Transcription factor localization dynamics and DNA binding drive distinct promoter interpretations

Abstract

Environmental information may be encoded in the temporal dynamics of transcription factor (TF) activation and subsequently decoded by gene promoters to enact stimulus-specific gene expression programs. Previous studies of this behavior focused on the encoding and decoding of information in TF nuclear localization dynamics, yet cells control the activity of TFs in myriad ways, including by regulating their ability to bind DNA. Here, we use light-controlled mutants of the yeast TF Msn2 as a model system to investigate how promoter decoding of TF localization dynamics is affected by changes in the ability of the TF to bind DNA. We find that yeast promoters directly decode the light-controlled localization dynamics of Msn2 and that the effects of changing Msn2 affinity on that decoding behavior are highly promoter dependent, illustrating how cells could regulate TF localization dynamics and DNA binding in concert for improved control of gene expression.

Keywords: CP: Molecular biology; gene expression; optogenetics; promoter decoding; signaling dynamics; transcription factors.

Figure 1.. Optogenetic control of Msn2 localization using CLASP

(A) A schematic of the Msn2-CLASP system and experiments in which time-varying light doses drove corresponding patterns of Msn2-CLASP nuclear localization and gene expression. (B) (Left) Micrographs showing Msn2-CLASP and Msn2*-CLASP localizing to the nucleus following 255 a.u. blue light (scale bar, 10 μm). Light-induced nuclear localization (middle) and reporter gene induction (right) by Msn2-CLASP (dashed lines) and Msn2*-CLASP (solid lines) following a 10 min pulse of 128 a.u. blue light (depicted by blue boxes). Data represent mean ± standard deviation (SD) for three biological replicates, each with ≥47 cells. (C) Schematic showing Msn2 functional domains—the transactivation domain (TAD), nuclear export signal (NES), nuclear localization signal (NLS), and zinc finger DNA-binding domain (DBD)—and residues mutated for improved optogenetic control. (D) (Left) Absolute nuclear Msn2-CLASP, Msn2*-CLASP, and Msn2-dCLASP following 15 min doses of blue light with intensities ranging from 0 to 255 a.u. (i.e., irradiances of 0–405 μW/cm², see Figure S1C). (Right) Nuclear Msn2-CLASP and Msn2*-CLASP in response to 128 a.u. blue light with varying degrees of pulse width modulation (PWM). Data represent mean ± SD of three biological replicates, each with ≥43 cells. Measurements were acquired by fluorescence microscopy.

Figure 2.. Light-sweep experiments and model-based characterization of Msn2 target genes

(A) Light-sweep experiments probe how promoters decode the nuclear localization dynamics of Msn2*. Each row represents a light program that drove Msn2 localization (left) and reporter expression (right). Msn2* localization measurements were pooled over many experiments and represent thousands of cells per condition. Expression measurements were normalized to maximum expression per reporter across all conditions and represent mean of ~100–600 cells per condition from three biological replicates. For comparison, figure layout is adapted from Hansen and O’Shea. See also Figure S2A. (B) Schematic of gene expression model. See STAR Methods for corresponding system of ordinary differential equations (ODEs) and Figure S2A for plots of fits. (C) Categorization of promoters based on how they decoded single pulses of nuclear Msn2*. (D) (Top) Predicted values of Kⁿ obtained from the gene expression model for top 0.1% and bottom 99.9% of parameter sets; two-sample Kolmogorov-Smirnov (KS) tests showed differences between these distributions (p = 0.0015 for RTN2 and p < 10⁻²⁴ for HSP12). (Bottom) Predicted affinity of each promoter for Msn2*. Data represent mean ±95% confidence interval (CI) for top 0.1% of parameter sets per promoter.

Figure 3.. Simulations demonstrate that changes in TF affinity for DNA have a strong effect on low-sensitivity promoters and a weak effect on high-sensitivity promoters

(A) Simulated maximum expression of hypothetical promoters for a 50 min 100% amplitude pulse of nuclear TF. Each kinetic parameter (k₁, d₁, k₂, d₂, and k₃) was varied, while other kinetic parameters were fixed to one (with n = 1). Simulations were repeated for varying values of K. Expression was normalized to case where K = 1 (bottom square of each column). (B) (Top) Modified gene expression model in which changes in Msn2 affinity for DNA are modeled by scaling K by α. Omitted model steps are shown in Figure 2B. (Bottom) Maximum predicted expression of hypothetical promoters with a range of baseline affinities (K) for a 50 min 100% amplitude pulse of TF (other kinetic parameters were fixed to one). To model the additional effect of a 2-fold increase or decrease in TF binding affinity, simulations were repeated while scaling K by α = 0.5 or 2, respectively. Expression of each hypothetical promoter was normalized to case where α = 1, which represents the affinity between the promoter and the wild-type (WT) TF. (C) Predicted reporter expression following a 50 min 100% amplitude pulse of nuclear Msn2* (shown in blue, assumes α = 1). To model the additional effect of a 2-fold increase or decrease in the affinity of Msn2 for DNA, simulations were repeated while scaling K by α = 0.5 or 2, respectively (as depicted in B, top). Data represent mean ±95% CI of predicted expression for top 10 parameter sets for each reporter. (D) Measured reporter expression following a 50 min 100% amplitude pulse of nuclear localization for Msn2* and high- and low-affinity mutants Msn2(A)* andMsn2(T)*. Data represent mean ± SD for three biological replicates. (E) Maximum predicted reporter expression for a 10 min 100% amplitude pulse of nuclear Msn2* (short), a 50 min 25% amplitude pulse (low), a 50 min 100%amplitude pulse (high; time courses shown in C), and six 5 min pulses with 100% amplitude and 5 min interpulse durations (pulsed). As above, simulations were repeated while scaling α to capture changes in Msn2 affinity. Data represent mean of maximum expression predicted for top 10 parameter sets per reporter. (Bottom) Predicted maximum reporter expression at each condition normalized to predicted expression for Msn2* (where α = 1).

Figure 4.. Light-sweep experiments with Msn2 DBD mutants reveal the differential effects of changing TF affinity on the signal decoding behaviors of promoters

Each row corresponds to a light program, which drove Msn2 localization (left) and subsequent gene expression (right). Msn2 localization measurements were pooled over many experiments and represent thousands of cells per condition. Gene expression measurements were normalized to the maximum expression level per reporter across all conditions and Msn2 DBD mutants and represent mean of ~100–600 cells from three biological replicates. See also Figure S2A.

Figure 5.. Analyzing the decoding behavior of high- and low-sensitivity promoters

(A) Maximum reporter expression following 50 min pulses of each Msn2 DBD mutant with amplitudes ranging from 0% to 100% (left) or 100% amplitude pulses with durations varying from 0 to 50 min (right). Data represent mean ± SD for three biological replicates. Gray circles denote conditions referenced in the main text. (B) Maximum reporter expression for pulsatile versus continuous doses of each Msn2 mutant. Circles represent maximum expression for 100% amplitude continuous pulses of nuclear Msn2 with durations of 0, 10, 30, and 50 min. Triangles represent maximum expression following 0, 2, 6, or 10 5 min pulses of nuclear Msn2 with 100% amplitude and 5 min interpulse durations. Data represent mean ± SD for three biological replicates. Solid and dashed lines show best fit lines for continuous and pulses conditions, respectively; shaded regions show 95% CI of best fit lines. Gray circles denote conditions referenced in main text. (C) Reporter slope ratios. Data represent mean ± SD of slope ratio per promoter and Msn2 DBD mutant. A two-way ANOVA test revealed significant differences in slope ratio between promoters and Msn2 DBD mutants, though differences between promoters were generally larger in magnitude. (D) Reporter expression for cells exposed to no light (dark), a 30 min 100% amplitude pulse of nuclear Msn2 (continuous), or six 5 min 100% amplitude pulses with 5 min interpulse durations (pulsed). Histograms represent single-cell fluorescence measurements for three biological replicates between 115 and 125 min. Dashed line represents the threshold above which RTN2 was considered active (calculated as 99^th percentile RTN2 level in dark).

Figure 6.. Changing Msn2 affinity can alter the ability of promoters to discriminate between stresses

Fluorescent reporter expression following 2 h of glucose starvation or hyperosmotic shock. All Msn2 mutants were expressed in the dCLASP system and had no mutations outside the DBD. Data represent mean ± SD for at least three biological replicates. See also Figure S6A.

Figure 7.. Concerted regulation of TF affinity and dynamics may facilitate improved control of gene expression

A TF that can transition between high- and low-affinity binding modes and continuous and pulsatile nuclear localization patterns could tune the expression of low-sensitivity genes while maintaining robust activation of high-sensitivity genes. Schematic is based on single-cell expression measurements of Figure 4C, where a pulsed dose of Msn2(T)* mutant minimally activated the low-sensitivity gene RTN2, while a continuous dose of Msn2(T)* weakly activated RTN2. In contrast, a pulsed dose of Msn2(A)* mutant moderately activated RTN2 and a continuous dose maximized RTN2 activation. Meanwhile, a dose of either Msn2(T)* or Msn2(A)* could robustly activate the high-sensitivity gene HSP12.