Low Affinity Orthophosphate Carriers Regulate PHO Gene Expression Independently of Internal Orthophosphate Concentration in Saccharomyces cerevisiae*

Phosphate is an essential nutrient that must be taken up from the growth medium through specific transporters. In Saccharomyces cerevisiae, both high and low affinity orthophosphate carriers allow this micro-organism to cope with environmental variations. Intriguingly, in this study we found a tight correlation between selenite resistance and expression of the high affinity orthophosphate carrier Pho84p. Our work further revealed that mutations in the low affinity orthophosphate carrier genes (PHO87, PHO90, and PHO91) cause deregulation of phosphate-repressed genes. Strikingly, the deregulation due to pho87Δ, pho90Δ, or pho91Δ mutations was neither correlated to impaired orthophosphate uptake capacity nor to a decrease of the intracellular orthophosphate or polyphosphate pools, as shown by 31P NMR spectroscopy. Thus, our data clearly establish that the low affinity orthophosphate carriers affect phosphate regulation independently of intracellular orthophosphate concentration through a new signaling pathway that was found to strictly require the cyclin-dependent kinase inhibitor Pho81p. We propose that phosphate-regulated gene expression is under the control of two different regulatory signals as follows: the sensing of internal orthophosphate by a yet unidentified protein and the sensing of external orthophosphate by low affinity orthophosphate transporters; the former would be required to maintain phosphate homeostasis, and the latter would keep the cell informed on the medium phosphate richness.

Phosphate is an essential nutrient that must be taken up from the growth medium through specific transporters. In Saccharomyces cerevisiae, both high and low affinity orthophosphate carriers allow this micro-organism to cope with environmental variations. Intriguingly, in this study we found a tight correlation between selenite resistance and expression of the high affinity orthophosphate carrier Pho84p. Our work further revealed that mutations in the low affinity orthophosphate carrier genes (PHO87, PHO90, and PHO91) cause deregulation of phosphate-repressed genes. Strikingly, the deregulation due to pho87⌬, pho90⌬, or pho91⌬ mutations was neither correlated to impaired orthophosphate uptake capacity nor to a decrease of the intracellular orthophosphate or polyphosphate pools, as shown by 31 P NMR spectroscopy. Thus, our data clearly establish that the low affinity orthophosphate carriers affect phosphate regulation independently of intracellular orthophosphate concentration through a new signaling pathway that was found to strictly require the cyclin-dependent kinase inhibitor Pho81p. We propose that phosphate-regulated gene expression is under the control of two different regulatory signals as follows: the sensing of internal orthophosphate by a yet unidentified protein and the sensing of external orthophosphate by low affinity orthophosphate transporters; the former would be required to maintain phosphate homeostasis, and the latter would keep the cell informed on the medium phosphate richness.
Phosphate is an essential nutrient required for biosynthesis of several cellular components such as phospholipids or nucleotide derivatives. Therefore, many organisms have developed mechanisms allowing them to adapt phosphate utilization to phosphate availability. The yeast Saccharomyces cerevisiae has been used extensively as a model to study how eukaryotic cells respond to variations of external orthophosphate concentrations. Yeast cells are able to take up orthophosphate from the external environment through at least five orthophosphate carriers catalyzing orthophosphate accumulation (Fig. 1) as follows: two high affinity permeases (Pho84p and Pho89p (1)) and three low affinity permeases (Pho87p, Pho90p, and Pho91p (2)). Combined deletion of these five genes is lethal, but viability can be rescued by overexpression of the GIT1 gene that encodes a glycerophosphoinositol permease that can also take up orthophosphate (2). In the presence of organic phosphate in the medium, secreted acidic phosphatases (Pho3p, Pho5p, Pho11p, and Pho12p (3)) are able to transform it into orthophosphate, which can in turn be transported by the orthophosphate permeases.
In response to high external inorganic orthophosphate concentrations, transcription of several genes is repressed (4,5). Among them are the genes coding for excreted phosphatases (PHO5, PHO11, and PHO12), the high affinity orthophosphate carriers (PHO84 and PHO89), GIT1, and a gene encoding a protein required for correct localization of Pho84p in the plasma membrane, PHO86 (6). The other PHO-regulated genes are the PHO81 family members (PHO81, SPL2, and YPL110C) encoding potential cyclin inhibitors and several genes coding for various phosphate metabolism enzymes such as the PHM genes ( Fig. 1) that are involved in polyphosphate (polyP i ) 1 synthesis or degradation (7).
Differential phosphorylation of the Pho4p transcription factor by the Pho80p-Pho85p cyclin-dependent kinase complex allows transcriptional regulation of the PHO regulon in response to variations of external orthophosphate concentration (8). When external concentration of orthophosphate is low, the cyclin-dependent protein kinase inhibitor Pho81p, which is constitutively bound to the Pho80p-Pho85p complex, becomes active resulting in low phosphorylation of Pho4p by the cyclincyclin-dependent protein kinase complex (9 -11). This low phosphorylated form of Pho4p has a high affinity for the nuclear importin Pse1p and a low affinity for the nuclear exportin Msn5p and is therefore preferentially located in the nucleus (10,(12)(13)(14), where Pho4p cooperates with Pho2p, another transcription factor (15), to activate expression of the target genes. When the external concentration of orthophosphate is high, Pho81p effect on Pho85p is lowered. Consequently, Pho4p is hyperphosphorylated by the Pho80p-Pho85p complex and is therefore mainly located in the cytoplasm, resulting in a decreased expression of the PHO regulon.
Although the late molecular events in the PHO pathway are relatively well characterized, little is known about how the external concentration of orthophosphate is sensed by the cells. Because PHO5 expression is constitutive in a pho84-defective mutant (16), the possibility of a coupling between orthophosphate transport and sensing has been investigated recently (2). However, it was clearly shown that Pho84p is not an inorganic orthophosphate sensor because derepression of PHO5 is only a consequence of a reduced orthophosphate uptake in a pho84⌬ mutant. Indeed, overexpression of any other orthophosphate carrier restored normal Pho5p levels (2). Consistently, it has been shown recently (17) that expression of PHO5 was tightly correlated to intracellular orthophosphate concentration as measured by 31 P NMR spectroscopy. Furthermore, these authors (17) have also established that internal orthophosphate concentration was low in a pho84⌬ mutant, thus confirming that Pho84p is most likely to affect phosphate signaling through its uptake capacity.
By screening a yeast knock-out collection for resistance to sodium selenite, we found a striking correlation between expression of PHO84 and selenite resistance. Furthermore, we show that mutants affected in the low affinity orthophosphate transporter genes (PHO87, PHO90, and PHO91) are unable to maintain PHO84 repression under high orthophosphate conditions. In addition, we demonstrate that derepression of PHO84 in the low affinity orthophosphate permease mutants does not correlate with an impaired orthophosphate uptake capacity nor with a decreased cytosolic orthophosphate and/or polyP i concentrations.
Growth Test-Yeast strains were grown at 30°C overnight on plates and then replica-plated on fresh medium and grown for an additional 6 h. Cells were then resuspended in sterile water to an A 600 (absorbance at 600 nm) ϭ 0.5 and submitted to 1/10 serial dilutions. Drops (5 l) of each dilution were spotted on freshly prepared plates containing low or high orthophosphate SDcasaWA or YPD media and supplemented or not with sodium selenite at different concentrations. Plates were incubated for 48 -72 h at 30°C, depending on the medium used.
Orthophosphate Uptake-Yeast strains were grown in SDcasaWA medium to A 600 ϭ 1. Cells harvested by centrifugation were washed twice in buffer A (50 mM potassium citrate, pH 4.5, 4% glucose) and resuspended in the same buffer at A 600 ϭ 1. Uptake measurement was started by addition of KH 2 [ 32 P]PO 4 (70 l, 10 -10,000 M, specific radioactivity 5-1,000 Ci/mol) to 630 l of cells suspension. Cells (200-l aliquots) were taken at 1-3 min (period of time during which the uptake was linear, see this work and Ref. 2), filtered immediately on 0.8-m filters (Supor®-800, Pall Life Science), and washed with 3 ml of ice-cold 0.5 M KH 2 PO 4 buffer, pH 4.5. Radioactivity was measured by scintillation counting. Uptake parameters were determined by a nonlinear regression of the saturation curve using the Graph-Pad software. 31 P NMR Spectroscopy-Yeast strains were grown overnight in 100 ml of YPD medium and were harvested at A 600 ϭ 1. Cells were centrifuged, and excess medium was removed to obtain a 500-l cellular suspension. D 2 O (100 l), and methylene diphosphonate, pH 6 (1 mol), was added to the yeast suspension. 31 P NMR spectroscopy analysis was performed with a Bruker Avance 500 narrow-bore spectrometer equipped with a 5-mm broad band probe. The 31 P NMR signal was recorded (512 scans) with the following parameters: 5-s pulse (ϳ53°f lip angle), 0.81-s acquisition time, 0.5-s relaxation delay, 10,080-Hz sweep width, 16 K memory size, Waltz 16 proton decoupling, and D 2 O lock. The free induction decays were converted by Fourier transformation after applying a 4-Hz line broadening and a 32 K zero filling. Chemical shift scale was calibrated using 0 ppm for 85% phosphoric acid. Peak areas were determined with the Bruker integration routine using the methylene diphosphonate peak as reference. Peak areas were corrected for partial saturation using T 1 values for the 31 P nuclei in the various compounds as already described (25). To calculate cellular orthophosphate concentration from NMR spectra, duplicates of 200-l aliquots of cells suspension used for NMR spectroscopy were dried at 80°C for 48 h to determine the dry weight of the cells. l/mg dry weight. 31 P NMR measurements for cytosolic orthophosphate and polyP i (contained in the PP 4-n peak, see Fig. 7) are given as millimolar of orthophosphate residues. (27), we had studied the effect of sodium selenite on yeast cells and isolated two gene dosage suppressors. To identify new yeast genes required for sodium selenite sensitivity, a collection of 4787 yeast haploid knock-out mutants were transferred to solid YPD medium supplemented or not with sodium selenite (5 and 10 mM). After 4 days at 30°C, pho87⌬ and yhr202w⌬ mutants were clearly more resistant to sodium selenite than the wild type. Serial dilution on sodium selenite-containing medium further indicated that the pho87⌬ mutant was slightly more resistant than yhr202w⌬ ( Fig. 2A). Because the function of the YHR202w gene product is unknown, we focused our attention on the PHO87 gene that encodes a low affinity orthophosphate carrier (2,28). We showed that the introduction of a plasmid carrying the PHO87 gene in the pho87⌬ strain restored a wild-type sensitivity to sodium selenite of this mutant (data not shown). Mutant strains for the two other low affinity orthophosphate transporters (Pho90p and Pho91p) also displayed an increased sodium selenite resistance (Fig. 2B). However, resistance of the double or triple mutants was not markedly increased compared with the single mutant strains (data not shown).

Screening a Collection of Yeast Knock-out Mutants for Sodium Selenite Resistance Reveals a Link between Orthophosphate Uptake and Selenite Sensitivity-In a previous work
To gain further insight into relationships between sodium selenite and orthophosphate assimilation, several mutants involved in various aspects of phosphate metabolism were tested for sodium selenite resistance. Clearly, regulators of the PHO pathway, pho2⌬, pho4⌬, pho81⌬, and pho86⌬, were highly sensitive to sodium selenite (Fig. 2C), as was the strain deleted for the high affinity orthophosphate carrier gene PHO84. Most important, addition of a high concentration (25 mM) of orthophosphate to the growth medium alleviated sodium selenite sensitivity of all the pho⌬ mutants and improved resistance of the wild-type strain to the drug (Fig. 2C), thus confirming the clear link between orthophosphate metabolism and sodium selenite sensitivity.
Sensitivity to Sodium Selenite Correlates with Expression of PHO84 -All the sodium selenite-sensitive mutants described in the previous section alter the expression or the localization of Pho84p. Indeed, it has been shown previously that PHO84 is not expressed in pho2⌬, pho4⌬, pho81⌬, and pho84⌬ single mutant strains (Fig. 3A) (29), and Pho84p is not correctly localized in the pho86⌬ mutant strain (6). The expression of PHO84 was monitored by Northern blot on yeast cells grown in high orthophosphate YPD medium (repression condition) (Fig.  3A). We found that the low basal expression of PHO84 is dependent on the presence of PHO2, PHO4, and PHO81 genes (Fig. 3A), as reported previously (30). Strikingly, the PHO84 Resistance/sensitivity to sodium selenite is tightly correlated to PHO84 expression levels. A, PHO84 expression in various mutant strains was monitored by Northern blot analysis. Cells were grown in SDcasaWA medium (high phosphate condition) to an A 600 ϭ 1 and submitted (ϩ) or not (Ϫ) to 30 min of sodium selenite treatment (10 mM final concentration). Hybridizations were done independently and were assembled for the figure. B, serial dilutions of exponentially growing cells spotted onto standard YPD medium containing or not sodium selenite (5 mM). C, serial dilutions of BY4742 cells transformed with either pCM189 (vector), P2053 (tet-PHO84), or P2079 (tet-PHO89) plasmids were spotted on SDcasaWA medium supplemented or not with tetracycline (100 mg/liter) and containing or not sodium selenite (2,5 mM). D, PHO84 and PHO89 expression driven by the Tet-regulable promoter was monitored by Northern blot as in A. transcript was much more abundant in the sodium seleniteresistant pho87⌬ mutant than in wild-type cells (Fig. 3A). Most important, PHO84 expression was not affected by selenite itself (Fig. 3A).
We reasoned that if sodium selenite resistance in the pho87⌬ mutant is due to PHO84 overexpression, it should be abolished in a pho84⌬pho87⌬ double mutant. Indeed, such a double mutant behaved as the single pho84⌬ mutant on sodium selenite-containing medium (Fig. 3B). Thus, Pho84p is necessary for the sodium selenite resistance of the pho87⌬ mutant, and overexpression of PHO84 in the pho87⌬ mutant should be the primary cause for sodium selenite resistance. Indeed, the PHO84 gene placed under the control of a tetracycline-repressible promoter led to sodium selenite resistance in the absence, but not in the presence, of tetracycline (Fig. 3C). In contrast, overexpression of Pho89p, the other high affinity orthophosphate carrier (Fig. 3D), did not result in sodium selenite resistance (Fig. 3C). Therefore, our data reveal a tight and specific correlation between PHO84 expression and sodium selenite resistance.
Finally, several general transcription machinery mutants, such as snf2⌬, gcn5⌬, spt7⌬, and spt2⌬, in which the expression of PHO84 was reported previously (31)(32)(33) to be dramatically decreased, were found to be sensitive to sodium selenite (data not shown), further validating the correlation between PHO84 expression and sodium selenite resistance.
Expression of PHO-regulated Genes Is Derepressed in the Presence of Inorganic Orthophosphate in the pho87⌬, pho90⌬, and pho91⌬ Mutants-The data presented in the previous section revealed a role for PHO87 in the regulation of PHO84 expression under high orthophosphate conditions. To further characterize this effect, the expression of PHO84 was monitored in strains carrying the pho87⌬, pho90⌬, and pho91⌬ deletions, either alone or combined. In these mutants, levels of PHO84 transcript were unaffected when cells were grown on low orthophosphate medium, whereas a clear derepression of the PHO84 transcript was observed in the mutant cells under high orthophosphate conditions (Fig. 4A). Transcription of another phosphate-regulated gene, SPL2, was also derepressed in the low affinity carrier mutants, establishing that the derepression mechanism does not specifically affect PHO84 (Fig. 4A). Finally, PHO5 expression monitored in a pho3⌬pho90⌬ double mutant (PHO5 expression cannot be monitored by Northern blot in a PHO3 background due to cross-hybridization between the two genes and constitutive expression of PHO3) was much higher than in the pho3⌬ single mutant (Fig. 4B). We conclude that mutations in the low affinity orthophosphate carriers result in derepression of the PHO-regulated genes.
Phosphate Repression Is Set Up Normally but Cannot be Maintained in the Absence of the Low Affinity Orthophosphate Transporters-Our results suggest that the presence of Pho87p, Pho90p, and/or Pho91p is required for normal phos-phate repression of the PHO-regulated genes. To get a more dynamic vision of the phosphate repression phenomenon, expression of PHO84 in response to orthophosphate addition was monitored more precisely. When wild-type, pho90⌬, and pho87⌬pho90⌬pho91⌬ cells were shifted from low orthophosphate (Ͻ100 M) to high orthophosphate (5 mM) medium, the amount of PHO84 transcript dropped by more than 95%. This repression level was maintained after 2 h in the wild-type strain, whereas in the pho90⌬ and the pho87⌬pho90⌬pho91⌬ mutants, PHO84 repression was not conserved (Fig. 5). In fact, after 2 h, in the triple mutant strain the PHO84 gene expression was similar to the level observed in nonrepressible mutants such as pho80⌬ and pho85⌬ (Fig. 5).
Derepression of PHO84 in the Low Affinity Carrier Mutants Does Not Correlate with Decreased Orthophosphate Uptake-A simple explanation for derepression of PHO84 in the pho87⌬, pho90⌬, or pho91⌬ mutants could be that uptake capacity is lowered in the absence of the low affinity transporters, leading to constitutive orthophosphate starvation, as reported previously (2) for pho84⌬ mutants. Orthophosphate uptake measurements in wild-type, pho84⌬, and pho87⌬pho90⌬pho91⌬ mutant strains were performed on cells grown under high orthophosphate conditions (Fig. 6A). Uptake parameters obtained in these conditions were determined by nonlinear regression. Saturation curves obtained for the wild-type and the pho84⌬ mutant strains were correctly fitted with a single hyperbola regression with correlation coefficients Ͼ0.99. The kinetic parameters thus determined were K T ϭ 1.0 Ϯ 0.1 mM and V max ϭ 0.70 Ϯ 0.03 nmol of orthophosphate incorporated per min/10 7 cells for the wild-type strain and K T ϭ 0.86 Ϯ 0.17 mM and V max ϭ 0.40 Ϯ 0.02 nmol of orthophosphate incorporated per min/10 7 cells for the pho84⌬ mutant strain. In contrast, the pho87⌬pho90⌬pho91⌬ triple mutant curve could not be fitted with a single hyperbola but was correctly fitted with a double hyperbola regression curve with the following kinetic parameters: K T1 ϭ 12.0 Ϯ 0.1 M and V max1 ϭ 0.49 Ϯ 0.01 nmol of orthophosphate incorporated per min/10 7 cells; K T2 ϭ 3.6 Ϯ 1.5 mM and V max2 ϭ 1.5 Ϯ 0.7 nmol of orthophosphate incorporated per min/10 7 . Altogether, these results revealed that maximal uptake velocity (V max ) was twice lower in the pho84⌬ mutant as compared with wild type (Fig. 6A), in good agreement with results published previously (2). However, in the pho87⌬pho90⌬pho91⌬ triple mutant strain, the V max measured (ϭV max1 ϩ V max2 ϭ 2 nmol of orthophosphate incorporated per min/10 7 ) was higher than that of the wild-type strain, thus indicating that orthophosphate uptake capacity is clearly not diminished but rather increased in this strain (Fig. 6A). In the pho87⌬pho90⌬pho91⌬ triple mutant, the K T1 for orthophosphate uptake most probably reflects the strong transcriptional derepression of the PHO84 gene that encodes a high affinity permease.
Strikingly, expression of a PHO84-lacZ fusion was found more than three times higher in the triple mutant than in the pho84⌬ mutant (Fig. 6B), although the triple mutant had much higher orthophosphate uptake capacities (Fig. 6A). Therefore, although in the wild-type and pho84⌬ strains the PHO84-lacZ expression is correlated to orthophosphate uptake capacity, in the pho87⌬pho90⌬pho91⌬ triple mutant an orthophosphate uptake capacity comparable with that of the wild-type strain does not allow transcriptional repression of the PHO regulon. Altogether, these results show that derepression of PHO-regulated genes is not correlated to impaired orthophosphate uptake capacities in the absence of Pho87p, Pho90p, and/or Pho91p.

Absence of Low Affinity Orthophosphate Transporters Affects Expression of PHO-regulated Genes Independently of Internal
Orthophosphate Concentration-Because orthophosphate uptake might not tightly reflect internal orthophosphate concentration, 31 P NMR spectroscopy was used to investigate whether the derepression of PHO-regulated genes observed in the pho87⌬, pho90⌬, pho91⌬, and pho87⌬pho90⌬pho91⌬ mutants could be due to a modification of internal orthophosphate or polyP i concentrations.
We first confirmed that, in our strain background, a much lower internal orthophosphate concentration was measured in wild-type cells grown under low orthophosphate conditions compared with high orthophosphate medium (data not shown). A severe decrease of the internal orthophosphate and polyP i concentrations were also observed in the pho84⌬ mutant in high orthophosphate condition (Fig. 7), as reported previously (17). This severe drop of phosphate pools, which is probably responsible for PHO gene derepression in the pho84⌬ mutant strain, was fully reversed by reintroduction of the PHO84 gene on a centromeric plasmid (data not shown).
Measurements of intracellular orthophosphate and polyP i pools in the pho80⌬ mutant strain revealed that, under high phosphate conditions, the cytosolic orthophosphate concentration was similar to that of the wild type, whereas a large accumulation of polyP i was observed (Fig. 7). Therefore, as expected for such a nonrepressible signaling mutant, a disconnection between the variation of phosphate pools and the expression of PHO genes was observed.
Finally, measurements of the phosphate pools in single and triple low affinity orthophosphate carrier mutants showed that both cytosolic orthophosphate and polyP i concentrations were at least equivalent or higher in these mutants than those measured in wild-type cells when cells were grown in high orthophosphate medium. These results clearly show that the derepression of PHO-regulated genes observed in these mutants is not correlated to drastic changes in phosphate pools but rather is similar to the situation observed in the pho80⌬ regulatory mutant in which the expression and variation of the phosphate pools for the PHO genes are disconnected. Together, our results indicate that low affinity orthophosphate carriers Transformants were grown in SDcasaWA medium (high phosphate conditions) to A 600 ϭ 1, washed with transport buffer, and submitted to uptake measurement as described under "Experimental Procedures." Results correspond to the mean of initial rate of uptake measured in at least three independent experiments. B, yeast strains were transformed with the PHO84-lacZ fusion plasmid. Transformants were grown in SDcasaWA medium to A 600 ϭ 1, and ␤-galactosidase (␤gal) measurements were performed as described under "Experimental Procedures. "   FIG. 7. Derepression of PHO-regulated genes in the absence of low affinity orthophosphate transporters is not linked to impaired internal orthophosphate or polyP i pools. Yeast strains were grown in YPD medium to A 600 ϭ 1, and internal orthophosphate and polyP i pools were estimated by 31 P NMR spectroscopy as described under "Experimental Procedures." A, overlapping of 31 P NMR spectra, each one being a representative spectrum for each strain tested. Methylene diphosphonic acid (MDP) was used as a reference. PP n indicates phosphate residue at the nth position in polyphosphate chains. B, orthophosphate (white bars) and polyP i (gray bars) indicate intracellular levels determined by 31 P NMR spectroscopy. Results presented correspond to the mean of three independent experiments for each strain. Polyphosphate levels were expressed in terms of orthophosphate residues determined from the PP 4-n peak. WT, wild type. affect PHO84 expression by a mechanism independent of internal orthophosphate concentration.

Low Affinity Orthophosphate Carriers Act Upstream of Pho81p in Regulation of the PHO Pathway, and Their Effect Is
Independent of Intracellular Orthophosphate Sensing-To describe further the regulatory role of the low affinity orthophosphate carriers in the PHO pathway, we intended to place these new regulators in the previously characterized phosphate signal transduction pathway.
We then addressed the question of whether pho87⌬, pho90⌬, and pho91⌬ mutants are still able to respond to intracellular orthophosphate variations. This was done by taking advantage of the fact that a phm3 (vtc4) mutant is unable to synthesize polyphosphates (4). Consequently, this mutant accumulates orthophosphate in the cytoplasm and shows lower expression of the PHO regulon compared with the wild-type strain (17). The phm3⌬ mutant can thus be used to artificially increase cytosolic orthophosphate concentration.
The pho87⌬ and pho90⌬ mutations were individually combined to phm3⌬ mutation to determine whether an increased internal orthophosphate concentration would affect PHO84 expression in the absence of the low affinity orthophosphate carriers. PHO84 expression and cytosolic orthophosphate levels were measured in single and combined mutants by Northern blot and 31 P NMR spectroscopy, respectively (Fig. 9). Under high orthophosphate conditions, the phm3⌬ single mutant displayed a PHO84 transcript level lower than the wild type (Fig.  9, A and B), an increased level of cytosolic orthophosphate (Fig.  9B), and no detectable polyP i (data not shown) as shown previously (17). The combination of phm3⌬ with either pho87⌬ or pho90⌬ mutations led to PHO84 expression levels similar to or higher than that of the wild-type strain, respectively (Fig. 9, A  and B), whereas cytosolic orthophosphate concentrations were very close to those measured in the phm3⌬ single mutant (Fig.  9B). As expected, because Phm3p is essential for polyphosphate synthesis (see Fig. 1) (4,17), polyP i was not detectable in the phm3⌬pho87⌬ and phm3⌬pho90⌬ double mutants (data not shown). Therefore, high internal orthophosphate levels led to tight repression of PHO84 in the phm3⌬ mutant, whereas the same internal orthophosphate levels were unable to maintain the repression of PHO84 in the phm3⌬pho90⌬ double mutant. These results further confirm that signaling through the low affinity orthophosphate transporter is independent of internal orthophosphate concentration. DISCUSSION Our results revealed a tight correlation between PHO84 expression and sodium selenite resistance. However, sodium selenite toxicity was reversed by orthophosphate even in the absence of Pho84p. Altogether, our results indicate that orthophosphate internalization itself is critical for resistance. Most important, we found that sodium selenite does not compete with either low or high affinity orthophosphate uptake systems even in the presence of a 10-fold molar excess of selenite (data not shown). Surprisingly, the pho80⌬ and pho85⌬ mutants known to constitutively overexpress the PHO84 gene were only slightly more resistant to sodium selenite than the wild-type strain (data not shown). However, we observed that in the pho80⌬ and pho85⌬ mutants, oxidative stress-response genes (such as GLR1, GSH1, TRR1, and TRX2) were not induced in response to sodium selenite, 2 whereas these genes are strongly induced in the wild-type strain (27). Thus, the moderate resistance to sodium selenite of the pho80⌬ and pho85⌬ mutants could reflect a balance between increased resistance due to PHO84 overexpression and increased sensitivity due to poor induction of oxidative stress-response genes. How orthophosphate leads to selenite detoxification remains to be clarified. However, our selenite toxicity studies revealed new phenotypes for mutants in the phosphate utilization pathway. In particular, this work provides the first growth phenotype for the single pho87⌬, pho90⌬, and pho91⌬ mutants. In the future, this growth phenotype could be useful for identifying new mutants affecting PHO84 expression.
Here we show that PHO87, PHO90, and PHO91, which encode low affinity orthophosphate permeases (2), play an important role in the regulation of the PHO genes. Recently, similar results were reported by others; however, in this report the authors observed derepression of PHO5 in the low affinity orthophosphate carrier mutants only for intermediary orthophosphate concentrations (34), whereas we observed the derepression even at high orthophosphate concentrations. This discrepancy could be due to strain differences because Harashima and co-workers (17) found much higher polyphosphate levels in their wild-type or pho84⌬ mutant strains than we did in the BY4742 strain or derived pho84⌬ mutant. To clarify this discrepancy, it would be interesting to measure orthophosphate concentrations in the low affinity carrier mutants, used by Harashima and co-workers (34), under various external orthophosphate conditions. However, despite this difference, it is clear from both reports that the expression of PHO genes is affected in the low affinity carrier mutants.
The question thus arises as to how the pho87⌬, pho90⌬, and pho91⌬ mutations, either alone or in combination, affect PHO84 expression. Because Pho87p, Pho90p, and Pho91p are, presumably, integral membrane proteins, their effect on the transcription of PHO genes is likely to be indirect. In a simple model, the role of Pho87p, Pho90p, and Pho91p would principally be to sustain intracellular orthophosphate concentration when orthophosphate concentration is high in the medium and PHO84 expression is low. Consequently, in these mutant strains, internal orthophosphate concentration would be lower, leading to derepression of PHO84 despite a high extracellular orthophosphate concentration. Our measurements of internal orthophosphate concentration in the pho87⌬, pho90⌬, and pho91⌬ mutant strains do not support this model because internal phosphate concentration (in the form of either free orthophosphate or polyphosphates) was found equivalent or even higher in the mutants than in the wild type. Therefore, our results demonstrate that the derepression of PHO84 expression in the pho87⌬, pho90⌬, and pho91⌬ mutants is not the result of an uptake defect that leads to orthophosphate starvation. We rather favor the hypothesis of low affinity carriers participating in a signaling pathway independently of the previously documented internal orthophosphate signaling pathway.
Clearly, in the wild-type strain there is a tight correlation between internal free orthophosphate concentration and PHO84 expression (17). Consistently, mutations affecting internal free orthophosphate concentration such as pho84⌬ or phm3⌬ also affect PHO84 expression (see Ref. 17 and this work). However, in regulatory mutants such as pho80⌬, there is not any correlation between internal orthophosphate concentration and PHO84 expression. Strikingly, the pho87⌬, pho90⌬, and pho91⌬ mutants behaved more like regulatory mutants than uptake defective mutants. How could those membrane proteins affect PHO84 expression independently of their role in orthophosphate uptake? One attractive possibility would be that Pho87p, Pho90p, and Pho91p are both orthophosphate transporters and sensors. Indeed, as other membrane sensors (35), all three proteins have a long hydrophilic tail that could be involved in signaling (36). Strikingly, the N-terminal extremity of Pho91p shows similarities with Pho81p, a major regulator of the Pho85p cyclin-dependent kinase. Furthermore, all three low affinity transporters, like Pho81p, carry a socalled SPX domain (Syg1p, Pho81p, and XPR1) that could be involved in G-protein-associated signal transduction (37)(38)(39).
The reason for having three phosphate membrane sensors, which appear partially functionally redundant, is not yet clear. However, in all experiments the pho90 deletion had a stronger effect than deletion of pho87 or pho91, indicating that they are not strictly equivalent. Most important, this conclusion was only valid for cells in exponential growth; indeed during postdiauxic phase pho87 and pho91 deletions have a greater effect than pho90 mutation. 1 Therefore, each of the three sensors could play its role at a specific stage of growth.
Most important, we have shown that signaling through Pho87p, Pho90p, and Pho91p strictly requires Pho81p. Mutants lacking Pho81p are nonderepressible, indicating that they cannot respond to orthophosphate variations. Consistently, a pho81⌬ mutation is epistatic to a pho84⌬ mutation (16). Therefore, Pho81p is a critical protein for internal phosphate sensing. Our data further suggest that Pho81p is also critical for signaling through the low orthophosphate affinity carriers. Pho81p, is a large protein carrying a cyclin inhibitorlike domain. Although a minimal inhibitory domain has been characterized (11), other parts of the protein could be important for integration of various signals. Our repression kinetic data (Fig. 4) show that in the pho87⌬, pho90⌬, and pho91⌬ mutants, phosphate repression can take place in the absence of Pho87p, Pho90p, and/or Pho91p but cannot be maintained. We propose that abrupt addition of orthophosphate to the medium results in an intracellular orthophosphate burst leading to repression via the action of intracellular phosphate sensor(s) indicating that the internal phosphate sensing pathway is still functional in the triple mutant. Consistently, the effect of the pho87⌬ and pho90⌬ mutations was partially compensated by that of the phm3⌬ mutant. Therefore, we propose that the low affinity orthophosphate carriers somehow signal to Pho81p independently of the previously described internal phosphate response pathway. Pho81p would then integrate these two signals and modulate expression of the PHO regulon.
Why should there be an external phosphate signaling pathway in addition to the internal phosphate sensing pathway? The intracellular phosphate concentration cannot vary extensively without affecting essential cellular processes, and yeast cells were shown to mobilize polyphosphates when internal orthophosphate concentration drops (40). Therefore, some mutants such as pho84⌬, which have a low internal phosphate concentration, lack polyphosphates (see Ref. 4 and this work). It becomes clear that polyphosphates are used by the cell as a phosphate "stock" (40), and consequently internal free orthophosphate concentration might not be adequate to accurately signal phosphate medium richness. Because external orthophosphate concentration is an important parameter for the decision of the cell to enter or not to enter into a new cycle, an attractive hypothesis would be that the low affinity orthophosphate carriers could somehow sense external orthophosphate concentration and send a mitogenic signal mediated by Pho81p to the cyclin-dependent kinase Pho85p.