At the heart of the negative feedback response that controls Met4 function is the expectation that repression of the Met4-controlled MET (Fig. 1, sulfate -> methionine) and SAM (Fig. 1, methionine -> S-adenosyl-methionine) genes in response to an accumulation of methionine lowers the levels of sulfur metabolites both in the sulfate/methionine and in the cysteine/glutathione part of the pathway. Contrary to this expectation, recent quantitative proteome and metabolite profiling analyses by Lafaye and colleagues 20 reveal that the levels of the methionine-derived sulfur metabolites (Fig. 1, methionine-> S-adenosyl-methionine -> cysteine -> glutathione) do not drop under conditions inducing repression of the MET and SAM promoters.

The way this surprising finding has been explained is that the high concentration of methionine necessary for Met4 inhibition is sufficient to drive nearly normal sulfur assimilation even when the levels of Met4-controlled enzymes are compromised. This possibility is likely, because expression of the Sam1 and Sam2 enzymes that are necessary for the conversion of methionine into S-adenosyl-methionine, while reduced under the conditions of Met4 inhibition, is still considerable (Fig. 1, methionine -> S-adenosyl-methionine, purple arrow) and the levels of sulfur metabolites are typically well below the K_m values of the respective enzymes 20. This explanation also agrees with the observation that the MET4 gene can be deleted without a major phenotype when methionine is supplied in the growth medium 15, indicating that methionine is converted into cysteine and glutathione in the absence of Met4 function.

These observations have three general implications for how sulfur assimilation is regulated. First, if the biosynthetic enzymes are not saturated, changing the pools of precursors and products would be sufficient to drive rapid metabolite fluxes. The accumulation of cysteine in response to an uptake of extra-cellular methionine (Fig. 1, purple) and the variations in sulfur metabolite pools under different culture conditions could both be explained by this mechanism 20. Second, by uncoupling sulfur assimilation from transcriptional regulation the repression of Met4-controlled genes would make sulfur assimilation even more susceptible to regulation via the pools of precursors and products. Among its benefits, this type of regulation would be directly linked to changes in the rate with which the metabolites are utilized during growth (Fig. 1, turquoise) and could be buffered by the levels of glutathione, the end product of sulfur assimilation (Fig. 1, navy blue). Indeed, glutathione can accumulate more than any other sulfur metabolite (2–15 mM vs <4 μM homocysteine, 30–80 μM for most sulfur metabolites and 500 μM methionine), allowing sulfur accumulation when it is available in the environment and serving as sulfur reserve during starvation 2120.

Conversely, if the enzymes are not saturated, activation of Met4 controlled gene expression could modify the status quo established by the pools of precursors and products. An example of such an effect is the activation of Met4-controlled MET and SAM promoters during the entry into exponential growth observed in yeast cultured in the presence and absence of methionine 3. The way this observation could be explained is that upon growth activation, more methionine and cysteine is redirected to protein biosynthesis (Fig. 1, turquoise, growth up), modifying the status quo established at the previous growth rate. Another example is the yeast Cd²⁺ response driven by the necessity to produce large amounts of glutathione that chelates the metal ions and exports them to vacuole for detoxification 22. This response overcomes the inhibitory effect of high cysteine levels and reduces the synthesis of sulfur-rich proteins (Fig. 1, navy blue) concurrently inducing expression of Met4-controlled genes (Fig. 1, green; 232420). As a result, sulfur metabolism aimed at the production of methionine and cysteine for protein synthesis (Fig. 1, turquoise) is redirected to the synthesis of glutathione (Fig. 1, navy blue).

In summary, according to the ever-changing demands of growth and stress responses, transcriptional regulation of the Met4-controlled genes either increases the sensitivity to (via repression), or modifies the outcome of (via activation) the regulation by rapid sulfur metabolite fluxes. While the accumulation of cysteine driven by high levels of methionine controls the regulatory aspect associated with transcriptional repression, cell growth and/or stress responses regulate sulfur metabolism primarily via activation of the Met4-controlled promoters.

The initial interpretation of analyses regarding the role of SCF^Met30 in Met4 regulation was driven by the assumption that SCF^Met30 acts exclusively as a negative regulator during an accumulation of cysteine. However, recent evidence suggests that SCF^Met30 can also activate Met4 and that the activation represents the primary aspect of Met4 regulation by SCF^Met30. In this model, the effect of cysteine is viewed as a block in Met4 activation (Fig. 1, purple) that is normally dependent on methionine (Fig. 1, green), supporting the notion that SCF^Met30 can either inhibit or activate Met4 depending on the sulfur metabolite context 3.

The first implication for the model that transcriptional repression is not the only function of Met4 regulation by SCF^Met30 was the observation that SCF^Met30 binds Met4 under repressive and non-repressive conditions 93. Turnover of Met4 protein has been next implicated in the activation of Met4-controlled promoters upon the entry into exponential growth in a medium with or without methionine 3. The notion that cell division plays a positive role in the activation of Met4 via proteolysis (Fig. 1, orange up) agrees with the positive role of Met4 proteolysis in cell division (Fig. 1, orange down), which has long been suggested by the observation that met30Δ yeast arrest at the G1-S phase while simultaneous deletion of the MET4 gene bypasses the arrest 19. The mechanism by which proteolysis could activate the function of a transcriptional activator is unknown, but this phenomenon appears to have a broad significance, as turnover of several transcriptional activators, including Gcn4, Gal4, and Ino2/4, has been implicated in their activity 25, and the activation domain of many transcriptional factors overlaps with the domain required for their proteolysis 26.

In light of the possibility that Met4 is activated via proteolysis, how legitimate is the view that Met4 is 'fully' active under conditions preventing its regulation by SCF^Met30? Due to the lethal phenotype of met30Δ yeast 19, this view is based only on the observation that the in vivo active Met4^K163R mutant does not accumulate as polyubiquitinated protein and cannot be inhibited by SCF^Met30 during growth on a methionine-rich synthetic medium 16. Ironically, the same report shows that Met4^K163R protein still binds SCF^Met30 and is highly unstable in a manner dependent on Cdc34 and SCF^Met30 function. Rapid turnover in response to polyubiquitination of a lysine distinct from K163 could best explain this phenomenon, suggesting that the K163R substitution prevents the inhibition, but not the activation of Met4 by SCF^Met30.

If proteolysis is necessary for Met4 activation in the context of cell division, could it also be necessary for Met4 activation during Cd²⁺ response, which is associated with strong up-regulation of Met4-controlled genes, including MET30 and MET32 24? The accumulation of Met30 protein in Cd²⁺ treated cells indeed suggests an increased demand for SCF^Met30 function. On the other hand, under the conditions of Cd²⁺ response, most of Met30 is detectable as a SCF^Met30-free protein 2728. This effect could reflect SCF^Met30 inactivation, as it has been proposed, explaining how the Cd²⁺ response overcomes the inhibitory effect of high cysteine levels (Fig. 1, purple with a navy blue cross). In this view, SCF^Met30 would not be able to participate in either inactivation or activation of the Met4-controlled genes. A different possibility is suggested by the more recent observation that Met4 proteolysis leads to the ATP hydrolysis-dependent disassembly of the proteasome and of SCF^Met30 (Babbitt and Skowyra, unpublished), an effect first linked to proteolysis of Sic1, the prototype substrate of the SCF^Cdc4 ubiquitin ligase 29. In this view, disassembly of SCF^Met30 could reflect its vigorous involvement in Met4 activation via proteolysis (Fig. 1, green), linking Cd²⁺ response to Met4 activation by SCF^Met30.

The notion that Met4 function depends on activity of SCF^Met30 rather then on its inactivation is additionally supported by the finding that the GSH1 promoter can be activated by Cdc34 overproduction, explaining why Cdc34 has been identified as a high copy suppressor of the defect in glutathione biosynthesis associated with the gsh2 mutant (30; Fig. 1, Gsh1). Overproduction of the Cdc34 E2 necessary for SCF^Met30 function could not activate the GSH1 gene expression if the expression were dependent on SCF^Met30 inactivation.

While the cofactor-free Met4 protein can be polyubiquitinated and degraded in vitro 3 and in vivo 14, the essential role of SCF^Met30 is most likely associated with the regulation of Met4 function in the context of transcriptional complexes.

The observation that the cofactors Cbf1, Met28, and one of the related Met31 and Met32 proteins can stabilize the interaction between Met4 and SCF^Met30 3 suggested first that SCF^Met30 could regulate protein-protein interactions within the transcriptional complexes. This possibility is further supported by the observation that SCF^Met30 binding is sufficient to dissociate Met4 homodimers (Chandrasekaran and Skowyra, unpublished). During this process, one Met4 monomer remains bound to SCF^Met30, while the other is released as a free protein (Fig. 2B). Dissociation of Met4 homodimers to free monomers can be observed in yeast extracts lacking Cbf1, Met28 and Met31 or Met32, while the monomers readily associate into large complexes when the cofactors are present 3. The monomers do not re-dimerize spontaneously in the absence of the cofactors, suggesting that a major, not easily reversible change in Met4 structure accompanies its monomerization by SCF^Met30. That a major structural change accompanies Met4 activation has been independently proposed based on genetic analysis of the constitutively active Met4-1 and Met4-2 mutants, which display a dominant negative phenotype in the presence of wild type Met4 31.

These observations have two implications for the mechanism by which SCF^Met30 regulates Met4. First, the dissociation of Met4 homodimers by SCF^Met30 could be necessary to stabilize and/or rearrange the protein-protein interactions within the Met4/Met28/Cbf1 complex and thereby trigger its full functional potential (Fig. 3B), explaining the positive regulatory role of SCF^Met30. Indeed, Met4 dimerization represents an interesting conundrum, as it depends on the LZ domain also implicated in binding to the basic leucine zipper (bZIP) regulatory protein Met28 and the basic helix-loop-helix (bHLH) DNA binding protein Cbf1 (Fig. 2A, LZ). As such, Met4 dimerization could antagonize proper assembly and function of the Met4/Cbf1/Met28 complex.

Second, if monomers represent an activated form of Met4 that cannot spontaneously form homodimers, degradation of Met4 monomers could be necessary to alleviate the positive effect of SCF^Met30. As a result, 'two-stepping' with SCF^Met30 might be necessary for each round of Met4 activity at a promoter, with the first step facilitating Met4 activation (Fig. 3B, step 1; and 3B) and the second step removing the activated Met4 (Fig. 3C–D, step 2). In this view, turnover of Met4 protein would be an integral part of its function, with only a fraction of the total cellular Met4 degraded at a time, balancing the constitutive expression of the MET4 gene (Fig. 3A; 931).

Finally, while degradation of the cofactor-free Met4 molecule (Fig. 3B, step 1) would have no immediate consequence for Met4 function at the promoter (Fig. 3B), its proteolysis could be necessary to recycle SCF^Met30. The observation that over-expression of Met30 is sufficient to enhance repression of the Met4 controlled genes is the best illustration that SCF^Met30 activity is limiting in vivo 32. On the other hand, over-expression of Met30 protein induces cell cycle arrest, demonstrating that low levels of Met30 are a necessary compromise associated with some other aspect of SCF^Met30 function. How does the abundance of Met30, which restricts the abundance of SCF^Met30, compare with the abundance of Met4? Direct measurements of protein expression in yeast under prototrophic growth conditions suggest that Met30 is about 6-fold less abundant than Met4 (217 vs. 1300 molecules per cell, respectively; 33). Its levels are expected to drop even more during auxotrophic growth as a result of MET30 gene repression and Met30 protein instability 932. The significance of the fluctuations of Met30 protein and MET30 mRNA levels 9 still needs to be determined, but low levels of SCF^Met30 appear to be an intrinsic aspect of Met4 regulation. Recycling of SCF^Met30 molecules could thus have evolved as a mechanism controlling the regulatory potential of Met30 without the necessity to accumulate SCF^Met30.

While the proposed role of SCF^Met30 in activating Met4 via dissociation of Met4 homodimers remains to be verified, this scheme illustrates particularly well how Met4 proteolysis could regulate sulfur assimilation even if the Met4 molecules at an active promoter are not polyubiquitinated or degraded, and even if the steady-state levels of Met4 do not change.

Like in the case of all other known SCF substrates, polyubiquitination of Met4 has been linked to rapid proteolysis 6913143. However, Met4 degradation is unusual as it can be blocked in vivo 151617133 and in vitro 3, leading to an accumulation of Met4 protein modified with the type of polyubiquitin chain that normally serves as a proteolytic signal.

Two models have been suggested for the stabilization of polyubiquitinated Met4. In a model proposed by Kaiser's group, a

u

i

m

The possibility that assembly between Met4, cofactors and SCF^Met30 controls function of the UIM domain could easily unite the models (Fig. 3C; exposure of UIM). Both models also agree that Met4 phosphorylation is necessary for its recruitment to SCF^Met30 31718 and that polyubiquitination of residue K163 is sufficient to inhibit Met4 activity (Fig. 3C, OFF). It is the involvement of proteolysis and its role in Met4 regulation that distinguishes the models. In the Kaiser et al. 1718 model, the inhibition of Met4 by proteolysis-resistant polyubiquitination is the only meaningful aspect of SCF^Met30 function. The Chandrasekaran et al. 3 model points out that the stabilization of polyubiquitinated Met4 only delays proteolysis, keeping it in a "stand-by" mode for further regulation, and that depending on the sulfur metabolite context, the proteolysis can play different roles. As we discuss below, this mechanism appears to be key for defining whether Met4 protein turnover is part of its normal activity cycle (Fig. 3A–D), or reinforces the inhibitory scheme (Fig. 3D^REP).

The central premise of the model proposed by Chandrasekaran et al. 3 is that stabilization and destabilization of binding between Met4, cofactors and SCF^Met30 is key to modulating the fate of polyubiquitinated Met4. Stabilization could trigger Met4 inhibition by polyubiquitination (Fig. 3C) unless destabilization of the complex would promote degradation of polyubiquitinated Met4 (Fig. 3D and 3D^REP). The notion that Met4 proteolysis can play different roles is suggested by the finding that either free methionine (Fig. 3D) or cysteine (Fig. 3D^REP) can promote degradation of polyubiquitinated Met4 by destabilizing its interaction with cofactors and SCF^Met30, but yet the amino acids act in a different concentration range in vitro (5 and 500 μM, respectively) and would signal different situations in vivo.

The methionine-mediated proteolysis would play its role best in the standard Met4 activity cycle (Fig. 3A–D; Fig. 1, green and orange). The low K_i for methionine (5 μM) in triggering the proteolysis of polyubiquitinated Met4 in vitro suggests that either low or high methionine concentration would destabilize Met4 in vivo. In agreement with this prediction, a high concentration of methionine is insufficient for repression of the MET and SAM promoters in vivo - the biosynthesis of cysteine, with the intermediate production of S-adenosyl-methionine are necessary 12. Met4 protein is thus active at low and high methionine concentrations as long as cysteine does not accumulate, meaning that Met4 activation, while dependent on methionine, is not limited to low methionine levels. In contrast, the high K_i for cysteine (500 μM) established in the in vitro studies suggests that the cysteine-mediated mechanism would engage in Met4 regulation only when cysteine accumulates above its normal (~80 μM) level, in agreement with its role in transcriptional repression 1214. While in vivo both schemes depend on biosynthesis of S-adenosyl-methionine (Fig. 1, methionine -> S-adenosyl-methionine, green and purple arrows), a highly unstable methionine metabolite, S-adenosyl-methionine does not trigger Met4 proteolysis in vitro 3, suggesting either that its role is indirect, or that it controls a yet undefined aspect of Met4 regulation.

How could the roles of cysteine and methionine be different if they both function via destabilizing Met4 complexes? An answer to this question may come from determining whether the same or different cofactor(s) serves as a sensor for methionine and cysteine, and whether methionine and cysteine are equally potent in dissociating cofactors from Met4. A different potency of methionine and cysteine in dissociating Met4 from cofactors could play a major role in light of the observation that proteolysis of polyubiquitinated Met4 (Babbitt and Skowyra, unpublished), like proteolysis of polyubiquitinated Sic1 29, is associated with the robust, ATP-hydrolysis-dependent disassembly of the 26S proteasome and the Met4-interacting proteins, including SCF^Met30. If methionine only destabilizes the interaction between Met4 and cofactors, disassembly of the transcriptional complex via the proteasome-dependent mechanism could be part of the Met4 activation cycle (Fig. 3D). In contrast, cysteine-mediated dissociation prior to Met4 proteolysis would protect the cofactors from effects of the proteasome-mediated disassembly (Fig. 3D^REP). One major difference between these mechanisms is that only in the case of the proteasome-mediated disassembly could promoters be cleared of the DNA binding cofactors Met31/32 and Cbf1, which bind DNA even in the absence of Met4.

The finding that deletion of the MET30 gene leads to cell cycle arrest unless accompanied by deletion of the MET4 or MET32 gene, but yet the MET4 gene alone can be deleted without a major phenotype as long as methionine is supplied in the growth medium 19 creates an interesting regulatory conundrum. It suggests that some yet undefined, Met32-linked aspect of the SCF^Met30-Met4 interplay rather than the inactivation and/or degradation of the Met4 protein per se plays an essential role. Its link to Met32 19 has recently been verified by analysis of the Met32^Δ145-192 dominant suppressor of the met30Δ cell cycle defect, for the first time implicating Met32 in stabilization of the Met4-Cbf1 complex 5. However, no mechanistic insight regarding the role of Met32 in the link between Met4 regulation by SCF^Met30 and cell division has been revealed. Similarly, analysis of the met30ts yeast showed that SCF^Met30 plays multiple roles in the cell cycle 28, but did not reveal the mechanism responsible for the cell cycle roles.

A potential mechanism by which the process of Met4 proteolysis rather then Met4 removal per se could be linked to cell division is suggested by the possibility that during normal Met4 activity cycle (Fig. 3A–D) the proteasome disassembles the transcriptional complexes and clears promoters from the DNA binding cofactors Cbf1 and/or Met31 or Met32 (Fig. 3D). Cbf1 (

c

b

f

1

In support of this model, Met4 stabilizes the Cbf1-DNA interaction in a manner dependent on Met28 2 and Met32 5, making its destabilization dependent either on presence of Met30, which by promoting Met4 polyubiquitination would recruit the proteasome, or on absence of MET4 gene, which by eliminating Met4 would prevent the Cbf1-DNA stabilization by Met28 and/or Met32. The model also agrees with the observation that expression of the Met32^Δ145-192 suppressor of the met30Δ cell cycle arrest prevents recruitment of Met4 to promoters 5, preventing Met4-dependent stabilization of Cbf1-DNA, and, possibly, Met31-DNA binding.

In principle, this simple mechanism could explain the link between Met4 turnover and cell division in the case of any cofactor that has to be shared between different cellular functions. Whether Cbf1 "mobilization" from Met4-controlled promoters indeed explains the essential nature of the SCF^Met30-Met4 regulatory interplay in cell division and whether similar paradigm applies to the regulation of other transcriptional activators, will need to be determined.