TS (thymidylate synthase) is a key enzyme in the de novo biosynthesis of dTMP, and is indispensable for DNA replication. Previous studies have shown that intracellular degradation of the human enzyme [hTS (human thymidylate synthase)] is mediated by the 26S proteasome, and occurs in a ubiquitin-independent manner. Degradation of hTS is governed by a degron that is located at the polypeptide's N-terminus that is capable of promoting the destabilization of heterologous proteins to which it is attached. The hTS degron is bipartite, consisting of two subdomains: an IDR (intrinsically disordered region) that is highly divergent among mammalian species, followed by a conserved amphipathic α-helix (designated hA). In the present report, we have characterized the structure and function of the hTS degron in more detail. We have conducted a bioinformatic analysis of interspecies sequence variation exhibited by the IDR, and find that its hypervariability is not due to diversifying (or positive) selection; rather, it has been subjected to purifying (or negative) selection, although the intensity of such selection is relaxed or weakened compared with that exerted on the rest of the molecule. In addition, we have verified that both subdomains of the hTS degron are required for full activity. Furthermore, their co-operation does not necessitate that they are juxtaposed, but is maintained when they are physically separated. Finally, we have identified a ‘cryptic’ degron at the C-terminus of hTS, which is activated by the N-terminal degron and appears to function only under certain circumstances; its role in TS metabolism is not known.
INTRODUCTION
The pyrimidine biosynthetic enzyme TS (thymidylate synthase; EC 2.1.1.45) catalyses the reductive methylation of dUMP by N5N10-methylenetetrahydrofolate, to form TMP and DHF (dihydrofolate) [1,2]. In being the sole de novo source of TMP for cellular DNA replication and repair, the enzyme has been an effective target for chemotherapeutic drugs used in the treatment of cancer [1,2]. Anti-metabolites such as 5-fluorouracil and raltitrexed form TS substrate analogues that bind to and inhibit the enzyme, leading to reduced thymidylate pools and apoptosis. X-ray crystallographic studies of the TS polypeptide from an evolutionarily wide range of eukaryotic and prokaryotic species have shown that the enzyme is highly conserved in terms of the location of secondary structural elements and overall three-dimensional (3D) conformation [3–6].
In previous work, we demonstrated that the intracellular stability of TS is increased by the binding of inhibitory ligands [7], a phenomenon that potentially constrains the effectiveness of TS-directed anti-cancer agents. Subsequent studies showed that the human enzyme [designated hTS (human thymidylate synthase)] is degraded by the 26S proteasome in a ubiquitin-independent manner [8–11]. Most proteasomal substrates require ligation of ubiquitin moieties to be recognized and degraded by the proteasome [12]. However, an increasing number of substrates do not require ubiquitinylation, including ODC (ornithine decarboxylase) [13,14], tumour suppressor p21Waf1/Cip1 [15,16], the NF-κB (nuclear factor κB) regulator IκBα (inhibitor of κBα) [17,18], apomyoglobin [19] and the yeast transcription factor Rpn4 [20], among others. Thus, hTS is a model for ubiquitin-independent protein turnover.
Studies in our laboratory have shown that an N-terminal segment of hTS spanning the first 45 residues is required for degradation, defining the region as a degron [9–11]. The hTS degron is portable, in that it is capable of destabilizing a heterologous protein to which it is attached [9–11]. Furthermore, it is bipartite in nature, comprising of two co-operating subdomains: an IDR (intrinsically disordered region) spanning residues 1–27, and an amphipathic α-helix (designated hA) at residues 31–42 [9]. The IDR is hypervariable in amino acid sequence among mammalian species [11], although its disordered nature has been conserved [4–6,11]. A free, unmodified N-terminal amino group and an arginine–arginine motif at residues 10–11 are required for IDR activity in promoting proteasomal degradation [10,11]. The hA segment, which is much more highly conserved among species [11], requires maintenance of an α-helical conformation for its degradative function [9]. Interestingly, the hTS degron is active when placed at either the N- or C-terminus of the reporter substrate [9]. The specific roles of the two subdomains in regulating hTS proteolysis have yet to be determined.
In the present report, we have carried out further analysis of the structure and function of the hTS degron. We have examined the origin of inter-species hypervariability within the IDR, and find that its tolerance of amino acid substitution derives from relaxed purifying selection, as opposed to positive Darwinian selection. We also show that co-operation between the IDR and hA does not require that the two elements be juxtaposed, but is maintained when they are physically separated. Finally, we have identified a ‘cryptic’ degron at the C-terminus of the hTS polypeptide that appears to be active only under particular circumstances.
EXPERIMENTAL
Bioinformatics
Sequences corresponding to the coding regions of TS mRNAs from 16 mammalian species were downloaded from the Ensembl database [21] and aligned using ClustalW2 [22,23]. Species included human (Homo sapiens), chimpanzee (Pan troglodytes), orang-utan (Pongo abelii), marmoset (Callithrix jacchus), tarsier (Tarsius syrichta), rhesus monkey (Macaca mulatta), mouse (Mus musculus), rat (Rattus rattus), guinea pig (Cavia porcellus), cat (Felis catus), dog (Canis lupus), cow (Bos taurus), horse (Equus caballus), pika (Ochotona princeps), rabbit (Oryctolagus cuniculus) and dolphin (Tursiops truncatus). A 40-codon insertion within the cow and dog sequences, corresponding to an additional exon and likely representing a cDNA copy of an incompletely processed transcript, were removed from the analysis.
Extents of synonymous and non-synonymous substitutions within the coding regions of mammalian TS transcripts were determined for all pairwise comparisons among the 16 species using SNAP (Synonymous Non-synonymous Analysis Program), as available on the HIV database website. SNAP estimates the numbers of synonymous substitutions per synonymous site (dS) and non-synonymous substitutions per non-synonymous site (dN), corrected for multiple hits [24,25]. Cumulative behaviour plots for both types of substitutions along the coding regions were constructed using the ‘xyplot’ option of SNAP.
Identification of codons where dS≠dN was carried out by the FEL (fixed effects likelihood) and REL (random effects likelihood) methods, as provided on the Datamonkey webserver [26]. A P-value of 0.1 was set as the cut-off for FEL.
Prediction of secondary structure was done using PSIPRED, as provided on the Bioinformatics Group Server of University College London. PSIPRED is a two-stage neural network that predicts protein secondary structure based on position-specific scoring matrices [27].
Plasmid constructs
All constructs for mutant analysis (shown in Figure 1) were generated using standard techniques and were verified directly by DNA sequencing. The parental expression plasmid was pMP610, containing a full-length eGFP [enhanced GFP (green fluorescent protein)] cDNA under the control of a CMV (cytomegalovirus) promoter. For selection of transfectants, an Escherichia coli TS cDNA under the control of a murine stem cell virus LTR (long terminal repeat) promoter, optimized for expression in mammalian cells by modification of codon-usage, was included in the plasmid [28]. Mutagenesis was carried out by standard PCR-based protocols. Details of all plasmid constructions are available upon request.
Constructs used in the present study
Cell culture and transfection
Cell line RJK88.13, a TS-deficient derivative of V79 Chinese hamster lung cells [29], was maintained in DMEM (Dulbecco's modified Eagle's medium) (Cellgro) containing 4.5 g/l glucose and supplemented with 10% heat-inactivated fetal bovine serum (Atlanta Biologicals) and 10 μM thymidine. Cells were cultured at 37°C in a humidified 5% CO2 atmosphere.
For generation of stable transfectants, RJK88.13 cells were transfected with the indicated expression plasmids using Lipofectamine™ 2000 according to the manufacturer's instructions. Transfectants were selected in thymidine-free medium containing the nucleoside transport inhibitor dipyridamole (5 μM; Sigma–Aldrich), pooled and maintained in mass culture.
Determination of protein half-lives
Half-lives were assessed following the addition of 50 μg/ml CHX (cycloheximide; Acros Organic) to the growth medium for the indicated times. Cells were harvested by scraping, and lysed in M-Per® buffer (Pierce) containing 10 mM DTT (dithiothreitol), 2 mM 2-mercaptoethanol, 5 mM PMSF, 200 μg/ml aprotinin, 100 μg/ml pepstatin and 50 μg/ml leupeptin. Crude lysates were centrifuged at 15000 g for 1 h at 4°C, and protein concentrations in the resulting extracts were quantified using the Bio-Rad assay reagent with BSA as a standard.
Immunoblotting was performed by standard techniques. Following SDS/PAGE, fractionated proteins were transferred to nylon membranes, and probed with a monoclonal antibody to GFP (Santa Cruz Biotechnology; catalogue no. sc-9996). To control for equal loading, blots were reprobed with an anti-actin monoclonal antibody (Sigma–Aldrich, Clone AC-40). The antigen–antibody complexes were visualized using appropriate secondary antibodies with the ECL® (enhanced chemiluminescence) kit (Amersham Biosciences). Densitometry was carried out using ImageJ software maintained by the National Institutes of Health. A 2-fold dilution series of each extract was included on the blots for calibration and to correct for film exposure times. All values were normalized to actin concentrations on the same blots. Data were plotted according to first-order decay kinetics, and half-lives were determined from the slopes of the resulting lines. Each half-life determination was carried out at least twice with independently prepared cells.
RESULTS
Origin of amino acid sequence divergence within the IDR
The amino acid sequence of the IDR within the TS degron is highly divergent among mammalian species [11]. Despite such divergence, the region's role in degradation is maintained, so long as a free, unblocked N-terminus and an arginine–arginine motif at residues 10–11 are present [10]. This, along with the observation that both the IDR and hA can be replaced with segments from unrelated proteins, led us to suggest that it is the overall structure, rather than the specific primary sequence, that is critical to degron function [9].
In order to gain a deeper understanding of IDR hypervariability and its origin, we aligned and compared the coding regions of the TS mRNAs from 16 mammalian species (Supplementary Figure S1 at http://www.bioscirep.org/bsr/033/bsr033e015add.htm). The IDR (encoded by nucleotides 1–81 of the human transcript) spans a maximum of 31 codons, including small insertions/deletions in some species. The remainder of the transcript (encoded by nucleotides 82–939 of the human, which includes hA) spans 286 codons and contains no insertions/deletions. Within the IDR, only 16 nucleotide sites (17.8%, not including the initiator Met codon) are completely conserved among the species; most of these are located within the region encompassing residues 5–8 (Supplementary Figure S1). In contrast, 556 sites (64.8%) within the body of the polypeptide are conserved. The difference in degree of nucleotide conservation between the IDR and the remainder of the molecule is highly significant (P<10−16), and indicates a dramatically higher level of divergence for the IDR-encoding portion of the mammalian TS transcript.
Cumulative behaviour plots (Figure 2) indicated that synonymous substitutions are distributed equally throughout the length of the mRNA (Figure 2). They occur at a rate of 0.227±0.0079 per codon within the region encoding the IDR (R2=96.96%; 95% CI=0.211–0.243) and 0.233±0.00073 per codon within the region encoding the body of the polypeptide (R2=99.72%; 95% CI=0.231–0.234). Thus, no significant differences were observed between the two domains (P= 0.40). In contrast, non-synonymous substitutions are distributed in a biphasic manner (Figure 2), with the rate being 0.395±0.011 per codon within the IDR (R2=97.97%; 95% CI=0.372–0.418), as compared with 0.0660±0.0005 per codon within the body of the molecule (R2=98.09%; 95% CI=0.065–0.067). Thus, a 6–7-fold higher rate of accumulation of non-synonymous substitutions is observed for the IDR (P<10−16).
Cumulative behaviour of synonymous and non-synonymous substitutions along the coding region of TS mRNA
Within the body of the polypeptide, the occurrence of non-synonymous substitutions is 3.5-fold lower than for synonymous substitutions (i.e. 0.0660 compared with 0.233 per codon, respectively; P<10−16) (Figure 3). This indicates strong negative selection constraining amino acid substitutions throughout most of the TS polypeptide. In contrast, within the IDR-encoding region, the rate for non-synonymous substitutions is 1.7-fold higher than for synonymous substitutions (0.395 compared with 0.227 per codon, respectively; P<10−16) (Figure 3). This may reflect positive Darwinian selection, whereby amino acid substitutions within the IDR are beneficial or adaptive, and are therefore favoured. Alternatively, it could be a consequence of relaxed or weakened intensity of negative (or purifying) selection, indicating a greater tolerance of amino acid changes.
Rates of accumulation of synonymous and non-synonymous substitutions along the TS mRNA
To distinguish between these two possibilities, we determined the proportions of non-synonymous substitutions per non-synonymous site (dN) and synonymous substitutions per synonymous site (dS), and calculated dN/dS ratios. Purifying, or negative, selection is indicated when such ratios are <1, while positive selection is indicated when they are >1; dN/dS ratios equal to 1 indicate neutrality, or an absence of selection. The dN and dS values for each of the 120 pairwise combinations among the 16 mammalian species were determined for both the IDR and the body of the polypeptide (see Supplementary Table S1 at http://www.bioscirep.org/bsr/033/bsr033e015add.htm). Average dN/dS ratios were 0.5953±0.4433 for the IDR and 0.0745±0.0334 for the body of the molecule. Thus, the IDR exhibited an 8-fold higher ratio (P<10−16) that was still significantly less than 1.0 (P<10−16). This indicates that amino acid substitutions within the disordered segment are constrained by negative selection, albeit at a relaxed or milder intensity as compared with the remainder of the polypeptide. No evidence for positive Darwinian selection was detected.
As an independent test of this conclusion, we utilized an approach that identifies individual codons under positive or negative selection. REL is a maximum likelihood strategy that determines if dN and dS are significantly different at each codon along a mRNA [26]. The approach failed to detect any positively selected sites in either the IDR or the body of the TS polypeptide. In contrast, 10.0% of sites within the IDR (3 of 30, not including the initiator Met codon) and 36.4% of sites within the body of the polypeptide (104 of 286) were identified as being under negative selection. The difference in the fraction of negatively selected codons between the IDR and the body of the polypeptide is statistically significant (P<10−4), and indicates that the intensity of negative selection is diminished in the former.
In all, our analysis indicates that positive Darwinian selection is not a major contributor to high levels of sequence divergence within the IDR of mammalian TS. Rather, the region is constrained by negative selection, although such selection is relaxed as compared with that exerted upon the rest of the polypeptide.
The IDR and hA are both required for degron activity
In previous investigations [11], we found that a fusion protein containing the IDR of hTS (residues 1–30, denoted N30) appended to an eGFP reporter is resistant to degradation, as is the parental eGFP reporter itself. Inserting hA into N30–eGFP between the IDR and the reporter results in a protein (N30–hA–eGFP; see Figure 1) that is significantly destabilized (Figure 4 and [11]). On the basis of these findings, we concluded that both the IDR and hA are required for degradation. However, the possibility that hA on its own functions as a degron was not formally tested. To do this, we measured the half-life of a polypeptide, hA–eGFP, in which hA alone was ligated to eGFP (Figure 1). A stably transfected cell line expressing hA–eGFP was generated (see the Experimental section), and its half-life was measured by a CHX chase assay. As seen in Figure 4, hA–eGFP, like the parental eGFP substrate, is quite stable, having a half-life >24 h. Thus, hA is inactive on its own as a degron. We conclude that neither the IDR nor hA is capable of acting alone in promoting degradation; both segments are required, and must co-operate for full degron function.
hA is inactive on its own as a degron
A helical conformation is required for degron function
Earlier, we observed that insertion of helix-disrupting proline substitutions at residues 40 and 41 within the hA segment of fusion protein N30–hA–eGFP stabilized the molecule, indicating that a helical conformation is required for maximal degron function [9]. This was the case regardless of whether the degron was at the N- or C-terminal end of the reporter [9]. As a more rigorous test of this requirement, we examined an eGFP-fusion protein containing a very strong degron derived from a mutant hTS molecule. Polypeptide N30–hAI/I–eGFP (see Figure 1) contains isoleucine substitutions at positions 39 and 42 within hA, rendering the protein very unstable, with a half-life of about 0.8 h [9]. We used the secondary structure prediction tool PSIPRED [27] to assess the potential effect of proline substitutions on the helical propensity of the hA segment in both N30–hAI/I–eGFP and in its wild-type counterpart N30–hA–eGFP. As seen in Figure 5, for both the wild-type and mutant segments, an α-helical domain spanning residues 31–42, which is the precise location of hA as detected by X-ray crystallography [4,5], was predicted with high confidence for both segments. In contrast, loss of the helical conformation was predicted when proline residues were introduced at positions 40 and 41 (Figure 5). Thus, the secondary structure of hA is likely to be significantly disrupted by insertion of proline substitutions at positions 40 and 41.
Predicted impact of amino acid substitutions on the helical propensity of hA segment
With this information, we measured the half-life of polypeptide N30–hAI/P/P/I–eGFP, which is a derivative of N30–hAI/I–eGFP containing proline substitutions at sites 40 and 41 within hA (see Figure 1). As shown in Figure 6(A), N30-hAI/P/P/I-eGFP had a half-life >24 h, indicating it to be significantly more stable than its parent, N30–hAI/I–eGFP. This provides strong verification of the notion that promotion of degradation by hA segment requires a helical conformation.
Degron activity requires a helical segment
To assess the generality of this conclusion, we tested the requirement for α-helix in a completely different system. We have shown that the N-terminal domain of the Escherichia coli lac repressor, which contains an HLH (helix-loop-helix) motif spanning residues 9–29, functions as a degron [9]. Deletion of the first helix of the HLH motif (residues 9–17) stabilizes the polypeptide [9], whereas deletion of the second (residues 22–29) has no effect (K. Barbour, unpublished work); thus, the first helix drives the region's degron activity. To test if an α-helical conformation is required, we introduced proline substitutions at residues 13 and 14 of the first helix, resulting in LACP/P–eGFP (see Figure 1). PSIPRED correctly predicted the wild-type LAC region to have an α-helical conformation spanning residues 9–17, which was abolished by insertion of proline at sites 13 and 14 (results not shown). Figure 6(B) shows that LACP/P–eGFP exhibited a half-life of ~7 h, which is significantly longer than the parental LAC–eGFP molecule. Thus, as with hA of hTS, full degron function requires maintenance of a helical conformation.
Juxtaposition of the IDR and hA is not required for co-operative degron function
The hTS degron exerts its degradative function when placed at either the N-terminal or C-terminal end of the substrate [9]. Given its bipartite nature, it was of interest to ask if the two co-operating segments, i.e. the IDR and hA, must be juxtaposed. We generated and tested a fusion construct (denoted N30–eGFP–hA) having the IDR at the N-terminus and hA at the C-terminus (see Figure 1). As shown in Figure 7, N30–eGFP–hA had a half-life of 1.4 h, which is considerably shorter than that exhibited by the N30–hA–eGFP polypeptide. Thus, the two segments of the hTS degron co-operate even when placed at opposite ends of the target substrate. We also tested a protein (termed hA–eGFP–IDR; see Figure 1) having the opposite orientation, i.e. with hA at the N-terminus and the IDR at the C-terminus. Interestingly, hA–eGFP–N30 was very stable, exhibiting a half-life >24 h (Figure 7). Thus, it may be that the ability of the two degron components to co-operate when separated is position-dependent, although further studies are necessary to verify this interpretation.
The IDR and hA co-operate when physically separated
The C-terminal end of TS contains a ‘cryptic’ degron
Previously, we observed that blocking the N-terminal amino group of hTS, either with a histidine-tag or by mutations that promote N-α-acetylation, results in a very stable polypeptide [8]. We interpreted this as indicating that initiation of proteasomal degradation occurs at the N-terminal end, and proceeds in the N→C direction [8]. Of interest was the finding that certain mutant hTS molecules (e.g. those with a P303L substitution) are stabilized by blocking the C-terminal end, suggesting that degradation of these mutant molecules may occur from the C-terminal end and proceed in a C→N direction [8,10]. Thus, the C-terminal region of hTS may contain a ‘cryptic’ degron that is activated in certain mutant polypeptides.
To test whether the C-terminal domain of hTS functions as a degron, we generated a polypeptide, termed eGFP–C38, containing the C-terminal 88 amino acids of hTS (residues 276–313) appended to eGFP (see Figure 1). The region includes a short, two-stranded β-sheet formed from residues 277–282 and 295–300. Figure 8 shows that the half-life of this molecule was 5.2 h, indicating that the hTS domain promotes degradation of the reporter, perhaps to a greater extent than the N-terminal degron (see Figure 4). Ligation of the N-terminal degron (i.e. residues 1–45) to eGFP–C38 results in a polypeptide, denoted N30–hA–eGFP–C38 (see Figure 1), that has a half-life of 1.5 h, indicating it to be even more unstable than eGFP–C38 (Figure 8). Thus, the C-terminal region of hTS functions as a degron, the activity of which is enhanced by the presence of the N-terminal degron at the other end of the substrate.
A ‘cryptic’ degron at the C-terminus of hTS
To determine if one or the other of the two subdomains comprising the N-terminal degron enhances the activity of the C38 element, we analysed polypeptides containing either the IDR or hA, neither of which is active on its own, appended to eGFP–C38. Polypeptide N30–eGFP–C38 (see Figure 1) exhibited a half-life of 2.9 h (Figure 8), indicating that the IDR modestly destabilizes eGFP–C38. Polypeptide hA–eGFP–C38 (see Figure 1) had a very short half-life of 0.6 h (Figure 8), indicating that hA induces potent destabilization of the substrate. We conclude that both the IDR and hA element, which by themselves are inactive in promoting degradation, promote the degron activity of the C38 element.
DISCUSSION
In earlier studies, we showed that the intracellular stability of TS is increased by the binding of inhibitory ligands [7], a finding that prompted an examination of the mechanism of TS degradation. We showed that proteolysis of the human enzyme (designated hTS) is carried out by the 26S proteasome, and occurs in a ubiquitin-independent manner. Furthermore, it is mediated by a portable, bipartite degron comprised of an IDR and an amphipathic α-helix located at the enzyme's N-terminal end [8–11]. A striking characteristic of the IDR is the occurrence of extensive sequence divergence among mammalian species [11]. Despite a high rate of amino acid substitution during evolution, the IDR has maintained its disordered character and its role in degron function. Thus, it is overall structure (or lack thereof) that is critical to degradation [9]. The specific amino acid sequence is less important, so long as it does not promote order.
It is well-recognized that IDRs typically evolve more rapidly than ordered regions, resulting in greater extents of amino acid substitution [30–32]; however, the forces driving such high substitution rates, specifically the role of positive against negative selection, are not well defined. Our analysis showed that the mean ratio of rates of non-synonymous to synonymous substitutions among pairwise comparisons of TS polypeptides from a variety of mammalian species is about 8-fold higher for the IDR as compared with the rest of the polypeptide; however, the ratio is still significantly <1 (Supplementary Table 1). Thus, although the disordered domain is more tolerant of amino acid substitutions than is the remainder of the molecule, it is still subject to negative selection. This conclusion is corroborated by the use of codon-based methods of analysis, which showed that the fraction of negatively selected codons is significantly reduced in the IDR as compared with the body of the polypeptide, and that few, if any, codons within the IDR are positively selected. In all, positive Darwinian selection appears to play little if any role in the high extents of amino acid variation within the IDR of mammalian TS.
Neither the IDR nor hA on its own is capable of promoting degradation (Figure 4). Interestingly, degron activity does not require that the two components be juxtaposed, in that they maintain the ability to co-operate when the IDR is at the N-terminus and hA is at the C-terminus (Figure 7). It is of interest that degron activity is lost when the two elements are in the opposite orientation, i.e. when the IDR is at the C-terminus and hA is at the N-terminus (Figure 7). This may indicate a position-dependency between the two segments when physically separated. However, other interpretations are possible. For example, it may be that when placed at the N-terminus, hA is blocked or modified in some fashion that inhibits its degradative role. The fact that the helical element promotes degradation of the newly identified ‘cryptic’ degron at the C-terminal end of TS (see Figure 8) suggests that it is functional. However, additional studies will be required to delineate whether or not the TS degron components do exhibit position-dependency when separated.
The ‘cryptic’ degron spanning residues 276–313 at the C-terminus of hTS (Figure 8) was predicted from the properties of certain mutant hTS molecules analysed in earlier studies [8,10]. This degron appears to be active only when large deletions or certain amino acid substitutions are introduced into the hTS polypeptide [8,10]. The region has a small β-sheet composed of two strands spanning residues 277–282 and 295–300. The importance of this structural element is unknown. The ‘cryptic’ degron may play a role in regulating hTS degradation under circumstances where the N-terminal degron is either blocked or inactivated. Definition of such circumstances and their physiological implications will require further studies.
It is formally possible that some of the unstable chimaeric proteins produced and examined in the present study have acquired a ubiquitin-dependent mode of degradation, resulting in shortened half-lives. We consider this unlikely, since we find no evidence of high-molecular-mass ubiquitinylated isoforms in any of the Western blots we have analysed. This holds true even when cells were grown in the presence of the proteasome inhibitor MG132, which generally causes accumulation of ubiquitinylated forms.
The requirement for disordered domains (and, in some cases, co-operating helical elements) may be a general theme among ubiquitin-independent substrates. Degradation of mammalian ODC is governed by an intrinsically disordered 37-residue region at the C-terminal end of the polypeptide [13,33]. Yeast ODC contains a disordered 45-residue domain at its N-terminal end that functions as a ubiquitin-independent degron and that is followed by a predicted α-helix [14]. The ubiquitin-independent degron of Rpn4, which is 80 amino acids in length, is located at the molecule's N-terminus and is predicted to consist of a 15–20-residue disordered domain followed by a structured region [20]. Finally, analysis of apomyoglobin degradation in a purified system in vitro has shown that a disordered element in the middle of the polypeptide, in co-operation with helical segments located at the N-terminus, is required for proteolytic breakdown [19]. In all, these studies indicate that ubiquitin-independent degradation requires flexible, disordered regions.
Our finding that TS proteolysis requires both a disordered and an ordered segment is consistent with the so-called ‘two-component’ model of proteasomal function [34,35]. In this model, degradation is viewed as occurring in two phases: proteasome recognition and binding followed by initiation of polypeptide insertion into the proteolytic chamber. For ubiquitin-dependent substrates, the first phase (recognition/binding) is mediated by covalently attached polyubiquitin moieties that recognize one or more subunits of the proteasomal complex [35–37]. For ubiquitin-independent substrates, residues within the target polypeptide chain are utilized [13]. The second phase (initiation) involves unfolding of the proteasome-bound substrate, ‘threading’ of the substrate through the pore that leads to the proteolytic chamber, and entry into the chamber itself, where proteolysis ensues [33,35]. These steps are typically governed by disordered segments that are often (although not always) located at one or the other end of the protein [33,37]. With regard to TS, the helical hA segment may be responsible for proteasome binding, whereas the IDR guides substrate unfolding and chamber entry. Specific amino acids within each domain (e.g. Arg10/Arg11 within the IDR, His39/Arg42 within hA) represent critical residues that mediate the respective functions of the degron sub-elements.
Co-operation between disordered and helical segments, such as in TS, has been observed for the PUP (prokaryotic ubiquitin-like protein) of actinobacteria. Similar to ubiquitin, PUP ligation to proteins targets them for degradation by a proteasome-like complex [38]. Detailed structural studies have shown that upon ligation to a target substrate, PUP forms what is essentially a ‘two-component’ degron in which a helical C-terminal region binds regulatory subunits of the proteasome, whereas a disordered N-terminal region governs initiation of chamber entry [39]. Sequence alignments indicate that the helix-forming region of the PUP degron is more conserved among bacterial species than is the disordered region [40]. Such differential rates of evolution for disordered against the ordered degron sub-regions was also observed for TS in the present study.
Abbreviations
- CHX
cycloheximide
- eGFP
enhanced green fluorescent protein
- FEL
fixed effects likelihood
- GFP
green fluorescent protein; hA, α-helix
- HLH
helix-loop-helix
- hTS
human thymidylate synthase
- IDR
intrinsically disordered region
- ODC
ornithine decarboxylase
- PUP
prokaryotic ubiquitin-like protein
- REL
random effects likelihood
- SNAP
Synonymous Non-synonymous Analysis Program
- TS
thymidylate synthase
AUTHOR CONTRIBUTION
Karen Barbour and Franklin Berger designed the research plan and analysed data; Karen Barbour and Yang-Yang Xing carried out the experiments; Edsel Peña carried out the statistical analysis of the data; and Franklin Berger wrote and prepared the paper with the assistance of Karen Barbour.
FUNDING
This work was supported by the National Institutes of Health [grant number CA44013, RR017698].