Quantifying and Engineering Protein Dynamics in Bacteria
MetadataShow full item record
CitationPotvin-Trottier, Laurent. 2017. Quantifying and Engineering Protein Dynamics in Bacteria. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractGenetically identical cells can display heterogeneity in their level of proteins due to stochastic gene expression. Such noise has been found to be widespread across biology, and can have a tremendous impact by allowing cells to access different phenotypes. However, the timescales of these fluctuations matter: slow fluctuations are potentially much more potent whereas if fluctuations are corrected rapidly they are relatively innocuous. Many simple mechanisms in cells – such as gene cascades, time-averaging to suppress fluctuations and epigenetic modifications – can create slow fluctuations. Quantifying protein dynamics remains a technical challenge as it requires measuring gene expression of long time series under constant growth conditions, and thus the timescales of fluctuations in gene expression remain largely unknown. By using a newly developed microfluidic device – to follow hundreds of cells for hundreds of cell divisions under constant growth conditions – we quantify the timescale of fluctuations of ~50 transcriptional reporters and ~10 translational reporters in Escherichia coli. All reporters show a strikingly similar and surprising behavior: an exponential decorrelation with a half-life of one generation. We show that the discrepancy with previous studies can be explained by artifacts arising from calculating autocorrelation functions with short time series or non-uniform growth conditions.
The general absence of slow fluctuations in cells opens up the question of how difficult it is to create epigenetic memory with fluctuations in protein copy number. To answer this question, we revisit a synthetic oscillator, the repressilator, to engineer oscillations correlated over hundreds of generations. Synthetic gene circuits typically have much lower accuracy than their natural counterpart, and our hypothesis for this difference is that they are usually designed without considering stochastic gene expression. We used principles from stochastic chemistry in single cells to reduce error propagation and information loss. By simply removing features from the circuit rather than adding feedback loops, we created highly regular and robust oscillations, with circuits keeping their 14-generation periods in a wide range of growth conditions. The phase was kept for hundreds of generations such that flasks of cells and bacterial colonies displayed synchronous oscillations, even without coupling between cells.
Proteins cannot adapt to upstream changes on a scale faster than their lifetime. In order to respond to changes in their environment more quickly than their division time, cells must degrade their proteins. Fluctuations in the proteolytic machinery could therefore cause fluctuations in the half-lives of the degraded proteins. We develop and validate the first tool measuring instantaneous proteolysis rates in single bacterial cells. By measuring the saturation curve of ssrA-tagged proteins, we show that these substrates exhibit Michaelis-Menten kinetics, with very high affinity and a half-life of ~45 seconds. We show that the SspB adapter protein increases the affinity of the ClpXP protease to ssrA-tagged substrates, which was previously demonstrated in vitro but not in vivo. However, depending on the substrate's local structure, degradation can be faster without the adapter. By measuring changes in degradation rates over time, we discover a proteolytic response: at high concentration of ssrA-tagged substrates, cells compensate by producing more proteases. We connect this phenomenon to the heat shock response and characterize the associated toxicity. We discuss the limitations of using fluorescent proteins as reporters for degradation – fluorescent proteins targeted with natural tags can be partially degraded such that there is no decay in fluorescence – and suggest potential alternative methods for measuring degradation in single cells.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:41140241
- FAS Theses and Dissertations