doi:10.1016/j.jmb.2007.03.008 J. Mol. Biol. (2007) 369, 168–176

Defining how Ubiquitin Receptors hHR23a andS5a Bind Polyubiquitin

Yang Kang1,2†, Xiang Chen1†, Jeffrey W. Lary3, James L. Cole3

and Kylie J. Walters1⁎

1Department of Biochemistry,Molecular Biology andBiophysics, University ofMinnesota, Minneapolis,MN 55455, USA2Department of Oral Sciences,University of Minnesota,Minneapolis, MN 55455, USA3Analytical UltracentrifugationFacility, Biotechnology-Bioservices Center,University of Connecticut,Storrs, CT 06269-3149, USA

† Y.K. and X.C. contributed equalAbbreviations used: hHR23, hum

Rad23; HSQC, heteronuclear singleUBA, ubiquitin-associated; UBL, ububiquitin interacting motif; GST,glutathione-S-transferase; VWA, vonE-mail address of the correspondi

walte048@umn.edu

0022-2836/$ - see front matter © 2007 E

Ubiquitin receptors connect substrate ubiquitylation to proteasomaldegradation. HHR23a binds proteasome subunit 5a (S5a) through a surfacethat also binds ubiquitin. We report that UIM2 of S5a binds preferentially tohHR23a over polyubiquitin, andweprovide amodel for the ternary complexthat we expect represents one of the mechanisms used by the proteasome tocapture ubiquitylated substrates. Furthermore, we demonstrate thathHR23a is surprisingly adept at sequestering the ubiquitin moieties of apolyubiquitin chain, and provide evidence that it and the ubiquitylatedsubstrate are committed to each other after binding.

© 2007 Elsevier Ltd. All rights reserved.

Keywords: ubiquitin receptors; proteasome subunit S5a; hHR23a; Rad23;proteasomal degradation

*Corresponding author

Introduction

Ubiquitin signaling regulates an astounding arrayof cellular events and remains essential throughoutthe life-cycle of a cell. In its most established role,ubiquitylation targets proteins for degradation bythe 26 S proteasome,1 a process important forcontrolling the lifespan of regulatory proteins,removing misfolded proteins,2 producing immuno-competent peptides,3 activating and repressing tran-scription,4,5 and regulating cell-cycle progression.6Ubiquitin-mediated protein degradation beginswithan enzymatic cascade that culminates in the attach-ment of polyubiquitin to a protein substrate.Substrates conjugated with chains linked by K487

orK638 of ubiquitin are degraded by the proteasome.Little is known of the pathway(s) that connectsubiquitylation to proteasomal degradation; how-

ly to this work.an homologue ofquantum coherence;iquitin-like; UIM,

Willebrand A.ng author:

lsevier Ltd. All rights reserve

ever, ubiquitin receptors undoubtedly have key rolesin this process. Two proteasomal ubiquitin receptorshave been identified including S5a9 and S6′.10 S5acontains two functional elements, an N-terminal 188amino acid residue von Willebrand A (VWA)domain and two ubiquitin-interacting motifs(UIMs) that bind ubiquitin (Figure 1).11 The VWAdomain is conserved in Rpn10, the yeast homologueof S5a, which is truncated after UIM1. Therefore, thisN-terminal domain is expected to mediate S5a/Rpn10 proteasome association.In addition to binding ubiquitin, the UIMs of

S5a bind members of the UBL/UBA ubiquitinreceptor family, which exist largely free of theproteasome.12,13 The UBL/UBA family of proteinsare defined by ubiquitin-associated (UBA) domainsthat bind ubiquitin,14–16 and ubiquitin-like (UBL)domains that bind the proteasome.12,13,17,18 Theyhave attracted much attention for their ability toregulate the lifespans of other proteins. The threeUBL/UBA family members in Saccharomyces cerevi-siae, Rad23 (hHR23a/b in humans), Dsk2 (hPLIC-1/2 in humans) and Ddi1, recruit ubiquitylatedsubstrates to the proteasome for degradation.19–24

Importantly, mice lacking the two Rad23 homolo-gues (mHR23a/b) die as embryos,25 suggesting thatit has an essential role in delivering one or moreubiquitylated substrates to the proteasome in mam-

d.

Figure 1. Domain architecture of ubiquitin receptors hHR23a and S5a. Regions that bind proteasome components orubiquitin are shaded in yellow or orange, respectively. The effect of the point mutations L198A and L355A are illustratedby the loss of the corresponding UBA domain. Each of these mutations destroys UBA domain structural integrity as wellas its capacity to bind (poly)ubiquitin.

169Defining how hHR23a and S5a Bind Polyubiquitin

mals. Experiments in yeast demonstrated that Rad23provides an Rpn10-independent pathway to theproteasome.20,23

In previous studies, we determined the structureof hHR23a to reveal that its UBL domain interactsdynamically with each of its UBA domains (Figure1).26 These interactions are abrogated in the pre-sence of ubiquitin,16 the preferred UBA domainbinding partner, or S5a,26 the preferred UBL domainbinding partner. Indeed, the binding constant forhHR23b binding to UIM2of S5a is 3.4 μM,27 andstructures of the hHR23a/UIM228 or hHR23b/UIM227 complexes demonstrate strong hydrophobicand polar contacts between these two proteins. S5aand hHR23a/b each have two ubiquitin bindingregions that are connected by flexible linkers, andthis shared attribute may contribute to their pre-ference for longer polyubiquitin chains.16

Very little is known of how polyubiquitin behavesin an environment of multiple receptors, yet suchknowledge is essential to determine how substratesare delivered to the proteasome. The structure ofK48-linked tetraubiquitin revealed that its ubiquitinsubunits pack against each other;29 however, K48-linked tetraubiquitin binds multiple Rad23 mole-cules,30 indicating that it adopts an opened con-formations when bound to this ubiquitin receptor.Here, we demonstrate that hHR23a is surprisingly

efficient at sequestering the ubiquitin moieties of achain such that the binding of additional receptors isstrongly disfavored, even for octaubiquitin. HHR23achanges its own conformation26 to bind S5a,12 butthrough a surface on S5a that also binds ubiquitin.S5a contains two UIMs (Figure 1), the second ofwhich (UIM2) exhibits a fivefold stronger bindingaffinity for monoubiquitin31 and binds hHR23a.12Intriguingly, we reveal that S5a binds hHR23a in thepresence of tetraubiquitin and we use NMR spectro-scopy to define the resulting ternary complex. Ourfindings yield a mechanistic model for how theproteasome captures its substrates.

Results

HHR23a UBA domains sequester polyubiquitin

By using a previously established assay,30 wedetermined that the UBA domains of hHR23a are

highly efficient at sequestering the ubiquitin moi-eties of K48-linked and K63-linked tetraubiquitin.Purified glutathione-S-transferase (GST)-taggedhHR23a bound to glutathione S-Sepharose resinwas exposed to K48-linked (Figure 2(a)) or K63-lin-ked (Figure 2(b)) chains followed by untaggedhHR23a protein (see Materials and Methods).Surprisingly, K48 and K63-linked tetraubiquitinsupported the binding of only one hHR23a mole-cule. This result was observed even with hHR23avariants containing only one intact UBA domain(Figure 2(a) and (b)); namely, those with the L355Aor L198A mutation incorporated (Figure 1).15,16

Furthermore, hHR23a binding to tetraubiquitinremained stoichiometric even when either proteinwas present at large molar excess, as measured byanalytical ultracentrifugation. In particular, themaximum value in the plot of sw versus tetraubiqui-tin:hHR23a molar ratio is close to 1:1 stoichiometry(Figure 3(a)). Furthermore, our data match closelywith a theoretical plot for 1:1 binding stoichiometry(Figure 3(a), red). The experimental data are slightlyabove the theoretical line at higher tetraubiquitin:hHR23a molar ratios due to the presence of smallamounts of aggregates under those conditions. Forthis analysis, we used a previously determinedsedimentation coefficient for hHR23a of 2.22 S (datanot shown). The sedimentation coefficient for tetra-ubiquitin was obtained from the DcDt+ analysis ofsamples containing tetraubiquitin:hHR23a at molarratios where tetraubiquitin was in excess, with aresulting value of 2.5 S. This value was supported byc(s) analysis (data not shown). Finally, c(s) analysisof each mixture revealed the presence of two peaks,a slower one in the region expected for the compo-nent in excess, and a faster one in the regionexpected for the 1:1 complex (∼3.8 S). A representa-tive c(s) profile is provided in Figure 3(b) for thesample with a tetraubiquitin:hHR23a molar ratio of0.64:1. The relative peak areas for this sample were18% for the peak at 2.26 S and 82% for the peak at 3.7S. These values are quite close to the expected valuesof 23% and 77% for tetraubiquitin and hHR23aforming a 1:1 complex. The SEDPHAT analysis ofthe data for the two samples containing tetraubi-quitin:hHR23a at molar ratios nearest to 1:1 yieldeda molecular mass of 76.2(±1.6) kDa and a correctedsedimentation coefficient, s20,w, of 3.8(±0.07) S. Thisresult matches closely with the sequence molecular

Figure 2. The UBA domains of HHR23a sequester theubiquitin subunits of polyubiquitin. Western blot analysisdemonstrates that two hHR23a molecules cannot bind acommon (a) K48-linked or (b) K63-linked tetraubiquitin.Glutathione S-Sepharose resin (20 μl) pre-incubated with0.1 nmol of GST-hHR23a wild-type, L355A, or L198Awasmixed with 0.03 nmol of (a) K48-linked or (b) K63-linkedtetraubiquitin without (−) or with (+) 0.1 nmol ofuntagged hHR23a wild-type (black), L355A (red), andL198A (blue), washed extensively and then probed withanti-hHR23a (top panel) or anti-ubiquitin (bottom panel).No ternary complex involving either (a) K48-linked or (b)K63-linked tetraubiquitin was observed, and greater than90% of the untagged protein was recovered in the flow-through (FT). As positive controls, Ni-NTA resin (20 μl)pre-incubated with 0.1 nmol of His6-tagged Rad23 wasmixed with 0.03 nmol of (a) K48-linked or (b) K63-linkedtetraubiquitin and 0.1 nmol of GST-tagged or untaggedhHR23a. (c) Octaubiquitin exhibits a strong preferencefor binding only one hHR23a molecule. 20 μl ofglutathione S-Sepharose resin pre-incubated with0.1 nmol of GST-hHR23a was mixed with 0.03 nmol ofK48-linked octaubiquitin and 0.1 nmol of untaggedhHR23a. The flow-through (FT) was loaded into lane 2and the resin washed extensively and loaded into lane 1.Western blot analysis was performed with anti-hHR23a(top panel) or anti-ubiquitin (bottom panel) to revealuntagged hHR23a largely in the flow-through rather thanretained in a ternary complex with GST-hHR23a andoctaubiquitin.

170 Defining how hHR23a and S5a Bind Polyubiquitin

mass for the 1:1 molar complex of hHR23a andtetraubiquitin, which is 74.3 kDa.We tested whether two hHR23a proteins bind

octaubiquitin and found that its binding to a second

hHR23a molecule is unexpectedly weak (Figure2(c)). In particular, only small quantities of untaggedhHR23a are retained in a ternary complex with GST-hHR23a and octaubiquitin (Figure 2(c), upper panellane 1). Altogether, these data suggest that thehHR23a UBA domains are highly efficient atsequestering the ubiquitin moieties of tetraubiquitin,and that two tetraubiquitins do not bind hHR23asimultaneously, suggesting that hHR23a commits toa single ubiquitylated substrate.The data of Figure 2(a) and (b) suggest that either

UBA domain of hHR23a can support binding topolyubiquitin, as the L198A and L355Avariants bindtetraubiquitin. It is worth noting that tetraubiquitinis not retained on GST-bound resin,30 and that theobserved interaction is therefore not a result of non-specific binding to the resin. To further validate thisresult, we acquired [1H,15N] HSQC spectra on 15N-labeled hHR23a alone and in the presence of twofoldmolar excess of K48-linked tetraubiquitin. The effectof adding tetraubiquitin to hHR23a is summarized inFigure 3(c), which was derived as described inMaterials and Methods. As expected, each of theUBA domains of hHR23a exhibited significantchemical shift perturbations in the presence oftetraubiquitin (Figure 3(c)). The UBL domain is alsoaffected, as UBL/UBA domain interactions aredisturbed. This result was observed also when theUBA domains of hHR23a bound monoubiquitin.16

Altogether, these data provide strong evidence thateach UBA domain of hHR23a binds K48-linkedtetraubiquitin.Earlier, we demonstrated that tetraubiquitin can

bind simultaneously to Rad23 and hHR23a.30 Thisexperiment is used as a positive control in Figure2(a) and (b). In these experiments, tetraubiquitin isincubated with his-tagged Rad23 bound to Ni-NTAagarose resin. After washing the resin extensively,GST-tagged or untagged hHR23a is added, and theresin is again washed extensively. In this experi-ment, each of the hHR23a species is retained on theresin only when His-tagged Rad23 is added.30 Thedata suggest that Rad23 uses a different mechanismto bind tetraubiquitin or changes the tetraubiquitinbinding mode of hHR23a. Indeed, UBA2 of Rad23differs significantly from the UBA domains ofhHR23a, as it mediates Rad23 dimerization.30,32

Further experiments are required to determinewhether hHR23a or Rad23 exhibit preference fordifferent ubiquitin subunits in a polyubiquitinchain.

S5a binds preferentially to hHR23a overtetraubiquitin

We used NMR spectroscopy to test whether it ismechanistically plausible for hHR23a to deliverpolyubiquitin to S5a. The hHR23a/S5a interactionoccurs via UIM2 of S5a,12 which also bindsubiquitin.11 Surprisingly, the hHR23a/S5a interac-tion was found to be unaffected by tetraubiquitin. Inparticular, the hHR23a UBL domain (Figure 4(a))and S5a UIM2 (Figure 4(b)) bind each other in an

Figure 3. The UBA domains of HHR23a are highly efficient at binding tetraubiquitin. (a) The weight averagesedimentation coefficient as a function of tetraubiquitin to hHR23a molar ratio reveals 1:1 binding stoichiometry.Experimentally determined values obtained by using sedimentation velocity analysis (black) and a theoretical plot(Materials andMethods) for 1:1 binding stoichiometry (red) are included. (b) Sedimentation velocity results for the samplecontaining a 0.64:1 molar ratio of tetraubiquitin:hHR23a. The c(s) distribution was obtained using SEDFIT, with the slowerpeak corresponding to excess hHR23a and the faster peak to the tetraubiquitin:hHR23a 1:1 complex. (c) Chemical shiftperturbation data analyzed according to equation reveals that each of the UBA domains of hHR23a bind K48-linkedtetraubiquitin. For this analysis, the amide chemical shift values of hHR23a alone were compared to its values in thepresence of a twofold molar excess of K48-linked tetraubiquitin.

171Defining how hHR23a and S5a Bind Polyubiquitin

identical manner whether tetraubiquitin is presentor not. This phenomenon is demonstrated in Figure4(a), which contains an expanded region of [1H,15N]HSQC spectra of 15N-labeled hHR23a by itself(black), mixed with equimolar quantities of S5a(196–306) (blue), or with equimolar quantities of S5a(196–306) and tetraubiquitin (red). HHR23a UBLdomain resonances exhibit identical chemical shiftperturbations due to the addition of S5a in thepresence of tetraubiquitin.Conversely, Figure 4(b) shows an expanded

region of [1H,15N] HSQC spectra acquired on15N-labeled S5a by itself (black), mixed withequimolar quantities of hHR23a (blue), or withequimolar quantities of hHR23a and tetraubiquitin(red). The comparisons reveal that the S5a UIM2resonances exhibit identical chemical shift changesupon addition of hHR23a when tetraubiquitin ispresent. Furthermore, when the mixture containingequimolar quantities of the three proteins issubjected to size-exclusion chromatography, thethree proteins co-elute with no higher or lowermolecular mass species (Figure 5). Altogether,

these results provide strong evidence that the S5aUIM2 binds preferentially to the hHR23a UBLdomain over tetraubiquitin.

HHR23a UBA2 and S5a UIM1 bindtetraubiquitin

Our data indicate that in the presence of S5a,hHR23a UBA2 residues bind tetraubiquitin, whereasthose of UBA1 do not (Figure 4(a)). In particular,resonances derived from UBA2 residues exhibitsevere broadening (such as E334 or E348) or chemicalshift perturbations (such as S335 or T317) in thesamples containing tetraubiquitin and S5a. In con-trast, UBA1 resonances exhibit only minor perturba-tions, as demonstrated in Figure 4(a) for R177, V181,and L184.S5a UIM1 binds ubiquitin11 but not hHR23a.12

Our NMR experiments reveal that as UIM2 bindshHR23a UBL domain, UIM1 residues bind tetra-ubiquitin (Figure 4(b), right panels). In particular,resonances originating from UIM1 residues exhibitsevere resonance broadening in the samples

172 Defining how hHR23a and S5a Bind Polyubiquitin

containing hHR23a and tetraubiquitin. No effect isobserved when tetraubiquitin is absent.

Discussion

Altogether, our results indicate that the hHR23aUBA2 domain binds tetraubiquitin as its UBLdomain binds S5a UIM2, which facilitates UIM1binding to the polyubiquitin. These findings lead toa model of the ternary complex that provides insightinto how ubiquitylated substrates are captured bythe proteasome (Figure 6). In particular, UIM1 isproximal to the S5a 188 residue VWA domain,which is expected to mediate its proteasome asso-ciation. Therefore, UIM1 is expected to be deeperwithin the proteasome and its interaction withpolyubiquitin, facilitated by the UIM2/UBL domain

interaction, offers a mechanism for moving sub-strates into the proteasome.We demonstrate that hHR23a binding to tetra-

ubiquitin is stoichiometric over a wide range ofmolar ratios, which include either protein being atgreater than fivefold molar excess over the other.There are two mechanisms by which hHR23a couldblock or dramatically reduce the binding of anothermolecule to tetraubiquitin or octaubiquitin, respec-tively. In particular, a preferred binding orientationcould exist whereby binding to certain subunits,such as the terminal subunits, is not desirable, ormultiple ubiquitin subunits could bind the UBAdomains simultaneously. These two possibilities aresupported by the significantly weaker bindingobserved to monoubiquitin33 and by the structureof the hHR23a UBA2 domain complexed withdiubiquitin.34 In this structure, the UBA domain issandwiched between the two ubiquitin subunits.The tendency of HHR23a to bind only one poly-ubiquitin chain and its exclusion of additionalproteins suggests that it becomes committed to asingle ubiquitylated substrate. Furthermore, thisproperty seems to have evolved in eukaryotic familymembers, as Rad23 (yeast ortholog of hHR23a/b)binds to Rad23-bound30 or hHR23a-bound (Figure2(a) and (b)) tetraubiquitin.S5a was identified for its role as a proteasome-

bound ubiquitin receptor, and later found to bindhHR23a12 and other ubiquitin receptors.13 The

Figure 4. Characterizing the hHR23a/S5a/tetraubi-quitin complex by NMR. (a) HHR23a binds S5a andtetraubiquitin simultaneously. An expanded region of[1H,15N]-HSQC spectra of 15N-labeled hHR23a alone(black), with equimolar quantities of S5a (196–306) (blue),or with equimolar amounts of S5a (196–306) and K48-linked tetraubiquitin (red) illustrates that hHR23a bind-ing to S5a is unaffected by tetraubiquitin. The resonancesof E334 and E348 (labeled in blue) are broadenedseverely and those of S335 and T317 shifted bytetraubiquitin, whereas those of R177, V181, and L184(labeled in red) are unaffected. The UBL, UBA1 andUBA2 domains of hHR23a span residues M1-A81, T159-T200, Q319-E363, respectively. For spectra acquired onhHR23a, S5a and tetraubiquitin (red), equimolar quantityof K48-linked tetraubiquitin was added to 15N-labeledhHR23a. Unlabeled S5a was added to this mixture andspectra recorded at 1:1:0 (not shown), 1:1:0.5 (notshown), 1:1:1 (red), and 1:1:2 (not shown) molar ratioof hHR23a: tetraubiquitin: S5a (196–306). The experi-ments were recorded on 0.32 mM 15N-labeled hHR23a in20 mM sodium phosphate (pH 6.5), 30 mM NaCl, 0.1%NaN3. (b) The UIM2 of S5a binds preferentially to theUBL domain of hHR23a. Expanded regions of [1H,15N]-HSQC spectra of 15N-labeled S5a (196–306) alone (black),with equimolar quantities of hHR23a (blue), or withequimolar amounts of hHR23a and K48-linked tetra-ubiquitin (red) indicates that UIM2 of S5a binds hHR23a(left panel) and that its UIM1 is attenuated due totetraubiquitin binding (right panels). The experimentswere performed with 0.25 mM 15N-labeled S5a (196–306)in 20 mM sodium phosphate (pH 6.5), 30 mM NaCl,0.1% (w/v) NaN3.

Figure 5. An expanded region of a chromatogram andaccompanying gel demonstrate that hHR23a, S5a andtetraubiquitin co-elute after size-exclusion chromatogra-phy. Equimolar quantities of hHR23a, S5a (196–306) andK48-linked tetraubiquitin were incubated for 1 h at 4 °C,and passed through an FPLC system equipped with aHiload 16/60 Superdex 200 preparation grade column(Amersham Pharmacia Biotech). Only one peak wasobserved, which was confirmed to contain all threecomponents by polyacrylamide gel electrophoresis andstaining with Coomassie brilliant blue. The retentionvolume of the fractions containing hHR23a, S5a andtetraubiquitin spanned 59.5–65.5 ml, which is reasonablefor the 87 kDa complex as albumin (67 KDa) elutes at∼75 ml. Lanes labeled I and M contain an aliquot of themixture before FPLC injection and a molecular massmarker, respectively.

173Defining how hHR23a and S5a Bind Polyubiquitin

amino acid residues used to bind ubiquitin, how-ever, overlap with those that bind hHR23a,11,12 andtherefore it was not known whether this interactionoccurs in the presence of a ubiquitylated substrate.Here, we demonstrate that the hHR23a UBA2domain binds tetraubiquitin as its UBL domainbinds the S5a UIM2. Within this complex, the S5aUIM1 binds polyubiquitin and, when S5a isassembled into the proteasome, this interactioncould have an important role in capturing substrates(Figure 6). In summary, we expect that the modelpresented in Figure 6 represents one mechanism bywhich ubiquitylated substrates are captured by theproteasome.

Materials and Methods

Sample preparation

Tagged and untagged versions of hHR23a wild-typeprotein as well as variants with the L198A or L355Amutations incorporated were expressed and purified fromEscherichia coli as described.16,26 K48-linked and K63-linked tetraubiquitin was purchased (Boston BiochemInc.) or synthesized using a previously publishedprotocol,35 which was also used to produce K48-linkedoctaubiquitin. S5a (196–306) was expressed and purifiedas described.31

Western blot analysis

The experiments to determine whether two hHR23aproteins bind a common polyubiquitin chain wereperformed as described.30 Purified GST-tagged hHR23a(0.1 nmol) or its L198A, L355A mutants were bound to20 μl of pre-washed glutathione S-Sepharose resin. Eachresin was allowed to mix at 4 °C overnight with K48-linked or K63-linked tetraubiquitin (Boston Biochem Inc.),or octaubiquitin, and then washed extensively with bufferA (20 mM sodium phosphate (pH 6.5), 50 mM NaCl, 0.5%(v/v) Triton X-100). The resin was next incubated withuntagged hHR23a wild-type or mutated protein for 1 h at4 °C. Each resin was pelleted by centrifugation and thenwashed extensively with buffer A. As a control, 0.1 nmolof purified His-tagged Rad23 was bound to 20 μl of pre-washed Ni-NTA resin (Qiagen), and mixed with K48-linked (Figure 2(a)) or K63-linked (Figure 2(b)) tetraubi-quitin and GST-tagged hHR23a or untagged hHR23aunder the same conditions, and washed extensively withbuffer B (50 mM Tris–HCl (pH 8.0), 100 mMNaCl, 20 mMimidazole, 10% (v/v) glycerol). In all cases, proteins thatwere retained on the resin were fractionated by electro-phoresis, whereas proteins in the supernatant wereprecipitated by 10% (w/v) trichloroacetic acid, resolvedby SDS-PAGE, transferred to a PVDF membrane, andprobed with anti-hHR23a antibody (Abcam) and anti-ubiquitin antibody (Invitrogen). Visualization was per-formed using anti-rabbit or anti-mouse horseradishperoxidase and ECL.

Sedimentation velocity analysis

Stock samples of hHR23a and K48-linked tetraubiquitinwere prepared at concentrations of 21.8 mg/ml and20.7 mg/ml, respectively, in 20 mM sodium phosphate(pH 6.5), 30 mM NaCl. Seven concentration ratios of theproteins were prepared, ranging from ∼12:1 to ∼1:6 at aconstant total protein concentration of ∼30 μM. Thedilutions were then subjected to sedimentation velocityexperiments at 20 °C and 50,000 rpm (An-60 Ti analyticalrotor) using a Beckman-Coulter XLI analytical ultracen-trifuge. Interference scans were acquired at 1 min intervalsfor 4.5 h. A g(s) analysis of the data for each mixing ratiowas performed using the program DcDt+.36,37 The weightaverage sedimentation coefficient, sw, for each mixturewas calculated by integration of the g(s) profile over therange of sedimentation coefficients covered by theanalysis. The plot of sw versus the molar ratio of the twocomponents will have a maximum value at the correctstoichiometric ratio of the complex. Data were alsoanalyzed by the c(s) method using SEDFIT.38

Figure 6. Proposed model for how hHR23a and S5a bind to tetraubiquitin. HHR23a recruits the chain via itsUBA2 domain, as its UBL domain binds proteasome component S5a, and UIM1 of S5a is able to receive thechain.

174 Defining how hHR23a and S5a Bind Polyubiquitin

Theoretical weight average sedimentation coefficient(sw) values were calculated by using equation (1), in whichthe c(i) and s(i) are the weight concentrations andsedimentation coefficients of each species, respectively:

sw ¼ ½ðcðAÞsðAÞÞ þ ðcðBÞsðBÞÞþ ðcðABÞsðABÞÞ�=½ðcðAÞ þ cðBÞ þ cðABÞÞ� ð1Þ

The data for the two samples with molar ratios closest to1:1 were also analyzed using the program SEDPHAT39

with a model that allows characterization of the pre-dominant species present in solution.

NMR spectroscopy

All NMR samples were dissolved in 20 mMNaPO4 (pH6.5), 30 mM NaCl, 0.1% (w/v) NaN3, and 10% 2H2O.Spectra were acquired at 25 °C on Varian NMR spectro-meters operating at 800 MHz with a cryogenically cooledprobe. Processing was performed in NMRPipe40 and theresulting spectra werevisualized in XEASY.41 Proteinconcentrations were calculated by using extinction coeffi-cients based on amino acid composition and absorbance at280 nm for protein dissolved in 6M guanidine-HCl.Chemical shift perturbation (CSP) data for hHR23a

binding to tetraubiquitin were obtained for each aminoacid residue by comparing the amide chemical shift valuesof hHR23a alone with those of hHR23a in the presence oftwofold molar excess tetraubiquitin. Values were calcu-lated according to:

CSP ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi0:2Dy2N þ Dy2H

qð2Þ

In this equation,ΔδN andΔδH represent the changes in theamide nitrogen and proton chemical shifts (in parts permillion), respectively.

Acknowledgements

Ultracentrifugation was performed at the Univer-sity of Connecticut's National Analytical Ultracen-trifugation Facility in Storrs, CT (James L. Cole,Director). We are grateful to Cecile Pickart for theE2-25k construct as well as to Hiroshi Matsuo andDeanna Koepp for helpful discussions. NMR datawere acquired in the UMN NMR facility (NSF BIR-961477), spectra were processed and interpreted inthe MSI BSCL, and the work was supported bythe National Institutes of Health CA097004 (toK.J.W.).

References

1. Ciechanover, A. (1994). The ubiquitin-proteasomeproteolytic pathway. Cell, 79, 13–21.

2. Schubert, U., Anton, L. C., Gibbs, J., Norbury, C. C.,Yewdell, J. W. & Bennink, J. R. (2000). Rapiddegradation of a large fraction of newly synthesizedproteins by proteasomes. Nature, 404, 770–774.

3. Rock, K. L. & Goldberg, A. L. (1999). Degradation of

175Defining how hHR23a and S5a Bind Polyubiquitin

cell proteins and the generation of MHC classI-presented peptides. Annu. Rev. Immunol. 17, 739–779.

4. Conaway, R. C., Brower, C. S. & Conaway, J. W. (2002).Emerging roles of ubiquitin in transcription regula-tion. Science, 296, 1254–1258.

5. Muratani, M. & Tansey, W. P. (2003). How theubiquitin-proteasome system controls transcription.Nature Rev. Mol. Cell Biol. 4, 192–201.

6. Yamaguchi, R. & Dutta, A. (2000). Proteasomeinhibitors alter the orderly progression of DNAsynthesis during S-phase in HeLa cells and lead torereplication of DNA. Expt. Cell Res. 261, 271–283.

7. Pickart, C. M. & Eddins, M. J. (2004). Ubiquitin:structures, functions, mechanisms. Biochim. Biophys.Acta, 1695, 55–72.

8. Hofmann, R. M. & Pickart, C. M. (2001). In vitroassembly and recognition of Lys-63 polyubiquitinchains. J. Biol. Chem. 276, 27936–27943.

9. Deveraux, Q., Ustrell, V., Pickart, C. & Rechsteiner, M.(1994). A 26 S protease subunit that binds ubiquitinconjugates. J. Biol. Chem. 269, 7059–7061.

10. Lam, Y. A., Lawson, T. G., Velayutham, M., Zweier,J. L. & Pickart, C. M. (2002). A proteasomal ATPasesubunit recognizes the polyubiquitin degradationsignal. Nature, 416, 763–767.

11. Young, P., Deveraux, Q., Beal, R. E., Pickart, C. M. &Rechsteiner, M. (1998). Characterization of two poly-ubiquitin binding sites in the 26 S protease subunit 5a.J. Biol. Chem. 273, 5461–5467.

12. Hiyama, H., Yokoi, M., Masutani, C., Sugasawa, K.,Maekawa, T., Tanaka, K. et al. (1999). Interaction ofhHR23 with S5a. The ubiquitin-like domain of hHR23mediates interaction with S5a subunit of 26 S protea-some. J. Biol. Chem. 274, 28019–28025.

13. Walters, K. J., Kleijnen, M. F., Goh, A. M., Wagner,G. & Howley, P. M. (2002). Structural studies ofthe interaction between ubiquitin family proteinsand proteasome subunit S5a. Biochemistry, 41,1767–1777.

14. Wilkinson, C. R., Seeger, M., Hartmann-Petersen, R.,Stone, M., Wallace, M., Semple, C. & Gordon, C.(2001). Proteins containing the UBA domain are ableto bind to multi-ubiquitin chains. Nauret Cell Biol. 3,939–943.

15. Bertolaet, B. L., Clarke, D. J., Wolff, M., Watson, M. H.,Henze, M., Divita, G. & Reed, S. I. (2001). UBAdomains of DNA damage-inducible proteins interactwith ubiquitin. Nature Struct. Biol. 8, 417–422.

16. Wang, Q., Goh, A. M., Howley, P. M. & Walters, K. J.(2003). Ubiquitin recognition by the DNA repairprotein hHR23a. Biochemistry, 42, 13529–13535.

17. Elsasser, S., Gali, R. R., Schwickart, M., Larsen, C. N.,Leggett, D. S., Muller, B. et al. (2002). Proteasomesubunit Rpn1 binds ubiquitin-like protein domains.Nature Cell Biol. 4, 725–730.

18. Saeki, Y., Sone, T., Toh-e, A. & Yokosawa, H. (2002).Identification of ubiquitin-like protein-binding sub-units of the 26S proteasome. Biochem. Biophys. Res.Commun. 296, 813–819.

19. Chen, L. & Madura, K. (2002). Rad23 promotes thetargeting of proteolytic substrates to the proteasome.Mol. Cell Biol. 22, 4902–4913.

20. Elsasser, S., Chandler-Militello, D., Muller, B., Hanna,J. & Finley, D. (2004). Rad23 and Rpn10 serve asalternative ubiquitin receptors for the proteasome.J. Biol. Chem. 279, 26817–26822.

21. Kleijnen, M. F., Shih, A. H., Zhou, P., Kumar, S.,Soccio, R. E., Kedersha, N. L. et al. (2000). The hPLICproteins may provide a link between the ubiquitina-

tion machinery and the proteasome. Mol. Cell, 6,409–419.

22. Saeki, Y., Saitoh, A., Toh-e, A. & Yokosawa, H. (2002).Ubiquitin-like proteins and Rpn10 play cooperativeroles in ubiquitin-dependent proteolysis. Biochem.Biophys. Res. Commun. 293, 986–992.

23. Verma, R., Oania, R., Graumann, J. & Deshaies, R. J.(2004). Multiubiquitin chain receptors define a layer ofsubstrate selectivity in the ubiquitin-proteasome sys-tem. Cell, 118, 99–110.

24. Kaplun, L., Tzirkin, R., Bakhrat, A., Shabek, N.,Ivantsiv, Y. & Raveh, D. (2005). The DNA damage-inducible UbL-UbA protein Ddi1 participates inMec1-mediated degradation of Ho endonuclease.Mol. Cell Biol. 25, 5355–5362.

25. Ng, J. M., Vrieling, H., Sugasawa, K., Ooms, M. P.,Grootegoed, J. A., Vreeburg, J. T. et al. (2002).Developmental defects and male sterility in micelacking the ubiquitin- like DNA repair gene mHR23B.Mol. Cell Biol. 22, 1233–1245.

26. Walters, K. J., Lech, P. J., Goh, A. M., Wang, Q. &Howley, P. M. (2003). DNA-repair protein hHR23aalters its protein structure upon binding proteaso-mal subunit S5a. Proc. Natl Acad. Sci. USA, 100,12694–12699.

27. Fujiwara, K., Tenno, T., Sugasawa, K., Jee, J. G.,Ohki, I., Kojima, C. et al. (2004). Structure of theubiquitin-interacting motif of S5a bound to theubiquitin-like domain of HR23B. J. Biol. Chem. 279,4760–4767.

28. Mueller, T. D. & Feigon, J. (2003). Structural determi-nants for the binding of ubiquitin-like domains to theproteasome. EMBO J. 22, 4634–4645.

29. Cook, W. J., Jeffrey, L. C., Kasperek, E. & Pickart, C. M.(1994). Structure of tetraubiquitin shows how multi-ubiquitin chains can be formed. J. Mol. Biol. 236,601–609.

30. Kang, Y., Vossler, R. A., Diaz-Martinez, L. A., Winter,N. S., Clarke, D. J. & Walters, K. J. (2006). UBL/UBAubiquitin receptor proteins bind a common tetraubi-quitin chain. J. Mol. Biol. 356, 1027–1035.

31. Wang, Q., Young, P. & Walters, K. J. (2005).Structure of S5a bound to monoubiquitin providesa model for polyubiquitin recognition. J. Mol. Biol.348, 727–739.

32. Bertolaet, B. L., Clarke, D. J., Wolff, M., Watson, M. H.,Henze, M., Divita, G. & Reed, S. I. (2001). UBAdomainsmediate protein-protein interactions betweentwo DNA damage-inducible proteins. J. Mol. Biol. 313,955–963.

33. Raasi, S., Orlov, I., Fleming, K. G. & Pickart, C. M.(2004). Binding of polyubiquitin chains to ubiquitin-associated (UBA) domains of HHR23A. J. Mol. Biol.341, 1367–1379.

34. Varadan, R., Assfalg, M., Raasi, S., Pickart, C. &Fushman, D. (2005). Structural determinants forselective recognition of a Lys48-linked polyubiquitinchain by a UBA domain. Mol. Cell, 18, 687–698.

35. Raasi, S. & Pickart, C. M. (2005). Ubiquitin chainsynthesis. Methods Mol. Biol. 301, 47–55.

36. Philo, J. S. (2006). Improved methods for fitting sedi-mentation coefficient distributions derived by time-derivative techniques. Anal. Biochem. 354, 238–246.

37. Philo, J. S. (2000). Amethod for directly fitting the timederivative of sedimentation velocity data and analternative algorithm for calculating sedimentationcoefficient distribution functions. Anal. Biochem. 279,151–163.

38. Schuck, P. (2000). Size-distribution analysis of macro-

176 Defining how hHR23a and S5a Bind Polyubiquitin

molecules by sedimentation velocity ultracentrifuga-tion and lamm equation modeling. Biophys. J. 78,1606–1619.

39. Schuck, P. (2003). On the analysis of protein self-association by sedimentation velocity analytical ultra-centrifugation. Anal. Biochem. 320, 104–124.

40. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G.,

Pfeifer, J. & Bax, A. (1995). NMRPipe: a multidimen-sional spectral processing system based on UNIXpipes. J. Biomol. NMR, 6, 277–293.

41. Bartels, C., Xia, T.-H., Billeter, M., Güntert, P. &Wüthrich, K. (1995). The program XEASY for compu-ter-supported NMR spectral analysis of biologicalmacromolecules. J. Biomol. NMR, 6, 1–10.

Edited by P. Wright

(Received 5 January 2007; received in revised form 10 February 2007; accepted 5 March 2007)Available online 12 March 2007