The kinetics of Snf1 nucleocytoplasmic shuttling are driven by carbon source availability

To understand how Snf1 mechanistically mediates glucose de-repression, we employed FRAP. Exponentially grown yeast cells expressing GFP-tagged Snf1 were exposed to YNB supplemented with either 2% glucose (sustaining a fermentative metabolism), 0.05% glucose or 2% glycerol (two conditions in which metabolism is mainly respirative) for at least 1 h before the onset of measurements. Fluorescently tagged Snf1 in the nucleus was bleached, and the subsequent recovery of fluorescence in the nucleus was observed. The obtained data were analyzed by fitting a single exponential model using a nonlinear mixed-effects (NLME) framework (SI data files 2-4 as well as methods section for details, data available via https://github.com/cvijoviclab/Snf1-localization ) 36. The diagnostic plots in Figure 1 show that the model has good predictive power for all conditions tested, although values at the later time points might be underestimated, as can be seen by the outliers in the observed data in comparison to the prediction intervals. The good fit is indicated by the weighted residuals clustering randomly around the x-axis (Figure 1, right panel), and when simulated the fitted model can reproduce the observed data (for detailed plot of observed FRAP curves and predicted values see supplementary information 2-4). Cell size and nucleus size measurements in the cells tested showed that the fraction of the bleached nucleus was unaffected by the conditions applied, indicated by a non-significant Kruskal Wallis test (cell area p = 0.72, nucleus area p = 0.13) and spot tests indicated that Snf1-GFP is functional (Figures S3-4).

–

FIGURE 1: Diagnosticplots of the NLME model used to predict population parameters. Left panel: visual predictive check to determine predictive power of the model. Datapoints were binned using least squares criteria and smoothed using linear interpolation. Solid line determines the observed data in the 10th,50th and 90th percentile. Gray/red areas are the prediction intervals for the respective percentile in which 90% of the simulated data lie, determined by 400 Monte Carlo simulations. Right panel: Scatterplots of individual residuals. Dots represent the individual weighted residuals over time represented by conditional mode of the distribution of residuals.

–

The kinetic coefficient (τ) for Snf1-GFP when cells were exposed to 2% glycerol indicates a faster nuclear-cytoplasmic shuttling in comparison to the cells exposed to 2% or 0.05% glucose (Table 1). Activating kinases are typically contained in the cytoplasm and Snf1 nuclear localization is dependent on both, its phosphorylation and on the interaction with Gal83 11. Faster recovery for cells exposed to glycerol indicates a mechanism that is independent on glucose concentration but rather on the presence of a fermentable carbon source. The parameters furthermore indicated a large immobile fraction. This could hint towards a large entity of bound Snf1-GFP, for example to the β-subunit Gal83.

To elucidate the steady-state accumulation of Snf1-GFP in the nucleus under these conditions, we compared the fraction of the total Snf1-GFP intensity in the nucleus by employing confocal microscopy. Snf1 steady-state nuclear fractions were comparable under high (2%) and low (0.05%) glucose concentrations (Figure 2A). A non-significant mean fold change of 1.06 (p-value = 0.5479), indicates that under steady-state conditions Snf1 localization was not modulated by glucose concentration.

–

FIGURE 2: Snf1 localization and Snf1 phosphorylation. (A) The nuclear-cytosolic ratio of Snf1-GFP was calculated by dividing the mean of the fluorescence of Snf1-GFP in the nucleus with the mean of the fluorescence in the cytosol. Horizontal lines indicate the mean, the boxplot has as lower and upper hinge respectively the 25thand 75th percentile and the whiskers denote the 95% confidence interval. Dots denote individual cells. (B) Mean of relative Snf1 phosphorylation by western blot quantification 5 minutes after upshift from ethanol to glucose, fructose or mannose, dots represent biological replicates.(C) Exemplary images of the western blot showing phosphorylated Snf1.

–

Together with results from the FRAP measurements, this could suggest that Snf1 nuclear localization is not strongly influenced by changes in concentration of glucose and that the relative difference in glucose concentration has a smaller effect than the type of carbon source itself.

Snf1 phosphorylation status correlates with Mig1 localization upon carbon source upshift

The data from FRAP experiments indicated that Snf1 localization and nucleo-cytoplasmic shuttling parameters depend on the type of carbon source rather than on concentration. We set out to find if this relationship was also present in the short-term phosphorylation pattern of Snf1 in different fermentable sugars at various concentrations. To evaluate the activation state of Snf1 during the shift-up, cells were grown in YNB containing ethanol until mid-exponential phase. Then the fermentable carbon source (glucose, fructose or mannose) at the indicated concentrations was added to the culture. Samples were taken after 5 minutes to measure the pT210-Snf1 level (pSnf1) (Figure 2B and C, Figure S2). When the fermentable carbon source was added to the cells, a reduction of Snf1 phosphorylation was observed, with a magnitude proportional to the concentration added. Glucose was the carbon source which reduced phosphorylation of Snf1 more strongly since it was reduced to 20% already at a concentration of 0.005% and reached maximal reduction at 1% concentration. Fructose and mannose present similarly, with an inhibition of 50% at 0.005%, and a maximal inhibition at 0.5% concentration (Figure 3A).

FIGURE 3: Mig1 nuclear localization after upshift. (A) Mig1 nuclear localization ratio in cells 5 minutes after the shift from ethanol to the indicated concentration of glucose, fructose, or mannose. Horizontal lines indicate the mean, the boxplot has as lower and upper hinge respectively the 25thand 75th percentile and the whiskers denote the 95% confidence interval. Dots denote individual cells. (B) Microscopic images of single cells, representative for the tested cells in panel B, 5 minutes after shift from ethanol to the indicated conditions. Mig1-GFP is displayed in cyan and Nrd1-mCherry in magenta. Brightness and contrast have been adjusted for better visibility.

–

With the same approach, the initial spatial Mig1 response towards different carbon sources was characterized through fluorescent time-lapse microscopy. Mig1 localization to the nucleus was observed at 0.005% glucose, while in upshift to mannose and fructose, Mig1 nuclear localization was only observed at concentrations above 0.05% (Figure 3A and Figure 3B for further exemplary images, Figure S1). These results suggest that the Mig1 nuclear import is more sensitive to glucose than to mannose and fructose, similarly, observed for Snf1 phosphorylation.

In order to better investigate the Snf1-Mig1 pathway, expression of two glucose repressed genes (i.e. SUC2 and HXK1) was analyzed by qPCR. Cells were grown in ethanol-containing YNB medium until mid-exponential phase. Cells were either left in ethanol, or glucose was added to the cultures at 0.05% or 4% final concentration. Samples were taken after 15 and 30 minutes to measure SUC2 and HXK1 expression (Figure 4). When the fermentable carbon source was added, the expression of both SUC2 and HXK1 was strongly reduced, similarly in the presence of low (0.05%) and high (4%) glucose concentration. This suggests that although glucose repression can be seen, expression of glucose repressed genes is not a sensitive readout to evaluate the dose-dependent Snf1 response to carbon sources. This could likely be the result of both transcriptional repression and rapid mRNA degradation that takes place upon glucose addition 37. The glucose concentration independent recovery time during FRAP further enforces this result by indicating that glucose concentration in the short term does not influence the nuclear import of Snf1 while it does influence the nuclear accumulation of Mig1.

–

FIGURE 4: Expression of HXK1 and SUC2 measured via qPCR, 15 minutes after shift to glucose. Data are shown as ratio change of gene expression. Control: baseline expression of cultures grown in ethanol at start of measurements. Dots represent biological replicates.

Strain maintenance and growth conditions

If not stated otherwise, yeast strains were grown at 30°C in synthetic complete medium, containing 1.7 g/l yeast nitrogen base, 5 g/l ammonium sulfate, 670 mg/l complete supplement mix here referred to as YNB with appropriate drop out where applicable; supplemented with carbon source as indicated by the specific experiments. All media components except for carbon source were obtained from Formedium. Strains used in this study are listed in Table 2.

To ensure functionality of Snf1-GFP constructs, we conducted a spot test on non-feasible carbon sources that would lead to a decrease in growth if Snf1 functionality is impaired (see Figure S3).

Fluorescence Recovery After Photobleaching (FRAP)

Strain BY4741 SNF1-GFP was grown in YNB supplemented with 2% glucose to exponential phase, OD/ml ≈ 0.3, and immobilized on an 8-well chambered coverglass (Ibidi) coated with poly-L-lysine (Sigma). Media were switched to YNB with either 2% glucose, 0.05% glucose or 2% glycerol and incubated at 30°C for 1 h before imaging to ensure adaptation to the new carbon source. At least 21 cells distributed over three separate experiments per condition were imaged on a LSM700 laser scanning confocal microscope (Zeiss) using a Plan-Apochromat 63x /1.4 oil immersion objective, pinhole set to 1 AU for the longest wavelength and kept constant over all channels. FRAP was performed on GFP-tagged Snf1, excitation 488 nm, emission 518 nm, 2x zoom with framesize 596×596, pixel size of 0.09µm and pixel dwell time of 0.9 µs. The cells were continuously imaged for 100 frames, and bleaching was done in 100 bursts at 100% at 488 nm after 5 pre-scans using a circular ROI covering the nucleus. To determine fluorescence decay by acquisition, a reference area of adjacent cells was captured, as well as a circular ROI for baseline determination. Determination of bleach area covering the nucleus was achieved by imaging the nuclear marker mCherry-tagged Nrd1 (excitation: 555 nm, emission: 572 nm) and drawing a circular ROI around the determined area.

For image processing, the average fluorescence intensity as well as diameter of the imaged areas was extracted from the time-lapse image series for each image using the software ZEN2012 Black (Zeiss). Given the values for background intensity, the intensity for the nuclear region, as well as a reference region containing adjacent cells in the same frame, the measured intensities were processed by first removing background signal and normalizing prebleach phase to 1 by calculating the ratio of measured intensity at a given timepoint over the mean of intensities measured in the frames pre-bleach for both, the reference and the bleached region. Measurements that were aberrating due to cell movement were removed from the dataset after visual inspection of the acquired time series. Correction for acquisition bleaching was done in RStudio, version 1.4.1106 48, by calculating the median of the normalized reference regions and fitting the curve according to the following equation:

With I_0ref=0 representing the baseline as fluorophores become bleached completely and data have been corrected for background noise.

With the parameters A_ref and τ_ref estimated by nonlinear least squares regression, measured FRAP intensities were interpolated and further used for nonlinear mixed effect modeling.

Data available at: 10.6084/m9.figshare.25397809 (FRAP movies) and 10.6084/m9.figshare.25397803 (FRAP Raw data).

Non-linear mixed effect model

Non-linear mixed-effects modelling is typically used for longitudinal data exhibiting both within- and in-between-subject variability 49. This method has been widely used in pharmacokinetic and pharmacodynamic studies 50, 51, but in recent years it is utilized in single-cell time-lapse data and facilitating our understanding of cell-to-cell variability 32, 36, 52, 53, 54. When analyzing fluorescence measurements of a tagged protein in single cells over time, the observed intensity will differ between measurements even if the cells are in a steady-state, due to measurement error. Moreover, owing to extrinsic variability, imaged fluorophores within cells will display different intensity levels. Using a mixed-effects framework, the observed cell-to-cell variability can be accounted for in the analysis by letting the rate parameters vary between cells according to a probability distribution. Furthermore, a mixed-effects framework allows the assessment of potential correlations between parameters in different cells.

To analyze dynamic measurements for Snf1-GFP from the FRAP experiment, a non-linear mixed-effect regression was implemented and simulated in Monolix (version 2023R1) 55, 56. The data, project files and models are available at the following github repository: https://github.com/cvijoviclab/Snf1-localization .

Assuming reaction dominant recovery, FRAP curves were fitted to a single exponential equation, summarizing experiments across conditions tested:

I₀ represents the degree of bleaching, A is the mobility constant, τ is the kinetic constant of bound/unbound species and t is the time after bleach.

Speed of recovery can be calculated by determining the recovery half max time where 50% of the plateau of estimated fluorescence recovery is reached: