|Home | About | Journals | Submit | Contact Us | Français|
Saccharomyces cerevisiae cells possess a plasma membrane sensor able to detect the presence of extracellular amino acids and then to activate a signaling pathway leading to transcriptional induction of multiple genes, e.g., AGP1, encoding an amino acid permease. This sensing function requires the permease-like Ssy1 and associated Ptr3 and Ssy5 proteins, all essential to activation, by endoproteolytic processing, of the membrane-bound Stp1 transcription factor. The SCFGrr1 ubiquitin-ligase complex is also essential to AGP1 induction, but its exact role in the amino acid signaling pathway remains unclear. Here we show that Stp1 undergoes casein kinase I-dependent phosphorylation. In the yck mutant lacking this kinase, Stp1 is not cleaved and AGP1 is not induced in response to amino acids. Furthermore, we provide evidence that Ssy5 is the endoprotease responsible for Stp1 processing. Ssy5 is significantly similar to serine proteases, its self-processing is a prerequisite for Stp1 cleavage, and its overexpression causes inducer-independent Stp1 cleavage and high-level AGP1 transcription. We further show that Stp1 processing also requires the SCFGrr1 complex but is insensitive to proteasome inhibition. However, Stp1 processing does not require SCFGrr1, Ssy1, or Ptr3 when Ssy5 is overproduced. Finally, we describe the properties of a particular ptr3 mutant that suggest that Ptr3 acts with Ssy1 in amino acid detection and signal initiation. We propose that Ssy1 and Ptr3 form the core components of the amino acid sensor. Upon detection of external amino acids, Ssy1-Ptr3 likely allows—in a manner dependent on SCFGrr1—the Ssy5 endoprotease to gain access to and to cleave Stp1, this requiring prior phosphorylation of Stp1 by casein kinase I.
Saccharomyces cerevisiae can detect the presence of amino acids in its extracellular environment and then induce transcription of a set of genes encoding permeases that mediate the uptake of these amino acids into the cell (12, 24). One such gene is AGP1, encoding a broad-specificity amino acid permease (31). Several proteins play an essential role in this nutritional signaling pathway. One is Ssy1, an amino acid permease homologue apparently devoid of transport activity and likely involved in recognition of external amino acids. Ssy1 differs from classical amino acid permeases by the presence of an unusually large cytosolic N-terminal domain that plays an important role in signaling (19, 31, 33, 35). Consistent with a direct role of Ssy1 in amino acid detection, a mutant form of Ssy1 (encoded by ssy1-23) has selectively lost the ability to respond to several amino acids (7). Furthermore, a constitutive mutant form of Ssy1 (encoded by SSY1382K) has recently been described (25). Ptr3 and Ssy5, two peripheral membrane proteins without any predicted transmembrane domain, are also essential to induction of amino acid permeases in response to external amino acids (7, 33, 35). Although these proteins have been shown to interact in two-hybrid assays (7), their exact biochemical role in amino acid signaling remains unclear. Several lines of evidence indicate that the Ssy1, Ptr3, and Ssy5 proteins are associated in a plasma membrane-associated complex named SPS (23). Induction of AGP1 depends on the synergistic action of two transcription factors, Stp1 and Uga35/Dal81 (1, 3, 31, 34). In the absence of amino acids, Stp1 is located at the cell surface. When amino acids are present, Stp1 undergoes Ssy1-, Ptr3-, and Ssy5-dependent endoproteolytic processing (3). The released C-terminal domain then translocates into the nucleus (3) and acts, together with Uga35/Dal81, through a common GC-rich upstream sequence named UASAA, to activate AGP1 transcription (1). Similar UASAA elements promote SPS-dependent induction, by amino acids, of other amino acid permease genes, e.g., BAP2 and BAP3 (16, 17, 45). Recent work indicates that Uga35/Dal81 is also activated in response to amino acids, but the mechanism involved remains unknown (1). Finally, transcriptional induction of AGP1 mediated by processed Stp1 and Uga35/Dal81 may be amplified severalfold by the Gln3 protein (1). This GATA family transcription factor acts through 5′-GATA-3′ core sequences and is specifically active in cells grown under limiting nitrogen supply conditions (15). Other key components of the external amino acid signaling pathway are ubiquitin and the SCFGrr1 ubiquitin-ligase complex (7, 8, 31). Transcription of AGP1 is largely defective in the doa4/npi2 mutant, where the internal pool of ubiquitin is severely reduced. This deficiency can be compensated by overproduction of ubiquitin. Furthermore, when the F-box protein Grr1 is absent, induction of AGP1 transcription is totally abolished. Thermosensitive mutations affecting other components of the SCF complex also markedly reduce AGP1 transcription at the nonpermissive temperature (7, 8, 31).
Here we report that casein kinase I (CKI) is involved in Stp1 phosphorylation and that this modification is a prerequisite to Stp1 activation by endoproteolytic processing. We further provide evidence that Ssy5 is the endoprotease that mediates Stp1 processing. Ssy5 indeed shares sequence similarity with several serine proteases and is self-processed. Its overproduction triggers Stp1 cleavage even when amino acids are not available and irrespective of whether the Ssy1, Ptr3, and SCFGrr1 elements are functional. We also describe a particular ptr3 mutant which has lost the ability to respond to some but not all amino acids, a phenotype consistent with Ptr3 acting with Ssy1 in amino acid detection and signal initiation. A model integrating these novel data is discussed.
Most S. cerevisiae strains used in this study derive from the wild-type Σ1278b strain (9), except for the yckts, cdc4-1, and cdc34-1 mutants, for which the corresponding wild types were used (Table (Table1).1). Cells were grown in a minimal buffered (pH 6.1) medium with 3% glucose, raffinose, or galactose (when mentioned) as the carbon source. To this medium, urea (5 mM), (NH4)2SO4 (10 mM), proline (5 mM), another amino acid (1 to 10 mM), or combinations of these compounds were added as a source(s) of nitrogen. Assays for resistance to the toxic amino acid analogue d,l-ethionine (20 μg/ml) were carried out on plates with (NH4)2SO4 as the sole nitrogen source.
All procedures for manipulating DNA were standard ones (4, 52). The Escherichia coli strain used was JM109. Plasmid pCA047 bearing Stp1-hemagglutinin (HA) was a generous gift of P. Ljungdahl (3). Plasmid YCpAGP1-lacZ has been described elsewhere (31). The YCp-ptr3-35 plasmid was isolated by inserting into BamH1-EcoRI-cleaved plasmid pFL38 a 3,288-bp fragment bearing the ptr3-35 allele. This fragment was obtained by PCR amplification, using genomic DNA of strain FB35 as template and primers 5-PTR3-C and 3-PTR3-C (Table (Table2).2). Plasmid pFA153 is a centromere-based URA3 vector bearing the SSY5 gene (with 954 bp of upstream and 2,364 bp of downstream sequences) cloned from strain Σ1278b. The pFA138SSY5 plasmid was constructed by cotransforming cells of strain 23344c with the SacI-cleaved pFA153 plasmid (SacI cuts upstream from the insertion site of SSY5 sequences) and a 641-bp DNA fragment made of the GAL1 gene promoter followed by the triple HA tag-coding DNA. This DNA fragment (flanked by appropriate 40-bp recombination sequences) was obtained by PCR with plasmid pFA6a-KanMX6-PGAL1-3HA as template (57) and primers 5GALHASSY5 and 3SSY5HA (Table (Table2).2). Plasmid pFA150 is the same as pFA153 except that the encoded Ssy5 protein is tagged at its N terminus by the triple HA tag. It was constructed by cotransforming cells of the 23344c strain with Pme1-cleaved pFA153 plasmid (Pme1 cuts just downstream from the ATG initiation codon of SSY5) and a 100-bp DNA fragment made of the triple HA tag-coding DNA. This DNA fragment (flanked by appropriate 40-bp recombination sequences) was obtained by PCR using plasmid pFA6a-KanMX6-PGAL1-3HA as template (57) and primers 5SSY5NtHA and 3SSY5HA (Table (Table2).2). Plasmid pFA154 is the same as pFA153 and plasmid pFA144 is the same as pFA138, except that the codon corresponding to the serine at position 640 has been replaced by the GCA triplet corresponding to an alanine (Ssy5S→A). Both plasmids were constructed with the QuikChange site-directed mutagenesis kit (Stratagene) and primers 5-SSY5-mutS-A and 3-SSY5-mutS-A (Table (Table2).2). Plasmid pFA148 is a derivate of pFA138 in which the DNA region spanning positions 1560 to 1578 relative from the ATG initiation codon was deleted using primers 5-SSY5ΔPISML and 3-SSY5ΔPISML (Table (Table2).2). The accuracy of each mutagenized plasmid was checked by sequencing.
The gap1Δ ssy1Δ,, gap1Δ grr1Δ, and pdr5Δ strains were constructed by the PCR-based gene deletion method (57). The DNA segments used to introduce these mutations were generated by PCR using the kanMX2 gene from plasmid pFA6a-kanMX2 as a template and the following PCR primers (Table (Table2):2): ssy1Δ:kanMX2, 5D-SSY1 and 3D-SSY1; grr1Δ:kanMX2, 5D-GRR1 and 3D-GRR1; pdr5Δ:kanMX2, 5D-PDR5 and 3D-PDR5. The yeast strains EK007 (ura3 gap1Δ) and 23344c (ura3) were transformed with the PCR fragments by the lithium method as previously described (26). Transformants were selected on complete medium containing 200 μg of Geneticin (Gibco-BRL)/ml. A similar transformation and selection procedure was used to construct strains GAL1-HA3-SSY5 gap1Δ, GAL1-HA3-SSY5 gap1Δ ssy1Δ, GAL1-HA3-SSY5 gap1Δ ptr3Δ, GAL1-HA3-PTR3 gap1Δ, GAL1-HA3-PTR3 gap1Δ ssy1Δ, GAL1-HA3-PTR3 gap1Δ ssy5Δ, GAL1-HA3-PTR3 gap1Δ grr1Δ, and GAL1-HA3-SSY5 yck1-Δ1 yck2-2ts. For these strains, insertion DNA cassettes made of the GAL1 promoter, the sequence encoding the triple HA tag, and the first codons of the SSY5 or PTR3 gene were generated by PCR with plasmid pFA6a-kanMX6-PGAL1-3HA (57) as template DNA and with the following PCR primers: 3-SSY5-HA and F4-GAL1-SSY5 or 3-PTR3-HA and F4-GAL1-PTR3 (Table (Table22).
All β-galactosidase assays were performed on cells that had reached the state of balanced growth. The β-galactosidase assays were performed as described earlier (2), and activities are expressed in nanomoles of o-nitrophenol formed per minute per milligram of protein. Protein concentrations were measured with the Folin reagent, bovine serum albumin being used as a standard (41).
Crude cell extracts were prepared as previously described (28). For Western blot analysis, equal quantities of proteins were loaded onto an sodium dodecyl sulfate-8% polyacrylamide gel in a Tricine system. After transfer to a nitrocellulose membrane (Shleicher & Schüell), the proteins were probed with anti-HA monoclonal antibodies 12CA5 (1:10,000; Roche Diagnostics), with polyclonal antibodies raised against the N-terminal tail of Agp1 (1:5,000) (1), or with polyclonal antibodies raised against Gic2 (1:2,500) (13). Primary antibodies were detected with horseradish peroxidase-conjugated anti-mouse or anti-rabbit immunoglobulin G secondary antibody (Amersham Pharmacia Biotech).
Crude cell extracts were prepared as previously described (28) except that the final pellets were dissolved in 4% sodium dodecyl sulfate-0.1 M Tris-HCl (pH 6.8)-20% glycerol-2% 2-mercaptoethanol. Alkaline phosphatase (20 U), 20 μl of dephosphorylation buffer (Roche Diagnostics), and 20 μl of 0.5 M Tris (pH 8.0) were added to 20 μl of extract, and the mixture was then incubated for 2 h at 37°C.
Previous work has shown that components of the SCFGrr1 ubiquitin-ligase complex are essential to transcriptional induction of the AGP1 gene in response to external amino acids (7, 8). Although SCF ubiquitin ligases most often catalyze ubiquitylation of protein substrates only once these have been phosphorylated (18), previous genetic screens failed to identify any protein kinase involved in amino acid signaling in yeast (7, 23, 33, 35). The SCFGrr1 complex is also essential to transmission of the glucose signal initiated by Snf3 and Rgt2, the transporter-like sensors of external glucose (22, 37). Recent work has shown that the yeast membrane-bound CKI is an essential component of Rgt2-mediated glucose signaling. Its role is to phosphorylate Std1 and Mth1, these modifications being essential to both proteins' Grr1-dependent degradation and to subsequent HXT gene induction in response to glucose (44). In Kluyveromyces lactis, the Snf3/Rgt2 homologue encoded by the RAG4 gene is essential to transcriptional induction of the RAG1 gene encoding a low-affinity glucose transporter (10). Interestingly, the K. lactis equivalent of CKI (Rag8) is also required for transcriptional induction of the RAG1 gene (11). Hence, CKI appears to play an important role in glucose signaling in both S. cerevisiae and K. lactis. These data prompted us to test whether CKI is also involved in amino acid signaling. For this, we used a strain with the YCK1 gene deleted and containing a thermosensitive yck2ts allele. This mutant grows normally at 24°C but fails to grow at the restrictive temperature of 37°C (50). We transformed the mutant and the congenic wild type with a lacZ reporter gene under the control of the AGP1 upstream region. Cells were grown at the permissive temperature of 24°C and then placed for 20 min at 37°C. Phenylalanine, a potent inducer of AGP1, was then added to the medium, and the culture was incubated for 2 h. The results (Fig. (Fig.1)1) showed that transcription of the AGP1-lacZ reporter gene was induced by phenylalanine in the wild type transferred to 37°C, but that this induction was totally abolished in the yckts mutant. At the permissive temperature of 24°C, AGP1-lacZ expression was induced in the yckts strain, but to a level significantly lower than in the wild type. Immunodetection of the Agp1 permease in extracts of wild-type and yckts cells confirmed that induction of Agp1 synthesis was totally abolished in the yckts strain transferred to 37°C (Fig. (Fig.1).1). These data clearly show that CKI plays an essential role in transcriptional induction of the AGP1 gene in response to amino acids.
An important event in the signaling pathway leading to AGP1 transcriptional induction is endoproteolytic processing of the membrane-bound Stp1 transcription factor (3). After cleavage, Stp1 translocates into the nucleus and acts together with the Uga35/Dal81 transcription factor, via the UASAA upstream element, to activate AGP1 transcription (1). To gain further insight into the role of CKI in amino acid signaling, we transformed the yckts mutant and the congenic wild type with a low-copy-number plasmid expressing an HA-tagged Stp1 protein under the control of the natural STP1 gene promoter. The cells were grown at 24°C and transferred for 20 min to 37°C, and then phenylalanine was added to the culture (Fig. (Fig.2A).2A). Western blot analysis of wild-type cell extracts revealed that Stp1 is normally cleaved at 37°C in response to phenylalanine and that this correlates with immunodetection of a high-intensity Agp1 signal. In the yckts mutant, cleavage was totally abolished and Agp1 was not induced. Further examination of the immunoblots revealed faster migration of the full-length Stp1 signal in extracts of the yckts mutant than in extracts of wild-type cells (Fig. (Fig.2B).2B). This suggests that Stp1 is phosphorylated in a CKI-dependent manner and that this modification occurs even when amino acids are not available in the medium. To test this, we examined the migration pattern of Stp1 immunodetected in wild-type and yckts cell extracts treated or not with alkaline phosphatase (Fig. (Fig.2C).2C). Stp1 migrated faster after treatment of wild-type extracts with alkaline phosphatase, confirming that Stp1 is phosphorylated. In extracts of the yckts strain, Stp1 migrated faster compared to the situation with untreated wild-type extracts but slower compared to the same extracts treated with alkaline phosphatase. Hence, CKI is clearly involved in Stp1 phosphorylation, and other protein kinases likely contribute to generate fully phosphorylated Stp1. The CKI-dependent phosphorylation occurs in media devoid of amino acids and seems essential to Stp1 endoproteolytic processing in response to amino acids.
Endoproteolytic processing of transcription factors often requires their prior ubiquitylation. For instance, yeast Stp23 and Mga2 are initially synthesized as inactive precursors anchored to the endoplasmic reticulum through their C-terminal tails. Activation of these factors involves their ubiquitylation and proteasome-mediated degradation of their C-terminal domains, whereas the N-terminal domains are left intact and migrate into the nucleus to activate gene transcription (30). The exact mechanism of Stp1 endoproteolytic processing remains unknown. Andreasson and Ljungdahl (3) reported that cleavage of Stp1 is normal in mutant cells lacking the F-box protein Grr1. They suggested that the SCFGrr1 ubiquitin-ligase complex is involved in later steps of the amino acid signaling pathway. In the course of our work, we repeated this experiment and obtained a completely different result: upon addition of phenylalanine to the growth medium, Stp1 was cleaved in the wild type but not in the grr1Δ mutant (Fig. (Fig.3A).3A). This result was observed in several independent experiments involving longer incubation times in the presence of diverse amino acids (data not shown). It has since also been observed by the group of P. Ljungdahl (personal communication). To determine if Grr1 were required for Stp1 cleavage as part of the SCF complex, we tested whether Stp1 processing occurs or not in the thermosensitive cdc34ts mutant strain (Fig. (Fig.3B).3B). CDC34 encodes the E2 ubiquitin-conjugating enzyme, which is normally part of SCF complexes and directly involved in transfer of ubiquitin to the target protein (18). The cdc34ts mutant and the congenic wild type were grown at 30°C without any amino acid, after which they were transferred to the restrictive temperature of 37°C and incubated for 60 min. Then, phenylalanine was added to the medium and the culture was incubated for 15 more minutes. The data presented in Fig. Fig.3B3B clearly show that Stp1 undergoes normal endoproteolytic processing in the wild type at 37°C. Processing, however, was totally abolished in the cdc34ts strain. This lack of cleavage is not an indirect consequence of an effect of the cdc34ts mutation on the cell cycle, since Stp1 was normally cleaved in the cdc4-1 strain (also defective in cell cycle progression at 37°C). Stp1 cleavage was also totally abolished when transfer of cells to 37°C and addition of phenylalanine were simultaneous (data not shown), further suggesting that the effect of the cdc34ts mutation is not the indirect consequence of some deficiency in another Cdc34-dependent cell function. In conclusion, our data show that at least two components of the SCFGrr1 ubiquitin-ligase complex, namely the F-box protein Grr1 and the E2 enzyme Cdc34, are essential to amino acid-induced endoproteolytic processing of transcription factor Stp1. These data corroborate previous results showing that transcriptional induction of AGP1 is impaired in grr1Δ and cdc34ts mutants as well as in strains with impaired function in other components of the SCFGrr1 complex (8).
Proteins ubiquitylated by SCF complexes generally undergo regional or complete degradation catalyzed by the proteasome (47). It has been reported, however, that the proteasome is not involved in Stp1 cleavage. During growth on synthetic complete medium, investigators readily detected the processed form of Stp1 in pre mutants impaired in three major activities of the proteasome (3). We sought to further investigate the role of the proteasome in Stp1 processing by using MG132, an inhibitor of the proteasome (36). As this toxic compound can be extruded from cells via the Pdr5 multidrug resistance pump (20), experiments were carried out in a pdr5Δ mutant. Cells were first grown in the absence of amino acids. MG132 was then added to the medium, and 2 h later, phenylalanine was added to induce Stp1 cleavage (Fig. (Fig.4).4). The results showed that Stp1 undergoes normal phenylalanine-induced cleavage regardless of whether MG132 is present in the medium (Fig. (Fig.4A).4A). Although the mobility of Stp1 seems to change over time in cells not treated with MG132 (Fig. (Fig.4A),4A), this was not reproduced in other experiments (data not shown). To confirm inhibition of proteasome function, we probed the same cell extracts with antibodies against Gic2, a protein previously reported to undergo SCFGrr1-dependent ubiquitylation and to accumulate as a ladder of high-molecular-weight forms in mutants impaired in proteasome function (32). MG132 did induce the appearance of high-molecular-weight forms of Gic2p, likely corresponding to ubiquitin-conjugated forms of the protein (Fig. (Fig.4B).4B). Hence, our data corroborate previous conclusions based on the use of pre mutants (3), and they indicate that Stp1 endoproteolytic processing is largely independent of proteasome function.
In all reported cases of ubiquitin-dependent modification and activation of transcription factors, the N- or C-terminal domain of the protein is degraded by the proteasome. Even in cases where degradation is initiated by an internal cut within a polypeptide loop of the protein, subsequent processive degradation of the polypeptide chain is mediated by the proteasome (38, 47). These situations differ from that described for Stp1, since the released N-terminal domain is apparently left intact after endoproteolytic cleavage (3). This may account for the fact that the proteasome does not intervene in Stp1 processing. Our observations are thus consistent with the intervention of another endoprotease activity.
One might predict that cells defective in the protease mediating Stp1 endoproteolytic processing will be unable to cleave Stp1 and induce AGP1 transcription in response to amino acids. Such a phenotype has been described for strains functionally impaired in Ssy1, Ptr3, or Ssy5, three factors proposed to form a complex (SPS) at the plasma membrane (3). We reasoned that overproduction of the endoprotease involved in Stp1 processing might induce Stp1 cleavage and thus activate AGP1 transcription even in a medium devoid of amino acids. We thus tested the effect of Ssy5 and Ptr3 overproduction on AGP1 transcription.
To overproduce Ssy5, we placed the chromosomal copy of the SSY5 gene under the control of the GAL promoter. The construct also introduced a triple HA tag at the extreme N terminus of the protein. This construct was introduced into various strains, including the gap1Δ mutant lacking the general amino acid permease. When the Ssy5 function is impaired in this strain, readily detectable growth phenotypes are produced (7). The HA-tagged Ssy5 protein proved functional, since the gap1Δ GAL-SSY5 strain behaved like a gap1Δ SSY5+ mutant when grown on galactose, i.e., cells could grow on leucine (1 mM) as sole nitrogen source and could not grow in the presence of the toxic amino acid analogue l-ethionine (Fig. (Fig.5).5). When grown on glucose, in contrast, the gap1Δ GAL-SSY5 strain did not grow on leucine and was resistant to l-ethionine, phenotypes typically displayed by gap1Δ strains additionally defective in any component of the SPS complex (7) (Fig. (Fig.55).
The gap1Δ GAL-SSY5 strain was grown on a minimal medium without any amino acid and with raffinose as a carbon source. At time zero of the experiment, galactose was added to the medium. Samples of the culture were withdrawn at time intervals, and crude cell extracts were prepared and probed with antibodies against HA (Fig. (Fig.6A).6A). As expected, Ssy5 was not detected in cells grown on raffinose. Thirty minutes after galactose addition, a single band migrating as an ~80-kDa protein was readily detected. This mobility corresponds with the calculated molecular mass of Ssy5 (76 kDa) tagged with a triple HA. After 120 min, remarkably, an additional band migrating as a ~50-kDa protein was also detected, indicating that the overproduced Ssy5 protein had undergone endoproteolytic cleavage (Fig. (Fig.6A).6A). The same immunoblot was then probed with antibodies against Agp1 (Fig. (Fig.6A).6A). As expected, no Agp1 signal was detected in cells growing on raffinose without any amino acid. The same was true 30 min after galactose addition, but 120 min after galactose induction a high-intensity Agp1 signal was readily detected. These results indicate that overproduction of Ssy5 leads to high-level synthesis of the Agp1 permease even if amino acids are not present on the medium. This conclusion was confirmed by assaying β-galactosidase activity in GAL-SSY5 cells transformed with lacZ reporter genes under the control of the entire AGP1 upstream region or the UASAA element alone (Table (Table3):3): growth of this strain on galactose coincided with β-galactosidase values at least as high as those measured in wild-type cells grown in the presence of amino acids. On glucose, however, no lacZ reporter gene was expressed, even when phenylalanine was present in the medium. These data clearly show that overproduction of Ssy5 leads to high-level constitutive activation of AGP1 transcription.
To determine if high-level constitutive activation of AGP1 transcription in Ssy5-overproducing cells correlates with constitutive Stp1 processing, we transformed the GAL-SSY5 strain with a low-copy-number plasmid expressing the HA-tagged Stp1 protein. Cells were first grown without any amino acid and with raffinose as a carbon source. As expected, Stp1 was readily detected on immunoblots as its full-length form (Fig. (Fig.6B).6B). Thirty minutes after addition of galactose, the ~80-kDa form of Ssy5 became detectable and Stp1 still migrated mainly as its full-length form. Ninety minutes after galactose addition, both forms of Ssy5 were readily detected and Stp1 migrated as its processed form. The cleaved form of Stp1 was still detected several hours after galactose addition. These data clearly show that overproduction of Ssy5 causes efficient inducer-independent endoproteolytic processing of Stp1.
The Ssy5 overproduction experiment was also performed in the CKI-deficient strain (Fig. (Fig.6C).6C). Synthesized Ssy5 undergoes normal endoproteolytic cleavage both in the wild-type and yckts strains transferred to the restrictive temperature of 37°C. Stp1, however, was cleaved in the wild type but not in the yckts strains upon Ssy5 expression. This shows that CKI is essential to Stp1 cleavage even under conditions of Ssy5 overproduction. Furthermore, Ssy5 seems to undergo normal processing in CKI-deficient cells.
Finally, we tested whether endoproteolytic processing of the same HA-tagged Ssy5 protein was also detectable in cells where the unique SSY5 gene was expressed under the control of its own promoter (Fig. (Fig.6D).6D). In cells grown without amino acids, the same two bands were detected, but the one corresponding to the unprocessed form of Ssy5 was of much lower intensity compared to experiments in which Ssy5 had been expressed under the control of the GAL promoter. Furthermore, addition of phenylalanine did not alter this migration pattern of Ssy5. We conclude that Ssy5 is efficiently processed even when synthesized at normal levels and that this reaction occurs independently of the presence or absence of amino acids in the medium.
We next sought to determine whether factors normally essential to amino acid-induced Stp1 processing and AGP1 transcription are also involved in constitutive Stp1 cleavage and AGP1 expression when Ssy5 is overproduced. The SSY1, PTR3, and GRR1 genes were thus deleted individually in the GAL-SSY5 strain. The resulting strains were then transformed with a plasmid bearing the AGP1-lacZ reporter gene. Table Table44 shows the β-galactosidase activities measured in transformed cells grown on glucose (SSY5 not expressed) or galactose (SSY5 overexpressed) as a carbon source. The data show that neither Ssy1 nor Ptr3 nor Grr1 is required for constitutive AGP1-lacZ expression in Ssy5-overproducing cells. Consistent with this result, overproduction of Ssy5 also relieved the growth phenotypes caused by lack of Ssy1, Ptr3, or Grr1 in the gap1Δ strain grown on leucine (1 mM) or in the presence of l-ethionine (Fig. (Fig.5).5). That similar effects of Ssy5 overproduction were observed in wild-type, ssy1Δ, ptr3Δ, and grr1Δ mutant strains also tallies with the observation that the Ssy5 protein is overexpressed to similar levels and undergoes similar endoproteolytic processing in these strains (data not shown). Taken together, these data clearly show that once Ssy5 is overproduced in cells, Stp1 is cleaved and AGP1 is transcribed to a high level, independently of the presence or absence of any external amino acid or functional Ssy1, Ptr3, or Grr1 factor.
In a previous study we showed that the Stp1 and Uga35/Dal81 transcription factors are activated in response to amino acids and that both contribute importantly to the high-level AGP1 transcription observed in cells grown in the presence of external amino acids. Namely, in stp1Δ and uga35/dal81Δ single mutant strains, AGP1 is induced, though weakly, but induction is abolished when both factors are lacking (1). The data presented in Table Table44 show that in Ssy5-overproducing cells, Stp1 was absolutely essential to the high-level constitutive expression of AGP1-lacZ observed in these cells. When Uga35/Dal81 was lacking, this constitutive expression was reduced by ~85%. This shows that while Stp1 is essential, Uga35/Dal81 also contributes importantly to the high constitutive expression of AGP1 caused by Ssy5 overproduction. It thus seems that Ssy5 overproduction specifically activates Stp1, which then acts together with Uga35/Dal81 to promote maximal constitutive activation of AGP1 transcription. This situation contrasts with that observed when amino acids are present. In the latter case, Uga35/Dal81 can stimulate AGP1 transcription to some extent even if Stp1 is not functional (1). Our results further show that when phenylalanine is added to the medium of stp1Δ mutant cells overexpressing SSY5, a significant level of AGP1-lacZ expression is observed. This activation is likely due to the Uga35/Dal81 factor, which under these conditions is activated by amino acids (1).
The above experiments indicated that Ssy5 is a major limiting factor of Stp1 endoproteolytic processing. Ssy5 might thus be the endoprotease that catalyzes Stp1 cleavage, this endoprotease function being dependent on endoproteolytic processing of Ssy5 itself (readily detected when Ssy5 is overproduced). Algorithms commonly used for sequence similarity searches (BLAST, PSI-BLAST) failed to reveal any significant resemblance between Ssy5 and any peptidase characterized to date or any other protein of known biochemical function (7, 23, 33, 35). The only proteins found to be similar to Ssy5 were in fact close Ssy5 orthologues of other fungal species. Furthermore, Ssy5 does not appear in the MEROPS database (49) (accessible at http://merops.sanger.ac.uk/), which inventories no less than 100 proteases in the proteome of S. cerevisiae, including 31 proteins of unknown function. We next turned to the InterProScan tool, which combines different protein signature recognition methods (59) to compare a query sequence against commonly used protein signature databases. Among these is Superfamily (27), a library of hidden Markov models (HMMs) constructed from the sequence alignments of 1,232 protein superfamilies inventoried in the Structural Classification of Proteins (SCOP) database (40). Remarkably, the InterProScan database (accessible at http://www.ebi.ac.uk/interpro/) assigned yeast Ssy5 as a possible member of the trypsin-like serine protease superfamily (superfamily code SSF50494). Further analysis using the Superfamily sequence search facility (http://supfam.org/SUPERFAMILY/) revealed that it is the C-terminal domain of Ssy5 (residues 459 to 687) that is significantly similar (E value of 2.5e−09) to HMMs corresponding to this serine protease superfamily. In particular, the triad of amino acids (His, Asp, and Ser) corresponding to the catalytic site of this protease superfamily, as well as surrounding residues, are highly conserved in the Ssy5 protein of S. cerevisiae (Fig. (Fig.7)7) and its orthologues in other fungal species. The analyses also suggest that the Ssy5 proteins are more closely related to different subfamilies included in the S1 family of serine proteases (49). Interestingly, only one yeast protein, the product of the NMA111/YNL123W gene, was assigned to the serine protease superfamily after analyses of the complete yeast proteome with sequences and tools of the Superfamily or MEROPS databases, and a recent study reported a role of this serine protease in apoptosis (21). We now propose that Ssy5 corresponds to another serine protease. However, contrary to Nma111 (a member of the Htra family of serine proteases) (14), Ssy5 is not highly similar to any previously described protease family. As described above, the Ssy5 protein undergoes endoproteolytic cleavage (Fig. (Fig.6).6). This tallies with the fact that many S1 proteases are expressed as inactive pro-forms characterized by an N-terminal inhibitory peptide and requiring proteolytic maturation for activation (48). The apparent molecular mass of the N-terminal peptide released after Ssy5 processing (~47 kDa) suggests that cleavage of Ssy5 takes place in the region just preceding the C-terminal S1 protease-like domain (Fig. (Fig.77).
These novel observations on the Ssy5 primary sequence, together with the data obtained with Ssy5-overproducing cells, strongly suggest that Ssy5 is the endoprotease catalyzing the endoproteolytic processing of membrane-bound Stp1 transcription factor in response to amino acids. Furthermore, as illustrated for many serine proteases (48), it is likely that endoproteolytic cleavage of the Ssy5 protein itself is required for its endoprotease function.
We next introduced mutations in the SSY5 gene and tested the influence of these mutations on Stp1 cleavage and Agp1 induction. The results in Fig. Fig.8A8A show that phenylalanine-induced Stp1 cleavage and Agp1 synthesis were totally defective in cells expressing an Ssy5 protein in which the serine residue of the predicted catalytic triad had been replaced by an alanine (Ssy5S→A). The same mutation was introduced into the plasmid-borne SSY5 gene expressed under the control of the strong GAL promoter. In this case, a triple HA tag at the N terminus allowed immunodetection of the Ssy5S→A mutant protein upon induction by galactose (Fig. (Fig.8B).8B). Remarkably, Ssy5S→A failed to undergo endoproteolytic processing after 2 h of induced synthesis. Furthermore, Stp1 was not cleaved in response to overproduction of this unprocessed form of Ssy5 (Fig. (Fig.8B).8B). The same result has been obtained when this Ssy5S→A form is expressed under the control of the natural SSY5 gene's promoter (data not shown). These results show that at least one of the residues of the predicted catalytic site of Ssy5 is essential to Ssy5 processing and Stp1 cleavage. These data are fully consistent with the above proposed model, i.e., Ssy5 is most likely the endoprotease responsible for Stp1 cleavage and, like many other proteases, this enzyme needs to be autoprocessed for being active. To immunodetect the C-terminal protease domain of Ssy5 released after processing, a triple HA tag was added to the C terminus of the protein. However, it turned out that this form of Ssy5 lost its ability to be autoprocessed and to mediate Stp1 cleavage (data not shown).
The apparent molecular mass of the N-terminal Ssy5 fragment released after processing suggests that Ssy5 cleavage takes place within a region lying around residue 420, i.e., just upstream from the protease homology region (Fig. (Fig.7).7). Remarkably, this Ssy5 putative cleavage region includes a short sequence highly conserved among Ssy5 orthologues in other yeast species (Fig. (Fig.8C).8C). A deletion of five residues in this sequence (PISMS) led to an Ssy5 mutant form (Ssy5ΔPISMS) that proved defective in both self-cleavage and Stp1 cleavage (Fig. (Fig.8D).8D). This strongly suggested that this conserved sequence corresponds to the Ssy5 self-cleavage site.
In another study (23), the Ssy5 protein tagged within its N terminus with the c-myc epitope and expressed under the natural SSY5 gene promoter was detected as a single band with an apparent molecular mass of 67 kDa, whereas the protein migrated as a 76-kDa protein in extracts of the ptr3 mutant. The authors concluded that Ssy5 is proteolytically modified in its C-terminal region in a Ptr3-dependent manner. These observations are not entirely consistent with ours, which suggest that Ssy5 cleavage takes place in a more-N-terminally located region (Fig. (Fig.77 and and8).8). However, in accord with our data (Fig. (Fig.6D),6D), these investigators observed this modification in cells growing in a medium without any amino acid, i.e., Ssy5 processing is not induced in response to amino acids. Furthermore, this processing was not observed in a ptr3 mutant (23), suggesting that autoprocessing of Ssy5 requires normal Ptr3 function. As indicated above, however, Ssy5 processing does not require Ptr3 if the endoprotease is overproduced in the cell.
Ptr3 (6) is yet another membrane-associated protein essential to Stp1 cleavage (3) and amino acid signaling (7, 33, 35). Furthermore, data from two-hybrid experiments indicate that Ptr3 interacts both with Ssy5 and with itself (7). We thus tested whether overproduction of Ptr3 also leads to Stp1 cleavage and AGP1 transcription in the absence of any amino acid. Using the same procedure as applied to SSY5, we isolated cells expressing the PTR3 gene under the GAL promoter. A triple HA tag was also inserted at the extreme N terminus of the protein. When the cells were grown on galactose, HA-tagged Ptr3 migrated as a protein of ~80 kDa (data not shown), a mobility consistent with the calculated molecular mass of the protein. Furthermore, analysis of extracts of GAL-SSY5 and GAL-PTR3 cells on the same immunoblot suggested that Ssy5 and Ptr3 were present at similar levels (data not shown). The AGP1-lacZ gene was not induced by phenylalanine in the GAL-PTR3 strain when the cells were grown on glucose (Table (Table3).3). This was expected, since Ptr3 is not synthesized under these conditions. In cells grown on galactose, AGP1-lacZ was expressed only if an amino acid was added to the medium. These data clearly show that the HA-tagged Ptr3 protein is functional and that its overproduction, contrary to that of Ssy5, does not lead to constitutive activation of the AGP1 gene. They further suggest that the biochemical function of Ptr3 in the amino acid signaling pathway differs from that of Ssy5.
Other data are consistent with a role of Ptr3 in early steps of the signaling pathway, namely, amino acid detection and/or signal initiation. In a previous study (7) our investigators performed a genetic screen aimed at isolating mutants defective in activation of the amino acid-sensing pathway. Most mutants fell into the SSY1, PTR3, and SSY5 complementation classes. We showed that mutants lacking both the general amino acid permease (Gap1) and any one of these genes failed to grow on several amino acids (e.g., leucine, isoleucine, and phenylalanine) used as sole nitrogen source (1 mM) because the AGP1 permease gene essential to growth under these conditions was not expressed. Among the many ssy1 mutants isolated, the ssy1-23 strain differed from the others by its ability to induce AGP1 transcription in response to leucine but not to several other amino acids. These observations reinforced the view that the Ssy1 protein is directly involved in amino acid recognition (7). Interestingly, among the ptr3 mutants isolated in the same screen, one (ptr3-35) behaved like the ssy1-23 strain (Fig. (Fig.9).9). In other words, a gap1Δ ptr3-35 mutant grew well on leucine at a low concentration used as sole nitrogen source but failed to grow on other amino acids like phenylalanine, methionine, or isoleucine. This phenotype differs from that of the gap1Δ ptr3Δ strain, which is unable to grow under any of these conditions. Furthermore, the gap1Δ ptr3Δ strain transformed with the cloned ptr3-35 allele displayed normal growth on leucine but not on the other tested amino acids. It thus seems that the ptr3-35 mutant has selectively lost the ability to induce AGP1 transcription in response to several external amino acids, while retaining the ability to respond to leucine. Accordingly, β-galactosidase assays performed on cell extracts of this mutant, or on extracts of the ptr3Δ strain transformed with the cloned ptr3-35 gene, showed high-level induction of AGP1-lacZ expression by leucine but no induction in response to phenylalanine, methionine, or isoleucine (Table (Table5).5). The apparently identical and specific phenotypes of the ptr3-35 and ssy1-23 mutants suggest that Ssy1 and Ptr3 contribute to the same molecular event required for the proper response of the cell to the presence of external amino acids. The ptr3-35 allele contains a single G→A substitution that results in a glutamate→lysine substitution at position 521 in the protein. Interestingly, this residue is located in a region previously proposed to share significant sequence similarity with transcription factor Gcn4 (35). Furthermore, the region a few amino acids downstream is suggested to be conserved not only in Gcn4 but also in amino acid permeases of the AAP/YAT family (35). It has been proposed that this region shared by Ptr3 and amino acid permeases may be involved in amino acid recognition (24, 35).
In this study we have pursued our investigation of the mechanisms by which the yeast AGP1 gene, encoding an amino acid permease, is transcriptionally induced in response to detection by the cell of external amino acids (12, 24). This induction is mediated by the concerted action of two transcription factors, Stp1 and Uga35/Dal81, acting via a short GC-rich upstream activating sequence named UASAA (1). In the absence of any external amino acid, Stp1 is present in an inactive pro-form somehow associated with the cell surface. In response to detection of external amino acids, Stp1 undergoes endoproteolytic processing in a region estimated to lie between residues 68 and 127 (3). The C-terminal domain of Stp1, containing a Kruppel-type zinc-finger domain, is thus released. It translocates into the nucleus where, in conjunction with Uga35/Dal81, it activates AGP1 transcription.
Our data provide evidence that endoproteolytic processing of Stp1 depends on its prior phosphorylation, mediated by CKI. This CKI-dependent phosphorylation of Stp1 does not require the presence of external amino acids—it was observed also in media devoid of them. It thus seems that CKI is not activated by external amino acids. The role of CKI seems to be, rather, to phosphorylate Stp1 to make it amenable to endoproteolytic processing in response to amino acids. The Yck1 and Yck2 isoforms are peripheral plasma membrane-associated proteins (56). Studies of Yck2p have shown that this kinase undergoes palmitoylation on C-terminal Cys-Cys sequences by the palmitoyl transferase Akr1 (51) and is targeted to the cell surface via the classical secretory pathway (5). The Yck1 and Yck2 isoforms are involved in numerous cell functions, including bud morphogenesis (50), trafficking of plasma membrane transporters (42, 43) and pheromone receptors (29, 46), and cytokinesis (50). Our data show for the first time a role of yeast CKI in phosphorylation of a plasma membrane-bound transcription factor. Further experiments will be needed to determine which of the several potential target sequences for CKI found in Stp1 is indeed phosphorylated in a CKI-dependent manner. Further work will also be required to determine whether failure to process Stp1 is alone responsible for the complete lack of AGP1 induction in the yckts mutant. For instance, we have previously reported that Uga35/Dal81 alone can, to some degree, activate transcription of AGP1 in response to amino acids (1). Our observation that AGP1 induction is totally abolished in the yckts strain at the nonpermissive temperature raises the possibility that activation of the Uga35/Dal81 branch of the amino acid signaling pathway might also be CKI dependent. In a recent paper, Moriya and Johnston (44) reported that CKI is also involved in glucose signaling. Their data support a model in which glucose binding to the Snf3 and Rgt2 transporter-like glucose sensors activates CKI, which then phosphorylates the Mth1 and Std1 transduction factors bound to the sensors. Phosphorylation of Std1 and Mth1 apparently leads to their Grr1-dependent degradation, a step required for relieving repression of HXT gene expression by the Rgt1 repressor (44). CKI thus appears to play a central role in nutrient signaling in yeast. Our data, however, do not support a model in which CKI would be activated directly by the permease-like Ssy1 protein in response to amino acids, since Stp1 was found to be phosphorylated also in the absence of amino acids.
In the present work we have also focused on Ssy5, a protein reported to be essential to AGP1 induction (7, 23, 33) and Stp1 processing (3) but whose exact biochemical role in the signaling pathway has remained obscure thus far. We provide evidence that Ssy5 is the endoprotease catalyzing cleavage of Stp1 in response to amino acids. First, we showed that overexpression of Ssy5 results in efficient Stp1 cleavage and high-level AGP1 transcription in a medium without any amino acid. Second, we have conducted sequence comparisons revealing, in the C-terminal region of Ssy5, the protein signature typical of serine proteases, including the amino acid triad (H, D, S) constituting the catalytic site of this class of proteases. Third, we showed that the Ssy5 protein itself undergoes endoproteolytic processing, a property shared by many if not all serine proteases (48). Fourth, a single amino acid substitution in the predicted Ssy5 catalytic triad prevented Ssy5 processing, as well as Stp1 cleavage and AGP1 transcription, indicating that Ssy5 is self-processed and that this reaction is essential to the endoprotease function of Ssy5. A previous study reported that Ssy5 is processed and that this processing is dependent on Ptr3 (23). On the other hand, Ptr3 was shown by two-hybrid experiments to interact with itself and with Ssy5 (7). Perhaps the Ssy5 pro-form interacts with a cell-surface Ptr3 dimer, this interaction favoring Ssy5-Ssy5 contacts and thus self-processing of Ssy5. We showed in this work that under conditions of Ssy5 overproduction, self-processing occurs independently of Ptr3, Ssy1, and Grr1. The high dosage of Ssy5 under these conditions might favor Ssy5-Ssy5 contacts and thus self-processing in a Ptr3-independent manner. A sequence in the Ssy5 region where self-cleavage occurs (just upstream from the protease domain [Fig. [Fig.7])7]) is particularly well conserved in fungal Ssy5 orthologues, and we showed that the integrity of this sequence is essential to Ssy5 self-processing and activity. Interestingly, a similar sequence is also conserved in Stp1 (Fig. (Fig.8C).8C). In the latter, the sequence is also located in a region that likely includes the protein's endoproteolytic processing site (between residues 68 and 127 ). Furthermore, Stp1 orthologues in other yeast species also show high conservation in the corresponding regions, and this sequence is also conserved in the Stp2 transcription factor reported to undergo amino acid-induced endoproteolytic processing (Fig. (Fig.8C)8C) (3). It is thus tempting to propose that Ssy5 self-processing and Ssy5-mediated Stp1 and Stp2 cleavage take place within these conserved sequences (Fig. (Fig.8C).8C). Further experiments based on saturating site-directed mutagenesis experiments will be required to test the validity of these predictions and to further delineate the Ssy5 and Stp1 cleavage sites.
Our data thus imply that in response to detection by the cell of external amino acids, the Ssy5 endoprotease catalyzes endoproteolytic processing of Stp1. This suggests that Ssy5 itself is activated by external amino acids. As mentioned above, many serine proteases are synthesized as inactive pro-forms characterized by an N-terminal inhibitory peptide and thus require proteolytic maturation for activation (48). In the case of Ssy5, however, we showed that processing occurs independently of the presence or absence of amino acids in the medium. Hence, some other mechanisms must account for Stp1 cleavage—catalyzed by the processed Ssy5 form—to occur specifically in response to amino acids. Our group's previous genetic data provided evidence that Ssy1 and Ptr3 act upstream from Ssy5 in the amino acid signaling pathway (7). In keeping with this model, we showed in this study that Ssy1 and Ptr3 are entirely dispensable for Stp1 cleavage and AGP1 transcription if Ssy5 is overproduced. Also in support of the model that Ptr3 acts early in the signaling cascade in tight conjunction with Ssy1, we described in this work a mutant form of Ptr3 (ptr3-35) that fails to respond to several amino acids but still responds efficiently to leucine. A similar phenotype has previously been described for the ssy1-23 mutant (7). Previous work has also shown that leucine is the amino acid to which cells appear able to respond with the highest affinity (25). It is thus probable that both the ssy1-23 and the ptr3-35 mutants respond to external amino acids with a globally reduced affinity. That the same partial phenotypes may be caused by particular mutations in the SSY1 and PTR3 genes strongly suggests that Ssy1 and Ptr3 conjunctly act in the same molecular event early in the amino acid signaling pathway, i.e., amino acid detection and signal initiation. Furthermore, a systematic two-hybrid search for protein-protein interactions has revealed that Ssy1 interacts with Ptr3 (55). We thus propose that the sensor for external amino acids mainly consists of a complex made of Ssy1 and Ptr3. The Ssy5 endoprotease would associate with Ptr3, and this interaction likely permits Ssy5 self-processing. Upon detection of amino acids, the Ssy1-Ptr3 proteins would undergo some kind of modification allowing the associated Ssy5 endoprotease to gain access to and to cleave Stp1. The role of Ssy1-Ptr3 in Stp1 cleavage, however, becomes entirely dispensable if Ssy5 is overproduced.
A still-unresolved question is: what is the precise role of the SCFGrr1 ubiquitin-ligase complex in the amino acid signaling pathway? In this work we have shown that two key components of this complex, namely the F-box protein Grr1 and the ubiquitin-conjugating enzyme Cdc34, are essential to amino acid-induced Stp1 processing. Yet Grr1 is no longer essential to Stp1 cleavage if Ssy5 is overproduced. These data support the view that SCFGrr1 acts upstream from Ssy5 in the pathway, i.e., at the level of the Ssy1-Ptr3 sensor complex. Other observations are consistent with this model. For instance, overproduction of the cytosolic N-terminal tail of Ssy1 overcomes the amino acid utilization defects of the grr1Δ mutant (unpublished data). Furthermore, it has been reported previously that the migration pattern of Ptr3 is altered upon addition of amino acids to the medium and that this modification depends on Ssy1 but not on Ssy5 (23). This amino acid-induced modification of the Ptr3 migration profile is also Grr1 dependent (unpublished data). Hence, upon recognition of external amino acids by the Ssy1-Ptr3 complex, the latter might undergo some SCFGrr1-dependent modification essential to subsequent activation of the Ssy5-dependent Stp1 cleavage reaction. Interestingly, the intervention of the SCFGrr1 ubiquitin-ligase complex does not seem to involve proteasome-catalyzed protein degradation: Stp1 cleavage is not deficient in pre mutants impaired in proteasome function (3) and, as shown here, it is insensitive to a proteasome inhibitor. It is known that the proteasome specifically mediates the degradation of proteins to which polyubiquitin chains linked through lysine-48 of ubiquitin have been attached. Perhaps SCFGrr1 catalyzes monoubiquitylation of one factor of the amino acid signaling pathway, this modification being essential to subsequent Ssy5 activation and Stp1 processing. There is indeed growing evidence that monoubiquitin may signal changes in protein location, activity, and interaction with binding partners (53). Our ongoing experiments thus aim to elucidate the protein targets(s) of SCFGrr1 in amino acid signaling and to better understand the molecular cascade from amino acid detection to Stp1 cleavage. We also are attempting to decipher how Uga35/Dal81, the other transcription factor responsible for high-level AGP1 transcription, is activated in response to amino acids.
We thank Catherine Jauniaux for her excellent technical assistance and all other members of the laboratory for fruitful discussions and encouragements. We also thank Florent Bernard for his contribution to the original description of the ptr3-35 mutant and Patrice Godard for his help in the analyses of Ssy5 sequences. We acknowledge Mike Tyers and Per Ljungdahl for providing cdc strains and plasmid pCA047, respectively. We also thank M. Peter for anti-Gic2 antibodies.
This work was supported by grants FRSM 3.4597.00.F and 3.4.605.05.F from the National Funds for Scientific Research, Belgium.