Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Nov 16;7(11):2618-2626.
doi: 10.1021/acssynbio.8b00279. Epub 2018 Oct 24.

A System for Gene Expression Noise Control in Yeast

Affiliations
Free PMC article

A System for Gene Expression Noise Control in Yeast

Max Mundt et al. ACS Synth Biol. .
Free PMC article

Abstract

Gene expression noise arises from stochastic variation in the synthesis and degradation of mRNA and protein molecules and creates differences in protein numbers across populations of genetically identical cells. Such variability can lead to imprecision and reduced performance of both native and synthetic networks. In principle, gene expression noise can be controlled through the rates of transcription, translation and degradation, such that different combinations of those rates lead to the same protein concentrations but at different noise levels. Here, we present a "noise tuner" which allows orthogonal control over the transcription and the mRNA degradation rates by two different inducer molecules. Combining experiments with theoretical analysis, we show that in this system the noise is largely determined by the transcription rate, whereas the mean expression is determined by both the transcription rate and mRNA stability and can thus be decoupled from the noise. This noise tuner enables 2-fold changes in gene expression noise over a 5-fold range of mean protein levels. We demonstrated the efficacy of the noise tuner in a complex regulatory network by varying gene expression noise in the mating pathway of Saccharomyces cerevisiae, which allowed us to control the output noise and the mutual information transduced through the pathway. The noise tuner thus represents an effective tool of gene expression noise control, both to interrogate noise sensitivity of natural networks and enhance performance of synthetic circuits.

Keywords: gene expression noise; mating pathway; regulation; synthetic biology; yeast.

Conflict of interest statement

The authors declare no competing financial interest.

Figures

Figure 1
Figure 1
Design and benchmarking of a noise tuning system. (a) Schematic depiction of the noise tuner. In a Tet-ON system doxycycline activates transcription of the fluorescent mNeongreen reporter gene by binding to the constitutively expressed reverse tetracycline trans-activator (rtTA, not shown). Addition of theophylline prevents mRNA degradation by disrupting the ribozyme cleavage activity in the 3′-UTR. (b) Comparison of the noise tuner with constructs harboring native 3′ regulatory regions measured by flow cytometry. All constructs are identical in their Tet promoter and mNeongreen fluorescent reporter gene and differ only in their 3′ RRs that confer low (FZF1t, GIC1t) or high (TPS1t, ADH1t) expression. Top panel: Dose–responses with 0 to 12.8 ng/μL doxycycline. The noise-tuner transcript was either unstable (0 mM theophylline, “NT”) or fully stabilized (12.8 mM theophylline, “NT + theo”). Bottom panel: The noise tuner exhibits similar inverse noise-median correlation as native 3′ RRs. Noise is given as the robust coefficient of variation (robust CV, see Methods). Median fluorescence intensities normalized to a constitutively expressed mTurquoise2 reporter gene are given in arbitrary units (a.u.). Higher median fluorescence intensities correspond to induction with higher doxycycline concentrations.
Figure 2
Figure 2
Gene expression noise is decoupled from mean expression by orthogonal control of transcription rate and mRNA degradation rate. (a) Median-noise landscape for noise-tuner expression at different combinations of doxycycline and theophylline concentrations. Colors indicate noise; lines indicate identical median normalized fluorescence, interpolated from the measured data (“isomedian lines”). At intermediate expression range, noise levels can be adjusted for a given median expression by use of different combinations of doxycycline and theophylline. The measured robust CV decreases when a median is reached with higher transcription rates and correspondingly lower mRNA degradation rates. (b) Example histograms of populations with similar median expression but different noise settings. Populations with higher mNeongreen transcription and mRNA degradation rates (red) display lower heterogeneity than populations with lower rates (blue). Insets indicate the compared populations from (a). (c) Deconvolution of measured noise. For populations shown in (b), noise is plotted against the radius of a circular gate around the median forward (FSC) and side scatter (SSC) values. Decreasing gate sizes lead to more homogeneous FCS-SSC populations, effectively filtering out the extrinsic component of the observed noise. In contrast to autofluorescence-subtracted raw data (solid lines), noise of autofluorescence-subtracted data normalized to a constitutively expressed mTurquoise2 reporter (dashed lines) is virtually independent of the gate size.
Figure 3
Figure 3
A simple mathematical model can reproduce the observed median-noise relationship. (a) Schematic depiction of the employed model. The protein number within single cells is defined by the synthesis and degradation rates of mRNA and protein (v0 and d0 and v1 and d1, respectively), which are used to calculate the coefficients a and b. (b) Simulated protein number and CV as a function of a and b. Lines indicate same mean protein number, colors indicate CV. Scaling of a and b axes was chosen to correspond to experimentally observed expression intensities with different doxycycline and theophylline concentrations. (c) Comparison of protein number distributions for simulated populations with different values for a and b. Insets indicate the compared populations from (b). Populations that reach a certain mean protein number with higher a and lower b show a smaller variance in protein number.
Figure 4
Figure 4
Calibration of noise in a signaling pathway component alters noise of pathway output. (a) Schematic depiction of the yeast pheromone signaling (mating) pathway. Yeast cells sense pheromones secreted by the opposite mating type via a G-protein coupled receptor. The signal is transduced through a mitogen-activated protein kinase (MAPK) cascade and ultimately induces the expression of pheromone response genes (PRG), such as the upstream negative feedback regulator Sst2. As a pathway activity readout, mNeongreen (mNG) driven by PRG promoter PFUS1 was genomically integrated. (b) Dose response curves of the pathway reporter. Blue and red indicate high noise and low noise condition of SST2 expression, respectively. Cells were stimulated for 180 min with different concentrations of pheromone. Lines indicate medians of normalized fluorescence, shaded areas show the corresponding median absolute deviations. Low noise (8.5 ng/μL doxycycline) and high noise (1 ng/μL doxycycline, 10 mM theophylline) conditions for SST2 expression were chosen in order to achieve similar pathway responses. (c) Pathway reporter noise at different pathway activity levels for high and low noise SST2 conditions. Points from left to right indicate stimulation with increasing pheromone concentrations as in (b). Over the whole pathway activity range, the population of yeast cells with low noise in SST2 expression displays a lower robust CV in the pathway reporter output than the population with high noise in SST2 expression. (d) Pathway information transmission is more precise with lower noise in SST2 expression. Mutual information between pheromone input and pathway output plotted against the pathway output for better comparison (see Methods for calculation). Precision is highest at intermediate reporter activity and overall higher with lower noise in Sst2. Points indicate stimulation with different pheromone concentrations as in (b) and (c).

Similar articles

See all similar articles

Cited by 1 article

References

    1. Elowitz M. B.; Levine A. j.; Siggia E. D.; Swain P. S. (2002) Stochastic Gene Expression in a Single Cell. Science 297, 1183–1186. 10.1126/science.1070919. - DOI - PubMed
    1. Shah P.; Ding Y.; Niemczyk M.; Kudla G.; Plotkin J. B. (2013) Rate-limiting steps in yeast protein translation. Cell 153, 1589–1601. 10.1016/j.cell.2013.05.049. - DOI - PMC - PubMed
    1. Chalancon G.; et al. (2012) Interplay between gene expression noise and regulatory network architecture. Trends Genet. 28, 221–232. 10.1016/j.tig.2012.01.006. - DOI - PMC - PubMed
    1. Newman J. R. S.; et al. (2006) Single-cell proteomic analysis of S. cerevisiae reveals the architecture of biological noise. Nature 441, 840–846. 10.1038/nature04785. - DOI - PubMed
    1. Wang Z.; Zhang J. (2011) Impact of gene expression noise on organismal fitness and the efficacy of natural selection. Proc. Natl. Acad. Sci. U. S. A. 108, E67–E76. 10.1073/pnas.1100059108. - DOI - PMC - PubMed

Publication types

MeSH terms

Feedback