Controlled Synthesis of Polyubiquitin Chains

https://doi.org/10.1016/S0076-6879(05)99002-2Get rights and content

Abstract

Many intracellular signaling processes depend on the modification of proteins with polymers of the conserved 76‐residue protein ubiquitin. The ubiquitin units in such polyubiquitin chains are connected by isopeptide bonds between a specific lysine residue of one ubiquitin and the carboxyl group of G76 of the next ubiquitin. Chains linked through K48‐G76 and K63‐G76 bonds are the best characterized, signaling proteasome degradation and nonproteolytic outcomes, respectively. The molecular determinants of polyubiquitin chain recognition are under active investigation; both the chemical structure and the length of the chain can influence signaling outcomes. In this article, we describe the protein reagents necessary to produce K48‐ and K63‐linked polyubiquitin chains and the use of these materials to produce milligram quantities of specific‐length chains for biochemical and biophysical studies. The method involves reactions catalyzed by linkage‐specific conjugating factors, in which proximally and distally blocked monoubiquitins (or chains) are joined to produce a particular chain product in high yield. Individual chains are then deblocked and joined in another round of reaction. Successive rounds of deblocking and synthesis give rise to a chain of the desired length.

Introduction

Ubiquitin modifies a broad spectrum of proteins in eukaryotic cells (Peng et al., 2003). All of these modified proteins share a common structural feature: at least one molecule of ubiquitin is linked through its C‐terminus (the carboxyl group of G76) to an amino group of the target protein. Usually, the linkage site is a lysine side chain of the target protein; less frequently, it is the α‐amino group (Chen 2004, Ciechanover 2004). The covalently linked ubiquitin(s) modulates the stability, location, or function of the target protein. Such regulation can follow directly from protein modification with a single ubiquitin, as in many instances of ubiquitin‐dependent endocytosis and trafficking (Hicke and Dunn, 2003). In other cases, appropriate functional regulation requires the initially conjugated ubiquitin to be extended into a polymer (a polyubiquitin chain) through the repeated use of a lysine residue of ubiquitin as the conjugation site. Abundant evidence indicates that monoubiquitin and polyubiquitin can be functionally distinct signals (Pickart and Fushman, 2004).

Ubiquitin has seven lysine residues. Polyubiquitin chains linked through K48 and K63 have well‐characterized and distinct roles; at least some other polyubiquitin chains probably serve novel signaling functions (Peng 2003, Pickart 2004). Recent appreciation of the structural and functional diversity of polyubiquitin chain signals, in conjunction with the discovery of multiple ubiquitin‐interacting domains, has generated a demand for polyubiquitin chains in quantities necessary for biochemical, structural, and biophysical studies. In this chapter we outline methods for the preparation of milligram amounts of K48‐ and K63‐linked polyubiquitin chains of defined length.

Ubiquitin's best‐characterized function is to direct cellular proteins to the 26S proteasome for degradation. The first studies of this role showed that proteasomal targeting requires target protein modification with a polyubiquitin chain linked through K48‐G76 isopeptide bonds (Chau et al., 1989). Because such chains are the principal proteasomal targeting signal and proteasomal proteolysis is an essential function, the K48R mutation in ubiquitin is lethal in budding yeast (Finley et al., 1994). The extensive scope of this function is further emphasized by the recent finding that ubiquitin itself is the most abundant ubiquitinated protein in budding yeast (Peng et al., 2003). Although this mass spectrometric study used a nonquantitative method, the data suggested that K48 was the predominant linkage (Peng et al., 2003). This conclusion is consistent with the failure of K‐to‐R mutations in ubiquitin, except K48R, to strongly alter the pattern or abundance of ubiquitinated proteins in the same species (Spence et al., 1995).

The Ub‐K63R mutation has no observable effect on proteasome function. Instead, this mutation causes hypersensitivity to DNA damage (Spence et al., 1995). This phenotype reflects the modification of proliferating cell nuclear antigen (PCNA) with a K63‐linked polyubiquitin chain, which in turn promotes a specific mode of DNA lesion bypass (Hoege et al., 2002). K63‐linked chains also act as signals in ribosomal translation, certain endocytosis events, and (in mammals) kinase activation [reviewed in Pickart and Fushman (2004) and Sun and Chen (2004)]. Although the receptors that transduce K63‐chain signals remain to be identified in most cases, it is unlikely that the proteasome is the destination of most substrates modified by K63‐linked chains. A recent study detected all seven ubiquitin–ubiquitin isopeptide linkages in the budding yeast proteome (Peng et al., 2003), and although definitive functional information is lacking for the remaining linkages, some hints are available. Chains containing K29 and K11 linkages have been implicated in the targeting of certain proteins to proteasomes (Baboshina 1996, Johnson 1995). Chains containing K6 and K27 linkages seem more likely be nonproteolytic signals (Morris 2004, Nishikawa 2004, Okumura 2004).

At least two levels of structure are relevant when considering polyubiquitin chains. The first is chemical structure; that is, which of ubiquitin's seven lysines is/are used within the polymer? The significance of chemical structure has been rigorously analyzed for canonical K48‐linked chains. Here, in vitro analyses of the chain ligated to a substrate by its cognate E3 enzyme showed that only K48‐G76 bonds were present (Chau et al., 1989), whereas studies that use preassembled chains of defined structure demonstrated the signaling competence of K48‐linked homopolymers (Thrower et al., 2000). Certain noncanonical chain signals are probably also homopolymers. The factors that catalyze K63‐chain synthesis in DNA damage tolerance and kinase activation produce homopolymers in vitro (Deng 2000, Hofmann 1999), and enzymes that produce K6‐ and K29‐linked homopolymers have been described (Nishikawa 2004, Wu‐Baer 2003, You 2001). Does a single chain ever contain more than one linkage? Proteomic evidence shows that the answer can be yes (Peng et al., 2003), but it is uncertain whether such chains arise purposefully or adventitiously. Very little is known concerning the signaling properties of heteropolymeric chains (Pickart and Fushman, 2004).

The second level of chain structure is conformational. Because ubiquitin is a globular protein, chains with different chemical structures might possess distinctive ubiquitin–ubiquitin interfaces. Solution structural studies indicate that K48‐ and K63‐linked chains indeed have different conformations in solution. At neutral pH, there are extensive ubiquitin–ubiquitin contacts in K48‐linked Ub2 (Varadan et al., 2002), whereas K63‐Ub2 adopts an extended conformation in which the covalent bond is the only significant inter‐subunit contact (Varadan et al., 2004).

Because polyubiquitin chains are assembled through isopeptide (versus peptide) bonds, enzymatic synthesis is necessary. Therefore, one's ability to make a given chain presupposes the availability of a conjugating enzyme(s) that is (1) linkage‐specific and (2) displays robust activity toward free ubiquitin. So far, we have identified such factors for K48‐ and K63‐linked chains. Although a number of other linkage‐specific factors have been reported (see preceding), the activity of most of these factors toward free ubiquitin has not been carefully investigated. Consequently the suitability of these enzymes for the method outlined in the following remains uncertain.

The conjugating enzyme should be highly linkage‐specific. This issue is often addressed using mutant forms of ubiquitin (see Chapter 1). However, mass spectrometry provides the most rigorous criterion of linkage specificity (Peng 2003, Pickart 2004). Application of a semiquantitative version of this method to K48‐ and K63‐linked chains synthesized by the methods described in the following has shown that only the desired linkage is detectable (J. Peng and C. Pickart, unpublished data). In contrast, this criterion was not met when we used a specific E3 enzyme to synthesize K29‐linked chains (M. Wang, J. Peng, and C. Pickart, unpublished data).

For the synthesis of K48‐linked chains, we use the mammalian enzyme E2‐25K (Chen et al., 1991). The biological function of E2‐25K is uncertain, but its activity is well characterized in the in vitro setting. Of particular importance, although E2‐25K binds free ubiquitin weakly, it is very active at the high concentrations of ubiquitin used in chain synthesis (Haldeman et al., 1997). The same is true of the yeast Mms2/Ubc13 complex (a UEV/E2 heterodimer), which participates in K63‐chain synthesis in DNA damage tolerance in vivo (Hofmann and Pickart, 2001). These conjugating factors are conveniently expressed in Escherichia coli.

The past several years have seen the discovery of a small group of protein domains that bind ubiquitin and/or polyubiquitin chains, including the ubiquitin interacting motif (UIM), ubiquitin‐associated domain (UBA), NPL4 zinc finger (NZF), and CUE domain (coupling of ubiquitin to endoplasmic reticulum degradation) [see Buchberger 2002, Hicke 2003, Pickart 2004]. Each domain occurs in a modest number of proteins in a given species. If polyubiquitin chains with different structures represent unique signals, there should be downstream binding factors that can discriminate among different chains. Recent studies identified a UBA domain and a zinc finger domain that preferentially bind K48‐ and K63‐linked chains, respectively (Kanayama 2004, Raasi 2004). The ability to generate large quantities of structurally defined polyubiquitin chains will aid in the discovery of new linkage‐specific binding proteins and should facilitate structural biology aimed at explaining the molecular basis of such linkage specificity. Enabling such studies has been an important motivation for developing the methods described in the following.

The method is outlined in Fig. 1 (Piotrowski et al., 1997). It involves a series of enzymatic reactions catalyzed by the linkage‐specific enzymes discussed previously. We refer to the end of the chain that would normally carry unconjugated G76 as the proximal end, whereas the end that would normally carry unconjugated K48 is the distal end (see Fig. 1). In each round of reaction, proximally and distally blocked monoubiquitins (or chains) are joined to produce a doubly blocked chain. The proximal block consists of an extra C‐terminal residue (D77) that is labile to ubiquitin carboxyl–terminal hydrolases (UCHs). The distal block consists of a cysteine residue, placed at the normal conjugation site, that is later converted to a lysine mimic (S‐aminoethylcysteine) through alkylation. Successive rounds of deblocking and conjugation give rise to a chain of any desired length. The method is presented in detail for K48‐linked chains and in outline form for K63‐linked chains. All of the plasmids mentioned in this chapter are available to academic researchers on request. When protein reagents are commercially available, we mention current suppliers.

Section snippets

Expression and Purification of Proximally and Distally Blocked Ubiquitin Monomers

Ub2 linked through K48 or K63 is synthesized from a matched pair of monomeric ubiquitin reactants. One monomer, which cannot conjugate through its C‐terminus, carries an extra residue (D77). The other monomer cannot conjugate through its lysine, because it carries a mutation to cysteine (or arginine) at this position. The blocked ubiquitins are produced in E. coli. We use pET3a plasmids that specify untagged ubiquitins, which yield 50–100 mg of purified ubiquitin per liter of culture. The

Expression and Purification of Conjugating Enzymes

For K48‐chain synthesis catalyzed by E2‐25K, we use untagged recombinant enzyme purified by conventional chromatography (Haldeman et al., 1997). However, most researchers will find it easier to produce a GST‐tagged version of E2‐25K using pGEX*E225K (Haldeman et al., 1997). In this case, release bound E2‐25K from GSH beads (Sigma) using thrombin. Use 1 U thrombin (Amersham‐Pharmacia Biotech) per 100 μg of fusion protein, incubating for several hours at room temperature. Cleavage is recommended,

Protocol for Synthesis and Purification of K48‐Ub2

The experienced protein chemist can also consult a more concise version of the protocol in the following (Raasi and Pickart, 2005).

  • 1

    Prepare PBDM8 buffer containing 250 mM Tris‐HCl (50% base, pH 8.0), 25 mM MgCl2, 50 mM creatine phosphate (Sigma P7396), 3 U/ml inorganic pyrophosphatase (Sigma I1891), and 3 U/ml creatine phosphokinase (Sigma C3755). Also prepare a neutral 0.1 M ATP solution (Sigma A2383). Both solutions are stable at −20°.

  • 2

    To synthesize K48‐Ub2, combine Ub‐D77 and Ub‐K48C at 7.5

Synthesis of K63‐Linked Chains

The principles and procedures are similar to those outlined for K48‐linked chains, with several differences. First, the synthetic reaction contains 8 μM each of yeast Mms2 and Ubc13 in place of E2‐25K. Second, the buffer is PBDM7.6. (PBDM7.6 is the same as PBDM8, except that (1) Tris, pH 7.6 (24% base), is substituted for Tris, pH 8, and (2) 10 mM ATP is included in PBDM7.6. Accordingly, it is not necessary to add ATP independently to the conjugation reaction). Third, yeast or mammalian E1 can

Factors Contributing to Yield and Recovery

In our original procedures, we removed the conjugating enzymes by subtractive anion exchange before acidifying the reaction mixture (Hofmann 2001, Piotrowski 1997, Raasi 2005). This step is dispensable for chains up to n = 4 but may be necessary to ensure maximum purity of longer chains. In general, good normalization of the molar concentrations of the chain reactants will maximize conversion, simplify purification, and optimize the yield. However, when adding a single ubiquitin to a

Specialized Applications

It is straightforward to modify the preceding protocols for selected purposes. For example, one can make structurally defined heteropolymers. We used the Mms2/Ubc13 complex to conjugate a preassembled K48‐chain to Ub‐K63 in a ubiquitin‐dihydrofolate reductase fusion protein (M. Ajua‐Alemanji and C. Pickart, unpublished data). Here we synthesized the K48‐linked chain from monomers carrying the K63R mutation to avert chain–chain conjugation in the final step. There are other instances in which

Acknowledgments

We thank Matt Steele for providing the data shown in Fig. 2A and members of the Pickart laboratory for comments on the manuscript. Our research on polyubiquitin chains is funded by grants from the NIH.

References (35)

Cited by (0)

View full text