Using the AKAR3-EV biosensor to assess Sch9p- and PKA-signalling in budding yeast

Abstract Budding yeast uses the TORC1-Sch9p and cAMP-PKA signalling pathways to regulate adaptations to changing nutrient environments. Dynamic and single-cell measurements of the activity of these cascades will improve our understanding of the cellular adaptation of yeast. Here, we employed the AKAR3-EV biosensor developed for mammalian cells to measure the cellular phosphorylation status determined by Sch9p and PKA activity in budding yeast. Using various mutant strains and inhibitors, we show that AKAR3-EV measures the Sch9p- and PKA-dependent phosphorylation status in intact yeast cells. At the single-cell level, we found that the phosphorylation responses are homogenous for glucose, sucrose, and fructose, but heterogeneous for mannose. Cells that start to grow after a transition to mannose correspond to higher normalized Förster resonance energy transfer (FRET) levels, in line with the involvement of Sch9p and PKA pathways to stimulate growth-related processes. The Sch9p and PKA pathways have a relatively high affinity for glucose (K0.5 of 0.24 mM) under glucose-derepressed conditions. Lastly, steady-state FRET levels of AKAR3-EV seem to be independent of growth rates, suggesting that Sch9p- and PKA-dependent phosphorylation activities are transient responses to nutrient transitions. We believe that the AKAR3-EV sensor is an excellent addition to the biosensor arsenal for illuminating cellular adaptation in single yeast cells.


Introduction
One universal aspect of life is change, and the ability to adapt to it is a major determinant of r epr oductiv e success. For the unicellular organism Saccharomyces cerevisiae (budding or baker's yeast), there is no exception. In the wild, this yeast lives on fruit and tree bark, wher e it endur es feast-famine cycles (Liti 2015, Jouhten et al. 2016. In industry, domesticated yeast also experiences changing conditions when used in large-scale fermenters with inoculation transitions and poor stirring (Lara et al. 2006, Wehrs et al. 2019. For bud ding yeast, n utrient availability is a major environmental parameter that sets the investment in metabolism, stress resistance, and pr olifer ation (Br oac h 2012, Conr ad et al. 2014. Understanding the logic of these circuitries is a k e y challenge in cell biology . Also for industry , contr ol ov er this r esponse , e .g. by selection of certain pr eferr ed subpopulation or r emov al of undesired populations, could incr ease pr oduction efficiencies (Xiao et al. 2016 ). Nutrient adaptations can be captured on population level using bulk assa ys , but these methods ma y mask the true r esponses pr oduced by the intracellular signalling circuits inside single cells. T hus , a mor e pr ofound c har acterization of the cellular ada ptation of budding yeast to nutrient changes at the single-cell level is highly desired.
Regar ding carbon sour ces, fruit sugars are preferred by yeast, and ther efor e it has de v eloped v arious pathways to sense and ada pt to c hanges in the av ailability of these substrates (Rolland et al. 2002, Santangelo 2006, Rødkaer and Faergeman 2014. The cAMP-PKA pathway is one of the major signalling cascades that get activated when cells encounter fermentable sugars (Eraso and Gancedo 1985, Botman et al. 2021. cAMP production is activated via two routes (Casperson et al. 1985, Kataoka et al. 1985, Broek et al. 1987, Munder and Küntzel 1989, Engelberg et al. 1990, van Aelst et al. 1990, van Aelst et al. 1991, Pardo et al. 1993, Colombo et al. 1998, Yun et al. 1998, Kraakman et al. 1999, Rolland et al. 2000, Rolland et al. 2001, Lemaire et al. 2004, Kim et al. 2013: via import and metabolism of sugars and via extracellular sensing of glucose and sucrose by the G-protein coupled receptor Gpr1p. These two inputs give a transient increase in cAMP, which causes the dissociation of Bcy1p (a PKA regulator) from the PKA subunits Tpk1-3. This finally results in an increase of PKA activity; PKA is a major effector kinase in yeast, accounting for 75%-90% of the cellular changes during a transition from a glucose-derepressed (respiratory) state to a fermentable glucose-r epr essed (fermentativ e) state (The v elein 1994 , Rolland et al. 2002, Winderic kx et al. 2003, Santangelo 2006, Zaman et al. 2009, Br oac h 2012 ). The evoked transition gives a radical change in yeast physiology; cells change their metabolism from respiratory to fully fermentativ e, r epr ess metabolic pathways for other carbon sources, decr ease their str ess r esistance, and make lar ge inv estments in ribosomal biogenesis.
T he T ORC1-Sch9 cascade is a second major signalling cascade in yeast cells (Crauwels et al. 1997, Roosen et al. 2004, Slattery et al. 2008, v an Zeebr oec k et al. 2021. In contr ast to PKA, whic h gets activated by mostly fermentable sugars, Sch9p is activated by TORC1p when a complete palette of nutrients for growth (such as amino acids , nitrogen, phosphate , and a fermentable carbon source) is available (Crauwels et al. 1997, Conrad et al. 2017, van Zeebr oec k et al. 2021. Although the two pathways can operate independently (Zurita-Martinez and Cardenas 2005 ), they have man y positiv e inter actions and can r escue eac h other's activities, pr obabl y via the lar ge ov erla p in their tar gets: Sc h9 and PKA pr oteins both phosphorylate the RRxT motif (Reinders et al. 1998, Plank et al. 2020, Plank 2022. Activ ation of Sc h9p activates 90% of the genes that PKA also activates (Zaman et al. 2009 ), making Sch9p as important as PKA for proper cellular decision making in yeast.
Curr entl y, PKA and Sc h9p activities ar e difficult to measur e in (single) cells: the most common method is measuring the activity of trehalase, a PKA and Sch9p target, or using kemptide as a substrate (Crauwels et al. 1997, Roosen et al. 2004 ). Howe v er, these bulk assays lack single-cell information and show only static activity le v els of PKA and Sc h9p activity. Studies suggest that the PKA activity in yeast is more a transient phenomenon activated during (mostly) sugar transitions and that the TORC1-Sch9p axis dictates the steady-state growth mode of a yeast cell (Crauwels et al. 1997. Dynamic readouts are needed to substantiate these interesting hypotheses . Moreo ver, single-cell dynamics allows to test for heterogeneity during nutrient transitions-a phenomenon that we did not find for cAMP dynamics (Botman et al. 2021 ), but it is unknown if heterogeneity exists more downstream. Furthermore, the sensitivity of Sch9p and PKA activity with respect to glucose remains to be c har acterized. Finally, the basal RRxT phosphorylation status of the cell at v arious gr owth r ates is also poorl y c har acterized.
Here, we implemented and tested the mammalian PKA sensor AKAR3-EV in yeast to provide a tool that can help to enlarge our understanding of PKA and Sch9 signalling. The sensor allo w ed us to measure the single-cell dynamics of the PKA and Sch9p phosphorylation status in a robust and accurate manner. We found lar ge heter ogeneous r esponses of yeast cells for some nutrient tr ansitions, whic h we did not pr e viousl y find in cAMP dynamics. The detected heterogeneity potentially affects the overall cellular state since these two kinases constitute the vast majority of the cellular transition during a transition to a fermentable carbon source . Furthermore , our data implies that the phosphorylation status is not related to growth rate. How the two kinases regulate the cellular transitions at a single-cell le v el can be studied in more depth using this sensor.

Sensor construction
AKAR3EV with YPET-eCFP as a FRET pair was kindl y pr ovided by Dr. Aoki (Komatsu et al. 2011 ). The sensor was amplified using KOD One™ PCR Master Mix (To y obo, Osaka, J apan) with 5 -A TGCT A GCA CGGA GCTCA CTGAA TTCGGCA TGGTGAG-3 and 5 -A TGGA TCCA CGGTCGA CA CTTT AA TC CA GA GTCA GGCG-3 as forw ar d and r e v erse primers, r espectiv el y. Next, the PCR pr oduct and the pDRF1-GW plasmid were digested using BamHI-HF and NheI-HF (New England Biolabs, Ipswich, MA, USA), and the PCR product was ligated into the plasmid using T4 DNA ligase (New England Biolabs), which yield pDRF1-GW AKAR3-EV, containing the AKAR3-EV sensor under PMA1 promotor expression.
AKAR3-EV-NR was constructed by performing a PCR with KOD One™ PCR Master Mix on pDRF1-GW AKAR3-EV with 5 -T A TTCCGGA TTGA GGCGCGCGGCCCTGGTTGA CGGCGGCCGCATG  GTGA GCAA GGGC-3  as  a  forw ar d  primer  and  5 -A TGGA TCCA CGGTCGA CA CTTT AA TCCA GA GTCA GGCG-3 as a  r e v erse primer. The PCR product and pDRF1-GW AKAR3-EV  were digested using Kpn2I and XbaI (Thermo Fisher Scientific,  Waltham, MA, USA). Afterw ar ds, the PCR product was ligated in the digested pDRF1-GW AKAR3-EV, replacing the sensor domain and eCFP with the non-responding (T506A) sensor domain and eCFP.
pDRF1-GW eCFP w as made b y performing a PCR with KOD One™ PCR Master Mix on AKAR3-EV with FW primer 5 -A TGCT AGCA TGGTGAGCAAGGGCG-3 and RV primer 5 -TAGCG-GCCGCTT ACTTGT AC AGCTCGTCC ATGCCG -3 , after which the PCR product and pDRF1-GW were digested using NheI-HF and NotI-HF (New England Biolabs). Finally, the PCR product was ligated into pDRF1-GW using T4 DNA ligase.

Yeast tr ansforma tion
Strains used in this study are listed in Table 1 . Strains were transformed by resuspending yeast cells from either a YPAD plate or a selective plate in a tr ansformation mixtur e containing 240 μL PEG 3350 (50% w/v), 40 μL 1 M LiAC, 10 μL Salmon Sperm DNA (10 mg/mL, Sigma-Aldrich, Stl. Louis, MO, USA), and 500-1000 ng plasmid DNA. Next, cells were incubated for 15-20 minutes at 42 • C. Afterw ar ds, the cells w ere centrifuged at 13 000 g for 30 seconds , the supernatant was remo ved, and 150 μL w ater w as added. Finally, the cells w ere plated on selective plates.
For the experiments that involved S25-31C, cells were grown on 1x YNB medium containing 1% ethanol, 5 mM glucose (Boom BV, Meppel, The Netherlands), 20 mg/L adenine hemisulfate, 20 mg/L L-tryptophan, 20 mg/L L-histidine, and 60 mg/L L-leucine until glucose was exhausted. Next, cells were k e pt on this medium for 2 more da ys , after which they were visualized under the microscope.

Concanavalin (ConA) coverslips
ConA cov erslips wer e made as described by Hansen et al. ( 2015 ). To pr epar e the cov erslips, the ConA was diluted to 200 μg/mL and put on co verslips . T he co v erslips wer e dried ov ernight in a 6-well plate.

Rapamycin experiments
Cells were grown as described and incubated with 200 nM rapamycin or a solvent (100% ethanol) for at least 2 h, after which the perturbations (addition of 10 mM glucose) were performed.

Microscopy analysis
Cells were segmented using an in-house script. In brief, this script stabilizes any drift using the image stabilizer plugin (Li 2008 ). Next, bac kgr ound corr ection was performed, and cells were seg-mented using the Weka Segmentation plugin (Ar ganda-Carr er as et al. 2017b , Ar ganda-Carr er as et al. 2017a ), and the mean fluorescence for each cell was calculated for each frame . T he resultant text files wer e anal ysed using R 4.1.3. For all cells, 40% bleedtrough correction was performed and the FRET ratio (i.e. bleedtroughcorrected YFP divided by the CFP signal) was calculated. Finally, baseline normalization was performed for time-lapse data. pHluorin ratios were calculated by dividing the fluorescence at 380-420 nm excitation over the fluorescence at 460-500 nm. For the dose-response fit, the final FRET le v els (i.e. the mean FRET value of the last 3 frames) after the glucose additions were fitted (using the nls function in R) against the final glucose concentration according to equation 1 , with [glucose] as the glucose concentration in mM, max as the maximal change in normalized FRET, and K 0.5 as the glucose concentration giving 50% of the maximal response.
Clustering was performed using the factoextra package in R. The optimal amount of clusters was determined by eye.
For the long-term ethanol to mannose tr ansition, micr oscopy ima ges wer e segmented with a convolutional neur al network with customized weights (P ac hitariu and Stringer 2022 ). Frame-to- frame association of segmented objects was done through maxim um matc hing based on inv erse centr oid distance. Fluor escence bac kgr ound was estimated fr om fluor escence ima ges masked with the dilated segmentation images using the Background2D class from the photutils python package (Bradley et al. 2022 ). FRET r atios wer e calculated as the bleedthr ough-corr ected YFP signal, divided by the CFP signal. For growth type classification, cells that were in or that r eac hed G2-phase (i.e. cells that had or did get a bud) during the experiment (between 2 and 6 h after the transition) wer e manuall y identified and classified as growing if there was a perceptible increase in bud volume, and as non-growing if there was no such perceptible increase. Analysis was restricted to cells that were present at the start of the experiment.

Flow cytometry
Cells wer e gr own as described with YNB medium containing either 1% ethanol, 100 mM glucose, 100 mM galactose, or 100 mM mannose. Next, samples were measured using a CytoFLEX S Flow Cytometer (Beckman Coulter, Brea, CA, USA). Cells were excited using a 405 and a 488 nm laser, and emission fluorescence was passed through a 470/20 and 525/40 nm filter and recorded by av alanc he photodiodes. Ev ents with a saturating forw ar d or side scatter wer e filter ed, after whic h the median fluor escence signal of the cells expressing the empty pDRF1-GW plasmid was subtr acted fr om all samples. Next, bleedtr ough was calculated using the eCFP-expr essing str ain, and cells with at least a acceptor fluorescence signal of 2500 (arbitrary units) were k e pt. Lastly, FRET r atios wer e calculated for all r emaining cells.

Growth assays
Yeast strains were grown on 1% ethanol medium as described. Cells were washed twice by centrifuging at 3500 g for 3 minutes and resuspending in YNB medium without a carbon source. Afterw ar ds, cells w er e r esuspended in YNB medium without carbon source to an OD 600 of 1. Next, 20 μL of cells were put in a well of a 48-well microtiter plate having 480 μL YNB medium containing either 10 mM glucose, 10 mM galactose, or 0.1% ethanol. OD 600 was measur ed e v ery 5 minutes with a CLARIOstar plate reader (BMG LABTECH, Ortenber g, German y) at 30 • C and 700 rpm orbital shaking. Gr owth r ates wer e calculated by calculating a moving av er a ge ov er eac h gr owth curv e. Next, a sliding windo w w as used in which a linear r egr ession was fitted for eac h window, whic h giv es the gr owth r ate during this window. Next, to r educe the effect of out-lier gr owth r ates of the sliding window fits, the tenth fastest found slope was selected as the determined growth rate.

AKAR3-EV shows a FRET response in budding yeast.
To visualize the cellular kinase activities of Sch9p and PKA, we tested the use of the AKAR3-EV sensor, which consists of YPET-FHA1-EVlinker -RRA T motif -eCFP (Zhang et al. 2001, Allen and Zhang 2006, Komatsu et al. 2011. This sensor, de v eloped for mammalian PKA assa ys , should also w ork in y east as PKA and Sch9p also have RRxT as recognition site (Reinders et al. 1998, Plank et al. 2020 ). We constructed a nonr esponsiv e AKAR3-EV-NR sensor as a control by mutating the threonine in the RRxT motif to alanine (T506A), and expressed both AKAR3-EV and AKAR3-EV-NR in the W303-1A and SP1 strains. Both W303-1A and SP1 were used as some mutant W303-1A str ains hav e difficulties to expr ess sensors. We pr e viousl y also observed this for the yEPAC cAMP sensor or e v en single fluor escent pr oteins, whic h indicates this is a more general problem for some W303-1A mutants and not a sensor-specific issue.
In these two strains, we assessed the FRET response from a 1% ethanol to a 100 or a 10 mM glucose transition (Figs. 1 A, B, and Supplementary Fig. S1). Furthermore, we tested the older generation AKAR3 sensor for performance comparison (Colombo et al. 2017, Colombo et al. 2022 ). The AKAR3-EV sensor gave a clear increase in FRET after glucose addition, which was 3-4 times higher compared to the AKAR3-EV -NR. Furthermore, we found only a marginal response of the original AKAR3 sensor, showing that the AKAR3-EV sensor has an impr ov ed ability to visualize the phosphorylation activity of PKA and Sch9p in yeast. Cellular expression was more than sufficient, and the sensor sho w ed a uniform distribution in cells ( Fig. 1 C). Of notice, the AKAR3-EV-NR sensor shows a significant incr ease shortl y after glucose addition in the W303-1A str ain. This incr ease is not caused by osmotic c hanges since 100 mM sorbitol addition did not incr ease FRET le v els (Supplementary Fig. S2A). Conclusions about the FRET responses of AKAR3-EV dir ectl y after a transition should therefore be taken car efull y. Furthermor e, the AKAR3-EV shows no basal drift in FRET, which we found for AKAR3. The original AKAR3 sensor also shows a small dip in FRET response after the glucose ad- FRET response of SP1 WT and SP1 TPK1 wimp cells expressing AKAR3EV or AKAR3-EV NR to 10 mM glucose addition. Cells were grown on 1% ethanol, incubated for at least 2 h with r a pamycin, moc k (ethanol), or without any addition and pulsed with 10 mM glucose. Lines show the mean FRET value, normalized to the baseline, shades indicate SD, colours indicate the expressed sensor, strain, and/or treatment. All data are obtained from at least two biological replicates. dition in W303-1A, which could come from the differential pH sensitivity of the fluor escent pr oteins in this sensor. AKAR3-EV uses YPET-eCFP as a FRET pair, which shows more pH robustness compared to the eCFP-Venus FRET pair used in AKAR3 ( Supplementary Fig. S2B) (Botman et al. 2019 ). Lastl y, expr ession le v els of AKAR3-EV did not affect basal FRET le v els, and gro wth w as not affected on various carbon sources (Supplementary Figs. S2C and S2D). In conclusion, the AKAR3-EV can be used in yeast to assess cellular phosphorylation of the RRxT motif.

AKAR3-EV responses are dependent on sch9 and PKA signalling, but not ypk1
To assess the whether the AKAR3-EV sensor is influenced by cAMP-PKA and TORC1-Sch9 signalling, we compared normalized FRET responses of W303-1A and SP1 WT to strains carrying the TPK1 wimp m utation (whic h has low PKA activity) and the sc h9 strain upon a shift from 1% ethanol to 10 mM glucose (Figs. 2 A cell but only report relative changes . T he TPK1 wimp shows a lower initial response but does gradually obtained a similar change in normalized FRET at the end of the timelapse (normalized FRET le v els of 1.16 and 1.14 for WT and TPK1 wimp , r espectiv el y). In contr ast, sc h9 deletion gav e a r eduction in the r eac hed plateau of 25% (normalized FRET le v els of 1.24 and 1.18 for WT and sch9 , r espectiv el y). To note, for both the TPK1 wimp and the sch9 , we found increased absolute FRET levels during the complete transition, indicative for a higher phosphorylation state. We also tested whether W303-1A CYR1 K1876M , which lacks the classical cAMP peak upon glucose addition shows an alter ed AKAR3-EV r esponse. This strain sho w ed a normalized FRET response that r eac hed a plateau similar to the sch9 strain (Figs. 2 C and Supplementary  Fig. S2C), but with different dynamics. Its absolute FRET le v els wer e compar able to WT (Supplementary Fig. S3C). Finall y, the RRxT motif can potentially also be phosphorylated by YPK1p and YPK2p (Chen et al. 1993 ) with YPK1p being the dominant protein. This kinase is regulated by TORC2, and this signalling br anc h is involved in lipid metabolism and can potentially also be activated by sugars (Jacquier andSchneiter 2010 , Plank et al. 2020 ). We tested the FRET response of AKAR3-EV in a W303-1A ypk1 strain and found a negligible decrease in the maximal response (1.28 versus 1.26 normalized FRET le v els for WT and ypk1 , r espectiv el y, Fig. 2 C and Supplementary Fig. S3C). Since both the cAMP-PKA and TORC1-Sch9p cascades affect the AKAR3-EV response, we assessed whether a str ain m utated in both cascades sho w ed any incr ease in FRET le v els upon glucose addition. For this, we used the S25-31C strain (tpk2 , tpk3 , bcy1 , and Sch9 ), which is known to hav e impair ed phosphorylation activity assessed by trehalase activity (Crauwels et al. 1997 ). As expected, 100 mM glucose addition to this strain sho w ed no increase in FRET levels of AKAR3-EV, indicative that AKAR3-EV FRET (and hence RRxT phosphorylation) le v els r el y on PKA and Sc h9p signalling (Fig. 2 D and Supplementary Fig. S2D). In contr ast, ther e is a slight decrease in FRET signal, which may be explained by glucose-induced phosphatase activity on the RRxT motif. The phosphorylation state of the sensor is in the end a steady-state balance between phosphorylation and dephosphorylation activity. Finally, we confirmed the dependency of RRxT phosphorylation on Sch9p and PKA by the simultaneous suppression of both pathwa ys . For this , we used the TORC1 inhibitor r a pamycin to inhibit this signalling cascade and combine this with the SP1 WT and TPK1 wimp str ains, whic h hav e decr eased (but not completel y absent) PKA activity. In wild-type SP1, we found a decrease of 27% in the FRET response of r a pamycin-tr eated cells compared to untreated cells after a 10 mM glucose pulse (Fig. 2 E and Supplementary Fig. S2E). SP1 TPK1 wimp cells already sho w ed a 27% decr eased maximal r esponse (compar ed to WT cells), whic h further decreased to 60% when treated with rapamycin. This response was still higher compared to the response of the non-responsive sensor, which most likely is attributed to the remaining activity of TPK1 wimp .
In conclusion, we show that the AKAR3-EV sensor can be used to measure TORC1-Sch9p and cAMP-PKA activity by visualizing the cellular RRxT phosphorylation status. Deletion of SCH9 or removal of the classical cAMP peak results in a decreased phosphorylation status of the cell dir ectl y after the carbon source transition. In contr ast, decr eased PKA acti vity gi v es a decr eased initial r esponse, but e v entuall y r eac hes the same plateau as wild-type cells within the short timeframe measured. T hus , the individual influence of Sch9p and PKA on the phosphorylation dynamics can be further elucidated by the AKAR3-EV sensor.

Sugar transitions shows distinct phosphorylation dynamics with single-cell heterogeneity
In our pr e vious study, we found that cAMP signalling in yeast is different for different sugars, and it was homogeneous with no clear subpopulations or non-responders (Botman et al. 2021 ). We used the AKAR3-EV sensor to measure the downstream response to different sugars. W303-1A cells grown on 1% ethanol medium were pulsed with either 100 mM glucose or sucrose (both Gpr1p agonists), 100 mM fructose (no Gpr1p agonist or antagonist), or 100 mM mannose (an antagonist of Gpr1p) (Lemaire et al. 2004, Botman et al. 2021. Glucose , sucrose , and fructose gave a clear tr ansient incr ease in the FRET ratio after addition, wher e sucr ose and glucose gave a slightly faster and more sustained response. In contrast, mannose sho w ed a decline in FRET signal after which the FRET r esponse incr eased to higher le v els, whic h was sustained until the end of the timelapse recording (Fig. 3 A, Supplementary  Fig. S4 for SP1 responses). For fructose , glucose , and sucrose , we found no clear and significant subpopulations . T he most striking response at the single-cell level, ho w ever, w as found for mannose ad dition. Mannose ad dition gav e a highl y heter ogenous r esponse (Figs 3 B-E, Supplementary movie S1), which was not found for the non-r esponsiv e sensor ( Supplementary Fig. S5) and was not caused by sensor expression levels, as the sum of the donor and acceptor emission (used as a proxy for sensor expression) was similar among the clusters (Fig. 3 D, second panel). After the dip in FRET le v els (whic h all cells do seem to hav e), we identified thr ee discr ete r esponses, whic h we cluster ed using k-means clustering. The first cluster consisted of cells that have a broad timeframe in which cells increase in FRET levels . Furthermore , final FRET levels are slightly increased (mean normalized FRET value of 1.08), and the slope of the FRET increase was high (mean normalized FRET increase of 0.26 per minute). The second cluster sho w ed clear s witchers , which obtained a final normalized FRET value of 1.25. Finally, the third cluster consisted of cells that did not incr ease but decr eased their RRxT phosphorylation le v els (a mean normalized FRET value of 0.90 at the end of the timeframe). Interestingly, for the mannose transitions, cells that had a lo w er baseline FRET value also seem to end in a lo w er FRET state, and vice versa (Fig. 3 D).
We further explored the heterogeneous FRET response after mannose addition by e v aluating the cellular growth of these yeast for at least 2 h after mannose addition. For accurate growth assessment, analysis was restricted to cells that had a bud at the start of the experiment or de v eloped a bud during the experiment. (Fig. 4 ). Single-cell FRET curves were also determined ( Fig. 4 A), and k-means clustering was performed (Fig. 4 B). We found again the 3 clusters, and when we overlayed gr owth v ersus non-gr owth in these clusters (Fig. 4 B and C), we found one cluster with mostly gro w ers, one cluster with non-gro w ers, and a mixed cluster. Growing cells had a significant lo w er basal FRET le v el (0.84 ± 0.16 versus 0.94 ± 0.16 for growing and non-gr owing, r espectiv el y. Student's t -test, P = 0.006, Fig. 4 D), obtained a larger normalized FRET change (0.20 ± 0.09 versus 0.07 ± 0.20 for growing and non-gr owing, r espectiv el y. Student's t -test, P < 0.001, Fig. 4 D), and sho w ed a faster normalized FRET increase per minute (0.12 ± 0.04 versus 0.09 ± 0.05, P = 0.001).
In conclusion, the AKAR3-EV sensor shows multifarious responses after sugar ad dition, de pendent on the specific sugar. For mannose, we found a highl y heter ogenous r esponse, and its clustering suggests that the r a pid dynamics of the RRxT phosphorylation status during a carbon-source transition correlates with the onset of r a pid gr owth.

The nutrient-induced phosphorylation system has a high affinity for glucose
When we tested the response of the AKAR3-EV sensor in W303-1A to different levels of glucose (Fig. 5 A), we found a graded response that could be fitted by a single binding curve with a relativ el y high affinity, with a K 0.5 of 0.27 mM for glucose (Fig. 5 B). A dose-r esponse curv e of str ain SP1 gav e similar par ameters (K 0.5 of 0.26 mM), indicating that this high affinity is not a strain-specific feature ( Supplementary Fig. S6). In addition, no clear subpopulations or non-responders were detected ( Supplementary Fig. S7). This is in line with pr e vious r esults found for cAMP (Botman et al. 2021 ).
In conclusion, we found that the affinity of the RRxT phosphorylation system (i.e . T ORC1-Sch9p and cAMP-PKA) has a high affinity for glucose-in comparison, the affinity we found for cAMP peak height was 3.0 mM, ten times higher, ther efor e. Furthermore, this system appears to be homogeneous since no clear non-responders or subpopulations were found.

AKAR3-EV shows no growth-rate dependent FRET status
After c har acterizing the TORC1-Sc h9p and cAMP-PKA signalling response during nutrient transitions, we studied the steady state phosphorylation le v els during (balanced) gr owth on differ ent carbon sources. W303-1A cells expressing either the AKAR3-EV or the AKAR3-EV-NR sensor wer e gr own to mid-exponential phase on 1% ethanol (growth rate of 0.16 h −1 ), galactose (growth rate of 0.25 h −1 ), glucose (growth rate of 0.36 h −1 ), or mannose (gr owth r ate of 0.34 h −1 ), and the FRET le v el distributions acr oss a population were measured using a flow cytometer (Fig. 6 A). We found significant differences between the conditions tested (Kruskal-Wallis, P < 0.01), except between galactose and mannose (Wilcoxon signed-rank test, P = 0.6). Ho w ever, w e also found such differences in the non-responsive sensor, and no clear relation w as found betw een gro wth rate and the FRET levels of AKAR3-EV after correction for aspecificity (Figs. 6 B and C).

Discussion
In the present study, we implemented the mammalian-optimized AKAR3-EV sensor in yeast and tested whether this sensor can also be applied to study Sch9p and PKA signalling in yeast. In yeast, the sensor has sufficient expression and a homogenous distribution in cells . Furthermore , gro wth rates w ere not affected by the sensor, indicating the sensor is harmless to cells (Supplementary Fig. S1C). The AKAR3-EV sensor sho w ed the expected response to an ethanol-to-glucose transition and outperformed a pr e viousl y published AKAR sensor in yeast (Fig. 1 A). The baseline drift of the original AKAR sensor can be caused by the photoc hr omic behaviour of one of the FPs (Botman et al. 2019 ), or differ ential FP r esponses to c hanges in the intr acellular composition. We also made a non-r esponsiv e v ersion of AKAR3-EV by mutating the RRxT motif to RRXA (Plank et al. 2022 ). This AKAR3-EV NR sensor indeed sho w ed significantly lo w er responses, although w e found that some transitions evoked a rather large response immediately after the transition. The origin of this response is not known and may be caused by nonspecific phosphorylation of the sensor domain, although this domain does not contain any other known phosphorylation sites of PKA or Sch9p. This response can also originate from changes of the intracellular composition (e.g. changes in redox potential, pH, and ion concentration). If needed, the AKAR3-EV NR can be used to corr ect for these aspecific r esponses . T his sensor can be potentially improved for use in yeast by using phosphorylation sites that ar e mor e common for yeast (i.e. RRXS) (Plank et al. 2022 ), although this could also increase the basal phosphorylation status of the sensor, decreasing its dynamic FRET range. In addition, (y)mTurquoise2 could be used instead of eCFP since this fluorescent protein is a better FRET donor (Goedhart et al. 2012, Mastop et al. 2017. The AKAR3-EV FRET response is indeed dependent on both TORC1-Sch9p and cAMP-PKA signalling (Fig. 2 ). Glucose addition to ethanol-grown cells showed that impaired PKA signalling (using the TPK1 wimp strain) decreased the rate of phosphorylation, although the maximal response remains similar. An Sch9 deletion, on the other hand, resulted in a lo w er normalized response after a glucose transition. Yet, these transitions were performed in different strains (SP1 and W303-1A, r espectiv el y). Inter estingl y, we also found that the W303-1A CYR1 K1876M mutation in which the transient cAMP peak is absent (Botman et al. 2021 ) sho w ed a decreased maximal response. Further research should clarify whether and why missing the transient cAMP peak has a long-term effect on cell signalling status and fitness . T he S25-31C strain with impaired signalling in both Sch9 as PKA, sho w ed a decrease in RRxT phosphorylation, proving that the AKAR3-EV sensor indeed measur es solel y PKA and Sc h9p activity. This was further confirmed by adding r a pamycin in SP1 WT and SP1 TPK1 wimp cells, where the r a pamycin-tr eated TPK1 wimp str ain sho w ed a se v er e r eduction in FRET response after glucose addition. Lastly, we found no significant effect of ypk1 deletion on the FRET response of AKAR3-EV in W303-1A, indicating that the AKAR3-EV sensor measures specifically PKA and Sch9p activity. As mentioned, some mutant strains had higher absolute FRET le v els, suggesting that the signalling arc hitectur e can compensate for a decreased kinase activity. This can occur, for example, via negativ e feedbac k patterns known for PKA , Dong and Bai 2011 and TORC1 (Péli-Gulli et al. 2017 ) or altered activity of RRxT phosphatases. Another possibility is that an altered cellular composition confounds FRET le v els (Moussa et al. 2014 ), for which the non-r esponsiv e sensor can function as a control. T hus , as for every FRET sensor, quantitative conclusions about normalized and absolute FRET le v els should be taken with car e. In this study, we did not wish to mak e quantiti ve conclusions about signalling mutants on the absolute phosphorylation status, but rather use these mutants to show that the AKAR3-EV sensor indeed measures the activity of these two kinases.
A major strength of biosensors is the ability to measure singlecell responses . T herefore , we assessed single-cell responses during tr ansitions fr om ethanol-gr own cells to glucose, sucr ose, fructose, and mannose (Fig. 3 ). As pr e viousl y found for cAMP signalling (Botman et al. 2021 ), hardly any heterogeneity or subpopulations were found for the glucose , sucrose , and fructose transitions. In contrast, we did find a heterogenic response upon mannose addition. Mannose is known as an antagonist of cAMP signalling (Lemaire et al. 2004, Botman et al. 2021 ), but is metabolized at rates similar to glucose as cells grow comparable on these sugars (growth rate of 0.34 h −1 for mannose and 0.36 h −1 for glucose). Gl ycol ytic startup, determined by pH measurements, also indicates that mannose is transported and metabolized at least as fast as glucose ( Supplementary Fig. S8). The conflicting signals between the signalling and metabolism of mannose may be the reason for the heterogenic response. Cells obtaining a higher RRxT phosphorylation state, compared to the pre-transition state, seem to start growth whereas cells with a (s)lo w er response seem to halt growth. This confirms that PKA and Sch9p activity, and the RRxT phosphatases have an effect on the cellular decision to start growth, in line with previous studies showing that cAMP-PKA signalling and Sch9p are involved in cell cycle pr ogr ession (Hubler et al. 1993, Müller et al. 2003, Jorgensen et al. 2004, Futcher 2006, Cocklin and Goebl 2011. The heterogenous Sch9p and PKA signalling dynamics potentially transmit further downstream, where it is shown that different temporal nuclear localization patterns of transcription factors (such as msn2p, a phosphorylation target of Sch9p and PKA) can result in differential transcription and tr anslation r esponses potentiall y r esulting in an alter ed gr owth response (Hao and O'Shea 2012, Zid and O'Shea 2014, Hansen and O'Shea 2016. The AKAR3-EV sensor r e v ealed a r elativ el y high affinity (K 0.5 = 0.24 mM) of the phosphorylation system for glucose (Fig. 4 ). This is far below the affinity of the high-affinity hexose transporters in yeast with a lowest K m of ∼1-2 mM, for HXT6 and 7 (Reifenberger et al. 1997, Maier et al. 2002, which are expressed in these ethanol-grown cells . Furthermore , the affinity is lo w er compared to our previously obtained glucose affinity of cAMP peak le v els in yeast (Botman et al. 2021 ), which confirms that the RRxT phosphorylation status of the cell is not solely determined by cAMP-PKA signalling. The high affinity of the RRxT phosphorylation system is, on the other hand, not too far from the Monod constant for glucose-limited gr owth, whic h is ar ound 0.5 mM, admittedly for another strain (CEN.PK) (Canelas et al. 2011 ). Moreov er, these v alues ar e in line with pr e vious observ ations for Mig1 translocation after glucose additions (Bendrioua et al. 2014 ). The fact that we did not find clear subpopulations at any glucose concentrations shows that, at least for glucose, yeast cells sense its concentration in a highly accurate and robust manner.
One lar ge unanswer ed question is whether the basal signalling status of the Sch9p and PKA signalling pathways depends on the specific gr owth r ate . T his ma y be expected as transcription of ribosomal genes is an important target of the pathway, and ribosomal content does scale with growth rate (Metzl-Raz et al. 2017 ). The obtained flow cytometry data ar e statisticall y significantl y differ ent between most conditions, but giv en the small size of the difference relative to the spread of the distributions (Figs. 6 B and C), their biological r ele v ance is disputable . The small differences could be caused by a maximal phosphorylation status of the cell under the conditions tested, especially since the RRxT is a pr eferr ed motif of Sc h9p and PKA (Plank et al. 2020 ). Yet, w e sho w ed that the FRET le v els can incr ease for > 20% in cells grown in 1% ethanol medium, and the basal FRET le v el of this condition is comparable with the other conditions tested. This implies that the sensor is not maximally phosphorylated. Ther efor e, we belie v e that the basal phosphorylation status of the RRxT motif is unchanged and not maximally phosphorylated between various growth rates. Unchanged RRxT phosphorylation le v els do not necessaril y impl y that PKA and Sch9p activities are also unchanged across growth conditions, as the phosphorylation le v el is a r esult of both phosphorylation and dephosphorylation. Furthermore, PKA and Sch9p phosphorylate also other motifs, potentially with different rates across conditions (Plank et al. 2022 ). We hypothesize that the RRxT phosphorylation response upon sugar transitions steers cells to the right cellular physiology, after which this signalling system returns to its basal level again. The dynamics of Fig. 3 A also suggest a temporal impact of sugars on the phosphorylation state.
In summary, the AKAR3-EV pr ov ed to be a robust sensor to measure the nutrient-induced RRxT phosphorylation status in y east cells. Ho w e v er, since the phosphorylation status is an integrated output of the activity of both Sch9p and PKA, but also on phosphatases, its inter pr etation is mor e c hallenging than for other sensors . Nonetheless , we belie v e that the AKAR3-EV sensor is a useful addition to the toolbox that can help to elucidate how yeast cells respond and adapt to nutrient changes.