Biochemical Journal

Research article

Differential regulation of glucose transport activity in yeast by specific cAMP signatures

Clara Bermejo, Farzad Haerizadeh, Mayuri S. C. Sadoine, Diane Chermak, Wolf B. Frommer


Successful colonization and survival in variable environments require a competitive advantage during the initial growth phase after experiencing nutrient changes. Starved yeast cells anticipate exposure to glucose by activating the Hxt5p (hexose transporter 5) glucose transporter, which provides an advantage during early phases after glucose resupply. cAMP and glucose FRET (fluorescence resonance energy transfer) sensors were used to identify three signalling pathways that co-operate in the anticipatory Hxt5p activity in glucose-starved cells: as expected the Snf1 (sucrose nonfermenting 1) AMP kinase pathway, but, surprisingly, the sugar-dependent G-protein-coupled Gpr1 (G-protein-coupled receptor 1)/cAMP/PKA (protein kinase A) pathway and the Pho85 (phosphate metabolism 85)/Plc (phospholipase C) 6/7 pathway. Gpr1/cAMP/PKA are key elements of a G-protein-coupled sugar response pathway that produces a transient cAMP peak to induce growth-related genes. A novel function of the Gpr1/cAMP/PKA pathway was identified in glucose-starved cells: during starvation the Gpr1/cAMP/PKA pathway is required to maintain Hxt5p activity in the absence of glucose-induced cAMP spiking. During starvation, cAMP levels remain low triggering expression of HXT5, whereas cAMP spiking leads to a shift to the high capacity Hxt isoforms.

  • cAMP
  • fluorescence resonance energy transfer (FRET)
  • glucose
  • signalling
  • starvation
  • yeast


Micro-organisms experience rapid and extreme changes in nutrient availability in their natural habitats. A major factor for such changes in soil is the sudden variation in water content caused by sun and rain cycles [1]. Nutrient acquisition and osmolality must therefore be adjusted rapidly. Saccharomyces cerevisiae outcompetes other microbes during colonization of grapes, indicating that it can rapidly acclimate to new nutrient environments [2].

The signalling pathways that contribute to the acclimation to starvation have been studied extensively in yeast. During glucose starvation, Snf (sucrose nonfermenting) 1p becomes phosphorylated by one of the three Snf1p kinases [Sak1p (Snf1-activating kinase 1), Elm1p (elongated morphology 1) or Tos3p (target of Sbf3)], creating a complex with a γ-like Snf4p regulatory subunit and three possible β-subunits [3,4]. The Snf1p complex regulates several transcription factors that affect the expression of genes required for metabolism of alternative carbon sources. However, much less is known about how yeast is primed for exposure to new nutrient resources during starvation. The exposure to glucose or sucrose after starvation triggers a spike in the level of cAMP, triggering activation of PKA (protein kinase A), which in return activates reserve carbohydrate mobilization and induction of growth-related genes, e.g. genes involved in cell division [5]. Yeast uses an extracellular G-protein-coupled sensor Gpr1p (G-protein-coupled receptor 1) to monitor extracellular sugars during the early growth phase and trigger cAMP spiking [6].

Previous work has shown that the HXT5 (hexose transporter 5) gene and Hxt5p transport activity are specifically induced in the absence of glucose to fulfil functions during future exposure of cells to glucose [711]. HXT5 expression is under the control of STRE (stress response elements) and HAP (haem activator protein) elements in the HXT5 promoter and appears to be regulated by the PKA pathway [9]. Hxt5p is rapidly endocytosed after the early growth phase and replaced by other Hxts [9]. Measurement of the capability of starved cells to accumulate glucose using genetically encoded FRET (fluorescence resonance energy transfer) sensors demonstrated that Hxt5p is responsible for a large fraction of glucose uptake activity in starved cells [11,12]. On the basis of the relevance of the Snf1p kinase under starvation conditions and observations by Verwaal et al. [9] regarding the role of the PKA pathway in Hxt5p regulation, we hypothesized that other protein kinases/phosphatases are involved in transmitting responses to starvation signals. To test the relevance of protein kinases and phosphatases in S. cerevisiae for glucose starvation signalling, FRET sensors were used in the present study in a systematic screen of the yeast protein kinase/phosphatase mutant collection. Using this systematic screen several pathways were found to be involved in maintaining the glucose uptake capacity during starvation (the ‘Ajar’ pathway, as in keeping the door to glucose accumulation slightly open as opposed to requiring full induction of the uptake system by glucose), including the Snf1 pathway, but surprisingly also the PKA pathway, which is normally involved in the response to the addition of sugar. Transcriptional induction/repression of many genes in response to glucose is independent of PKA activation, but dependent on the intracellular levels of glucose/cAMP [13]. This regulation of the response to glucose by alternative pathways to the PKA pathway is primarily related to genes involved in glucose metabolism. The finding that adenylate cyclase was critical for Hxt5 activity in the absence of glucose indicated that base cAMP levels were necessary. The cAMP FRET sensor Epac2-camps300 was used to show that the base cAMP level is critical for Hxt5 activity during glucose starvation [14].


Strains and plasmids

All strains used for the kinase/phosphatase screen were isogenic to BY4743 [MATahis3Δ1/his3Δ1 leu2Δ0/leu2Δ0 LYS2/lys2Δ0 met15Δ0/MET15 ura3Δ0/ura3Δ0] and were purchased from Open Biosystems (Supplementary Table S1 at The additional yeast strains used in the present study are described in Supplementary Table S2 (at The LRA85 and LRA86 strains were provided by Dr Kelly Tatchell (Louisiana State University Health Sciences Center, Shreveport, LA, U.S.A.). The MSY858 strain was provided by Dr Martin Schmidt (University of Pittsburgh, Pittsburgh, PA, U.S.A.). The HXT5 ORF (open reading frame) was cloned into the pDR-R1R2-his3 vector. The glucose sensor FLII12Pglu-700μδ6 [15] and cAMP sensor Epac2-camps300 [14] were cloned into the pDRf1GW-ura3 vector [16] for sensor expression in yeast.

Media and growth conditions

Yeast cells were grown in YPD medium (10 g/l yeast extract, 20 g/l peptone and 20 g/l dextrose) if not specified otherwise. When carrying plasmids, yeast cells were grown in synthetic complete medium [17] containing 2% glucose (SCglc) or 2% ethanol (SCethanol) and lacking the appropriate selection requirements. Cultures were maintained at 30°C and harvested in the log phase (D600=0.6–0.8) for all of the analyses described. When indicated, non-transformed yeast cells where grown in synthetic media (SCglc or SCethanol) including all of the essential amino acids.

Monitoring yeast growth

Yeast growth was monitored in 96-well plates by light scattering using a microplate reader (Tecan GENios™ at 595 nm) using a synthetic medium with all of the essential amino acids. Selected strains were grown overnight in SCglc medium. The cultures were then divided into two parts and the D600 was adjusted to 0.1; one part was grown in the SCglc medium, whereas the other part was glucose-starved in the SCethanol medium. After 16 h in either SCglc or SCethanol medium, 5×104 cells were inoculated to grow in a final volume of 100 μl of SCglc medium in the 96-well microplates. Plates were incubated at 30°C with constant shaking; light scattering was measured at 20 min intervals.

Analysis of the FRET responses by fluorimetry

Analyses were essentially performed as described previously, either using a microplate fluorimeter or a microscopic approach using microfluidic trapping chambers [11,18]. For determination of the response to glucose, transformed yeast cells were grown at 30°C in the SCglc medium to D600=0.8–0.9. Cells were collected and resuspended in SCethanol medium and grown for 16 h for glucose starvation. Cells were washed twice in 20 mM Mes (pH 6.1), the D600 value was adjusted to 0.5 and 180 μl were transferred into 96-well microplates (Greiner PS, F-bottom) for analysis in a microplate reader (Tecan Safire™; excitation at 428/12 nm; emission at 485/12 nm and 530/12 nm). Two measurements were taken before the addition of 20 μl of glucose. Background fluorescence intensity was measured for CFP (cyan fluorescent protein) and YFP (yellow fluorescent protein) channels recorded in parallel (8 cycles; each cycle ~100 s). Measurements were repeated at least three times independently. The emission ratio is defined here as the uncorrected fluorescence intensity at 528 nm divided by the intensity at 485 nm. Background fluorescence from untransformed cells in each channel was subtracted. Emission ratios were normalized to the initial ratio.

For single cell FRET analysis cells transformed with FLII12Pglu-700μδ6 or the Epac2-camps300 sensor were grown in SCglc medium to D600=0.8. Cells were washed in 20 mM Mes (pH 6.1) and starved in the SCethanol medium where indicated. Cells were trapped in a Y2 microfluidic plate (CellASIC) and subjected to reversible changes of the external sugar supply. Changes in single cells were recorded by FRET imaging as described previously [15]. Two genotypes (wild-type/mutant) were analysed in parallel in the two trapping areas. Cells were perfused with glucose or 2-deoxyglucose at the indicated concentrations. After each pulse, sugar was removed from the cell trapping area by perfusion with buffer [20 mM Mes (pH 6.1)]. Images were acquired every 10 s when the emission levels of the Epac2-camps300 cAMP sensor were under study and 20 s when assaying any of the glucose sensors. The background signal was subtracted for each channel for each time point and the ratio was calculated. Data were analysed using Slidebook 4 (Intelligent Imaging Innovations) by comparing emission ratios (535/470 nm) for >30 cells.

Zero-trans glucose uptake assays

The cells were harvested by centrifugation at 4°C, washed three times in ice-cold potassium phosphate buffer (100 mM, pH 6.5) and resuspended in the same buffer to a concentration of D600=4 (dry mass 5 g/l) and kept on ice. Zero-trans influx of glucose was determined at 30°C [19]. For each reaction, 50 μl of cells were mixed with 12.5 μl of 2 mM glucose labelled with D-[U-14C]glucose (590 KBq/μmol; Amersham) and incubated for 5 s. Uptake was terminated by transfer to 10 ml of quench buffer [ice-cold 100 mM potassium phosphate (pH 6.5) and 500 mM D-glucose] and the cells were harvested by vacuum filtration on to a glass microfibre filter (GF/C, Whatman). Filters were washed twice with quench buffer, transferred into scintillation vials containing 10 ml of Ultima Gold XR Scintillator liquid (PerkinElmer) and analysed by scintillation counting (Beckman). Triplicate determinations were performed for each strain from at least three independent cultures.

Real-time PCR

Total RNA was isolated from cells (5×107) using a mechanical disruption protocol and an RNeasy MINI kit following the manufacturer's instructions (Qiagen). RNA concentrations were determined by measuring absorbance at 260 nm. RNA purity and integrity were assessed by electrophoresis using denaturing agarose gel. Q-RT–PCR [real-time quantitative RT (reverse transcription)–PCR) assays were performed using a LightCycler® 480 System (Roche). For quantification, the abundance of each transcript was determined relative to the standard transcript of ACT1 (actin 1) following the 2−ΔΔCT method [20] and the ratio was calculated. The following primers were used: ACT1 forward, 5′-ATCACCGCTTTGGCTCCAT-3′ and ACT1 reverse, 5′-ACCAATCCAGACGGAGTACTTTCTT-3′; and HXT5 forward, 5′-TGGAAGGGTCTGCTACTGTGA-3′ and HXT5 reverse, 5′-ATGTAACTTGAGAGACGGGTTTAGCTT-3′.


Identification of signalling pathways involved in glucose uptake capacity during starvation

To identify the signalling networks responsible for anticipatory Hxt5p activity, an unbiased screen was carried out including all viable protein kinase/phosphatase-deficient strains in the yeast knockout collection (108 protein kinases and 42 protein phosphatases) by testing for effects of mutations on steady-state glucose levels in glucose-starved cells. The mutants and the corresponding wild-type were grown in glucose-containing medium, starved in ethanol-containing medium and analysed for glucose-induced FRET responses in three independent screens. Candidates showing effects in at least one screen were retested using a minimum of three independent transformants and compared with the glucose accumulation in the wild-type strain BY4743 (Figure 1A). Reproducible differences were observed in 19 out of the 150 mutants (Supplementary Figures S1 and S2 at The deficiency in glucose accumulation was quantified as the difference in sensor output (maximal YFP/CFP ratio at 100 mM glucose) between the wild-type and mutant for the highest glucose concentration used in the screening (addition of 100 mM glucose; Table 1). The largest reduction in glucose accumulation was observed for the snf1Δ and snf4Δ mutants defective in AMP kinase activity (Figure 1B and Table 1, and Supplementary Figure S1). A specific effect of ethanol on Snf1p-mediated control of sugar acquisition can be excluded because analysis of cells that had been starved in medium lacking any carbon source responded in an indistinguishable way (Figures 1E and 1F). On the basis of the horizontal screen of all kinases and phosphatases the analysis was expanded vertically, testing components linked to the respective kinases/phosphatases. First, the role of the upstream kinases Sak1p, Tos3p and Elm1p in the Snf1 pathway was evaluated. Whereas the single mutants did not show significant differences compared with the wild-type, a triple sak1Δ tos3Δ elm1Δ mutant was characterized by an almost complete lack of glucose accumulation, demonstrating that Snf1p activation is essential for Hxt5p activity (Figures 1C and 1D). Moreover, Gal83p (galactose metabolism 83), which encodes a regulatory subunit of the Snf1 complex triggering nuclear targeting, is also involved (Supplementary Figure S1). The Snf1 pathway is known to play a role in detecting a low energy status in yeast cells [21]. In agreement with this role, the screen identified components from the Snf1 pathway expected to be involved in the anticipation of future exposure to glucose by induction of Hxt5p activity.

Figure 1 Reduced cytosolic glucose accumulation in yeast mutants

Yeast cells expressing FLII12Pglu-700μδ6 were glucose-starved in SCethanol medium, washed and transferred on to microplates. Glucose was added into the wells after time point 2, and the time-dependent glucose response was analysed in the wild-type strains BY4743, BY4742, W303 and LRA27 (A, C, G and I). The mutants tested were snf1Δ (B), sak1Δ elm1Δ tos3Δ (MSY858) (D), tpk1M164G tpk2M147G tpk3M165G (LRA85) (H), cdc35-10 (J) and cdc25-10 (K). The glucose concentration was varied between 0.5 and 100 mM. Error bars represent the S.D. for at least three to six different transformants. The absolute ratio change in the wild-type cells was 0.4–0.55 depending on the background. The x-axes are broken to indicate an interruption of the time course during the addition of glucose. (E and F) BY4743 and snf1Δ cells expressing FLII12Pglu-700μδ6 were glucose-starved in SC medium with no carbon (no C) source and tested the same way as described above. (L) Glucose accumulation in the wild-type and gpr1Δ cells after exposure to different external glucose concentrations (2, 5, 10 and 100 mM). Error bars represent the S.D. for six independent transformants.

View this table:
Table 1 Glucose accumulation in protein kinase/phosphatase-deficient strains

Response of FLII12Pglu-700μδ6 in protein kinase or phosphatase mutants to exposure to 2, 5, 10 or 100 mM glucose was analysed using microplate fluorimetry. The Table shows hits deficient in glucose accumulation. Quantification of the relative cytosolic glucose accumulation in the kinase/phosphatase-deficient strains screened (Supplementary Table S1 at was done using the wild-type value set to 100% as a reference. Accumulation in the mutants is calculated relative to the wild-type strain at 100 mM external glucose. MAPK, mitogen-activated protein kinase.

Deletion of either of two catalytic subunits of PKA affected the potential to accumulate glucose in the cytosol of starved cells (tpk1Δ and tpk2Δ; Table 1 and Supplementary Figures S1 and S2). Interestingly, reduced glucose accumulation was also observed in the strains lacking the kinases Pho85p and Mck1p (meiotic and centromere regulatory serine, tyrosine-kinase), factors controlling glycogen metabolism and glycolysis [22,23]; ptc1Δ, lacking a phosphatase type 2C, which has been associated with HOG, TOR (target of rapamycin) and other cellular processes [2426]; and bck1Δ, which lacks the MAPKKK (mitogen-activated protein kinase kinase kinase) of the cell integrity pathway. The results obtained with the pho85Δ and ptc1Δ cells were confirmed by analysis of FRET responses in individual yeast cells trapped in a microfluidic platform (Figure 2A and Supplementary Figure 3A at Since Pho85p has a role in the repression of hexokinases [27], the observed decreased rate of glucose accumulation in the pho85Δ cells is consistent with increased glucose metabolism. Thus, although FRET sensor analysis does not provide actual flux rates for transport and metabolism, the differential can provide insights into whether the transport or metabolic rates are affected. Other mutants with limited accumulation of glucose include gcn2Δ, rts1Δ, ctk1Δ, and kps1Δ (Table 1). Thus either additional pathways contribute to Hxt5 activation, or these factors could exert their action through the Snf1, PKA or Pho85 pathways.

Figure 2 Responses of mutants to glucose starvation

(A) ptc1Δ cells were grown in glucose and then starved in ethanol medium. Cells were trapped and perfused with different glucose (Glc) concentrations (2, 5, 10 and 100 mM, represented with grey shading, 4 min pulses). Each pulse was followed by a Mes buffer wash (black trace, wild-type; grey trace, ptc1Δ). (B) The yeast strains snf1Δ, ptc1Δ, mck1Δ, bck1Δ, pho85Δ and tpk1M164G tpk2M147G tpk3M165G were grown at 30°C in synthetic glucose medium followed by glucose starvation. Glucose uptake assays were performed using 14C-labelled D-glucose with 2 mM glucose using the same amount of cells (5×106 cells in 50 μl of medium). The error bars represent the S.D. for at least three independent yeast suspensions (duplicate measurements; wild-types as indicated in light grey).

Figure 3 Schematic representation of the ‘Ajar’ pathways

(A) The ‘Ajar’ pathways responsible for the anticipatory glucose uptake capacity in starved yeast cells. The results shown in the present paper suggest that the G-protein-coupled receptor Gpr1p, Gpa2p, the Ras branch with CDC25, adenylate cyclase (CYR1) and PKA are necessary for the ability of starved cells to accumulate glucose. How this pathway measures lack of sugars is not understood at present. In addition, the AMP kinase pathway with Tos3p, Elm1p, Sac1p, Snf1p, Snf4p and Gal83p is also required. Either reduced AMP or glucose 6-phosphate levels, lead to derepression of Snf1p. (B) For comparison, glucose responsive signalling pathways were active after yeast cells have been exposed to glucose. Both Gpr1p and Ras2p are thought to measure glucose levels and produce a transient cAMP peak that activates PKA and induces, e.g. cell division. The extracellular glucose sensors Snf3p and Rgt2p lead to induction of Hxt1p and other members of the Hxt family. Gal83, galactose metabolism 83; Mth1, MSN three homologue 1; Sac1, suppressor of actin; SCF, Skp1–Cullin–F-box; STD1, suppressor of Tbp deletion 1; Rgt, restores glucose transport; Tpl, Takashi's protein kinase; Yck1, yeast casein kinase I homologue.

Role for other elements of the Gpr1 and Ras/PKA pathway

The involvement of PKA was confirmed by the finding that a triple mutant carrying point mutations in each of the catalytic subunits of PKA (tpk1M164G tpk2M147G tpk3M165G) was impaired in its ability to accumulate glucose during starvation (Figures 1G and 1H), indicating that PKA can exist in at least two states: one state functioning during starvation (before the addition of glucose) and a second state when hyperactivated by a glucose-triggered increase in cAMP [28]. One would expect that a triple knockout mutant of the PKA subunits, similar to snf1Δ, would show complete absence of glucose accumulation; however, this cannot be verified experimentally because the triple knockout mutant is lethal.

The Snf1p and PKA pathways may act independently or interact with each other to control sugar uptake capacity. The relative role of the Snf1 and PKA pathways was tested in a tpk2Δgal83Δ double mutant. After starvation, the single mutants tpk2Δ and gal83Δ showed reduced glucose accumulation compared with the wild-type (Supplementary Figures S4A–S4C at and this effect is slightly enhanced in the double mutant tpk2Δgal83Δ (Supplementary Figure S4D).

PKA is vertically connected to the extracellular G-protein-coupled glucose response pathway and is activated by increases in cAMP levels triggered by adenylate cyclase Cyr1p [cAMP requirement 1; also known as Cdc (cell division cycle) 35]. An increase in cAMP promotes dissociation of the inhibitory subunit Bcy1p (bypass of cyclase mutations 1) from the Tpk (Takashi's protein kinase) subunits activating PKA. Adenylate cyclase can be activated either via a putative sugar receptor, the G-protein-coupled Gpr1/Gpa2 (G-protein α subunit 2) pathway, or by activation of Ras by the membrane-bound guanine nucleotide-exchange factor Cdc25p. Since the cells had not been exposed to glucose prior to analysis, the upstream components of the Gpr1/Ras/cAMP/PKA pathway were not expected to be involved (Figure 3). However, most of the components of the Gpr1/cAMP/PKA pathway that were tested were also necessary for glucose accumulation (Table 1 and Supplementary Figure S1). Mutation of the adenylate cyclase gene CYR1 (cdc35-13) caused no significant accumulation of glucose in the cells at external glucose concentrations ≤10 mM (Figure 1J), demonstrating its importance for glucose acquisition potential in the absence of glucose. Also, CDC25 mutations (cdc25-10 and cdc25-5 [29]) led to a significant decrease in glucose accumulation (Figure 1K) compared with the wild-type (Figure 1I). Finally, the most unexpected result was that the putative sugar receptor Gpr1p, which acts upstream of Cyr1p, was also necessary for the preformed glucose transport activity in the absence of its putative ligand glucose (Figure 1L). Thus both branches for the cAMP pathway and adenylate cyclase activity must play important roles for ‘measuring the absence of glucose’. This raises an interesting question, namely why two pathways are required and how they communicate. The Snf1 pathway is supposed to respond to either a reduction in cytosolic glucose 6-phosphate or an increase in AMP levels [21,29a]. Regarding the downstream PKA substrates, the tyrosine kinase Rim11p [regulator of IME2 (inducer of meiosis 2) 11] appears to be also involved in the regulation of Hxt5p by PKA since its deletion also reduces glucose accumulation (Table 1 and Supplementary Figure S1). This is in agreement with Rim11 transcriptional regulator kinase activity being inactive in the presence of glucose and becoming active when it is depleted, potentially inducing Hxt5 transport activity during starvation [30].

The Snf1 and PKA pathways induce glucose uptake capacity in glucose-starved cells

Steady-state metabolite levels depend on the rates of net influx and metabolic conversion. Although the single cell analysis had provided hints on whether uptake was inhibited or whether metabolism was increased in the mutants, it was necessary to determine independently whether the reduced steady-state levels of glucose were caused by a decrease in uptake activity. Glucose uptake in yeast can be determined using zero-trans influx experiments, which measure the initial rate of radiolabelled glucose uptake within very short time scales. Zero-trans [14C]-D-glucose uptake assays were performed [19] for the mutants snf1Δ, bck1Δ, mck1Δ, pho85Δ, ptc1Δ and tpk1M164G tpk2M147G tpk3M165G. Glucose uptake was severely reduced in the snf1Δ, bck1Δ, ptc1Δ and tpk1M164G tpk2M147G tpk3M165G mutants, whereas no reduction in uptake was observed in pho85Δ and mck1Δ cells (Figure 2B). This result suggests the participation of Bck1p (bypass of C kinase 1), Snf1p, Ptc1p (phosphatase type 2C 1) and PKA in the regulation of Hxt5p activity in starved cells, whereas Mck1p and Pho85p appear to play a role in glucose metabolism. In agreement with a role for Snf1p and the PKA pathway in the transcriptional regulation of HXT5, the transcript levels of HXT5 were severely reduced in snf1Δ cells and the triple mutant tpk1M164G tpk2M147G tpk3M165G carrying point mutations in each of the catalytic subunits of PKA (Supplementary Table S3

Multiple states of activation of the PKA pathway correlate with different levels of cAMP

cAMP peaks trigger activation of PKA when starved yeast cells are exposed to glucose [31,32]. Since PKA is involved in a glucose-response pathway it is surprising that it would also play a critical role in a glucose-deprivation pathway. To assess the role for cAMP in the regulation of glucose transport in starved yeast, cAMP levels were monitored directly using a genetically encoded cAMP sensor (Epac2-camps300) [14]. Epac2-camps300 bound cAMP with a Kd value of 482±104 nM (Supplementary Figure S5A at, which is in the range of concentrations thought to be relevant for PKA activation [14]. Epac2-camps300, when analysed in vitro, did not show a detectable response to glucose and 2-deoxyglucose even at levels greater than those used to trigger cAMP responses in vivo (Supplementary Figures S5B and S5C). Using single-cell analysis in microfluidic chambers, cAMP levels were measured in cells that were starved as well as after re-addition of glucose. A decrease in the cAMP levels was observed ~4 min after exposing the yeast cells to medium with no sugar (Figures 4A and 4B). Yeast that were starved from glucose for several minutes (20–30 min) and then were perfused with glucose (10–100 mM) or medium containing 2% glucose (Figures 4A and 4B), showed a transient cAMP accumulation. This spike returned to a new basal level within ~1 min, higher than the original cAMP levels observed during starvation. When glucose was completely eliminated from the medium, cytosolic cAMP levels returned to the starvation levels after ~5–7 min. Perfusion with the glycolysis inhibitor 2-deoxyglucose reduced cAMP levels further; below that of the starved cells. The observation that 2-deoxyglucose-triggered reduction in cAMP leads to loss of Hxt5 activity, together with the finding that Hxt5 is functional during starvation at intermediate cAMP levels, indicates that the PKA pathway exists in at least two activation states that are defined by at least two different levels of cAMP. This indicates that cAMP is being produced during starvation to maintain the pathway in an active state.

Figure 4 In vivo dynamics of cytosolic cAMP in starved yeast

(A and B) Yeast cells expressing Epac2-camps300 were grown in glucose medium and trapped in a microfluidic device. Wild-type cells were perfused first with SCglc medium (grey shaded area) and then switched to same SC medium with no glucose. (C) Wild-type cells were trapped, washed with Mes buffer and perfused with 100 mM glucose solution (~8 min), 100 mM 2-deoxyglucose (~15 min) and 5 mM bromoadenosine 3′,5′-cAMP (~12 min). Each pulse was followed by a Mes buffer wash. Mutant cdc35-13 (LRA86) was trapped in the microfluidic device, washed with Mes buffer and perfused with 2-deoxyglucose (100 mM) and different concentrations of glucose followed by Mes buffer washes (D and E). Error bars represent the S.D. for at least 30 cells.

Specific point mutations such as cdc35-13 in CYR1 adenylate cyclase impair growth [29] in different carbon sources at intermediate and restrictive temperatures. This point mutation cdc35-13 also limits the ability of starved yeast to accumulate glucose in the cytosol (Figure 1J). To assess a possible role for cAMP levels in controlling glucose transport activity the levels of cAMP were measured in cdc35-13 cells. Interestingly, the cAMP peak was severely reduced compared with the corresponding wild-type, even at the permissive temperatures (Figure 4C). This result indicates that different levels of cAMP regulate the glucose transport capabilities in starved yeast, most probably through Hxt5.


Nutrient supply fluctuates widely, thus all organisms need to adjust their metabolism to increasing or decreasing concentrations of each essential or beneficial nutrient. Regulatory systems have evolved to adjust the kinetic properties and capacity of transporters and enzymes to the various supply levels to optimize the usage efficiency [33]. Cells thus need to be able to measure supply levels over a wide range, and they have to be able to detect increases or decreases in availability. Over the past decade, much progress have been made regarding how cells measure increasing supply or demand [3436]. Conceptually it is more challenging to understand how cells measure loss or absence of nutrients.

Can yeast cells measure the absence of external glucose, and which regulatory networks are involved in signalling ‘famine’ conditions? Yeast induces specific transporters, namely Hxt5p, when glucose is absent. The presence of such a transporter in the absence of external glucose appears counterintuitive, since its production requires energy, which is limiting, and the transporter is inactive since the substrate is missing. The expense is only useful if the transporter provides a competitive advantage, e.g. if it allows the cells to take up glucose immediately after new resources become available. Cells having such a preformed metabolic pathway can import glucose immediately and use it for its energy demands as well as an intracellular inducer of genes required under ‘feast’ conditions. They are thus in a position to outcompete other organisms.

The critical question was thus whether yeast measures the lack of sugars in the medium to induce the capacity to take up glucose or whether the respective transporters are constitutively present. In the present study it was found that the induction of the capacity to take up sugars in the absence of those sugars depends on the activity of at least three signalling pathways: the Snf1 AMP kinase pathway and both branches of the Gpr1/cAMP/PKA pathway. The Snf1 pathway acts intracellularly, thus being dependent on glucose import, and requires the activation of Snf1p by the Snf1-activating kinases, Sak1p, Tos3p and Elm1p. Therefore the ‘Ajar’ pathway is necessary to allow the Snf1 pathway to respond to an increase in external availability. Snf1p activity is low under high glucose supply conditions, but is derepressed when glucose levels drop. The intracellular receptor and the nature of the molecule leading to derepression in yeast are not fully understood; it has been speculated that either falling glucose 6-phosphate levels or increasing AMP levels lead to activation [3,21,37]. The results of the present study show that Snf1p is an essential part of the ‘Ajar’ pathway that prepares yeast cells during starvation for conditions of ample sugar supply. Apparently, the Snf1 pathway needs to activate Hxt5p in order to fulfil functions in the phase after glucose re-addition.

The involvement of the Gpr1/cAMP/PKA pathway is more puzzling. One simple scenario would be that PKA represses glucose uptake activity. Thus in the absence of glucose, inactive PKA would lead to de-repression of the transporter genes. If this hypothesis were correct, it should be possible to delete upstream parts of the pathway without affecting glucose accumulation. However, the results of the present study show that all of the upstream key players in this pathway are necessary. The Gpr1/cAMP/PKA pathway is well known to regulate the expression of genes involved in the response to exposure of starved cells to high levels of glucose. Although certain aspects of ligand interaction with the receptors are still unresolved, the overall concept is that glucose binds to the G-protein-coupled receptor Gpr1p, which in turn activates the Gα Gpa2p, which activates adenylate cyclase to produce a spike in cAMP, which in turn activates PKA. However, our measurements were carried out with cells that had not seen glucose for many hours, nevertheless glucose accumulation occurred immediately after the addition of glucose, demonstrating a role for this pathway in detecting the absence of glucose. The simplest hypothesis of how the absence of glucose triggers the induction of glucose uptake is that Gpr1p is constitutively active, creating a base state of the signalling pathway. This basic activity leads to the production of low levels of cAMP, creating an activation ‘state I’ of PKA. ‘State I’ could be instated by phosphorylation of one (or one set) of phosphorylation sites at low levels of cAMP. In this ‘state I’, PKA activates Hxt5p and possibly other genes involved in the ‘Ajar’ pathway activity (Figure 3A). After the addition of glucose the same pathway is highly active producing a spike in cAMP, which in turn converts PKA into ‘state II’ leading to activation of a different set of genes involved in acclimation to the new condition and promoting cell division. ‘State II’ may be a different conformation of PKA produced by phosphorylation of a different set of phosphorylation sites. It is noteworthy that Verwaal et al. [9] observed a very weak induction of Hxt5 transcript in ras2 mutants; however, no significant difference in glucose accumulation was observed in a ras2 mutant, possibly implying post-translational control in Hxt5 (Supplementary Figure S6 at Further work will be required to test this hypothesis, specifically whether the pathway is constitutively active in the absence of glucose, whether PKA exists in different phosphorylation states and whether the two states regulate different sets of genes. An interesting question to address will also be how the two pathways co-operate to trigger the ability to accumulate glucose and whether they measure distinct signals, such as lack of extracellular glucose or reduced energy status. Interestingly, both the Ras/PKA and Snf1 pathways are also involved in control of pseudohyphal growth under conditions of nitrogen starvation [38,39]. Since both pathways are predominantly affected by sugars and energy status, the two pathways may act as an input of low carbon supply/energy status into the checkpoint that controls entry into pseudohyphal growth.

In addition, the screen identified several protein kinases playing a role in glucose accumulation, e.g. Pho85p as well as its binding partners Plc6p and Plc7p [40], suggesting that these cyclins could be regulating Pho85p in conditions of limited cytosolic glucose accumulation. Glucose transport assays showed no difference in glucose uptake rates, indicating that Pho85p plays a more important role in controlling metabolism of imported glucose (Figure 2B). Pho85p has been shown to function as an inhibitor of glycogen synthesis, and thus its deletion may induce glycogen accumulation. Alternatively, Pho85p may directly control glycolytic flux, since chemically repressed Pho85p has been shown to up-regulate expression of GLK1 (glucokinase 1) and HXK1 (hexokinase 1) [27]. The lack of Pho85p activity may lead to up-regulation of Glk1p and Hxk1p activity, thus decreasing steady state levels of free glucose in the cytoplasm.

Relevance of ‘Ajar’ pathways for ecology

Colonization of new habitats typically requires an organism to have a competitive edge over other organisms. Whenever competition is intense, specific properties of the individual organisms will determine the final population structure. Outpacing the growth of others allows micro-organisms or plants to successfully colonize new habitats. Population growth, especially in micro-organisms, shows a lag phase controlled by checkpoints in the cell cycle that determine the energy status; only if sufficient nutrients are available will cell division proceed. Thus the earlier checkpoint requirements are fulfilled, the earlier the organism can start to divide and grow exponentially. In its natural habitat yeast experiences cycles of extreme sugar supply and subsequent exhaustion. Yeast cells typically are transferred from carbon-poor environments, such as soil, to new habitats rich in nutrients, such as grapes or other fruit. Saccharomyces is often extremely successful in outpacing other microbes also present on grapes or other fruits [2]. Sudden exposure to high glucose levels is especially important for the industrial inoculation of dry yeast in fresh medium. A substrate-induced pathway, as in the case of the lac operon in Escherichia coli, appears less suited for outcompeting other organisms that have a constitutive lactose uptake capacity. The ‘Ajar’ pathways may be important for the competitiveness of wild strains. It will thus be interesting to determine whether the same ‘Ajar’ pathways exist and are important in isolates from grapes in wineries.

In summary, the use of optical sensors allowed us to explore how yeast cells starved for glucose prepare for exposure to a high sugar supply. S. cerevisiae expresses sugar transporters even in the absence of glucose with three major roles: (i) faster ability to compete for resources, (ii) rapid signalling through intracellular receptors for triggering acclimation to the new status, and (iii) potentially protection from osmotic shock by rapid equilibration of the osmoticum [41]. The screening approach can be used to identify transcription factors that control Hxt5 activity, identify the Hxts responsible for the remaining activity in hxt5Δ, or to identify unknown players in the pathway. Moreover, double mutant screens may help in identifying the connection between the Snf1 and PKA pathways and place the kinases that have not yet been integrated into the network.


Clara Bermejo, Farzad Haerizadeh and Wolf Frommer designed the research; Clara Bermejo, Farzad Haerizadeh, Mayuri Sadoine and Diane Chermak performed the research; Clara Bermejo, Farzad Haerizadeh, Mayuri Sadoine, Diane Chermak and Wolf Frommer analysed the data; and Clara Bermejo, Farzad Haerizadeh and Wolf Frommer wrote the paper.


This work was supported by the National Institutes of Health [grant number 1RO1DK079109].


We thank Professor Martha Cyert (Stanford University, Stanford, CA, U.S.A.) and Dr David Ehrhardt (Carnegie Institution for Science/Stanford University, Stanford, CA, U.S.A.) for comments, advice and several mutants, and Dennis Kunkel Microscopy for the yeast electron microscopy image.

Abbreviations: ACT1, actin 1; Cdc, cell division cycle; CFP, cyan fluorescent protein; Cyr1, cyclic AMP requirement 1; Elm1, elongated morphology 1; FRET, fluorescence resonance energy transfer; GLK1, glucokinase 1; Gpa2, G-protein α subunit 2; Gpr1, G-protein-coupled receptor 1; HXK1, hexokinase 1; Hxt5, hexose transporter 5; Mck1, meiotic and centromere regulatory serine, tyrosine-kinase; Pho85, phosphate metabolism 85; PKA, protein kinase A; Plc, phospholipase C; Q-RT–PCR, real-time quantitative reverse transcription–PCR; Rim11, regulator of inducer of meiosis 2 11; Sak1p, Snf1-activating kinase 1; Snf, sucrose nonfermenting; Tos3p, target of Sbf3; YFP, yellow fluorescent protein


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 29a.
  31. 30.
  32. 31.
  33. 32.
  34. 33.
  35. 34.
  36. 35.
  37. 36.
  38. 37.
  39. 38.
  40. 39.
  41. 40.
  42. 41.
View Abstract