The Molecular Basis of N-end Rule Recognition (original) (raw)

• Journal List
• HHS Author Manuscripts
• PMC3114436

Mol Cell. Author manuscript; available in PMC 2011 Jun 14.

Published in final edited form as:

PMCID: PMC3114436

NIHMSID: NIHMS79167

Kevin H. Wang

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

Giselle Roman-Hernandez

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

Robert A. Grant

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

Robert T. Sauer

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

Tania A. Baker

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

*Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

†Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

*Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA 02139

Summary

The N-end rule targets specific proteins for destruction in prokaryotes and eukaryotes. Here, we report a crystal structure of a bacterial N-end rule adaptor, ClpS, bound to a peptide mimic of an N-end rule substrate. This structure, which was solved at a resolution of 1.15 Å, reveals specific recognition of the peptide α-amino group via hydrogen bonding and shows that the peptide’s N-terminal tyrosine side chain is buried in a deep hydrophobic cleft that preexists on the surface of ClpS. The adaptor side chains that contact the peptide’s N-terminal residue are highly conserved in orthologs and in E3 ubiquitin ligases that mediate eukaryotic N-end rule recognition. We show that mutation of critical ClpS contact residues abrogates substrate delivery to and degradation by the AAA+ protease ClpAP, demonstrate that modification of the hydrophobic pocket results in altered N-end rule specificity, and discuss functional implications for the mechanism of substrate delivery.

Introduction

Targeted proteolysis is essential for regulation of cellular systems, for balancing the composition of the proteome, and for protein-quality control in all cells (reviewed in Gottesman, 1996). In bacteria, archaea, and eukaryotic cells, ATP-dependent AAA+ proteases such as ClpAP, ClpXP, HslUV, Lon, and the proteasome play major roles in these processes (reviewed in Gottesman, 2003; Sauer et al., 2004; Hanson and Whiteheart, 2005; Bukau et al., 2006). Because degradation is an irreversible process, highly specific recognition is required to avoid unwarranted destruction of the wrong proteins. Some protein substrates are recognized directly by AAA+ proteases, whereas other substrates are recognized and delivered to the protease by specialized adaptor proteins (reviewed in Baker and Sauer, 2006). Although outlines of recognition strategies are emerging, there are almost no cases in which the detailed molecular basis of specific recognition by these proteases or adaptors is known.

The N-end rule is a highly conserved mechanism that targets specific proteins to the proteolytic machinery of the cell based on the identity of the substrate’s N-terminal amino acid (Bachmair et al., 1986). In bacteria, for example, aromatic and large hydrophobic residues (Tyr, Trp, Phe, and Leu) are primary N-end rule degradation signals (Tobias et al., 1991). In eukaryotes, these same N-terminal residues also serve as degradation cues, as do Ile, and basic amino acids (Bachmair et al., 1986; Gonda et al., 1989). In addition, certain secondary N-terminal amino acids are recognized by specific amino-acyl transferases, which add a primary residue to the N-terminus and convert the protein into a good N-end-rule substrate (Balzi et al. 1990; Shrader et al. 2003). Importantly, N-end rule substrates are not generally generated by translation and standard N-terminal processing but by endoproteolytic cleavage, making N-end rule degradation especially suited for regulated proteolysis in diverse cellular processes (Mogk et al., 2007).

Although the principle is the same, the pathways used for N-end rule recognition are different in prokaryotes and in eukaryotes. In bacteria, for example, the ClpS adaptor protein recognizes N-end rule substrates, tethers them to the ClpAP protease, and then, in a poorly understood step, actively transfers the substrate to ClpAP for degradation (Dougan et al., 2002; Erbse et al., 2006; Wang et al., 2007; Hou et al., 2008; Wang et al., 2008) (Fig. 1A). N-end rule substrates are initially transferred from ClpS to the AAA+ ClpA hexamer, which harnesses the energy of ATP hydrolysis to denature the substrate and then to translocate it into the proteolytic chamber of the ClpP peptidase for degradation (Thompson et al., 1994; Weber-Ban et al., 1999). The structure of ClpS in complex with the N-terminal domain of ClpA is known (Guo et al., 2002; Zeth et al., 2002), but how N-end rule substrates are recognized and transferred to ClpA has remained elusive. In eukaryotes, by contrast, a specific family of E3 ubiquitin ligases recognizes and, together with an E2 enzyme, covalently modifies N-end rule substrates by polyubiquitin addition, thereby marking them for subsequent energy-dependent degradation by the proteasome (Bartel et al., 1990; Tasaki et al., 2005) (Fig. 1B). Interestingly, ClpS is homologous to a region of these E3 enzymes, suggesting that a common module has evolved to mediate N-end rule recognition in organisms ranging from bacteria to mammals (Lupas and Koretke, 2003).

N-end rule recognition in bacteria and eukaryotes

(A) N-end degron recognition. In bacteria, ClpS recognizes the substrate N-end rule signal (purple star) and directs substrate degradation by the ClpAP protease. (B) In eukaryotes, a specific E3 ligase recognizes this signal and mediates addition of ubiquitin (Ub), which then leads to recognition and degradation by the proteasome.

Here, we determine the structure of the ClpS adaptor bound to an N-end rule substrate mimic. This atomic-resolution structure reveals how binding specificity for the N-terminal residue of the substrate and for the normal set of bacterial N-end rule side chains is established. Model building based on the cocrystal structure also suggests how β-branched hydrophobic side chains are excluded from the wild-type ClpS binding pocket, which facilitated the design of a novel altered-specificity adaptor that efficiently targets substrates bearing N-terminal Ile or Val residues for ClpAP degradation. Finally, the structure suggests a plausible mechanism by which N-end substrates could be transferred from ClpS to ClpA by reprogramming contacts between the adaptor and the -amino group of the substrate.

Results

Cocrystal structure

To determine how N-end rule substrates are recognized, we expressed and purified variants of ClpS from Escherichia coli, Pseudomonas aeruginosa, and Caulobacter crescentus and screened for cocrystals with a decapeptide containing an N-terminal tyrosine, one of the universally recognized N-end-rule residues. The peptide sequence was designed based on a high-affinity N-end rule degradation tag isolated by genetic selection (Wang et al., 2007). Crystals that diffracted to atomic resolution were obtained for a C. crescentus ClpS variant consisting of residues 35-119. Using molecular replacement to obtain phases, we initially solved the structure at 2.0-Å resolution and subsequently collected higher resolution data and extended the structure to 1.15-Å resolution (pdb code 3DNJ; Rwork 13.5%; Rfree 15.9%). There were two ClpS monomers and two N-end-rule peptides in the asymmetric unit of the crystal (space group P21). Each adaptor molecule bound the N-end-rule peptide in the same fashion, but only the first few residues of each peptide were visible in the electron-density map. As expected for a structure at very high resolution, both the quality of the map (Fig. 2A) and the geometry of the final structure (Table 1) were excellent. The folds of C. crescentus ClpS and E. coli ClpS were essentially the same (Fig. 2B), with an r.m.s.d. of 0.5 Å for main-chain atoms. The E. coli ClpS structure was determined without an N-end rule peptide and in complex with the N-terminal domain of ClpA (Guo et al., 2002; Zeth et al., 2002). Thus, neither binding to a peptide substrate nor to the ClpA N-domain appears to change the adaptor’s conformation substantially. As shown in Fig. 2B, the binding surfaces for the N-end-rule peptide and the ClpA N-domain are located on opposite sides of ClpS.

Cocrystal structure and N-end recognition

(A) Electron density, contoured at 1.25, from a 2Fo–Fc map for residues 110 and 111 of ClpS chain A in the cocrystal structure. (B) Alignment of the ClpS portions of the E. coli ClpS•N-domain complex (1R6C) and the Caulobacter ClpS•N-end-peptide complex. (C) Key hydrogen-bond contacts (dotted lines) between ClpS and the N-end rule peptide. Atom colors — oxygen (red); nitrogen (dark blue); carbon (purple for peptide; light blue for ClpS); hydrogen (white). Only a subset of hydrogen atoms is shown. (D) The tyrosine side chain (purple, stick representation) of the N-end rule peptide binds in a deep hydrophobic pocket on ClpS (surface representation —oxygen (red); nitrogen (blue); carbon and hydrogen (white); sulfur (yellow). (E) Cutaway view of hydrophobic specificity pocket. The N-end rule peptide is shown in stick representation. Caulobacter ClpS residues 45, 51, 53, 56, 75, 78, 79, and 112 are shown in surface representation. Atom colors — oxygen (red); nitrogen (dark blue); sulfur (yellow-orange); carbon (purple for peptide; white for ClpS); hydrogen (white). (F) Electrostatic potential of Caulobacter ClpS calculated without bound peptide using GRASP (Nicholls et al., 1991). Negative surface (red); positive surface (blue); neutral surface (white). Molecular graphics were prepared using PYMOL (DeLano, W.L. http://www.pymol.org) or GRASP.

Table 1

Crystallographic data and refinement statistics

Recognition of the α-amino group and N-end side chain

The cocrystal structure reveals a simple mechanism by which ClpS recognizes the first residue of N-end rule substrates. The peptide α-amino group, which is a unique chemical signature of the N-terminal residue, forms hydrogen bonds with the side chains of Caulobacter ClpS residues Asn47 and His79 and with a water molecule that hydrogen bonds to the Asp49 side chain and the carbonyl oxygen of the N-terminal peptide residue (Fig. 2C). In principle, the peptide α-amino group (pKa ~7) could be charged or neutral under physiological conditions. In our structure, the α-amino group is involved as a hydrogen-bond donor in three interactions with good geometries (N-H•••O or N-H•••N distances 1.92 ± 0.03 Å; angles 158 ± 4°). This result is expected if the α-amino group is positively charged, as only the −NH3+ form has the capacity to donate three hydrogen bonds (the −NH2 species could donate two hydrogen bonds). Consistently, we found that binding of an N-end rule peptide to ClpS was much weaker at pH 9, where the α-amino group would be largely deprotonated, than at pH 6.5, and circular-dichroism spectra confirmed that ClpS had similar secondary structures at both pH’s (not shown).

ClpS recognizes N-end-rule substrates with N-terminal Tyr, Phe, Trp, and Leu side chains (Erbse et al., 2006; Wang et al., 2007). In the cocrystal, the N-terminal tyrosine ring of the peptide packs into a deep hydrophobic pocket on the surface of ClpS (Fig. 2D–F). At the bottom of this cavity, the tyrosine hydroxyl group donates a hydrogen bond to the main-chain oxygen of Leu46 (Fig. 2C). This specificity pocket of ClpS—which is formed by atoms from residues Ile45, Leu46, Asn47, Asp48, Thr51, Met53, Val56, Met75, Val78, His79 and Leu112—could easily accommodate the side chains of Phe and Leu and appears to be large enough to allow the binding of a Trp side chain. Hence, we anticipate that ClpS will bind the N-terminal residues of all bacterial N-end rule substrates in a generally similar manner, with the bulky hydrophobic side chain fitting into the specificity pocket of ClpS and hydrogen bonds between ClpS and the α-NH3+ group pinning this side chain in place.

ClpS also makes hydrogen bonds with the amide bond connecting the second and third peptide residues, but specific side-chain contacts were not observed at these peptide positions, and electron density for additional residues was poor or absent. These results are consistent with studies showing that peptide residues past the N-terminus play only modest roles in determining affinity for ClpS, with the main effect being weakening of the binding interaction by acidic residues (Erbse et al., 2006; Wang et al., 2008). Indeed, Glu and Asp side chains at these positions would interact unfavorably with the negative electrostatic potential of the N-end-rule binding site (Fig. 2F), which would help to stabilize binding of the positively charged α-amino group.

Although the molecular contacts between ClpS and the N-end-rule peptide appear to be exquisitely specific, their total number is small. For example, the binding interaction buries only about 600 Å2 of surface area. Nevertheless, the peptide binds ClpS with an affinity of ~400 nM (see below). For comparison, ClpS binding to the N-domain of ClpA buries far more surface area (~1600 Å2) but results in a similar binding constant (~330 nM; Guo et al., 2002; Zeth et al., 2002). Our cocrystal structure provides a reasonable explanation for the relatively tight binding of the N-end-rule peptide to ClpS despite limited contacts. First, the N-end-rule binding site is preformed in free ClpS. For example, the hydrophobic specificity pocket is present in unliganded ClpS and the side chains that will contact the α-amino group of the substrate have the same rotamer conformations as in the cocrystal structure. Second, the peptide can adopt the correct binding geometry simply by choosing one of four possible rotamer conformations for its N-terminal tyrosine side chain and by fixing the dihedral angles of the first two peptide bonds. Thus, the entropic cost of reorganization of the two macromolecules that accompanies binding should be relatively modest.

Contact residues are highly conserved and functionally important

In the PFAM multiple sequence alignment of 432 ClpS orthologs (PF02617), the positions corresponding to Asn47 and His79 in C. crescentus ClpS are ~90% conserved. Moreover, in the remaining sequences Asn47 is always replaced by Asp, and His79 is almost always replaced by Asp or Asn. Thus, the residues at these positions in all ClpS orthologs could accept hydrogen bonds from the α-amino group of the N-end-rule residue. Similarly, the ClpS side chains that form the hydrophobic pocket for the N-end-rule side chain are also highly conserved in orthologs (Fig. 3A). This strong evolutionary conservation of the ClpS side chains that make specific contacts with the N-end rule peptide supports the functional relevance of our Caulobacter cocrystal structure.

Conserved residues in ClpS are functionally important

(A) Sequence homology suggests conserved recognition of N-end rule substrates. The top eight ClpS sequences are bacterial. A plant ortholog (CLPS_ARATH) and two eukaryotic E3 ligases (UBR1_YEAST; UBR2_MOUSE) are also shown. Sequence numbering, secondary structure, and the positions of contacts with the N-end rule peptide are from the C. crescentus cocrystal structure. (B) Caulobacter ClpS binds N-end rule peptides beginning with Tyr, Phe, Trp, and Leu with sub-micromolar affinities. The curve shows ClpS binding to the Tyr peptide, assayed by changes in fluorescence anisotropy. (C) Mutations in E. coli ClpS (ClpSEc) that should alter direct or indirect contacts with the N-end rule α-amino group cause defects in ClpAP degradation of Tyr-GFP. The N34A, D36A, and H66A mutations in ClpSEc correspond to N47A, D49A, and H79A in Caulobacter ClpS. Reactions contained ClpSEc or variants (900 nM), ClpAP (100 nM), Tyr-GFP (100 nM), and ATP (4 mM). (D) Michaelis-Menten plots of Tyr-GFP degradation by ClpAP (100 nM) and ClpSEc variants (900 nM). Representative plots for each variant and average KM and Vmax values are shown.

We also tested the importance of key ClpS residues directly. Because prior studies of the bacterial N-end rule have all been carried out with the E. coli protein, we first established that C. crescentus and E. coli ClpS have similar binding specificities. Indeed, like E. coli ClpS, the Caulobacter adaptor bound tightly to peptides with Tyr, Trp, Phe, or Leu at the N-terminus (Fig. 3B). The KD values measured in these experiments were 150 to 500 nM. For comparison, KD values between 260 nM and 1.5 µM were reported for binding of E. coli ClpS to substrates with an N-terminal Phe (Erbse et al., 2006). By contrast, Caulobacter ClpS bound at least 20-fold more weakly to an otherwise identical peptide with Ile at the N-terminus (not shown). Next, based on the structure, we mutated E. coli ClpS residues predicted to contact the α-amino group of N-end-rule substrates and assayed ClpS-mediated ClpAP degradation. The E. coli N34A mutation (corresponding to N47A in Caulobacter ClpS) eliminated detectable ClpAP degradation of a model N-end rule substrate at the concentration tested (Fig. 3C). Similarly, this mutant variant of ClpS did not detectably bind N-end rule peptides at the highest concentration tested (1.2 µM; not shown). Thus, the contacts altered by this mutation are very important for recognition of N-end-rule substrates.

The E. coli D36A and H66A mutations (corresponding to D49A and H79A in Caulobacter ClpS) also compromised substrate recognition (Fig. 3C & 3D), albeit to lesser degrees than the N34A mutation. The D36A mutation caused a ~2-fold increase in the Michealis constant (KM) for degradation of an N-end-rule substrate, whereas the H66A mutation increased KM about 5-fold and also lowered Vmax substantially (Fig. 3D). We conclude that the structural interactions observed between ClpS and the α-amino group of the “substrate-mimic” N-end-rule peptide play important roles in the recognition and delivery of N-end rule substrates for degradation.

Altered specificity mutation relieves restriction against Ile and Val

Lupus and Koretke (2003) originally reported sequence homology between ClpS and a subset of eukaryotic E3 ligases. Our cocrystal structure establishes that the amino acids that form the ClpS N-end-rule binding site are highly conserved in these ligases (Fig. 3A). Thus, essential features of N-end-rule substrate recognition are also very likely to be preserved. Interestingly, however, at least some of the E3 ligases accept Ile as an N-end-rule residue in addition to Tyr, Phe, Trp, and Leu (Gonda et al., 1989), whereas ClpS only recognizes the latter four side chains.

In our structure, the side chain of the N-terminal peptide tyrosine packs tightly against the side chain of Met53 in Caulobacter ClpS (Fig. 4A). When Ile was modeled at the peptide N-terminus, its β-branched methyl group chain clashed sterically with Met53, suggesting a mechanism for exclusion of Ile and the β-branched Val side chain from the N-end-rule specificity pocket. To explore this possibility, we constructed, purified, and assayed a variant of E. coli ClpS in which the corresponding methionine was replaced by alanine (M40A). The wild-type E. coli adaptor mediated efficient ClpAP degradation of green fluorescent protein with the N-terminal sequence Leu-Leu-Phe-Val-Gln-Glu-Leu (Leu-GFP), but showed little activity toward otherwise identical substrates with Val, Ile, or Thr at the N terminus (Fig. 4B & 4D). By contrast, the M40A mutant delivered Ile-GFP to ClpAP as efficiently as Leu-GFP, and delivered Val-GFP better than either of these substrates (Fig. 4C–E). Importantly, the M40A variant retained the ability to recognize specific features of the N-terminal amino acid. This mutant adaptor efficiently delivered Val-GFP for ClpAP degradation but failed to deliver the isosteric Thr-GFP protein (Fig. 4D–E). Thus, the methionine side chain at position 40 of E. coli ClpS serves as a specificity gatekeeper by excluding β-branched amino acids in N-end rule recognition. Although methionine is the most common residue at this position in ClpS orthologs (72%), alanine is present in a few bacterial adaptors. This observation suggests that the repertoire of hydrophobic N-end-rule residues may be different in some bacteria.

Altered N-end rule degradation specificity

(A) The Met53 side chain of Caulobacter ClpS packs against the N-terminal Tyr1 side chain of the N-end rule peptide. Replacing Tyr1 with Val or Ile would introduce a methyl or ethyl group at the side-chain atom marked by an asterisk, leading to clashes. (B) Wild-type ClpSEc delivered Leu-GFP but not Val-GFP for ClpAP degradation. (C) M40A ClpSEc delivered Val-GFP for ClpAP degradation even better than Leu-GFP. (D) Rates of ClpS-mediated ClpAP degradation of GFP variants with different N-terminal residues were determined from experiments like those in panels A and B and normalized to the Leu-GFP rate for each ClpS variant. Ile-GFP and Val-GFP were delivered efficiently by M40A ClpSEc but not by wild-type ClpSEc. Thr-GFP was delivered poorly by both adaptors. (E) Michaelis-Menten plot of Val-GFP degradation by ClpAP and ClpSEc M40A. KM and Vmax are similar to the values for degradation of Tyr-GFP by ClpAP and wild-type ClpSEc (Fig. 3D). In all panels, the ClpS concentration was 900 nM and the ClpAP concentration was 100 nM. Substrate concentrations in panels A–C were 500 nM.

Our results show that modest changes in the N-degron binding pocket could easily account for differences in recognition of bulky hydrophobic N-end rule residues in prokaryotes and eukaryotes. Arg, Lys, and His also serve as primary N-end rule residues in eukaryotes, but these basic side chains are recognized by a distinct binding site in the E3 enzyme (Gonda et al., 1989; Xia et al., 2008). In a recent study, Xia et al. (2008) identified mutations in the yeast UBR1 E3 ligase that prevent recognition of hydrophobic N-end rule residues. Three of these loss-of-function mutations mapped to the ClpS-homology region of UBR1 and altered residues corresponding to residues 43, 48, and 51 in Caulobacter ClpS, which are in or near the N-end-rule binding pocket.

Discussion

Structural implications for the mechanism of substrate delivery

Our cocrystal structure establishes the molecular basis of N-end-rule recognition by ClpS. Prior structural studies have shown how ClpS binds to the N-terminal domain of ClpA (Guo et al., 2002; Zeth et al., 2002). Although these interactions are necessary for efficient ClpAP degradation of N-end-rule substrates, biochemical and genetic studies reveal that they are not sufficient. For example, deletion of unstructured residues at the N-terminus of ClpS does not interfere with its binding to N-end-rule substrates or to ClpA, but these deletions prevent ClpAP degradation of the tethered substrate (Hou et al., 2008). Moreover, by itself, ClpA has weak affinity for N-end-rule substrates, suggesting that adaptor-mediated substrate delivery involves transfer of the substrate from its initial binding site in ClpS to a site in ClpA (Wang et al., 2007).

Modeling studies suggest that ClpS binds near the outer periphery of the hexameric ring of ClpA (Guo et al., 2002; Zeth et al., 2002), which would place the N-end-rule substrate far from the central pore of the hexamer where substrate engagement must ultimately occur (Hinnerwisch et al., 2005). However, the N-terminal domains of ClpA, which bind ClpS, appear to be highly mobile, and it has been suggested that this flexibility may allow these domains to relay substrates to an interaction site in the AAA+ ring (Ishikawa et al., 2004). Because simple ClpS-mediated tethering of N-end-rule substrates to ClpAP is not sufficient to ensure degradation (Hou et al., 2008), it seems likely that an N-domain•ClpS•substrate complex must eventually move close enough to the pore to allow efficient transfer of the substrate from the ClpS adaptor to the central pore of ClpA (Fig. 5). Proximity, however, would not ensure transfer if the substrate remained tightly bound to ClpS (Fig. 5B). We propose that residues in or near the ClpA pore facilitate hand-off by interacting with ClpS residues close to the binding pocket. For example, such interactions could reposition one of the ClpS side chains that contacts the -amino group of the substrate (Fig. 5C). This interaction, in turn, would destabilize adaptor-substrate binding in a manner analogous to the N34A and H66A mutations in E. coli ClpS, allowing substrate release and capture by the binding site in ClpA (Fig. 5D), and eventual initiation of degradation by translocation of the substrate through the pore of ClpA and into ClpP (Fig. 5E).

Model for facilitated transfer from ClpS to ClpA

(A) Initial binding of a complex of ClpS (light blue) and the N-domain of ClpA (dark brown) to an N-end-rule substrate (magenta). Hydrogen bonds between ClpS and the α-amino group of the substrate pin the side chain of the N-terminal residue (Phe) into the hydrophobic pocket. The N-domain is flexibly linked to the AAA+ domains of ClpA (salmon) in this state. (B) The N-domain•ClpS•substrate complex moves closer to the central pore of the ClpA hexamer, in a process involving the N-terminal residues of ClpS. (C) A ClpS residue that formerly pinned the α-amino group of the substrate now interacts with a ClpA side chain instead, weakening binding of the N-end-rule side chain to ClpS. (D) The N-terminal Phe of the substrate is transferred from the hydrophobic pocket of ClpS to a binding site near the ClpA pore. (E) The N-domain•ClpS complex moves back to a mobile position, allowing ClpA to denature the substrate by translocation through the pore and to initiate degradation.

The facilitated transfer model depicted in Fig. 5 needs to be tested directly. However, several prior observations are consistent with this general idea. For example, Zeth et al. (2002) found that ClpS dissociated more slowly from ClpA when ATPS was present. Because nucleotide binds to the AAA+ domains of ClpA, it would not directly alter the interaction of the ClpA N-domain with ClpS. Nucleotide binding could, however, create a second binding site for ClpS near the central pore of the hexamer, as required by the facilitated transfer model. Moreover, Erbse et al. (2006) found that the Y37A and E41A mutations in E. coli ClpS decrease the efficiency of substrate delivery to ClpAP. They speculated that these residues formed part of the binding site for the N-end residue. However, the cocrystal structure shows that the corresponding side chains in Caulobacter ClpS do not contact the N-end-rule peptide directly. It is possible, therefore, that these adaptor residues, which are close to the substrate binding site, are involved in ClpS binding to a site near the pore of ClpA as part of the transfer process. We anticipate that the cocrystal structure will aid in designing experiments to probe the mechanism of downstream transfer of N-end-rule substrate from ClpS to ClpAP.

Exclusion of potential substrates

The cocrystal structure provides simple explanations for N-end-rule recognition, but also raises an important question. Why, for example, doesn’t methionine serve as an N-end-rule residue? This is a critical biological issue, as most bacterial proteins and many eukaryotic proteins start with methionine. Modeling using the cocrystal structure shows that a methionine side chain fits nicely into the specificity pocket of ClpS. Because the side chain of methionine is somewhat less hydrophobic than that of leucine and can adopt more rotamer conformations, substrates with N-terminal methionines would be expected to bind less well than substrates with N-terminal leucines (Lipscomb et al., 1998). Indeed in protein folding studies using T4 lysozyme, Gassner et al. (1999) found that Leu→Met substitutions in different parts of the hydrophobic core decreased stability by 0.4 to 2.0 kcal/mol. However, even destabilization by 2 kcal/mol would only decrease the ClpS affinity for a substrate with an N-terminal Met by about 30-fold compared to the same substrate with an N-terminal Leu. Moreover, because roughly 70% of all bacterial proteins have an N-terminal methionine (Frottin et al., 2006), these potential substrates would greatly outnumber authentic N-end-rule substrates and higher concentration would offset reduced affinity. There is, however, no evidence that proteins starting with methionine are targeted for N-end rule degradation in vivo (Bachmair et al., 1986; Wang et al., 2007). We conclude that ClpS affinity for proteins with N-terminal methionines is either much lower than expected based on theoretical considerations (Lipscomb et al., 1998) and modeling based on the cocrystal structure, or that another mechanism prevents the wasteful degradation that would otherwise occur.

In addition to exceptionally weak binding, multiple mechanisms might prevent ClpS-mediated degradation of proteins with N-terminal methionines. First, the N-terminal methionine could be inaccessible because it is part of the native protein structure. Second, some proteins with good N-end-rule residues are not degraded by ClpS•ClpAP if the N-terminal residue is too close to the body of the folded protein molecule (Erbse et al., 2006; Wang et al., 2008). In this instance, ClpS still binds the protein efficiently but engagement by ClpAP does not occur. A third mechanism involves a second step of substrate discrimination. For example, the ClpS•substrate complex might need to have sufficient kinetic stability to allow transfer to the pore of ClpA. This type of kinetic “editing” could also explain our observation that the H66A mutation in E. coli ClpS, which weakens apparent binding to N-end-rule substrates by removing a contact with the α-amino group, also decreases the rate of ClpAP degradation at saturating substrate concentrations about 4-fold. It is also possible that the secondary N-end-rule binding site in or near the ClpA pore discriminates against methionine, providing a second step of thermodynamic selection.

ClpS recognition of additional substrates

N-end rule proteins are the best-characterized substrates for ClpS mediated degradation by ClpAP, but additional substrates are also likely to be degraded by this adaptor-dependent system. For example, only an aggregated form but not the soluble form of malate dehydrogenase is degraded in a ClpS-dependent fashion by ClpAP (Dougan et al., 2002). Moreover, in peptide-array experiments, ClpS binds relatively strongly to some sequences without N-terminal Tyr, Phe, Trp, or Leu residues (Erbse et al., 2006). It seems likely, therefore, that the site that ClpS uses to bind N-end-rule substrates will also be utilized to bind other substrates, for example ones with internal Tyr, Phe, Trp, or Leu side chains. Indeed, starting from the cocrystal structure, we were able to add extra residues to the N-terminus of the peptide with good geometry and without steric clashes. Thus, the structure of the binding site would allow binding to peptides with internal Tyr, Phe, Trp, or Leu side chains. In this case, however, there would be no appropriately positioned α-amino group to maximize hydrogen bonding with ClpS, which would remove potential stabilizing interactions. Indeed, it is known that acetylation of the α-amino group of a good N-end-rule peptide diminishes but does not eliminate binding (Wang et al., 2008). Nevertheless, favorable interactions between residues upstream of an internal Tyr, Phe, Trp, or Leu side chain and ClpS could easily restore this lost binding energy. It will be important to test directly for internal recognition of substrates by ClpS and to determine how sequence context influences such recognition.

Perspective

The N-end rule for protein degradation can be stated simply and concisely. Nevertheless, the cellular machinery that executes and influences this degradation is complex and is only beginning to be understood in molecular detail. The cocrystal structure of ClpS in complex with an N-end rule peptide provides the first atomic-level view of a critical step in which an N-end-rule substrate is initially targeted for degradation by binding to a specific adaptor protein and suggests potential mechanisms for downstream recognition of the N-end degron that leads to eventual degradation. Homology between ClpS and the E3 ubiquitin ligases suggests that many of the detailed features of N-degron recognition are conserved in organisms ranging from bacteria to mammals.

Experimental Procedures

Proteins and peptides

Residues 20-119 of C. crescentus ClpS were fused to the C-terminus of His6-SUMO-Tyr-Gly-Arg- using a pET23b vector (Novagen). Residues 1–19 were not included in this construct because the corresponding residues in E. coli ClpS are largely unstructured (Guo et al., 2002; Zeth et al., 2002). Following fusion-protein purification by Ni++-NTA chromatography (Qiagen), cleavage with SUMO protease resulted in an insoluble ClpS fragment. However, cleavage with thrombin (Novagen) produced a mixture of soluble products. After dialysis into 10 mM HEPES (pH 7.5), 80 mM KCl, 1 mM DTT, and 5% glycerol, His6-SUMO and small peptides were removed by chromatography on Ni++-NTA and Mono-Q (GE Healthsciences) columns. The purified ClpS fragment was shown to consist of residues 35–119 by mass spectrometry. ClpS35–119 was exchanged into a final buffer of 10 mM HEPES (pH 7.5), 200 mM KCl, and 1 mM DTT by gel filtration on Superdex-75 (GE Healthsciences) and concentrated to 10 mg/ml using Millipore spin filter columns with a molecular weight cutoff of 5 kDa. Full-length variants of E. coli ClpS were constructed by PCR mutagenesis and purified as described (Dougan et al., 2002). E. coli ClpA, E. coli ClpP-His6, and GFP substrates with the N-terminal sequence Xxx-Leu-Phe-Val-Gln-Glu-Leu (where Xxx is a variable position) were purified as described (Maurizi et al., 1990; Kim et al., 2000; Wang et al., 2008). Peptides were synthesized by standard FMOC techniques using an Apex 396 solid-phase synthesizer.

Crystallography

Crystals in space group P21 were obtained after 1 week at 20 °C in hanging drops containing 2 µl of a solution containing C. crescentus ClpS35–119 (8 mg/ml) and the peptide Tyr-Leu-Phe-Val-Gln-Arg-Asp-Ser-Lys-Glu (2 mM) in 10 mM HEPES (pH 7.5), 200 mM KCl, and 1 mM DTT mixed with 1 µl of reservoir solution (0.1 M bis-Tris (pH 5.5), 0.2 M MgCl2, and 19% PEG 3350). Crystals were frozen without additional cryoprotection. X-ray diffraction data to 2.0-Å resolution were collected on a Rigaku MicroMax007-HF rotating anode source equipped with Varimax-HR mirrors, an RAXIS-IV detector, and an Oxford cryo-system, and were processed using HKL-2000 (Otwinowski and Minor, 1997). Initial phases were obtained by molecular replacement using E. coli ClpS (1R60 chain C) as a search model in PHASER (Storoni et al., 2004). Iterative model building using COOT (Emsley and Cowtan, 2004) and refinement using PHENIX (Adams et al., 2002) resulted in a structure with good refinement statistics (Rwork 17.7%; Rfree 22.7%). A data set that extended to 1.15-Å resolution was subsequently collected at the NE-CAT 24-ID-C beamline at the Argonne National Labs Advanced Photon Source. These data were processed and the structure was refined as described above. The final structure (Rwork 13.5%; Rfree 15.9%) was refined with macromolecular hydrogen atoms, with variable occupancy of water molecules, and with anisotropic b-factors for all non-hydrogen and non-water atoms. Changes in accessible surface area were calculated with the ccp4 program AREAIMOL (Collaborative Computational Project, Number 4, 1994; Lee and Richards, 1971).

Biochemical assays

ClpAP degradation assays were performed and analyzed as described (Wang et al., 2008). Briefly, ClpA6, (ClpP-His6)14, and ClpS were pre-incubated in reaction buffer (50 mM HEPES (pH 7.5), 300 mM NaCl, 20 mM MgCl2, 0.5 mM DTT, 10% glycerol) with GFP substrate for 2 min at 30 °C before adding ATP regeneration mix (4 mM ATP, 50 mg/ml creatine kinase, 5 mM creatine phosphate) to initiate the degradation reaction (Weber-Ban et al., 1999; Wang et al., 2007). For ClpS binding assays, synthetic peptides (Xxx-Leu-Phe-Val-Gln-Tyr-His6-Cys) were labeled by modification with fluorescein maleimide (Thermo Scientific) (Wang et al., 2008). Binding was assayed by changes in fluorescence anisotropy using a Photon Technology International instrument. Experimental data were fit to a hyperbolic binding isotherm or to the Michaelis-Menten equation to obtain KD, KM, and Vmax values. Cited KD values were averages (± SD) of three independent experiments; KM and Vmax values were averages of two experiments.

Acknowledgements

We thank S. Bissonnette, S. Brohawn, P. Chien, J. Davis, S. Glynn, J. Hou, M. Laub, I. Levchenko, J. Partridge, T. Schwartz, and E. Spear for help and advice. Supported by HHMI and NIH grant AI-16892. K.H.W. was supported by a National Science Foundation graduate fellowship. T.A.B. is an employee of the Howard Hughes Medical Institute. Studies at the NE-CAT beamlines of the Advanced Photon Source were supported by NIH-NCRR award RR-15301 and by the DOE Office of Basic Energy Sciences under contract DE-AC02-06CH11357.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

• Adams PD, Grosse-Kunstleve RW, Hung LW, Ioerger TR, McCoy AJ, Moriarty NW, Read RJ, Sacchettini JC, Sauter NK, Terwilliger TC. PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr D Biol Crystallogr. 2002;58:1948–1954. [PubMed] [Google Scholar]
• Bachmair A, Finley D, Varshavsky A. In vivo half-life of a protein is a function of its amino-terminal residue. Science. 1986;234:179–186. [PubMed] [Google Scholar]
• Baker TA, Sauer RT. ATP-dependent proteases of bacteria: recognition logic and operating principles. Trends Biochem Sci. 2006;31:647–653. [PMC free article] [PubMed] [Google Scholar]
• Balzi E, Choder M, Chen WN, Varshavsky A, Goffeau A. Cloning and function analysis of the arginyl-tRNA-protein transferase gene ATE1 of Saccharomyces cerevisiae. J Biol Chem. 1990;13:7464–7471. [PubMed] [Google Scholar]
• Bartel B, Wunning I, Varshavsky A. The recognition component of the N-end rule pathway. EMBO J. 1990;9:3179–3189. [PMC free article] [PubMed] [Google Scholar]
• Bukau B, Weissman J, Horwich A. Molecular chaperones and protein quality control. Cell. 2006;125:443–451. [PubMed] [Google Scholar]
• COLLABORATIVE COMPUTATIONAL PROJECT, NUMBER 4. `The CCP4 Suite: Programs for Protein Crystallography'. Acta Cryst. 1994;D50:760–763. [PubMed] [Google Scholar]
• Davis IW, Leaver-Fay A, Chen VB, Block JN, Kapral GJ, Wang X, Murray LW, Arendall WB, Snoeyink J, Richardson JS, Richardson DC. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 2007;35:W375–W383. [PMC free article] [PubMed] [Google Scholar]
• Dougan DA, Reid BG, Horwich AL, Bukau B. ClpS, a substrate modulator of the ClpAP machine. Mol Cell. 2002;9:673–683. [PubMed] [Google Scholar]
• Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60:2126–2132. [PubMed] [Google Scholar]
• Erbse A, Schmidt R, Bornemann T, Schneider-Mergener J, Mogk A, Zahn R, Dougan DA, Bukau B. ClpS is an essential component of the N-end rule pathway in Escherichia coli. Nature. 2006;439:753–756. [PubMed] [Google Scholar]
• Frottin F, Martinez A, Peynot P, Mitra S, Holz RC, Giglione C, Meinnel T. The proteomics of N-terminal methionine cleavage. Mol Cell Proteomics. 2006;5:2336–2349. [PubMed] [Google Scholar]
• Gassner NC, Baase WA, Lindstrom JD, Lu J, Dahlquist FW, Matthews BW. Methionine and alanine substitutions show that the formation of wild-type-like structure in the carboxy-terminal domain of T4 lysozyme is a rate-limiting step in folding. Biochemistry. 1999;38:14451–14460. [PubMed] [Google Scholar]
• Gonda DK, Bachmair A, Wunning I, Tobias JW, Lane WS, Varshavsky A. Universality and structure of the N-end rule. J Biol Chem. 1989;264:16700–16712. [PubMed] [Google Scholar]
• Gottesman S. Proteases and their targets in Escherichia coli. Annu. Rev. Genet. 1996;30:465–506. [PubMed] [Google Scholar]
• Gottesman S. Proteolysis in bacterial regulatory circuits. Annu Rev Cell Dev Biol. 2003;19:565–587. [PubMed] [Google Scholar]
• Guo F, Maurizi MR, Esser L, Xia D. Crystal structure of ClpA, an Hsp100 chaperone and regulator of ClpAP protease. J Biol Chem. 2002;277:46743–46752. [PubMed] [Google Scholar]
• Hanson PI, Whiteheart SW. AAA+ proteins: have engine, will work. Nat Rev Mol Cell Biol. 2005;6:519–529. [PubMed] [Google Scholar]
• Hinnerwisch J, Fenton WA, Furtak KJ, Farr GW, Horwich AL. Loops in the central channel of ClpA chaperone mediate protein binding, unfolding, and translocation. Cell. 2005;121:1029–1041. [PubMed] [Google Scholar]
• Hou JY, Sauer RT, Baker TA. Distinct structural elements of the adaptor ClpS are required for regulating degradation by ClpAP. Nat Struct Mol Biol. 2008;15:288–294. [PubMed] [Google Scholar]
• Ishikawa T, Maurizi MR, Steven AC. The N-terminal substrate-binding domain of ClpA unfoldase is highly mobile and extends axially from the distal surface of ClpAP protease. J Struct Biol. 2004;146:180–188. [PubMed] [Google Scholar]
• Kim YI, Burton RE, Burton BM, Sauer RT, Baker TA. Dynamics of substrate denaturation and translocation by the ClpXP degradation machine. Mol Cell. 2000;5:639–648. [PubMed] [Google Scholar]
• Lee B, Richards FM. The interpretation of protein structures: estimation of static accessibility. J Mol Biol. 1971;55:379–400. [PubMed] [Google Scholar]
• Lipscomb LA, Gassner NC, Snow SD, Eldridge AM, Baase WA, Drew DL, Matthews BW. Context-dependent protein stabilization by methionine-to-leucine substitution shown in T4 lysozyme. Protein Sci. 1998;7:765–773. [PMC free article] [PubMed] [Google Scholar]
• Lupas AN, Koretke KK. Bioinformatic analysis of ClpS, a protein module involved in prokaryotic and eukaryotic protein degradation. J Struct Biol. 2003;141:77–83. [PubMed] [Google Scholar]
• Maurizi MR, Clark WP, Kim SH, Gottesman S. Clp P represents a unique family of serine proteases. J Biol Chem. 1990;265:12546–12552. [PubMed] [Google Scholar]
• Mogk A, Schmidt R, Bukau B. The N-end rule pathway for regulated proteolysis: prokaryotic and eukaryotic strategies. Trends Cell Biol. 2007;17:165–172. [PubMed] [Google Scholar]
• Nicholls A, Sharp KA, Honig B. Protein Folding and Association: Insights From the Interfacial and Thermodynamic Properties of Hydrocarbons. Proteins Stuc Func and Genet. 1991;11:281–296. [PubMed] [Google Scholar]
• Otwinowski Z, Minor W. Processing of X-ray Diffraction Data Collected in Oscillation Mode. In: Carter CW Jr, Sweet RM, editors. Methods in Enzymology, Volume 276: Macromolecular Crystallography, part A. New York, USA: Academic Press; 1997. pp. 307–326. [PubMed] [Google Scholar]
• Sauer RT, et al. Sculpting the proteome with AAA(+) proteases and disassembly machines. Cell. 2004;119:9–18. [PMC free article] [PubMed] [Google Scholar]
• Shrader TE, Tobias JW, Varshavsky A. The N-end rule in Escherichia coli: cloning and analysis of the leucyl, phenylalanyl-tRNA-protein transferase gene aat. J Bacteriol. 1993;175:4364–4374. [PMC free article] [PubMed] [Google Scholar]
• Storoni LC, McCoy AJ, Read RJ. Likelihood-enhanced fast rotation functions. Acta Crystallogr D Biol Crystallogr. 2004;60:432–438. [PubMed] [Google Scholar]
• Tasaki T, Mulder LC, Iwamatsu A, Lee MJ, Davydov IV, Varshavsky A, Muesing M, Kwon YT. A family of mammalian E3 ubiquitin ligases that contain the UBR box motif and recognize N-degrons. Mol Cell Biol. 2005;25:7120–7136. [PMC free article] [PubMed] [Google Scholar]
• Thompson MW, Singh SK, Maurizi MR. Processive degradation of proteins by the ATP-dependent Clp protease from Escherichia coli. Requirement for the multiple array of active sites in ClpP but not ATP hydrolysis. J Biol Chem. 1994;269:18209–18215. [PubMed] [Google Scholar]
• Tobias JW, Shrader TE, Rocap G, Varshavsky A. The N-end rule in bacteria. Science. 1991;254:1374–1377. [PubMed] [Google Scholar]
• Wang KH, Sauer RT, Baker TA. ClpS modulates but is not essential for bacterial N-end rule degradation. Genes Dev. 2007;21:403–408. [PMC free article] [PubMed] [Google Scholar]
• Wang KH, Oakes ESC, Sauer RT, Baker TA. Tuning the strength of a bacterial N-end rule degradation signal. J Biol Chem. 2008 (June 11; Epub ahead of print). [PMC free article] [PubMed] [Google Scholar]
• Weber-Ban EU, Reid BG, Miranker AD, Horwich AL. Global unfolding of a substrate protein by the Hsp100 chaperone ClpA. Nature. 1999;401:90–93. [PubMed] [Google Scholar]
• Xia Z, Webster A, Du F, Piatkov K, Ghislain M, Varshavsky A. Substrate-binding sites of UBR1, the ubiquitin ligase of the N-end rule pathway. J Biol Chem. 2008 (Epub ahead of print). [PMC free article] [PubMed] [Google Scholar]
• Zeth K, Ravelli RB, Paal K, Cusack S, Bukau B, Dougan DA. Structural analysis of the adaptor protein ClpS in complex with the N-terminal domain of ClpA. Nat Struct Biol. 2002;9:906–911. [PubMed] [Google Scholar]