Negative feedback that improves information transmission in yeast signalling

Abstract

Haploid Saccharomyces cerevisiae yeast cells use a prototypic cell signalling system to transmit information about the extracellular concentration of mating pheromone secreted by potential mating partners. The ability of cells to respond distinguishably to different pheromone concentrations depends on how much information about pheromone concentration the system can transmit. Here we show that the mitogen-activated protein kinase Fus3 mediates fast-acting negative feedback that adjusts the dose response of the downstream system response to match the dose response of receptor-ligand binding. This 'dose-response alignment', defined by a linear relationship between receptor occupancy and downstream response, can improve the fidelity of information transmission by making downstream responses corresponding to different receptor occupancies more distinguishable and reducing amplification of stochastic noise during signal transmission. We also show that one target of the feedback is a previously uncharacterized signal-promoting function of the regulator of G-protein signalling protein Sst2. Our work suggests that negative feedback is a general mechanism used in signalling systems to align dose responses and thereby increase the fidelity of information transmission.

Figure 1. The pheromone response system

Proteins are indicated by labeled ovals, translocation by dotted lines, protein activation by arrows, inhibition by T-bar arrows, and protein association by double-headed dashed arrows. Pheromone binding by receptor Ste2 causes dissociation of the heterotrimeric G-protein (1) into G_α subunit (Gpa1) and the G_βγ dimer (Ste4-Ste18). GTP-activating protein (GAP) function of the Regulator of G-protein Signaling (RGS)-protein Sst2 promotes re-association of Gpa1 with Ste4-Ste18. Upon dissociation of the G-protein, Ste4 helps recruit the MAP kinase scaffold Ste5 to the membrane (2). Ste5 recruitment activates of the MAP kinase cascade, in which Ste20, Ste11, Ste7, and the MAP kinases Fus3 and Kss1 phosphorylate one another in sequence. Phosphorylated Fus3 (3) translocates to the nucleus and phosphorylates Dig1 and Ste12, eliminating Dig1 repression of Ste12, a transcriptional activator (4). Ste12 activates transcription of pheromone responsive genes (PRGs) (5,6).

Figure 2. DoRA increases response distinguishability

a) Dose-responses of receptor occupancy (calculated from reported receptor-pheromone binding affinity measurements 43,44) and reporter gene expression output corrected for known sources of cell-to-cell variation (pathway output P 5), align closely. b) Relationship between receptor occupancy and downstream response (from panel a) is essentially linear. Evenly-distributed receptor occupancies (20 %, 40 %, 60 %, and 80 %, red vertical lines) corresponded to evenly-spaced downstream responses (blue horizontal lines). c) Example of dose-response misalignment, in which the downstream output is 20-fold more sensitive than that in panel a (i.e., the EC50 is reduced 20-fold). d) Dose-response misalignment makes transfer function non-linear, which compresses the downstream responses (blue horizontal lines) corresponding to the majority of receptor occupancies (red vertical lines), reducing downstream response distinguishability. e,f) Dose-response misalignment results in noise amplification. Receptor occupancy (red vertical line) with some noise (pink spread) yields downstream responses (horizontal dotted lines) with associated noise (spread around horizontal blue bars). In system with DoRA (e), linear transfer function yields less noise in downstream response than in system with misaligned dose-responses and non-linear transfer function (f).

Figure 3. Initial system dynamics indicate negative feedback

a) Loss of G-protein FRET. Corrected median (+/− SE) loss of Gpa1-Ste18 FRET values (relative to maximum change measured in pheromone-stimulated cells; see Fig. S2) in RY2062b cells stimulated with pheromone (open purple triangles; n=262) quickly peaked and declined to a plateau relative to unstimulated cells (black circles; n=143). b) YFP-Ste5 recruitment. Corrected median (+/− SE) YFP-Ste5 membrane recruitment (relative to maximum change measured in pheromone-stimulated cells; see Fig. S3e) in RY2013 cells stimulated with pheromone stimulation (open cyan triangles, n=361) quickly peaked and declined to a plateau compared to unstimulated cells (black circles, n=223). c) Fus3 activity. Mean ratios (+/− S.E., n=3–5) of activated (phospho-Y180 and phospho-T182) Fus3 to total Fus3, normalized to the peak measured ratio (see Fig. S4 for representative immunoblot images). Fus3 activity levels peaked 2.5 minutes after pheromone stimulation and declined to a plateau within 5 minutes of stimulation. Total Fus3 levels, compared to levels of non-pheromone regulated proteins GAPDH and PGK1, remained constant over this time period (data not shown). New protein synthesis is not required for the observed peak and decline (Fig. S4c). d) Loss of Dig1-Ste12 FRET. Median (+/− SE) loss of Dig1-Ste12 FRET (scaled to minimum and maximum values for measured in pheromone-stimulated cells; for raw values, see Fig. S7) in RY1130b cells peaked about 3 minutes following pheromone stimulation, and then declined to a plateau (open red triangles, n=246) relative to unstimulated cells (black circles, n=138). e) FUS1 mRNA. Average ratio (high/low values indicated) of FUS1 mRNA probe band intensity to loading control (ACT1 mRNA probe band intensity) after pheromone stimulation (filled squares, n=2) (See Fig S8 for raw image). f) Composite timing plot shows persistent peak-and-decline toward a plateau for all system responses, suggesting action of negative feedback. YFP-Ste5 recruitment and Dig1/Ste12 FRET (from panels b and d) were smoothed using a moving window of five data points.

Figure 4. Fus3 mediates negative feedback

Values scaled to peak signal measured in cells stimulated with only pheromone. Error bars indicate +/− SE. For all panels: P, stimulated with 100 nM pheromone; P+I, stimulated with 100 nM pheromone + 10 μM 1-NM-PP1; I, 10 μM 1-NM-PP1; U, untreated. a) Fus3 mediates negative feedback. In fus3-as2 cells (RY1134b), mean (n=4) Fus3 phosphorylation peaked and declined, as in FUS3 cells (Fig. 3b), after pheromone stimulation (black circles), but did not decline when we stimulated cells simultaneously with Fus3-as2 inhibitor (open green circles). Treating cells with only inhibitor (gray circles) caused the signal to slowly rise, indicating cells actively regulate basal signal level. b) Kss1 does not mediate negative feedback. In kss1-as2 cells (RY1133b), mean (n=4) Fus3 phosphorylation in pheromone-stimulated cells without (black circles) or with (open green circles) simultaneous treatment with Kss1-as2 inhibitor were identical. Treating cells with only inhibitor caused no significant increase in Fus3 phosphorylation (gray circles). c) Fus3-mediated feedback acts on or upstream of Ste5 membrane recruitment. In fus3-as2 cells (RY2013), median YFP-Ste5 membrane recruitment peaked and declined after pheromone stimulation (filled circles; n=361), but did not decline after simultaneous treatment with Fus3-as2 inhibitor (n=196, open blue squares). There was no relative Ste5 recruitment in cells treated with inhibitor alone (open diamonds; n=134) or in completely untreated cells (gray triangles; n>100). The small increase in Fus3 phosphorylation measured in cells treated with inhibitor only (black circles in Fig. 4a) suggests that additional Fus3-independent mechanisms maintain low basal levels of Ste5 recruitment. d) Fus3-mediated negative feedback acts downstream of G-protein dissociation. In fus3-as2 cells (RY2062b, derived from TMY101 6), median Gpa1-Ste18 loss of FRET peaked and declined in pheromone-stimulated cells (filled circles; n=262) with the same dynamics as in pheromone-stimulated cells simultaneously treated with Fus3-as2 inhibitor (open purple squares; n=263). Unstimulated cells in the presence (open diamonds; n=229) or absence (gray triangles n=143) of inhibitor showed no loss of Gpa1-Ste18 FRET. e) One target of Fus3-mediated negative feedback is a novel Sst2-dependent increases in YFP-Ste5 recruitment. Median YFP-Ste5 membrane recruitment in pheromone-stimulated fus3-as2 Δsst2 cells (RY2024) peaked and declined both in the absence (black circles; n=188) or presence (open cyan squares; n=300) of Fus3-as2 inhibitor, similar to SST2 cells with active Fus3 (black circles, panel (c)). Unstimulated cells, gray triangles (n>100). f) Mutation of predicted Fus3/MAPK phosphorylation site in Sst2 DEP1 domain eliminated Sst2 promotion of YFP-Ste5 recruitment. Median (+/− SE) YFP-Ste5 membrane recruitment in pheromone-stimulated sst2-T134A (RY2066) cells peaked and declined both in the absence (filled circles; n>150) and presence (open cyan squares; n>150) of inhibitor, similar to Δsst2 cells (Fig. 4e).

Figure 5. DoRA requires Fus3-mediated negative feedback

a) Model of negative feedback regulation of Ste5 membrane recruitment. Sst2 promotes (thick red arrow) Ste5 recruitment to the membrane (blue dashed arrow), and Fus3 negatively regulates this signal promotion (thick T-bar arrow). b) Fus3 inhibition disrupts dose-response alignment. In pheromone-stimulated fus3-as2 (RY2052b) cells, inhibition of Fus3 kinase activity (open green circles) reduced the sensitivity (EC50) of the dose response of mean Fus3 phosphorylation (+/− SE; n=3–4) relative to cells not treated with inhibitor (black filled circles). Fus3 phosphorylation measured after 15 minutes of pheromone stimulation, after the signal reaches the dose-dependent plateau (see Fig. 3c). Black lines show fits to Hill functions. Fus3 inhibition reduced the EC50 of the dose-response by greater than 20 fold without affecting the gradedness (cooperativity) of the average response (see Supplementary Information for details).