Many Zn(II)2Cys6 transcriptional regulators exhibit changes in phosphorylation that are coincident with their roles in transcriptional activation. It is, however, unclear whether these changes occur as a cause of, or as a result of, transcriptional activation. In this paper, we explore the relationship between these two events and collate data available on the phosphorylation state of those transcriptional regulators where the relationship has not been clearly identified.
- transcriptional activation
- transcriptional regulator
- Zn(II)2Cys6 protein
The Zn(II)2Cys6 family of transcriptional regulators contains upwards of 50 members in Saccharomyces cerevisiae alone. This family of proteins, characterized by the presence of an N-terminal DNA-binding domain containing two zinc atoms bound by six cysteine residues, is involved in diverse functions, ranging from carbon metabolism (Gal4p and Rtg1p) and nitrogen metabolism (Put3p and Leu3p) to stress responsiveness (War1p and Hal9p) . Like other transcriptional regulators, many Zn(II)2Cys6 proteins exhibit changes in phosphorylation concurrent with changes in transcriptional activation (see Table 1). It is, however, unclear whether these changes occur as a cause or as a result of activation. As with many other aspects of eukaryotic transcriptional control, Gal4p provides a paradigm for activation-related phosphorylation with which other Zn(II)2Cys6 family members can be compared.
Gal4p is responsible for the transcription of genes involved in galactose metabolism and is the archetypal Zn(II)2Cys6 transcription factor . In the absence of galactose, the GAL genes are transcriptionally inert; DNA-bound Gal4p is physically associated with the transcriptional inhibitor Gal80p. When galactose is the sole carbon source, the inducer, Gal3p, binds galactose in an ATP-dependent manner, resulting in its association with Gal80p. This interaction relieves the repressing effect of Gal80p, allowing Gal4p-mediated activation of GAL gene expression. Concurrent with the activation of GAL gene transcription, Gal4p exhibits an increase in phosphorylation, as detected by a mobility shift when analysed by Western blotting or immunoprecipitation [2,3]. Non-induced, non-repressed Gal4p migrates on SDS/PAGE as two distinct forms (Gal4p I and Gal4p II). However, within 30 min of activation, a third form (Gal4p III) becomes apparent, at the expense of the other two. Phosphatase treatment of active Gal4p results in the disappearance of the two highest-molecular-mass species, and a corresponding increase in the lowest-molecular-mass form, indicating that they are phosphorylated forms of Gal4p [3–6].
Investigation into the sites of phosphorylation revealed that Gal4p is phosphorylated at four serine residues: 691, 696, 699 and 837. In addition, it could potentially be phosphorylated twice between residues 1 and 238 and once between 701 and 768, but these sites have not been characterized. While conversion of all known phosphorylation sites into alanine reduces the band shift observed in wild-type Gal4p greatly, it was found that only Ser837 is responsible for the formation of Gal4p III. However, exclusively, the S699A mutation caused a decrease in transcriptional activity [6,7]. Interestingly, this decrease in activity is not apparent in yeast strains in which the GAL80 gene was deleted, resulting in constitutively active Gal4p. This indicates that phosphorylation at Ser699 is necessary for efficient induction by galactose, but not for activation in the absence of Gal80p . Furthermore, Ser699 can be phosphorylated in gal80Δ cells under repressing conditions, but not in cells containing Gal80p, suggesting that Gal80p can inhibit Gal4p phosphorylation . This could indicate that Gal4p becomes competent for phosphorylation only when Gal80p repression is relieved.
Full Gal4p phosphorylation is wholly dependent on its DNA-binding ability [6,7]. This suggests a link between phosphorylation and other co-activators or GTFs (general transcription factors). Indeed, the appearance of Gal4p III and phosphorylation at Ser699 are dependent on Gal11p, which is required for Mediator recruitment to the GAL1 promoter [6–10]. Moreover, it has been demonstrated that the cyclin-dependent kinase associated with Mediator, Srb10p, is required for phosphorylation of Ser699 and Ser837, while the Kin28p component solely phosphorylates Ser837 . During GAL gene expression, Srb10p has also been shown to associate with SAGA (Spt-Ada-Gcn5 acetyl transferase) in an Spt3p-dependent manner, as part of the Srb8p–Srb11p complex . Further evidence supporting this association between Srb10p, GTFs and Gal4p is witnessed in the fact that Srb10p has been shown to physically associate, both in vivo and in vitro, with a Gal4p molecule composed of the DNA-binding domain (amino acids 1–100) and activating region II (amino acids 840–881), suggesting that wild-type Gal4p can associate directly with Srb10p . The fact that the Gal4p phosphorylation requires DNA binding and that the activating kinase is physically associated with Mediator, as well as SAGA, suggests that Gal4p phosphorylation is a result of activation, rather than a cause of it. Indeed, Gal4p does not become phosphorylated until it activates transcription; yet, it does not become fully active unless Ser699 is phosphorylated [11,14]. Thus it would appear that, while Ser699 phosphorylation is required for galactose induction of Gal4p, phosphorylation is dependent on GTFs, possibly representing a feedback loop between initial Gal3p-based activation and the full recruitment of transcriptional machinery.
A possible interpretation of these events suggests that there are two separate mechanisms operating in the activation of Gal4p, one that feeds through Srb10p and its associated cyclin, Srb11p, as well as another that feeds through the inducer Gal3p . This model is supported by the fact that srb10Δ and gal3Δ strains are individually able to grow with galactose as the sole carbon source; however, when combined, the double mutant is unable to do so, demonstrating that both pathways are required for GAL gene induction .
In summary, full Gal4p activation appears to require two signals: active Gal3p and active Srb10p. Based on this supposition, the following model can be proposed; under inducing conditions, Gal3p binding of Gal80p activates Gal4p to basal levels. Once the transcriptional machinery is recruited to the GAL promoters, Gal4p is then fully activated by Srb10p.
Rgt1p offers a different example of Zn(II)2Cys6 action to the GAL system, as Rgt1p is primarily a repressor and secondarily an activator . In the absence of glucose, Rgt1p represses HXT gene expression; however, it is also required for maximal HXT1 expression under high glucose conditions . In glucose-deficient environments, Rgt1p is bound to the promoters of HXT1, HXT2 and HXT3. When the glucose concentration increases above 0.1%, it dissociates from these promoters, allowing transcriptional activation to proceed . Rgt1p-dependent de-repression is directly proportional to its phosphorylation state; increasing the quality of the carbon source results in Rgt1p becoming increasingly phosphorylated . The sites of this phosphorylation are likely to be Ser753, Ser755 and Ser758, because an Rgt1p variant containing a deletion of amino acids 750–760 does not become phosphorylated, and constitutively represses HXT gene expression . The kinase responsible for this phosphorylation is not currently known.
Cat8p and Sip4p
Cat8p and Sip4p both bind to the carbon-source-responsive element and regulate gluconeogenic gene expression. The genes FBP1, PCK1 and ICL1 become active when the sole carbon source is non-fermentable (e.g. glycerol or ethanol) [17–19]. Cat8p, like Gal4p, migrates as two distinct bands (Cat8p I and Cat8p II) on SDS/PAGE when harvested under non-inducing, non-repressing conditions. However, when grown on a non-fermentable carbon source, a third form appears, Cat8p III, which corresponds to the expression of gluconeogenic genes. Moreover, Cat8p II and Cat8p III disappear when cells are provided with glucose, mirroring the physiological shift to a fermentable carbon source. In a manner analogous to Gal4p, phosphatase treatment of Cat8p results in the disappearance of Cat8p II and Cat8p III, with a corresponding increase in Cat8p I [17,19]. The appearance of these forms is, in part, dependent on the Snf1p kinase, as the appearance of the slower migrating forms and gluconeogenic gene expression are greatly reduced in snf1Δ strains . In addition, two Snf1p consensus sites are present in both Kluyveromyces lactis (Kl) Cat8p and S. cerevisiae (Sc) Cat8p. Mutating ectopically expressed ScCat8 Ser562 to alanine, in an snf1Δ background, abolishes the ability to grow on a non-fermentable carbon source, while the S562E mutation confers greater growth than seen in wild-type strains. Similarly, mutation of the corresponding serine residue in KlCat8p to alanine reduces ICL1 gene expression, whereas mutation to glutamic residue enhances it . Although Snf1p appears to be the Cat8p activating kinase, neither Klsnf1Δ nor Scsnf1Δ completely abolishes Cat8p phosphorylation .
Sip4p is activated and concurrently phosphorylated under glucose limiting conditions. When grown under repressing conditions (glucose concentrations of ≤0.05%), Sip4p migrates primarily as a singlet on SDS/PAGE, with a minor secondary band migrating slightly slower. However, under de-repressing conditions, most of the Sip4p migrates as the higher-molecular-mass species. This higher-molecular-mass Sip4p isoform is reduced to a singlet upon phosphatase treatment . Thus, like the previously discussed Zn(II)2Cys6 transcription factors, Sip4p shows a concurrence between phosphorylation and activation. This phosphorylation is required for a rapid response to glucose limitation and is dependent on Snf1p, in a Gal83p-dependent manner . Yet, the kinetics of this phosphorylation are only slowed in GAL83 mutants, indicating that full Sip4p phosphorylation is reliant on additional kinases . Indeed, Sip4p phosphorylation in response to non-fermentable carbon sources is also reduced in srb10Δ mutants. Moreover, Sip4p not only physically associates with Srb10p, but its activity is reduced in an srb10Δ background, which is subsequently rescued by ectopic Srb10p. It has been suggested that Snf1p phosphorylation primes Sip4p for Srb10p phosphorylation, which could represent a mechanism like that proposed for Gal4p, where Gal80p de-repression by Gal3p primes Gal4p for Srb10p phosphorylation, and thus full activation .
Put3p and Leu3p
Put3p is required for the full expression of the proline utilization genes, PUT1 and PUT2. Put3p exhibits full activation when proline is the sole nitrogen source; however, it also exhibits increases in basal activity under nitrogen-limiting conditions . As nitrogen source quality decreases, Put3p exhibits increases in activity concurrent with increases in phosphorylation, showing maximal phosphorylation when proline is the sole nitrogen source . Like the other Zn(II)2Cys6 transcription factors we have discussed, the apparent high-molecular-mass forms of Put3p are reduced to a single band upon phosphatase treatment . Put3p also exhibits hyperphosphorylation when cells are treated with the antibiotic rapamycin, which is known to mimic nitrogen de-repressing conditions, suggesting that Put3p could be responding to nitrogen-sensing networks . In vitro, Put3p will bind proline and activate transcription when it is in an unphosphorylated form . It is possible, however, that phosphorylation could increase the efficiency of both these events, although this has not yet been tested. Moreover, as the kinase and precise sites of phosphorylation have not been identified in Put3p, it is not possible to discern whether or not phosphorylation is a cause of transcriptional activation or an effect.
Leu3p is required for the transcription of the genes involved in branched chain amino acid synthesis . Leu3p activation, like that for Put3p, is dependent on the presence of a low-molecular-mass metabolite, in this case α-isopropylmalate, an intermediate of leucine biosynthesis . Western-blot analysis of highly purified Leu3p has demonstrated that the protein migrates as a doublet, which is reduced to a single band upon phosphatase digestion . No further data are available on the relationship between Leu3p phosphorylation and activation.
Prd1p, Pdr3p and War1p
Pdr1p, Prd3p and War1p all control the expression of ABC transporters (ATP-binding-cassette transporters). Pdr1p and Pdr3p co-regulate the expression of PDR5, SNQ2 and YOR1 and bind to the PDRE (pleotropic drug resistance element), while War1p governs the expression of PDR12 by binding to the WARE (weak acid response element) [28,29]. Although both Pdr1p and Pdr3p have been identified as phospho-proteins , with Pdr3p exhibiting phosphorylation at its C-terminal half, no function has been ascribed to these events . War1p (like Put3p, Leu3p and Gal4p) exhibits phosphorylation coincident with activation, yet the kinase responsible and sites of phosphorylation are not known .
Many Zn(II)2Cys6 transcriptional activators are known to show increases in phosphorylation coincident with transcriptional activation; however, the actual relationship between these two events is often unclear. Indeed, it is not readily apparent whether phosphorylation takes place as a cause of activation or as an effect of the activation process. The best-explored examples of this relationship are Gal4p and Sip4p, which regulate carbon metabolism, and do provide some insight as to how these events might operate. In the Sip4p and Gal4p systems, it is possible that priming for phosphorylation occurs upon initial activation: in the case of Gal4p, de-repression of Gal80p, and in the case of Sip4p, phosphorylation by Snf1p. Subsequently, full activation occurs upon recruitment of the transcriptional machinery, via phosphorylation by one of its associated kinases: an Srb10p-dependent event in the case of both Gal4p and Sip4p. Contrary to this assertion, only the mutation of one out of four identified phosphorylation sites in Gal4 reduced its transcriptional activity. Moreover, both Gal4p and Sip4p still exhibit band shifts when the known kinases and phosphorylation sites are ablated. It therefore seems likely that only key residues are required for priming and/or activation and subsequent residues are phosphorylated purely as a result of transcriptional activation.
Although Gal4p and Sip4p provide us with a potential model, it is possible that they share this prospective mechanism because they are both associated with carbon metabolism and repressed by glucose. Thus they are responding to similar signals and signalling pathways. It could be envisaged that genes responding to other stimuli may operate through an entirely different mechanism, potentially not interacting with the Mediator-associated Srb10p for post-priming activation. However, much more information regarding this relationship is required to make a broader assumption about the role phosphorylation plays in governance of this family of transcriptional regulators.
We thank Christopher Sellick (Faculty of Life Sciences, University of Manchester) for critically reading this paper. Work in our laboratory is funded by the BBSRC (Biotechnology and Biological Sciences Research Council) and The Wellcome Trust.
Information Processing and Molecular Signalling: A Focus Topic at BioScience2006, held at SECC Glasgow, U.K., 23–27 July 2006. Edited by M. Clague (Liverpool, U.K.), P. Cullen (Bristol, U.K.), S. Keyse (Dundee, U.K.), R. Layfield (Nottingham, U.K.), J. Mayer (Nottingham, U.K.), P. Newsholme (University College Dublin, Ireland), R. Porter (Trinity College Dublin, Ireland), R. Reece (Manchester, U.K.), S. Shears (NIH, U.S.A.), S. Shirazi-Beechey (Liverpool, U.K.), S. Urbé (Liverpool, U.K.) and M. Wymann (Basel, Switzerland). The first five papers featured in this section are from the Ubiquitin, Proteasomes and Human Diseases mini-symposium, which is dedicated to the memory of Cecile Pickart.
Abbreviations: GTF, general transcription factor; SAGA, Spt-Ada-Gcn5 acetyl transferase
- © 2006 The Biochemical Society