Recognition of Lys48-Linked Di-ubiquitin and Deubiquitinating Activities of the SARS Coronavirus Papain-like Protease

Summary

Deubiquitinating enzymes (DUBs) recognize and cleave linkage-specific polyubiquitin (polyUb) chains, but mechanisms underlying specificity remain elusive in many cases. The severe acute respiratory syndrome (SARS) coronavirus papain-like protease (PLpro) is a DUB that cleaves ISG15, a two-domain Ub-like protein, and Lys48-linked polyUb chains, releasing diUb^Lys48 products. To elucidate this specificity, we report the 2.85 Å crystal structure of SARS PLpro bound to a diUb^Lys48 activity-based probe. SARS PLpro binds diUb^Lys48 in an extended conformation via two contact sites, S1 and S2, which are proximal and distal to the active site, respectively. We show that specificity for polyUb^Lys48 chains is predicated on contacts in the S2 site and enhanced by an S1-S1′ preference for a Lys48 linkage across the active site. In contrast, ISG15 specificity is dominated by contacts in the S1 site. Determinants revealed for polyUb^Lys48 specificity should prove useful in understanding PLpro deubiquitinating activities in coronavirus infections.

Graphical Abstract

Highlights

• •A Lys48 linkage-specific diubiquitin activity-based probe selectively labels SARS PLpro
• •The structure of a diUb^Lys48∼SARS PLpro complex reveals an extended di-Ub conformation
• •S2-S1 and S1-S1′ interactions make SARS PLpro specific for K48-linked polyubiquitin
• •SARS PLpro recognizes Lys48-linked polyUb chains and ISG15 via distinct manners

Békés et al. present a high-resolution crystal structure of a SARS virus PLpro∼diUb^Lys48 complex that reveals an extended conformation of the Lys48-linked diUb unit and shows the biochemical basis for SARS PLpro’s preference for Lys48-linked polyUb chains.

Introduction

Viruses can dampen the host anti-viral response by hijacking the ubiquitin (Ub) system (Bhoj and Chen, 2009, Isaacson and Ploegh, 2009) by expressing factors such as viral deubiquitinating enzymes (DUBs) that antagonize Ub-dependent pro-inflammatory pathways (Bailey-Elkin et al., 2014b, Capodagli et al., 2011, Mielech et al., 2014). For instance, the severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) human coronaviruses encode viral polypeptide processing proteases that can also catalyze deubiquitinating and deISGylating activities. The relevant host substrates of these viral DUBs and their mechanisms of target selection remain poorly defined.

Ub-dependent signaling pathways are regulated by the type of Ub modification (mono-ubiquitin [monoUb] or polyubiquitin [polyUb]) or type of Ub chain linkage utilized (Pickart, 2001), as polyUb chains can be formed by conjugation to any of Ub’s seven lysine (Lys) residues or its N-terminal methionine (Komander and Rape, 2012). DUB-mediated cleavage of Ub chains or conjugates serves as a critical regulator or antagonist of Ub-driven signaling pathways (Reyes-Turcu et al., 2009).

While the molecular basis for monoUb specificity for many human and viral DUBs was revealed in structures of monoUb-bound DUB complexes, the molecular basis of Ub chain linkage specificity is understood only for a handful of DUBs (Keusekotten et al., 2013, Mevissen et al., 2013, Sato et al., 2015). In these cases, specificity is dictated by direct readout of a particular isopeptide-linkage via binding the primed (S1′) Ub and unprimed, or proximal, (S1) Ub across the DUB active site. In one case, Lys11-specificity of the OTUD2 catalytic core is achieved by recognizing S1′ Ub and S1 Ub across the DUB active site, an activity that is aided by another domain that contacts a third Ub in a distal S2 site (Mevissen et al., 2013). Lys48-linked Ub chains are highly abundant in cells (Kirkpatrick et al., 2006), and their conjugation to other proteins targets them for proteasomal degradation (Chau et al., 1989), but the molecular basis for DUB recognition of Lys48-linked Ub chains remains unclear.

Ub-based chemical probes have been widely used to profile DUB activities and to stabilize monoUb-bound DUB complexes for structure determination (Ekkebus et al., 2013, Hemelaar et al., 2004). Total chemical synthesis of di-ubiquitin (diUb) activity-based probes (Ub-ABPs), with Cys-reactive warheads at the isopeptide linkage of choice (Mulder et al., 2014) enabled profiling of DUB linkage-specificities across S1-S1′; however these ABPs are not suitable for characterizing DUBs that recognize Ub chain topology through alternative mechanisms.

We and others recently showed that a coronavirus DUB, SARS papain-like protease (PLpro), but not MERS PLpro, preferentially recognizes and releases diUb^Lys48 units during cleavage of polyUb chains by an alternative mechanism that relies on SARS PLpro recognition of diUb via at least two binding sites in S2-S1, rather than S1-S1′ (Békés et al., 2015, Ratia et al., 2014). The structure of SARS PLpro bound to monoUb revealed surfaces required for S1 recognition and plausible explanations for Ub chain specificity and potential surfaces important for S2 recognition (Ratia et al., 2014), but the molecular basis for SARS PLpro Lys48-Ub chain specificity remains unknown. We report the crystal structure of SARS PLpro bound to a diUb^Lys48-ABP. The structure reveals SARS PLpro DUB recognition of an extended Lys48-linked diUb chain via distinct ^S1Ub and ^S2Ub binding sites, and biochemical studies show that ^S2Ub binding is most important for polyUb processing. The molecular basis for diUb^Lys48 recognition by a Lys48-specific DUB has remained unclear, and in this case, SARS PLpro Ub chain specificity is dominated by indirect readout of a unique diUb chain conformation at a site distal from the active site.

Results

A DiUb^Lys48 Activity-Based Probe Preferentially Labels SARS PLpro

SARS PLpro and MERS PLpro are efficient deubiquitinating enzymes, on par with other human DUBs (Báez-Santos et al., 2014, Békés et al., 2015, Ratia et al., 2014); however, SARS PLpro rapidly removes Lys48-linked Ub chains from conjugated substrates, including polyubiquitinated IκBα stabilized by tumor necrosis factor α (TNF-α)/MG132 treatment (Figure 1 A), and is unique in its recognition and release of diUb^Lys48 units from chains of three or more Ubs (Békés et al., 2015). This unique activity is striking when compared to the related MERS PLpro, since they are structurally similar and share 52% amino acid sequence homology (Báez-Santos et al., 2014). While SARS PLpro efficiently cleaves higher-molecular-weight (HMW) polyUb^Lys48 conjugates, it exhibits poor activity in cleavage assays using free diUb chains or mono- or di-ubiquitinated substrates, such as IκBα (Figure 1A). These data supported a model whereby SARS PLpro uses distal Ub binding site (S2) to recognize diUb^Lys48 across S2-S1 (Figure 1B), rather than across S1-S1′, as is typical for most DUBs.

Figure 1.

Distal diUb^Lys48 ABP Labels SARS PLpro

(A) Ub-conjugate cleavage in TNF-α-treated HeLa cell lysates by SARS and MERS PLpro. Dotted lines added for clarity.

(B and C) Schematics of (B) Ub chain recognition by SARS PLpro and (C) recognition and cleavage of biotin-tagged triUb^Lys48.

(D) Cleavage of biotin-triUb^Lys48 by SARS PLpro. Cleavage intermediates detected by avidin-HRP reveal biotin on the diUb product.

(E) Qualitative labeling of SARS PLpro by Ub-ABPs (cartoons at bottom with red stars indicate warhead positions).

(F) Quantitative labeling of SARS PLpro by Ub-ABPs indicating percent of SARS PLpro labeled as derived from gels in Figure S2C. Error bars represent ±SEM.

See also Figures S1 and S2.

To provide evidence for this model, a singly N-terminal biotin-tagged triUb^Lys48 chain (Figure 1C) was generated (Figures S1A–S1G) and cleaved using SARS PLpro. Analysis of cleavage intermediates shows that the N-terminal biotin-label is retained on the diUb product (Figure 1D, right), suggesting that tri-ubiquitin (triUb) recognition requires binding via S2-S1, in a distal-to-proximal direction (Figure 1C, top schematics). In contrast, cleavage intermediates produced by other USP-family DUBs contain mixtures of mono- and diUb products bearing the biotin tag (Figure S2A). USP21_CD and USP2_CD show little preference, while MERS PLpro displays a slight preference.

We next took advantage of linkage-specific diUb activity-based probes that place warheads at the isopeptide linkage (Figure 1E, “in-between”; Mulder et al., 2014) or proximal end (Figure 1E, “distal diUb^Lys48,” green, right cartoon; Flierman et al., 2016). The distal diUb-ABP bears an isosteric non-hydrolyzable triazole linker in lieu of the native isopeptide linkage (Figure S2B). In labeling assays with SARS PLpro (Figure 1E), the distal diUb^Lys48-ABP (green) reacted well, monoUb-ABP (orange) reacted slowly, and the in-between diUb^Lys48-ABP (red) reacted poorly, and quantification shows distal diUb^Lys48-ABP adduct forms most efficiently in comparison to other probes (Figures 1F and S2C). The observation that SARS PLpro formed adducts least efficiently with the in-between diUb^Lys48-ABP probe (compare to monoUb-ABP), suggests that it might bind the in-between diUb^Lys48-ABP via S2-S1, preventing it from binding and reacting via S1-S1′ interactions (Figure S2D). Importantly, the in-between diUb^Lys48-ABP efficiently labels other DUBs that do not exhibit diUb preferences (Figures S2E and S2F). With the ideal reagent in hand, we set out to determine the structural basis for diUb^Lys48 recognition by SARS PLpro.

Crystal Structure of SARS PLpro Bound to a DiUb^Lys48-ABP

SARS PLpro was cross-linked to the distal diUb^Lys48-ABP, purified and crystallized. Crystals diffracted to 2.85 Å, and a structure of SARS PLpro-diUb^Lys48-ABP (Figure 2 A) was determined by molecular replacement (Supplemental Experimental Procedures). Two SARS PLpro-diUb^Lys48-ABP complexes occupy the asymmetric unit. The model was refined to an R_work/R_free of 23.2/26.4 with good stereochemistry (Table 1 ). One of the two complexes exhibits continuous electron density, while the other is less ordered with some discontinuity. Electron density is evident for the propargyl warhead of diUb^Lys48-ABP and active site Cys112 of SARS PLpro in both complexes (Figure S3A), but the diUb Lys48 isopeptide-mimic triazole linkage is weaker in one complex (Figure S3B).

Figure 2.

Crystal Structure of SARS PLpro Bound to a diUb^Lys48-ABP

(A) Cartoon and surface representations of SARS PLpro covalently bound to diUb^Lys48-ABP. SARS PLpro shown in surface representation with the USP-family DUB domain colored blue-white, the N-terminal Ubl domain in gray, the ^S2Ub binding motif in green, and the active site cysteine (C112) in yellow. Cartoon representation of Lys48-linked diUb with proximal Ub (^S1Ub) in salmon and distal Ub (^S2Ub) in raspberry.

(B–E) Cartoon representation of interaction surfaces between SARS PLpro and diUb^Lys48; color as in (A). (B) Structure-based alignment of SARS-PLpro-diUb^Lys48 and a closed conformation of diUb^Lys48 (PDB 1AAR). Ub Ile44 (hydrophobic patch) that contact SARS PLpro shown as spheres. (C) View of the covalent triazole linkage between ^S1Ub-Lys48 and ^S2Ub-Gly75 indicating minimal contacts to SARS PLpro. (D) View of contacts between the SARS PLpro palm domain (M209 and Arg167/Asp168) and the ^S1Ub, highlighting both hydrophobic and polar interactions (^S1Ub-Ile44 and -Gln49/Arg42), respectively. (E) View of contacts between the SARS PLpro ^S2Ub binding motif (Phe70, Glu71, and His74) and the ^S2Ub-Ile44 hydrophobic patch (^S2Ub-Ile44, -Leu8, and -His68).

See also Figure S3.

Table 1.

Crystallographic Data and Refinement Statistics

Data Collectiona
Source	APS 24IDE
Wavelength (Å)	0.9791
Number of crystals	1
Space group	P21
Cell dimensions
a,b,c (Å)	72.98, 68.24, 119.02
α,β,γ (°)	90.0, 103.21, 90
Resolution (Å)	50–2.85 (2.95–2.85)
Completeness (%)	98.0 (100.0)
Total reflections	86,261 (7,690)
Unique reflectionsa	50,843 (5,048)
Wilson B-factor	66.9
Redundancy	3.3 (3.0)
Rmerge (%)	7.0 (53.5)
CC1/2 (%)	99.7 (57.0)
CCa (%)	99.9 (85.2)
< I >/σ(I)	13.99 (1.97)
Refinementb
Resolution (Å)	50–2.85 (2.95–2.85)
Reflectionsc (work/free)	50,864/2,590
Rwork/Rfree (%)	23.2 (36.5)/26.4 (40.3)
Number of atoms	7366
Protein	7298
Ligand	13
Water	55
Average B factors (Å2)	79.8
Protein	79.9
Ligand	73.6
Water	58.1
Rmsd
Bond lengths (Å)	0.003
Bond angles (°)	0.60
MolProbityd
Favored (%)	93.3 (856)
Allowed (%)	99.7 (853)
Outliers (%)	0.3 (3)
Clash score	100th percentile
MolProbity score	100th percentile
PDB code	5E6J

^aUnique reflections for data collected and refinement include anomalous data.

^bStatistics calculated with Phenix; highest shell in parentheses.

^cReflections includes Bijvoet pairs.

^dCalculated with the program MolProbity.

The diUb^Lys48-ABP-bound SARS PLpro structure reveals the basis for SARS PLpro catalytic domain recognition of proximal (^S1Ub) and distal (^S2Ub) Ub molecules within the context of a diUb^Lys48 unit. The SARS PLpro catalytic domain includes an N-terminal Ub-like (Ubl) domain that is dispensable for SARS PLpro activity (Békés et al., 2015, Mielech et al., 2014), followed by classical palm and finger DUB domains, as described for SARS PLpro (Ratia et al., 2006) and other USP-family member DUBs (Reyes-Turcu et al., 2009). The DUB catalytic module superposes well between diUb^Lys48-ABP-bound SARS PLpro and structures of apo (PDB: 2FE8) or monoUb-bound SARS PLpro (PDB: 4MM3; Figures S3C and S3D), with root-mean-square deviation (rmsd) values of 0.56 Å and 0.44 Å over 255 amino acids (Ser61-Ile315), respectively.

The orientation of diUb^Lys48 bound to SARS PLpro is different from prior structures of Lys48-linked Ub chains, whether bound or unbound; this is most apparent when our structure is compared to the “closed” conformation of diUb^Lys48 (PDB: 1AAR; Cook et al., 1992; Figure 2B). It appears that SARS PLpro stabilizes Lys48-linked Ub chains in an extended conformation, akin to conformations of Lys63- or Met1-linked diUb (Komander et al., 2009), yet distinct from those as well. Although SARS PLpro contacts ^S2Ub and ^S1Ub, it makes few contacts to the interface between ^S2Ub and ^S1Ub or the isopeptide analog (Figure 2C).

The position of ^S1Ub within SARS PLpro-diUb^Lys48-ABP is similar to the monoUb-SARS PLpro structure (Ratia et al., 2014), including contacts to the ^S1Ub C terminus; the ^S1Ub-Ile44 patch via Met209 of SARS, and polar contacts to ^S1Ub-Gln49 and ^S1Ub-Arg42 by the SARS PLpro palm domain via Arg167 and Asp168, respectively (Figure 2D). The related viral DUB, MERS PLpro, recognizes ^S1Ub in a similar manner, yet specific contacts to the ^S1Ub-Ile44 patch are not identical (Bailey-Elkin et al., 2014a). Additionally, SARS PLpro cradles ^S1Ub with its fingers domain, with ^S1Ub interaction surfaces comprising the largest buried interaction surface area (∼890 Å²). In comparison to apo SARS, both structures with monoUb and diUb bound reveal similar displacements of the BL2-loop (Figure S3E) that accommodates the Ub C-terminal tail in the active site (Ratia et al., 2014). When compared to the monoUb-bound SARS PLpro complex, a small conformational change is observed in ^S1Ub with respect to displacement of a loop between Ub amino acids 51–57 that is next to Lys48 and the triazole linkage (Figures S3F and S3G). To query if displacement could be due to the triazole linkage, our diUb structure was compared to Lys48-linked diUb (PDB 1AAR) revealing that amino acids 51–57 adopt a similar conformation to that observed in Lys48-linked diUb (Figure S3F), despite dissimilar ^S2Ub conformations (Figures 2B and S3G).

Recognition of ^S2Ub involves contacts centered on a hydrophobic interface between the ^S2Ub-Ile44 patch and a SARS PLpro α helix between the palm domain and N-terminal Ub-like (Ubl) domain that spans amino acids 62–74 (Figure 2E). The buried surface area in the ^S2Ub-SARS interface is smaller than the ^S1Ub-SARS interface (∼540 Å² and ∼890 Å², respectively) but includes contacts to ^S2Ub-Ile44, -His68, and -Leu8 by SARS PLpro residues Phe70, Glu71, and His74. ^S2Ub-Lys48 is exposed on the surface (Figure 2E), suggesting that additional Lys48-linked Ub molecules could be accommodated in the context of a polyUb chain. Other contacts to the ^S2Ub core include SARS PLpro residues Asn129, Asn178, and Glu180 from the palm domain, with the latter contacting ^S2Ub-Lys11 (Figure 2E).

Differential Contributions of ^S2Ub and ^S1Ub Binding Sites for PolyUb^Lys48 Cleavage

We next queried if proximal and distal Ub recognition are important for Ub chain processing by generating SARS PLpro mutants (Figure S4A) and assaying their activity on polyUb chains. Non-conservative substitutions of a cluster of residues in SARS PLpro that are in proximity to the ^S2Ub-Ile44 hydrophobic patch (e.g., SARS PLpro F70S/E71K/H74G) or individual substitutions F70S and H74G greatly reduce Ub chain cleaving activity by SARS PLpro, as assayed on pentaUb^Lys48 (Figure 3 A, top, green; Figure S4B) and tetraUb^Lys48 (Figure S4C). In contrast, SARS PLpro N178A/E180K and E180K substitutions, residues that contact ^S2Ub-Lys11, have a less pronounced effect on pentaUb^Lys48 (Figure 3A, top, green; Figure S4B) or tetraUb^Lys48 (Figure S4C). These data suggest the importance of distal ^S2Ub contacts, as the ^S1Ub binding surface remains intact in these mutants. Further supporting a dominant role for ^S2Ub interactions is the observation that mutation of residues surrounding the ^S1Ub-Ile44 patch have a modest effect compared to ^S2Ub-Ile44-disrupting mutations for pentaUb^Lys48 (Figure 3A, bottom, blue; Figure S4B) and tetraUb^Lys48 (Figure S4C) especially the M209S, R167S, R167S/E168R mutants. The E168R mutant, and a helix-swap mutant that replaces SARS residues with those in MERS PLpro (R167S/E168R+helix), have somewhat diminished activities. And while the N178A/E180K (^S2Ub-contacting via Lys11) and R167S/E168R+helix mutants (^S1Ub-contacting via Gln49/Arg42) display diminished polyUb cleaving ability, their defects are less than that observed for the F70S/E71K/H74G (^S2Ub-contacting via Ile44) mutant. The catalytic mutant (C112A) has no activity.

Figure 3.

Contribution of SARS PLpro ^S1Ub and ^S2Ub Sites to polyUb^Lys48 Cleavage

(A) Ub chain cleavage activities of SARS PLpro WT (black) and its mutants (^S2Ub, top, green; ^S1Ub, bottom, blue) on pentaUb^Lys48. Representative gels used to derive graphs shown in Figure S3B. Error bars represent ±SEM.

(B) Schematics of tetraUb^Lys48 and diUb^Lys48 recognition by SARS PLpro.

(C) Gel-based cleavage assays of SARS PLpro WT and mutants on tetraUb^Lys48 (top) and diUb^Lys48 (bottom) indicating differential effects of diUb ^Lys48 and tetraUb ^Lys48 cleaving activities for ^S1Ub and ^S2Ub mutants of SARS PLpro. Additional mutants analyzed in Figure S4C.

(D) Cartoon and surface representation of SARS-PLpro∼diUb^Lys48 indicating the location of the ^S1Ub and ^S2Ub mutants.

(E) Michaelis-Menten kinetics of WT (black) and selected SARS PLpro mutants (M209S, hydrophobic S1 mutant, light blue; R167S/E168R, polar S1 mutant, dark blue; F70S/E71K/H74G, S2 mutant, green) on diUb^Lys48-AMC substrates and comparison to SARS PLpro WT on monoUb-AMC (in red). Extracted kinetic parameters (k_cat and K_M) are in Table 2.

See also Figure S4.

The relative contribution of ^S1Ub and ^S2Ub interactions within SARS PLpro was further probed by monitoring cleavage activity using tetraUb^Lys48 where cleavage depends on binding via S2-S1 (Figure 3B, top) and comparing this to diUb^Lys48 cleavage reactions that depend on binding via S1-S1′ (Figure 3B, bottom). It is worth noting that diUb^Lys48 cleavage by SARS PLpro requires 5-fold higher concentration of enzyme compared to tetraUb^Lys48 to observe activity. SARS PLpro wild-type (WT) activities were also compared to enzymes carrying mutations in the ^S1Ub and ^S2Ub binding sites using Lys48-linked diUb^Lys48 and tetraUb^Lys48 substrates (Figures 3C and 3D). As expected, the SARS PLpro ^S2Ub mutant (F70S/E71K/H74G) exhibits diminished activity against tetraUb^Lys48 (Figure 3C, top) yet retains WT-level activity on diUb^Lys48 (Figure 3C, bottom). Thus, ^S2Ub recognition is dispensable for diUb^Lys48 cleavage, consistent with an S1-S1′ binding mode being relevant for diUb^Lys48 cleavage. In contrast, the ^S1Ub mutant (R167S/E168R+helix) has diminished diUb^Lys48 cleavage activity yet retains the ability to cleave tetraUb^Lys48 chains with characteristic accumulation of diUb intermediates (Figure 3C). Single point mutants of the composite mutants (F70S for ^S2Ub, E168R for ^S1Ub) exhibit similar cleavage profiles (Figure S4C). These results suggest that mutations predicted to disrupt ^S1Ub recognition do not prevent cleavage of tetraUb^Lys48 when an intact ^S2Ub binding surface is present. As earlier, mutations within SARS PLpro predicted to disrupt contacts to ^S2Ub near Lys11 have a modest effect on tetraUb^Lys48 cleavage, and combining mutations designed to disrupt both ^S2Ub-Ile44 and -Lys11 patches are not additive (data not shown). Additionally, mutations designed to disrupt contacts to ^S2Ub, alone or in combination, do not display gain-of-function activity toward diUb^Lys48, suggesting that disrupting ^S2Ub interaction does not convert SARS PLpro into a DUB with stronger preference for S1-S1′ binding.

To quantify the contribution of ^S2Ub and ^S1Ub contacts with respect to diUb^Lys48 recognition by SARS PLpro, we turned to recently developed fluorogenic model diUb^Lys48-AMC substrates (containing a triazole-linker between Ub moieties), where the AMC fluorophore is conjugated to the proximal end of diUb^Lys48 (Flierman et al., 2016), enabling kinetic characterization of SARS PLpro by monitoring fluorescence during hydrolysis of the -AMC amide bond. Michaelis-Menten kinetic analysis of monoUb- and diUb^Lys48-AMC cleavage by SARS PLpro and its selected ^S2Ub and ^S1Ub mutants reveals that SARS PLpro cleaves the diUb^Lys48-AMC about ∼37-fold more efficiently than it cleaves monoUb-AMC (Figure 3E, black and red, respectively; Table 2 , compare apparent k_cat/K_M values of 3.3E+04 M⁻¹s⁻¹ for monoUb-AMC to 1.26E+06 M⁻¹s⁻¹ for diUb^Lys48-AMC). Moreover, kinetic comparison of SARS PLpro mutants reveals that loss of S2 interactions (F70S/E71K/H74G; Figure 3E, green; Table 2) results in an ∼33-fold loss of catalytic efficiency (k_cat and K_M could not be measured independently, and the S2 mutant could not be saturated by the diUb^Lys48-AMC substrate, indicative of a binding defect). Thus, the S2 mutant converts SARS PLpro into an S1-dependent, monoUb-based DUB. In contrast, mutation of either the hydrophobic interaction in the ^S1Ub site (M209S, Figure 3E, light blue; Table 2) or polar contacts to ^S1Ub (R167S/E168R, Figure 3E, dark blue; Table 2) results in a modest decrease in catalytic efficiency, an effect mainly driven by a 2- to 3-fold increase in K_M without a corresponding loss in k_cat (Table 2). Assaying additional point mutants at a single monoUb- or diUb^Lys48-AMC substrate concentration mirrors these trends (Figure S4D, top). Importantly, S2 mutants do not exhibit diminished monoUb-AMC cleavage rates (Figure S4D, bottom), consistent with gel-based experiments, as they maintained activity when processing diUb^Lys48 into monoUb.

Table 2.

Kinetic and Inhibition Parameters for SARS PLpro and Its Mutants on -AMC Substrates

SARS PLpro Mutant	Kinetic Parameter	MonoUb-AMC	Triazole Linked DiUbLys48-AMC	Native DiUbLys48-AMC	ISG15-AMC
WT	Apparent kcat/KM [M−1s−1]	3.33E+04	1.26E+06	1.01E+06a	5.98E+05
	kcat [s−1]	0.5042 ± 0.02839	42.02 ± 3.872	n/a	9.533 ± 1.218
	KM [μM]	15.12 ± 1.747	33.42 ± 4.869	n/a	15.94 ± 3.172
	Fold kcat/KM over monoUb-AMC	1.00	37.70	30.33a	17.93
	Michaelis-Menten curve fit (R2)	0.9845	0.9668	n/a	0.9411
F70S E71K H74G (S2 mutant)	kcat/KM [M−1s−1]	—	4.23E+04a	—	2.94E+05
	kcat [s−1]	—	n/a	—	2.748 ± 0.6693
	KM [μM]	—	n/a	—	9.359 ± 3.547
	% kcat/KM of WT (per substrate)	—	3.37	—	49.10
R167S E168R (S1 polar mutant)	kcat/KM [M−1s−1]	—	6.50E+05	—	3.64E+04
	kcat [s−1]	—	65.56 ± 22.38	—	0.318 ± 0.1184
	KM [μM]	—	100.8 ± 42.49	—	8.764 ± 5.614
	% kcat/KM of WT (per substrate)	—	51.63	—	6.08
M209S (S1 hydrophobic mutant)	kcat/KM [M−1s−1]	—	7.06E+05	—	4.20E+05
	kcat [s−1]	—	46.59 ± 10.29	—	4.774 ± 1.263
	KM [μM]	—	66.01 ± 19.61	—	11.88 ± 4.578
	% kcat/KM of WT (per substrate)	—	56.13	—	67.19
WT	Ki [μM] with monoUbb	NI	NI	NI	NI
	Ki [μM] with diUbLys48	2.26 (0.9265)	9.05 (0.8942)	12.07 (0.8892)	3.31 (0.8351)
	Ki [μM] with triUbLys48	—	10.57 (0.9509)	10.11 (0.9066)	4.08 (0.6709)
	Ki [μM] with ISG15	NI	NI	NI	NI

n/a, not applicable (k_cat and K_M cannot be independently calculated); NI, no detectable inhibition (IC₅₀ > 100 μM or data do not converge).

^aSubstrate not saturated, k_cat/K_M calculated from slope of linear graph.

^bK_i values were derived from IC₅₀ values based on the equation K_i = IC₅₀/(S/K_M+1), assuming competitive inhibition, where S is the concentration of the substrate (based on Cer et al., 2009). Brackets show goodness of fit (R²) of IC₅₀ values obtained from Prism’s log(inhibitor) versus normalized curve fit. Inhibition curves are shown in Figure S6A.

Structure-based alignment of the Lys48 linkage visible in the triazole-linked diUb^Lys48-ABP∼SARS-PLpro crystal structure to native Lys48 in free diUb^Lys48 (PDB: 1AAR) suggests that the triazole-linkage mimics the distance and geometry of a native isopeptide-bond (Figure S4E). To test if the triazole linkage is a good functional mimic of the native isopeptide-bond, we generated a native isopeptide-linked diUb^Lys48-AMC reagent (Figure S4F) and assayed initial cleavage rates for triazole-linked and native diUb^Lys48-AMC substrates (Figure S4G). Results indicate that the triazole linker is a faithful mimic of the isopeptide bond as initial cleavage rates are similar (despite native diUb^Lys48-AMC being contaminated with monoUb-AMC precursor; see Figure S4F).

SARS PLpro could not be saturated using the native substrate due to insufficient quantities, so inhibition studies were performed with triazole-linked or native isopeptide diUb^Lys48-AMC as substrates to calculate inhibition constants (K_i) using Lys48-linked Ub chains as inhibitors. Results in Table 2 (see also Figure S6A for inhibition curves) show that diUb^Lys48 and triUb^Lys48 inhibit triazole-linked or native diUb^Lys48-AMC cleavage with comparable K_i values (9 and 10 μM or 12 and 10 μM, respectively), values just ∼3-fold lower than the K_M for diUb^Lys48-AMC as determined by Michaelis-Menten kinetics. Additionally, diUb^Lys48 inhibited monoUb-AMC hydrolysis with a K_i of ∼2.2 μM, a result consistent with kinetic assays where diUb^Lys48 recognition is preferred over monoUb. We were unable to observe inhibition of cleavage of diUb^Lys48-AMC substrates using monoUb or free ISG15 (Table 2).

Collectively, these results support the conclusion that distal ^S2Ub and proximal ^S1Ub binding surfaces are important for SARS PLpro activity, with ^S2Ub interactions being dominant, and that polyUb contacts to S2-S1 surfaces are preferred over interactions with S1-S1′ when processing Lys48-linked Ub chains.

S2-S1 Recognition by SARS PLpro Underlies Lys48 Ub Chain Linkage Specificity

DiUB can be linked via seven Ub lysine residues as well as its N terminus to alter their topology. Although dynamic, diUb conformations can be stabilized by interactions between Ub molecules, sometimes templated by their interacting partners (Ye et al., 2012). Linkage specificity for most DUBs characterized thus far is determined by contacts across the protease active site with diUb occupying S1-S1’ sites (Keusekotten et al., 2013, Mevissen et al., 2013, Sato et al., 2008, Sato et al., 2015). As shown previously, SARS PLpro is poor at cleaving diUb and does not strictly require a specific linkage across S1-S1′ (Békés et al., 2015). Given the extended conformation of diUb^Lys48 observed in complex with SARS PLpro and the paucity of contacts to the diUb^Lys48 linkage (Figure 2C), we hypothesized that SARS PLpro might indirectly sense linkage specificity by requiring that diUb occupies both ^S2Ub and ^S1Ub sites, a requirement that would place limits on the type of chain that could be accommodated because of differences in chain topology and distance between individual Ub molecules relative to the two Ub binding sites.

To assess S2-S1 linkage specificity of SARS PLpro, we tested a panel of linkage-specific distal diUb-ABPs to covalently label SARS PLpro. Although not as efficient as diUb^Lys48-ABP, diUb^Lys27-ABP exhibited better labeling efficiency compared other linkage-specific diUb-ABPs, which reacted at levels similar to monoUb-ABP (Figure 4 A and S5A). Since covalent activity-based probes are very reactive, we also tested linkage-specific diUb-AMC substrates by analyzing initial cleavage rates by SARS PLpro. These results suggest that SARS PLpro activity is highly restricted to diUb^Lys48-AMC cleavage (Figures 4B and S5B) as diUb^Lys48-AMC is cleaved ∼100-fold faster compared to other linkages. Each chain-forming residue in ^S1Ub is available for conjugation in our diUb^Lys48-SARS PLpro complex (Figure 4C), but the distance between ^S1Ub-Lys48 and the SARS-Phe70 ^S2Ub binding site is closest (26 Å) with each of the other sites requiring an additional 10 to 15 Å to span between the conjugated lysine and ^S2Ub binding site (Figure S5C). This raised the possibility that linkage-specificity across S2-S1 is enforced by restricting access to other topologies by requiring that SARS PLpro read the distance between diUb Ile44 hydrophobic patches via S2-S1 (Figure 2B), a requirement that is only satisfied by chains carrying Lys48 linkages. Accordingly, assaying homotypic linkage-specific tetra-ubiquitin (tetraUb) chains in endpoint cleavage assays also confirmed SARS PLpro to be Lys48-specific (Figure 4D).

Figure 4.

SARS PLpro Activity Is Restricted to Lys48-Linked Ub Chains with Specificity Dominated by S2-S1 Interactions

(A and B) S2-S1 linkage specificity of SARS PLpro probed by (A) SDS-PAGE analysis of cross-linking to linkage-specific diUb-ABPs (SYPRO-stained) and (B) release of AMC using diUb-AMC fluorogenic substrates with initial linear cleavage rates (V_i) plotted as percent of diUb^Lys48-AMC cleavage rate by WT SARS PLpro as derived from curves in Figure S4B.

(C) View of SARS-PLpro∼diUb^Lys48-ABP highlighting the location of ^S1Ub-Lys48 (dark green, sticks) and its proximity to Phe70 (green) in the SARS PLpro ^S2Ub binding site. Other chain-forming ^S1Ub residues (Lys-6, -11, -27, -29, -33, and -63 and Met1) shown as sticks (red).

(D) Linkage specificity of SARS PLpro assayed using homotypic tetraUb chains. Representative gel shown and bar graph indicating ±SEM from duplicate experiments.

(E) Schematics of triUb^Lys48 chain recognition by SARS PLpro via S2-S1 (green) and S1′ (gray).

(F) Cartoon of tetraUb chains, indicating SARS PLpro preferred sites of recognition (curved line) and cleavage (dotted line).

(G) Time-course for cleavage of tetraUb chains by SARS PLpro and its S2 mutant.

(H) Quantification of V_i as determined from duplicate experiments in Figure 4G. Error bars represent ±SEM.

See also Figure S5.

SARS PLpro Lys48 Specificity Is Also Aided by S1-S1′ Interactions

Most linkage-specific DUBs characterized to date rely on S1-S1′ interactions to achieve linkage-specificity, although OTUD2 and OTUD3 have been shown to utilize both S2-S1 and S1-S1′ interactions to mediate Lys11- and Lys6/11-specificity, respectively (Mevissen et al., 2013). We noticed greater defects for diUb^Lys48-AMC cleavage for some SARS PLpro mutants (especially E180K, N178A/E180K, E168R, and R67S/E168R+helix) compared to gel-based assays using polyUb^Lys48 substrates. As diUb^Lys48-AMC requires only S2-S1 interactions for cleavage, we wondered if additional interactions outside S2-S1 might be responsible for the residual specificity and activity observed for cleavage of polyUb^Lys48 substrates.

To determine if a Lys48-linkage was also preferred across S1-S1′ (Figure 4E), we assayed SARS PLpro and its ^S2Ub-Ile44 mutant (F70S/E71K/H74G) for cleavage of homotypic Lys48- and Lys63-linked tetraUb chains and a mixed linkage tetraUb chain consisting of two Lys48-linked dimers linked by Lys63 (see schematics in Figure 4F). This latter mixed chain could be recognized by SARS PLpro in S2-S1, but its cleavage would require accommodation of a Lys63-linkage across S1-S1′. A time course reveals that SARS PLpro readily cleaves homotypic Lys48 tetraUb chains but is inactive on homotypic Lys63 tetraUb chains. Interestingly, the mixed chain is cleaved when the Lys63 linkage is presented across S1-S1′, but only when an intact S2 site is present (Figure 4G). Initial cleavage rates (Figure 4H) show that SARS PLpro is ∼5-fold slower in cleaving the mixed chain compared to the homotypic Lys48-linked chain, suggesting some specificity for a Lys48 linkage across S1-S1′. More importantly, the ^S2Ub-Ile44 mutant exhibits faster cleavage rates on homotypic Lys48 chains than on mixed or on homotypic Lys63-linked chains (Figures 4G and 4H). Taken together, these data suggest a measurable specificity for Lys48 across S1-S1′, even when diUb^Lys48 recognition via S2-S1 is compromised. Thus, Lys48-specificity of SARS PLpro is enforced by S2-S1 recognition but complemented by a preference for Lys48 linkages across S1-S1′, suggesting that SARS PLpro would be most active on polyUb^Lys48 chains. Consistent with the hypothesis of S2-S1-S1′ recognition of polyUb^Lys48 chains, di- and monoUb-conjugated IκBα accumulates during cleavage by SARS PLpro (Figure 1D, IκBα), while unmodified IκBα remains static (Figure 1D, lighter exposure IκBα). These data suggest that SARS PLpro acts efficiently on polyUb^Lys48 chains in a diUb-dependent manner yet generates mono- and diUb-conjugated substrate remnants that may no longer represent its preferred substrate.

In the absence of a crystal structure of a diUb^Lys48 unit occupying S1-S1′ sites of SARS PLpro, we analyzed our structure for loops predicted to be proximal to S1′ within SARS PLpro that differ in sequence from MERS PLpro, which displays little linkage-specificity across S1-S1′ (Békés et al., 2015). We identified two residues, W107 and A108, adjacent to the exit tunnel of the SARS PLpro active site and mutated them to residues observed in MERS PLpro (Figure S5D). Cleavage assays with this putative S1′ mutant reveals that W107L/A108S has diminished activity on triUb^Lys48, which requires interactions with S2-S1-S1′ (Figures S5E and S5F), but exhibits no significant loss of activity when cleaving diUb^Lys48-AMC, a substrate that is solely dependent on S2-S1 (Figure S5F). This mutant, however, loses activity on Ub-AMC and ISG15-AMC (whose recognition is primarily S1 dependent). Combined with the observation that a Lys48-linked isopeptide is preferred as a substrate, these data suggest that S1′-dependent recognition of Lys48-linkages by SARS PLpro is possible, although additional work will be required to explore this hypothesis.

Recognition of DiUb^Lys48 and ISG15 Requires Distinct Elements within SARS PLpro

ISG15 is a tandem ubiquitin-like (Ubl) molecule consisting of two Ubl folds linked by a flexible hinge (Narasimhan et al., 2005). ISG15 is implicated in anti-viral immunity, and it is a preferred substrate of SARS PLpro when compared to monoUb. Indeed, ISG15-AMC is cleaved ∼20-fold faster than Ub-AMC (Table 2) (Békés et al., 2015, Lindner et al., 2007, Ratia et al., 2014). DiUb^Lys48-AMC (apparent k_cat/K_M of 1.26E+06 M⁻¹s⁻¹) is only preferred by ∼2-fold compared to ISG15-AMC (apparent k_cat/K_M of 5.98E+05 M⁻¹s⁻¹), suggesting that the preferred substrate for SARS PLpro is Lys48-linked polyUb chains.

The activity of SARS PLpro on ISG15 has been shown to be dependent on the distal Ubl within ISG15 (Lindner et al., 2007). Because the S2 binding site is important for diUb^Lys48 cleavage as described here and as proposed previously (Ratia et al., 2014), we sought to directly compare mutations in S2 and S1 of SARS PLpro and their impact on ISG15 cleavage activity. SARS PLpro ^S1Ub and ^S2Ub mutants were used to cleave ISG15-AMC using Michaelis-Menten kinetics (Figure 5 A). Mutations have contrasting effects for ISG15-AMC cleavage (Figure 5A; Table 2). The S1 polar mutant (Figure 5A, dark blue) exhibits greater defects for ISG15-AMC cleavage with minimal effects diUb^Lys48-AMC cleavage. In contrast, the S2 mutant exhibits major defects for diUb^Lys48-AMC cleavage with less severe defects in ISG15-AMC cleavage. To confirm these differential effects, we utilized lysates prepared from IFNβ/MG132-treated cells that contained polyUb^Lys48 chains and ISG15-conjugated substrates and added recombinant SARS PLpro mutants (Figure 5B). Analyzing loss of HMW polyUb^Lys48-conjugates and the appearance of free ISG15, indicative of cleavage of ISGylated substrates, reveals contrasting effects for S2 and S1 mutants (Figure 5C). While the S2 mutant compromises polyUb^Lys48 chain, but not ISG15 cleavage, the S1 mutant has the opposite effect. Finally, we compared other SARS PLpro mutants for cleavage of diUb^Lys48-AMC and ISG15-AMC (Figure S6B). Overall, ^S2Ub site mutants (green) show minimal loss of ISG15-AMC cleavage activity, while ^S1Ub mutants (blue), particularly those containing the E168R mutation, have a pronounced loss-of-function effect. Thus, ISG15 recognition appears more dependent on interactions within S1 and perhaps an alternative S2, while diUb^Lys48 recognition is more dependent on contacts within S2. It is difficult to rationalize these effects in the absence of a structure of ISG15 bound to SARS PLpro, but it appears clear that ISG15 recognition differs in details when compared to diUb^Lys48 recognition (Figure S6C).

Figure 5.

Recognition of DiUb^Lys48 and ISG15 by SARS PLpro Appears Distinct

(A) Michaelis-Menten kinetics of WT (black) and selected SARS PLpro mutants (M209S, S1 mutant, light blue; R167S/E168R, S1 mutant, dark blue; F70S/E71K/H74G, S2 mutant, green) using ISG15-AMC. Extracted kinetic parameters (k_cat and K_M) in Table 2.

(B) Cleavage of HMW-Ub^Lys48 (top, WB anti-K48) and ISG15-conjugates (bottom, WB ISG15) in lysates prepared from IFNβ/MG132-treated cells by SARS PLpro WT and S2 and S1 mutants.

(C) Quantification of loss of HMW-Ub^Lys48 (left) and appearance of free ISG15 (right) from duplicate experiments shown in Figure 5B. Error bars represent ±SEM.

(D) Schematic representation of SARS PLpro substrate specificity. Dashed lines can indicate the -AMC substrate, a non-Lys48-linked Ub unit, or a protein substrate.

See also Figure S6.

Discussion

SARS PLpro appears unique among viral and human DUBs characterized thus far in its ability to recognize polyUb chains by reading units of a Lys48-linked diUb (Békés et al., 2015). Here, we reveal the structural basis for diUb^Lys48 recognition and specificity by SARS PLpro. Coupled with mutational, biochemical and kinetic data, our structure helps to explain the strict Lys48-linkage specificity exhibited by this viral DUB, which is primarily enforced by engaging the diUb module within S2-S1, and enhanced by a slight preference for Lys48-linked Ub across S1-S1′ (Figure 5D). To our knowledge, the diUb^Lys48-SARS PLpro structure represents the only available structure of a linkage-specific DUB bound to a Lys48-linked Ub chain, a result enabled by recently developed diUb activity-based probes (Flierman et al., 2016).

Instead of relying on diUb recognition across S1-S1′, as is common among other DUBs (Keusekotten et al., 2013, Mevissen et al., 2013, Sato et al., 2015) and endoproteases (Berger and Schechter, 1970), SARS PLpro recognizes diUb across S2-S1 binding surfaces that are tuned to recognize Lys48-linked Ub chains. This mode of Ub chain recognition has only been reported for the catalytic core of OTUD2, where an additional binding site mediates interaction with a Lys11-linked diUb (Mevissen et al., 2013). Through modeling, mutational analysis, and the crystal structure of a monoUb-bound SARS PLpro, the Mesecar group suggested that SARS PLpro is an S2-S1 mode DUB (Ratia et al., 2014) and that the distal Ub would be recognized by a hydrophobic surface in SARS PLpro, involving Phe70. Our current study illuminates the structural basis for this interaction. We provide biochemical evidence that the ^S2Ub interface has a dominant role in diUb^Lys48 recognition and polyUb chain cleaving activity, since mutating the distal ^S2Ub recognition surface is more detrimental than disrupting the proximal ^S1Ub recognition surface. We also provide evidence that SARS PLpro Lys48-specificity is complemented by a preference for Lys48-linked chains across S1-S1′.

The hydrophobic patches in Ub proteins are usually packed against each other in free Ub^Lys48 chains (Fushman and Wilkinson, 2011). In the diUb^Lys48-SARS PLpro complex, diUb^Lys48 exhibits an extended conformation, with Ub hydrophobic patches separated by ∼30 Å and recognized by contacts provided by the ^S1Ub and ^S2Ub sites in the diUb^Lys48-SARS PLpro complex. As Ub chains are dynamic in solution (Ye et al., 2012), and because SARS PLpro is structurally similar in apo, monoUb, and diUb^Lys48 complexes, it is likely that SARS PLpro captures diUb units in this conformation.

While SARS PLpro exhibits an ∼5-fold preference for a Lys48-linkage in the S1-S1′ binding mode when an S2 site is also occupied, a lax requirement at S1′ is consistent with its function as an endopeptidase for viral pre-protein processing and perhaps cleavage of ISG15 substrates. The biological targets of SARS PLpro remain unclear, but the preference for polyUb^Lys48 chain cleavage into units of diUb^Lys48 suggests it is likely targeted to substrates that are modified by Lys48-linked polyUb chains.

The diUb-based recognition exhibited by SARS PLpro suggests that SARS PLpro could stabilize monoUb-modified substrate products (Békés et al., 2015), as they are not preferred substrates for SARS PLpro. Indeed, cleavage of Lys48-linked polyUb chains from IκBα by SARS PLpro led to an increase in di- and monoUb-conjugated forms of IκBα. Whether SARS PLpro cleaves other polyUb^Lys48-conjugated substrates to di- or monoUb-conjugated forms remains to be determined. Mono-ubiquitination at membranes and at the endoplasmic reticulum (ER) has been shown to regulate endocytosis and vesicle trafficking (Clague et al., 2012), which are also involved in coronavirus propagation. Given that SARS PLpro is ER localized, it is possible that SARS PLpro functions to stabilize mono-Ub “stubs” on ER substrates to provide an unknown advantage for the virus.

It remains unclear if the anti-inflammatory properties of SARS PLpro require all or a combination of its endopeptidase, DUB, or deISGylating activities. With mutations described herein, which bias activities in deubiquitination versus deISGylation, it may be possible to discern if both activities are important during SARS infection. With that said, it is worth noting that SARS viral titer levels peak at 16–20 hr in cell culture and in mouse model infection studies (Channappanavar et al., 2016, Totura and Baric, 2012), while interferon (IFN)-responsive genes, such as ISG15, are only induced later during infection (Channappanavar et al., 2016). As such, it is likely that the virus has already achieved its full replicative potential before SARS PLpro would encounter ISG15. Additionally, the function of substrate-conjugated ISG15 in anti-viral immunity in humans is now in question (Bogunovic et al., 2012), as a non-conjugatable form of ISG15 was shown to have similar activities as WT ISG15, likely via stabilization of USP18 (Zhang et al., 2015). Given the preference for polyUb^Lys48 and diUb^Lys48 over ISG15, it appears likely that polyUb^Lys48-conjugated substrates are the primary cellular targets of SARS PLpro. These observations suggest that SARS PLpro activities against ISG15 targets may not be as relevant for coronavirus infection as previously thought.

Although the identity of true SARS PLpro substrates remains to be determined, they could include host factors involved in anti-viral signaling, such as IκBα, or viral proteins targeted for degradation by host anti-viral E3 ligases. We propose a model where the most favored substrates for SARS PLpro would be Lys48-linked polyUb chains (S2-S1-S1′ dependent), followed by diUb^Lys48-conjugates (S2-S1 dependent), followed by ISG15-conjugates, with the least favored substrates being monoUb-conjugated substrates and other polyUb chains (only S1 dependent) (Figure 5D).

Experimental Procedures

Synthesis of the Singly Biotinylated TriUb^Lys48 Substrate

Biotinylated triUb^Lys48 was generated using procedures based on previously reported protocols (El Oualid et al., 2010) with modification described in Supplemental Experimental Procedures.

Cloning, Protein Expression, Purification, Crystallization, and Structure Determination of the SARS-PLpro∼DiUb^Lys48-ABP Complex

The generation of recombinant SARS PLpro was described elsewhere (Békés et al., 2015), with modifications as described in Supplemental Experimental Procedures. SARS PLpro (45 μM) was reacted with diUb^Lys48-PRG (45 μM) for 30 min at 37°C in 20 mM Tris (pH 8.0), 150 mM NaCl, and 5 mM DTT; purified by size-exclusion chromatography; concentrated to 11 mg/ml; and frozen in liquid nitrogen for storage (−80°C). Diffraction-quality crystals (∼50–100 μm) grew for 1 month at 12°C by hanging drop vapor diffusion against 0.1 M MES (2-(N-morpholino)-ethanesulfonic acid) (pH 5.5), 0.1 M lithium-acetate, and 12%–20% PEGs 4000/6000/8000. Single crystals were cryo-protected by addition of 20% ethylene glycol and flash cooled in liquid nitrogen. Diffraction data were collected from a single crystal and processed, and the structure was determined using methods reported in Supplemental Experimental Procedures.

Ub-ABP Labeling Assays

For qualitative assays, DUBs (1 μM) were incubated with excess activity-based probes (2–5 μM, monoUb-PRG, diUb^Lys48-VME [“in-between” diUb-ABP]; Mulder et al., 2014) or diUb^Lys48-PRG (“distal” diUb-ABP; Flierman et al., 2016) for indicated times at 37°C in 20 mM Tris (pH 8.0), 150 mM NaCl, and 5 mM DTT. Reactions were performed at least in duplicate. For linkage-specific distal diUb-ABPs, TAMRA-labeled probes were used in a 30-s labeling assay at 37°C. Reactions were quenched with loading sample buffer (4× LDS [Invitrogen], with 5 mM DTT), and analyzed by SDS-PAGE and SYPRO-staining. Gels were scanned to visualize the TAMRA-label (488 nM), imaged using Bio-Rad Gel-Doc, quantified by ImageJ software, and graphed with Prism. Error bars represent ±SEM.

Kinetic Assays with -AMC Substrates

To determine apparent k_cat/K_M for SARS PLpro and its mutants, monoUb-AMC, diUb^Lys48-AMC, and ISG15-AMC were prepared as 2-fold serial dilutions (starting at 30 μM: monoUb and diUb^Lys48 [triazole-linked]; at 15 μM: ISG15-AMC) in 20 mM Tris (pH 8.0), 150 mM NaCl, and 5 mM DTT. SARS PLpro was used at 10 nM (diUb^Lys48- and ISG15-AMC) or 50 nM (monoUb-AMC), and the final reaction volume was 10 μl. Substrates and DUBs were preincubated at 25°C for 1 min, and cleavage of UBL-AMCs was performed at 30°C using a Spectramax fluorescence plate reader running SoftMax Pro 5 (Molecular Devices) operated in kinetic mode in black, round-bottom 384-well plates (Corning, #3698). AMC fluorescence was monitored by excitation at 355 nm and emission at 460 nm over time for 5–10 min. Initial linear cleavage rates (V_i) were fitted by the Michaelis-Menten equation by Prism based on a free AMC standard curve. Experiments were performed at least in triplicate, and error bars indicate ±SEM. To compare triazole-linked or native diUb^Lys48-AMC, substrates were prepared as 2-fold serial dilutions of 3.75 μM (limited by the concentration of the native diUb^Lys48-AMC). Assays were performed in duplicate. To compare individual SARS PLpro mutants, monoUb-AMC, diUb^Lys48-AMC, and ISG15-AMC were used at 400 nM final concentrations, using 10 or 50 nM SARS PLpro. Data were plotted as percent of WT cleavage rates for each substrate (n = 3), and error bars indicate ±SEM. To compare linkage-specific diUb-AMC substrates using SARS PLpro, diUb-AMC substrates were used at 200 nM final concentrations, using 5 nM SARS PLpro. Data were plotted as percent of the diUb^Lys48-AMC cleavage rate (n = 3), and error bars indicate ±SEM.

Gel-Based Ub Chain Cleavage Assays

Ub chains (1 μM; 20 mM Tris [pH 8.0], 150 mM NaCl, and 5 mM DTT) were cleaved at 37°C for indicated times by SARS PLpro WT and its mutants at 10 nM for penta-, tetra-, and triUb^Lys48 or 50 nM of diUb^Lys48 cleavage. Reactions were quenched with loading sample buffer (4× LDS, Invitrogen) and analyzed by SDS-PAGE and SYPRO-staining. Gels imaged using a Bio-Rad Gel-Doc, quantified by ImageJ, cropped where indicated by heavy dashed lines, and graphed using Prism. Assays using SARS mutants were performed in batches; mutants were always compared to cleavage by WT. Loss of uncleaved substrate is expressed as a percent value of uncleaved substrate over the total Ub signal per lane. Initial cleavage rates were calculated from linear portions of curves showing loss of uncleaved substrate over time and expressed as percent of WT rates in arbitrary units. Reactions performed at least in duplicate. Error bars represent ±SEM.

DUB Assay in Lysates

Lysates from human interferon beta (IFN-β; 500 U/ml, 48 hr) or TNF-α (10 ng/ml, 10 min) and MG132 (10 μM, 40 min) stimulated HeLa cells (10 μg total lysate per reaction) were incubated with 100 or 50 nM DUBs, as indicated, in 20-μl reaction volumes with 25 mM DTT for indicated times. Reactions terminated by heating in SDS loading buffer, analyzed by SDS-PAGE and western blotting with indicated antibodies. Blots developed by horseradish peroxidase (HRP) chemiluminescence. Films were scanned, cropped where indicated by heavy dashed lines, quantified and graphed as described above. Light dashed lines in all figures included for clarity.

Author Contributions

M.B. initiated and performed the study with guidance from T.T.H. and C.D.L. G.J.v.d.H.v.N. synthesized diUb^LysX-AMC substrates and biotin-triUb^Lys48; R.E. synthesized distal diUb^LysX-ABP probes in H.O.’s laboratory. M.B. and C.D.L. determined the structure and wrote the manuscript.

Published: May 19, 2016

Accession Numbers

The accession number for the coordinates and structure factors reported in this paper is PDB: 5E6J.