Chemistry & Biology
`
`Article
`
`The Differential Modulation of USP Activity
`by Internal Regulatory Domains, Interactors
`and Eight Ubiquitin Chain Types
`
`Alex C. Faesen,1,3 Mark P.A. Luna-Vargas,1,3 Paul P. Geurink,2 Marcello Clerici,1 Remco Merkx,2 Willem J. van Dijk,1
`Dharjath S. Hameed,2 Farid El Oualid,2 Huib Ovaa,2 and Titia K. Sixma1,*
`1Division of Biochemistry and Center for Biomedical Genetics
`2Division of Cell Biology
`The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands
`3These authors contributed equally to this work
`*Correspondence: t.sixma@nki.nl
`DOI 10.1016/j.chembiol.2011.10.017
`
`SUMMARY
`
`Ubiquitin-specific proteases (USPs) are papain-like
`isopeptidases with variable inter- and intramolecular
`regulatory domains. To understand the effect of
`these domains on USP activity, we have analyzed
`the enzyme kinetics of 12 USPs in the presence and
`absence of modulators using synthetic reagents.
`This revealed variations of several orders of magni-
`tude in both the catalytic turnover (kcat) and ubiquitin
`(Ub) binding (KM) between USPs. Further activity
`modulation by intramolecular domains affects both
`the kcat and KM, whereas the intermolecular activa-
`tors UAF1 and GMPS mainly increase the kcat.
`Also, we provide the first comprehensive analysis
`comparing Ub chain preference. USPs can hydrolyze
`all linkages and show modest Ub-chain preferences,
`although some show a lack of activity toward linear
`di-Ub. This comprehensive kinetic analysis high-
`lights the variability within the USP family.
`
`INTRODUCTION
`
`Since the 1980s, posttranslational modification of proteins by Ub
`has been the focus of many studies due to the important role of
`Ub in cellular processes (Hochstrasser, 2009; Pickart, 2004).
`Ubiquitination can mediate a multitude of signals due to its ability
`to form chains. It does so by using one of the seven lysine resi-
`dues (K6, K11, K27, K29, K33, K48, and K63) or the N-terminal
`amine (‘‘linear’’), with potentially a different signal
`for each
`linkage. To counteract the effects of ubiquitination, the differen-
`tial removal of Ub (chains) is carried out by deubiquitinating
`enzymes (DUBs).
`The human genome encodes nearly 100 putative DUBs
`belonging to at least five subfamilies of isopeptidases (Nijman
`et al., 2005). The ubiquitin-specific protease (USP) family is the
`largest class of DUBs, with more than 60 members (Komander
`et al., 2009a; Nijman et al., 2005). USPs are cysteine proteases
`that use a papain-like mechanism to hydrolyze the isopeptide
`bond between the carboxy terminus of Ub and the ε-amine of
`the target lysine.
`
`USPs are variable in both size and modular domain architec-
`ture, and these domains can include substrate-binding domains,
`ubiquitin-like (UBL) domains, and other protein-protein interac-
`tion domains (Nijman et al., 2005; Zhu et al., 2007) (Figure 1A).
`They share a common papain-like fold, but the catalytic domains
`can have large insertions (Ye et al., 2009), possibly directly
`affecting activity, Ub binding, or localization as seen in USP4
`(Luna-Vargas et al., 2011b), USP5 (Reyes-Turcu et al., 2008),
`USP14 (Borodovsky et al., 2001), and CYLD (Komander et al.,
`2008). In addition, some USPs need structural rearrangements
`to bind their substrate and catalyze hydrolysis (Avvakumov
`et al., 2006; Hu et al., 2002; Hu et al., 2005; Ko¨ hler et al., 2010;
`Samara et al., 2010).
`USPs are often found in large protein complexes, and many
`interaction partners of USPs have been identified (Sowa et al.,
`2009). Although the function of most interaction partners is still
`unclear, some play a role in the modulation of USP activity. For
`example, GMP synthetase (GMPS)
`interacts and activates
`USP7 (Faesen et al., 2011; Sarkari et al., 2009; van der Knaap
`et al., 2005), whereas the WD40 repeat containing UAF1
`(WDR48) activates USP1, USP12, and USP46 (Cohn et al.,
`2007, 2009).
`With its diversity of domain architectures, internal insertions
`within the catalytic domain, and external modulators, the USP
`family apparently requires different levels of regulation. This
`poses a number of unanswered questions. For instance, what is
`the variability of the activity between the catalytic domains and
`the full-length proteins? Are there preferences for Ub-chain types,
`and does this change in the presence of external modulators?
`To address these questions, we have developed and
`produced (El Oualid et al., 2010) chemical tools and used them
`to characterize a set of 12 USPs. This revealed variations of
`several orders of magnitude in catalytic turnover and Ub binding
`and allowed characterization of intra- and intermolecular activity
`modulation. We determined the chain preferences of all USPs
`against all eight topoisomers. This showed modest chain spec-
`ificity among the di-ubiquitin linkages that was variable between
`USPs. We observe activity toward all topoisomers, except for
`some USPs that are inactive toward linear di-ubiquitin. These
`preferences did not change in the presence of the modulators.
`Kinetic analysis of the hydrolysis showed that there is no addi-
`tional Ub binding site, suggesting that the chain preferences
`are achieved by steric hindrance or reduced catalytic turnover.
`
`1550 Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 1
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 1. Overview of the Characterized
`USPs
`(A) Domain architecture of the USPs used in this
`study. The constructs used in this manuscript are
`highlighted with corresponding residue numbers
`and expression system.
`(B) Final purification product of the USP constructs
`shown on SDS-PAGE gel. An asterisk indicates
`the expressed USP. USP7FL has an N-terminal
`GST tag.
`Related to Table S1.
`
`USP7CD, USP8CD, USP16CD, USP21CD,
`USP30CD, and USP39CD)
`(Figures 1A
`and 1B). In addition, we expressed and
`purified two known USP activity modula-
`tors: UAF1 (Cohn et al., 2007) and GMPS
`(van der Knaap et al., 2005). Cloning, ex-
`pression, and purification protocols are
`provided in the Materials and Methods
`section.
`
`Large Variations in Both Catalytic
`Turnover and Ub Binding
`Although USP family members share
`a homologous catalytic domain, many
`contain insertions within their catalytic
`domain or have additional domains with
`the potential to influence their activity
`(Luna-Vargas et al., 2011b; Ye et al.,
`2009) (Figure 1A). To study these effects,
`we determined the kinetic parameters of
`all the USPs we have available. To this
`end, we produced a minimal synthetic
`Ub substrate fused at its C-terminus to
`the small molecule 7-amino-4-methyl-
`coumarin (UbAMC) (Dang et al., 1998; El
`Oualid et al., 2010). The UbAMC
`substrate is a reagent widely used to
`assay DUB activity. Upon hydrolysis by
`the DUB, the free AMC reporter molecule
`produces a fluorescent signal that allows
`for a direct read-out of activity (Figure S1A
`available online). The presence of the
`AMC moiety instead of the endogenous
`target makes this into a minimal universal
`substrate.
`This assay is performed in the presence of EDTA to prevent
`inhibition by divalent cations (Ferna´ ndez-Montalva´ n et al.,
`2007). Since this might affect the structural
`integrity of the
`zinc-containing USPs (Figure 1), we also determined the relative
`activity without EDTA (Figure S1B). The activity of all USPs
`except USP30CD is unaffected. Here, the catalytic turnover is
`decreased 2.5-fold upon addition of EDTA (Figure S1C). The
`activity of USP30CD without EDTA is shown in Figure 2.
`Overall, we observed variations of several orders of magnitude
`in both KM and kcat between the USP constructs (Figure 2).
`Previously published kinetic parameters of USPs are listed in
`
`RESULTS
`
`Protein Cloning, Expression, and Purification
`Based on protein expression trials (Luna-Vargas et al., 2011a),
`we identified constructs suitable for large-scale protein expres-
`sion of 12 different USPs in either E. coli or Sf9 insect cells (Fig-
`ure 1A). In this study, we could therefore include 16 constructs
`containing either the (almost) full-length constructs (USP1DN,
`USP7FL, USP11FL, USP12FL, USP16FL, USP25FL, and
`USP46DN, with DN and DC denoting N- and C-terminal
`truncations, respectively), or the catalytic domain (USP4CD,
`
`Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved 1551
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 2
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 2. Kinetic Parameters Using UbAMC
`(A and B) Michaelis-Menten curves for the different USPs, obtained by determining the initial rates (V0) at different UbAMC concentrations, and for USPs with
`intramolecular modulation (B). The assay was performed in duplicate.
`(C) Overview of the kinetic parameters (kcat, KM, and kcat/KM) for the different USPs. Values for USP4 and USP7 are from Luna-Vargas et al. (2011b) and Faesen
`et al. (2011), respectively.
`(D) Activity classification of USPs, based on kinetic parameters, where group 1 represents the USPs with the lowest activity; group 2 contains USPs with
`intermediate activity, and group 3 contains the USPs with the highest activity. Dashed lines link the catalytic domains with the corresponding full-length USPs.
`Solid lines show the effect of intramolecular activating and inhibiting domains.
`Related to Figure S1.
`
`Table S1. This substrate allows direct comparison of relative
`activity among the USP family members. This resulted in a rough
`classification in three groups based on the kinetic parameters
`(Figure 2D). Group 1 represents the USPs whose activity is
`very limited due to a low kcat (USP1DN, USP4CD, USP7CD,
`USP12FL, USP39CD, and USP46DN). The ‘‘intermediate’’
`group, group 2, contains the USPs that show moderate activity
`
`(USP4-D1D2, USP11FL, USP16CD, USP16FL, USP21CD,
`USP25FL, and USP30CD), and group 3 contains very active
`USPs (USP7FL, USP7CD-HUBL, and USP8CD).
`As expected, group 1 contains USP39CD. It shows no activity,
`since it lacks the catalytic cysteine and histidine residues (Nijman
`et al., 2005). Group 1 also contains USP1DN, USP12FL, and
`USP46DN, all
`three known to have low activity, which is
`
`1552 Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 3
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 3. Di-Ub Topoisomer Preference for
`the Different USPs
`(A) Ub (1UBQ) Showing all Lysines.
`(B) Overview of a time-course using all eight
`different di-Ub topoisomers (5 mM) (Linear, K6,
`K11, K27, K29, K33, K48, and K63) for the active
`USPs (75 nM). Samples from each time point (0, 5,
`10, 30, 60, and 180 min) were analyzed on
`Coomassie-stained SDS-PAGE gels. The assay
`was performed twice, and representative gels are
`shown here.
`Related to Figure S2.
`
`domain, which is essential for both activity
`and Ub binding in vitro and in vivo (Faesen
`et al., 2011; Ferna´ ndez-Montalva´ n et al.,
`2007; Ma et al., 2010). The activity of
`USP16CD is modulated by the zinc-finger
`Ub specific protease (ZnF-UBP) domain.
`Surprisingly, the activity is enhanced by
`increasing catalytic turnover, rather than
`by the KM (Figures 2B and 2D). Since it is
`a Ub-binding domain, the effect of the
`zinc-finger could be more prominent in
`poly-Ub processing (Pai et al., 2007),
`which might add up to a bigger difference
`than observed here. USP39CD also
`contains a Znf-UBP domain, but it is
`unlikely that this will
`lead to enzymatic
`activation since USP39CD does not
`have the catalytic residues.
`Overall, this shows that several intramolecular domains are
`able to modulate USPs. The modulation can affect KM (USP4),
`kcat (USP16), or both (USP7), and both inhibitory and activating
`domains are found in USPs. Together, this creates an additional
`layer of regulation of the catalytic activity of USPs.
`
`Di-Ub Preferences of USPs
`Most studies of DUB specificity have focused on processing K48-
`and K63-linked poly-Ub. K48-linked ubiquitination targets
`a protein for active degradation by the proteasome (Chau et al.,
`1989), whereas ubiquitin chains using K63 have mostly nondegra-
`dative outcomes (Chen and Sun, 2009). Our knowledge of func-
`tions of the other linkages is growing. For example, linear ubiquitin
`chains play a role in the NFkB activation pathway and immune
`response and are structurally similar to K63-linked poly-Ub (Ger-
`lach et al., 2011; Komander et al., 2009b; Tokunaga et al., 2009).
`K11 is also a strong degradation signal and is involved in the cell
`cycle (Williamson et al., 2009). The roles of the other linkages
`remain elusive, but they have been implicated in DNA damage
`response (K6 by BRCA1/BARD1 (Wu-Baer et al., 2003)) or lyso-
`somal degradation (K29 (Chastagner et al., 2006, 2008)).
`Since the additional
`linkages serve important cellular func-
`tions, we synthesized all seven lysine-linked di-Ub topoisomers
`(El Oualid et al., 2010). Together with linear di-ubiquitin we used
`them in a qualitative assay to assess all linkage preferences of
`the panel of USPs (Figure 3 and Figure S2). Previously published
`
`enhanced by the external modulator UAF1 (Cohn et al., 2009;
`Cohn et al., 2007).
`In contrast, group 3 represents the most active USPs, and
`contains both USP8CD and the USP7 constructs with activating
`C-terminal Hausp UBL (HUBL) domain (Faesen et al., 2011).
`Interestingly, USP8CD has an unusually weak KM, possibly due
`to an inserted a-helix in the catalytic domain, which is suggested
`to stabilize the observed closed conformation (Avvakumov et al.,
`2006). However, this is compensated by a very high catalytic
`turnover, rendering it a very active USP overall.
`
`Intramolecular Modulation of USP Activity
`Not only do we observe differences in enzymatic behavior
`between the USPs, but we also observe differential effects of in-
`tramolecular domains on the activity of the (minimal) catalytic
`domains in USP4, USP7, and USP16 (Figure 2B).
`We recently showed that USP4 contains a UBL domain in-
`serted in its catalytic domain (USP4CD; Figure 1A), which inhibits
`the activity of USP4CD (group 1; Figures 2B and 2D) (Luna-Var-
`gas et al., 2011b; Zhu et al., 2007). The presence of this UBL
`domain in USP4CD increases the KM and is therefore less active
`than the minimal catalytic domain USP4-D1D2 (group 2; Fig-
`ure 2D) (Luna-Vargas et al., 2011b). In contrast, both kcat and
`KM are affected in USP7, where the minimal catalytic domain
`(group 1) shows far less activity than the full-length enzyme
`(group 3). Here, the activity of USP7 is modulated by its HUBL
`
`Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved 1553
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 4
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 4. Isopeptide-Linked Ubiquitin FP-
`Reagents
`(A) Schematic view of N-terminal TAMRA-labeled
`Ub peptide (K6) conjugated with Ub. Table shows
`the peptide sequences used with the corre-
`sponding residue numbers for the different types
`of Ub linkage. The conjugated lysine is highlighted.
`(B) Michaelis-Menten curves for USP4-D1D2 (top)
`and USP7FL (bottom) were obtained using the
`TAMRA-labeled Ub peptides in an FP hydrolysis
`assay. The curves for USP7 could not be fitted.
`The assay was performed in triplicate.
`Related to Figure S3.
`
`seen between catalytic domain and
`longer constructs, showing that
`the
`modulation effects are substrate-inde-
`pendent mechanisms.
`Overall, this shows that in contrast to
`other DUB families, USPs can hydrolyze
`all di-Ub topoisomers, albeit with differ-
`ences in catalytic efficiency. Also, some
`USPs show perturbed activity toward linear-linked ubiquitin.
`The differences in catalytic efficiency are preserved in the pres-
`ence of the intramolecular activity modulators.
`
`In the Case of USPs, Isopeptide-Linked Ub Is Not
`Representative for di-Ub
`To explain the Ub linkage preference, we might not need full-
`length di-Ub (Shanmugham et al., 2010). To test this in an activity
`assay, we designed and synthesized a panel of fluorescence
`polarization-based (FP) reagents that mimic the lysine-linked
`di-Ubs. In these reagents, TAMRA-labeled Ub peptides were
`linked via an isopeptide linkage to the carboxy terminus of
`wild-type full-length mono-Ub (Tirat et al., 2005) (Figures 4A
`and S3)). Therefore, in contrast to the peptide linkage in UbAMC,
`these FP reagents use the natural isopeptide linkage. The prox-
`imal Ub is represented by 14-mer peptides, each representing
`one of the seven lysines of Ub (Figures 3A and 4A). In addition,
`a di-peptide (KG) was prepared to serve as a minimal substrate.
`Mass spectrometry and SDS-PAGE analysis of these new Ub
`substrates showed that the synthesis was successful for all eight
`different TAMRA-labeled isopeptide-linked Ub FP reagents
`(Figures S3C and S3F).
`As a proof of principle, we used the minimal ‘‘KG’’ FP reagent
`to determine the kinetic parameters of USP4-D1D2 (Figures 4B
`and S4H). With this reagent we determined KM (293 nM) and
`1) values similar to the kinetic parameters obtained
`kcat (0.07 s
`using UbAMC. Only the kcat is higher, possibly due to the differ-
`ence in the chemical nature of the linkage, since the FP reagents
`contain a natural isopeptide linkage in contrast to the UbAMC
`reagent. However, since the KM values are similar, both repre-
`sent comparable Ub reagents.
`In the di-Ub time course assay, we observed linkage prefer-
`ences of USP4-D1D2 and USP7; e.g., USP7 prefers the hydro-
`lysis of K6- over K27-linked di-Ub, and USP4-D1D2 prefers
`K63- over K48-linked di-Ub (Figure 3). Although difficult to fit
`for USP7, with our FP reagents we observed no difference in
`
`preferences are corroborated (Song et al., 2010; Ye et al., 2011).
`Overall, the relative activities from the UbAMC assay are re-
`tained, with a few exceptions. For example, USP21CD shows
`only intermediate activity in the UbAMC assay, but it displays
`activities in the di-Ub assay almost matching the most active
`USP, USP8CD.
`The USP family seems to be rather promiscuous compared to
`other DUB families. For example, the OTU family displays strong
`linkage preferences for specific di-Ub topoisomers (Bremm
`et al., 2010; Edelmann et al., 2009; Virdee et al., 2010; Wang
`et al., 2009). Figure 3 shows that the differential activity of the
`USPs is smaller. Most of the active USPs from this study hydro-
`lyze all di-Ub topoisomers. Nevertheless, there are clear differ-
`ences in efficiency. For instance, although USP1DN, USP7,
`USP8CD, USP11FL, and USP25FL showed robust activity
`toward the lysine-linked di-Ub topoisomers, we observe no
`activity toward linear di-Ub. On the other hand, USP4, USP16FL,
`and USP21CD are active against linear di-Ub. Of these three,
`USP21CD is the only one that is less active against the linear
`di-Ub compared to the other topoisomers. The hydrolysis of
`linear di-Ub is unique for the USP family, since this feature is
`not observed in other DUB families (Komander, 2010).
`Also in the hydrolysis of the lysine-linked topoisomers we
`observe differential activity. For instance, most USPs have diffi-
`culties hydrolyzing K27- and, to a lesser extent, K29-linked di-
`Ub. USP7 has limited activity toward hydrolyzing K27- and
`K29-linked di-Ub. In contrast, the K6, K11, K48, and K63 Ub top-
`oisomers are hydrolyzed relatively efficiently. Another clear
`example is USP4,
`for which K63-linked di-Ub is a better
`substrate than K48-linked di-Ub (Luna-Vargas et al., 2011b;
`Song et al., 2010).
`We wondered whether the intramolecular modulating domains
`in USP4, USP7, and USP16 change the linkage preferences. The
`different USPs respond differently to modulation by internal
`domains, analogous to what was observed with UbAMC (Figures
`3, S2B, and S2C). However, no change in linkage preference was
`
`1554 Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 5
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`activity for either USP4-D1D2 or USP7 and therefore could not
`recapitulate the preferences observed in the di-Ub assay
`(Figures 4B, S3G, and S3H). This shows that these FP reagents
`do not contain the information required to mimic di-Ub for USPs.
`
`The Proximal Ub Hinders Binding to USP7 and USP21
`in Specific Linkages
`Since the FP reagents were not sufficient to reproduce the
`observed linkage preference, we used full-length di-Ubs to
`determine the kinetic parameters directly. We determined KM
`and kcat of the hydrolysis of all di-Ubs by USP7 and USP21, using
`gel-based initial rate experiments that monitored the appear-
`ance of mono-Ub (Figure 5). These assays reproduced the differ-
`ences observed in the time course assay (Figure 3). For USP7,
`the kinetic parameters were similar to the UbAMC assay
`1 in the UbAMC assay), but for USP21CD,
`(2.9 mM and 1.37 s
`1 with
`the kcat is 7- to 8-fold higher in the di-Ub assay (0.1 s
`UbAMC). The KM is not tighter in the di-Ub assay compared to
`the UbAMC (roughly 3 mM compared to 2.56 mM with UbAMC),
`suggesting that there is no induced binding or catalysis effect
`by the proximal Ub moiety.
`These experiments showed that the linkages that are most
`efficiently hydrolyzed by USP7 and USP21 (K6, K11, K33, K48,
`and K63) have similar kinetic behavior (Figures 5B and 5C). In
`the initial di-Ub assay, the K27, K29, and linear linkages showed
`a clear delayed hydrolysis by USP7 and USP21 (Figure 3). This
`was nicely reproduced in this kinetic di-Ub assay (Figures 5
`and S4). Interestingly, for K27 and K29 for both USPs, there
`was hardly any change in kcat; rather, the KM increased far above
`the concentrations used in our assays. This suggests that the
`preference for
`the di-Ub topoisomers arises from steric
`hindrance rather than an additional binding site for the proximal
`Ub moiety. Apparently, the binding of some linkages to the cata-
`lytic domain is impaired, resulting in lower activity.
`The linear di-Ub is a particularly bad substrate for USP7. Also,
`in the kinetic analysis, no hydrolysis is observed, even when
`using up to 15 mM of substrate (Figure 5A). On the other hand,
`USP21CD is active toward linear di-Ub. The kinetic analysis
`showed that both the kcat and the KM are reduced compared
`to those of
`the other di-Ub topoisomers. This suggests
`a decreased capacity to hydrolyze peptide bonds compared to
`isopeptide bonds, with possibly a reduced binding as well.
`
`Intermolecular Activation of USPs by UAF1 and GMPS
`Only Affects kcat
`Besides their intrinsic activity, some USPs are activated by inter-
`molecular modulation. For example, USP1, USP12, and USP46
`are activated by the WD40-repeat containing UAF1, and USP7
`is activated by GMPS (Cohn et al., 2007, 2009; Faesen et al.,
`2011; van der Knaap et al., 2005). Here, we used the UbAMC
`assay to quantify this activation (Figures 6A, 6B, and S4). In
`agreement with previous data, we observe mainly a kcat increase
`(7-fold) of USP1DN activity in the presence of UAF1. The USP1
`used in this work has a mutation in the self-cleavage site
`(Gly671,672Ala)
`(Cohn et al., 2007). UAF1 also activates
`USP12FL and USP46DN, where the kcat
`is increased by 66-
`and 70-fold,
`respectively. Also,
`in the case of USP7, we
`observed a kcat increase (5.5-fold) in the presence of its modu-
`lator GMPS. Interestingly, in contrast to variable modulation
`
`invoked by internal domains (Figure 2D), intermolecular modula-
`tion is achieved mainly by an increase in the catalytic turnover
`rather than in substrate binding (Figure 6B).
`To investigate whether this activation also induces new
`linkage preferences of these USPs, we repeated the di-Ub assay
`in the presence of UAF1 or GMPS (Figure 6C). As expected from
`the UbAMC kinetics, USP1DN shows limited activity in the
`absence of UAF1, and USP12FL and USP46CD show no activity.
`However, in the presence of UAF1, the activity of all three USPs
`is increased, albeit not to the same level. In complex with their
`activators, USP1DN and USP7CD-HUBL show the most activity,
`but no change in chain-type preference by UAF1 or GMPS. This
`agrees well with an activation mechanism that only increases
`kcat, but does not induce binding, which should translate to
`changing KM values.
`
`DISCUSSION
`
`In this study, we used novel reagents to determine the kinetic
`parameters of substrate-independent activity of 12 USPs, their
`di-Ub linkage preferences, and characteristics of both intra-
`and intermolecular activity modulation. We observe large varia-
`tions in both the catalytic turnover (kcat) and Ub binding (KM)
`between USPs. This variability in activity can be explained in
`several ways. First, the activity can be affected by structural re-
`arrangements in both Ub binding sites and active sites, as shown
`by structural studies (Avvakumov et al., 2006; Hu et al., 2002).
`Second, intramolecular domains of USPs can modulate the
`DUB activity, as seen here for USP4, USP7, and USP16. External
`modulator proteins can further regulate the activity of the USP by
`enhancing its activity, as seen for USP1, USP7, USP12, and
`USP46 (Figure 6).
`Here, we characterize a few cases where intramolecular
`modulators regulate the USP catalytic efficiency: either inser-
`tions within or additional domains outside the catalytic domain.
`For both USP7 and USP16 the enzymatic behavior is regulated
`by intramolecular domains (the HUBL and ZnF-UBP domains,
`respectively) outside the catalytic domain,
`resulting in the
`increase of the activity. In addition, variations in kinetics can be
`induced by (large)
`insertions in the catalytic domains them-
`selves, as demonstrated for USP4, where a UBL-containing
`insert
`inhibits the catalytic efficiency (Luna-Vargas et al.,
`2011b). These variations and intramolecular modulations result
`in the unique activity of each USP.
`For the last decade, the main focus on DUB specificity for Ub
`chains has been on K48- and K63-linked poly-Ub chains.
`However, different Ub linkage topoisomers can result in different
`cellular fates, some of which are very specific (Jin et al., 2008;
`Matsumoto et al., 2010; Wu et al., 2010) and require a minimal
`chain length to invoke its function (Cook et al., 1994; Thrower
`et al., 2000). Our study presents the first complete and compre-
`hensive study on di-Ub preference of all eight linkages for USP
`family members. Also, with our
`(mainly) synthetic di-Ub
`substrate, we confirm earlier reports on preferences (Song
`et al., 2010; Ye et al., 2011). Although none of the DUBs so far
`has been tested for all Ub linkages, some DUBs show remark-
`able specificity (Cooper et al., 2009; Edelmann et al., 2009; Kaya-
`gaki et al., 2007; McCullough et al., 2004; Wang et al., 2009).
`Next to CYLD (Komander et al., 2008), the USPs do not have
`
`Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved 1555
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 6
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 5. Michaelis-Menten Kinetics of di-Ub Hydrolysis by USP7FL and USP21CD
`(A and C) Representative western blots of the Michaelis-Menten analysis of di-Ub hydrolysis by USP7FL (A) and USP21CD (C). Assay was performed using 2-fold
`
`dilutions of the di-Ub starting at 15 mM for 5 min at 37
`C.
`(B and D) Michaelis-Menten analysis for USP7FL (B) and USP21CD (D) for di-ubiquitin hydrolysis. Initial rate (V0) of di-Ub conversion into mono-Ub was
`determined at different substrate concentration from western blots shown in (A). The conversion to mono-Ub was quantified using the unsaturated di-Ub signal
`corrected for conversion. The assay was perfomed in duplicate.
`
`1556 Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 7
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`Figure 6. Intermolecular USP Activity Modulation Is Achieved by Increasing kcat
`(A) Kinetic parameters (kcat, KM, and kcat/KM) using UbAMC as the substrate for USP1DN, USP12FL, and USP46DN in the presence of UAF1 and USP7CD-HUBL
`in the presence of GMPS. The assay was performed in duplicate.
`(B) Graphical comparison of the kinetic parameters comparing the USP activity between the USPs and in the presence of their modulator.
`(C) Activity modulation by UAF1 and GMPS toward all eight di-Ub topoisomers. The USP concentration used was 75 nM. Samples from each time point (0, 5, 10,
`30, 60, and 180 min) were analyzed on Coomassie-stained SDS-PAGE gels.
`Related to Figure S4.
`
`strict chain-type specificity, but rather have preferences. Kinetic
`analysis of the hydrolysis by USP7 and USP21 showed us that
`0
`Ub binding site to induce Ub topoisomer
`there is no proximal S1
`preference;
`rather,
`the proximal Ub moiety induces steric
`constraints for binding to the USP in the case of K27 or K29 link-
`ages. These linkage preferences might be increased when using
`longer Ub chains, since some might be ordered in higher-order
`structures (Bremm et al., 2010; Tenno et al., 2004; Varadan
`et al., 2002).
`Overall, the hydrolysis efficiency of the USPs toward K6-,
`K11-, K48-, and K63-linked Ub was higher than for K27- and,
`to a lesser extent, K29- and K33-linked di-Ub. These residues
`localize in distinct regions on Ub (Figure 3A). The lysine residues
`involved in the most easily hydrolyzed linkages (K6, K11, K48,
`and K63) are in the b-sheet or loops. In contrast, the lysine resi-
`dues of the more difficult linkages (K27, K29, and K33) are posi-
`tioned on the other side of the Ub molecule, and are all in the
`a1-helix. In addition, K27 is barely accessible, which possibly
`induces a steric constraint, resulting in the lower activity. This
`interesting property needs future investigation.
`Compared to the other DUB families, the USPs display a more
`promiscuous behavior and some are able to hydrolyze all Ub
`topoisomers with modest differences. For some USPs, the
`activity toward linear di-Ub is slower or even completely lost
`(Figures 3B and 5A). There are several possible explanations.
`First, there is a difference in chemistry due to the lower pKa of
`the N-terminal amine (9.2) compared to the ε-amine of lysine
`
`(10.5). Second, the peptide bond in linear di-Ub is conformation-
`ally more restrained compared to the more flexible isopeptide
`lysine linkage. Finally, the large side chain of the N-terminal
`methionine introduces steric hindrance in the (‘‘linear’’) peptide
`bond. Any of these aspects could influence binding and/or cata-
`lytic efficiency of the hydrolysis of the peptide bond. Still, DUBs
`from the other families are not able to hydrolyze linear-linked Ub
`chains, and therefore the USP family is the only known family
`with members that can process this linkage type.
`A previous study suggested that Ub-peptide reagents might
`be sufficient to discriminate between topoisomers in binding
`(Shanmugham et al., 2010). However, in our activity assays
`with the FP Ub-peptide reagents, we observed no difference
`between Ub linkages. This suggests that the peptides do not
`contain enough information to mimic the proximal Ub for the
`USPs. Nevertheless, they may be sufficient for DUBs from fami-
`lies with more pronounced Ub specificity and be useful tools in
`those cases.
`In our di-Ub assays, some USPs seemed more active
`compared to the UbAMC assay. For example, USP7 was one
`of the most active substrates in the UbAMC assay, whereas in
`the di-Ub assay, this activity was matched by USP11 and
`USP16.
`In our kinetic analysis of the di-Ub hydrolysis, we
`observe no changes in catalytic parameters for USP7, which
`shows that this enzyme does not differentiate between the two
`substrates. This subsequently shows that several other USPs
`are more active against di-Ub, which is an endogenous
`
`Chemistry & Biology 18, 1550–1561, December 23, 2011 ª2011 Elsevier Ltd All rights reserved 1557
`
`Post-Grant Review Petition for US 9,840,491
`EXHIBIT 1026
`Page 8
`
`EXHIBIT 14
`DELANSORNE DECLARATION
`
`214LT:20700:449517:1:ALEXANDRIA
`
`

`

`Chemistry & Biology
`
`Biochemical Characterization of 12 USPs
`
`substrate, compared to UbAMC. Therefore, our KG FP reagent
`could prove a good alternative for UbAMC, since it contains
`the natural isopeptide linkage, which is not present in UbAMC.
`Using the KG FP reagent with USP4-D1D2 and USP21CD shows
`a slightly higher kcat com