally, I demonstrated one application for utilizing bacterial signal encoding, the prediction of bacterial characteristics.

Driven by the development of new technologies and an ever expanding knowledge base of molecular and cellular function, Biology is rapidly gaining the potential to develop into a veritable engineering discipline - the so-called `era of synthetic biology' is upon us. Designing biological systems is advantageous because the engineer can leverage existing capacity for self-replication, elaborate chemistry, and dynamic information processing. On the other hand these functions are complex, highly intertwined, and in most cases, remain incompletely understood. Brazenly designing within these systems, despite large gaps in understanding, engenders understanding because the design process itself highlights gaps and discredits false assumptions.

Here we cover results from design projects that span several scales of complexity. First we describe the adaptation and experimental validation of protein functional assays on minute amounts of material. This work enables the application of cell-free protein expression tools in a high-throughput protein engineering pipeline, dramatically increasing turnaround time and reducing costs. The parts production pipeline can provide new building blocks for synthetic biology efforts with unprecedented speed. Tools to streamline the transition from the in vitro pipeline to conventional cloning were also developed. Next we detail an effort to expand the scope of a cysteine reactivity assay for generating information-rich datasets on protein stability and unfolding kinetics. We go on to demonstrate how the degree of site-specific local unfolding can also be determined by this method. This knowledge will be critical to understanding how proteins behave in the cellular context, particularly with regards to covalent modification reactions. Finally, we present results from an effort to engineer bacterial cell suicide in a population-dependent manner, and show how an underappreciated facet of plasmid physiology can produce complex oscillatory dynamics. This work is a prime example of engineering towards understanding.

Colorectal cancer (CRC) is the third most diagnosed type of cancer, and the 5-year survival rate drops significantly once the patient develops liver metastases. Notably, over the past decade, multiple patient-derived models of cancer (PDMC) have been developed and are widely accepted as preclinical models. Current chemotherapy does not distinguish the primary and metastatic loci, and there lack a direct comparison between different PDMC (e.g., patient-derived organoids (PDO), patient-derived xenografts (PDX) and PDO-derived xenografts (PDOX)) and the patient tumor (PT). Therefore, understanding the differences between the cells from metastasized CRC and the primary site, as well as the differences between the PDMC and the original patient specimen is of critical importance.In CRC, many conventional studies have focused on associating genetic mutations with clinical phenotypes. However, non-genetic alterations including changes in chromatin accessibility, transcriptome and histone modification markers provide an alternative and even faster way for the tumor cells to adapt to their microenvironment. In this dissertation, we first focused on how the liver microenvironment can affect the epigenetic transformations of the metastasized CRC. Using high-throughput sequencing such as ATAC-seq, RNA-seq and Mint-ChIP, we identified an HGF-PU.1-DPP4 epigenetic reprogramming axis that facilitates the metastatic tumor cells to adapt to the liver microenvironment. The results were validated by extensive numbers of patient samples and the precision epigenetic modification tools CRISPR/dCas9KRAB/HDAC. We identified several FDA approved drugs including Sitagliptin and Norleual, which can be repurposed to treat CRC liver metastases. Furthermore, using the similar set of tools, we revealed that each PDMC undergo distinctive epigenetic reprogramming following two modeling axes. The first axis delineates PDX and PDO from patient, while the second axis distinguishes PDX and PDO. We further identified that the transcription factors KLF14 and EGR2 are collectively more active in the PDOX than in PDO. Moreover, we demonstrated that the varied expression level of their common downstream targets EPHA4 led to distinct drug responses in PDO to 147 FDA approved compounds. We concluded that there are differences in growth and drug sensitivity between PDOX and PDO, which should be taken into consideration when using PDMC to predict clinical outcomes.