Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Sep 4.
Published in final edited form as: Structure. 2018 Jul 19;26(9):1226–1236.e3. doi: 10.1016/j.str.2018.06.004

A Structure-based Strategy for Engineering Selective Ubiquitin Variant Inhibitors of Skp1-Cul1-F-box Ubiquitin Ligases

Maryna Gorelik 1,2,Y, Noah Manczyk 3,4,Y, Alia Pavlenco 1, Igor Kurinov 5, Sachdev S Sidhu 1,2,*, Frank Sicheri 2,3,4,*,#
PMCID: PMC6415690  NIHMSID: NIHMS1015708  PMID: 30033217

Summary

Skp1-Cul1-F-box (SCF) E3 ligases constitute the largest and best characterized family of the multisubunit E3 ligases with important cellular functions and numerous disease links. The specificity of a SCF E3 ligase is established by one of the 69 human F-box proteins that are recruited to Cul1 through the Skp1 adaptor. We previously reported generation of ubiquitin variants (UbVs) targeting Fbw7 and Fbw11, which inhibit ligase activity by binding at the F-box -Skp1 interface to competitively displace Cul1. In the present study, we employed an optimized engineering strategy to generate specific binding UbVs against 17 additional Skp1-F-box complexes. We validated our design strategy and uncovered the structural basis of binding specificity by crystallographic analyses of representative UbVs bound to Skp1-Fbl10 and Skp1-Fbl11. Our study highlights the power of combining phage display with structure-based design to develop UbVs targeting specific protein surfaces.

eTOC blurb

We used custom ubiquitin variant library to generate selective inhibitors against 17 different members of the SCF E3 ligase family. Structures of two representative ubiquitin variants in complex with their targets reveal details of the binding mechanism validating our approach and suggesting further improvements to the library design.

Graphical Abstract

graphic file with name nihms-1015708-f0001.jpg

Introduction

Ubiquitin (Ub), a highly conserved 76-residue protein, is attached through the E1-E2-E3 enzymatic cascade in a variety of topologies to mark proteins for degradation or to alter their activity. Not surprisingly, the E3 ligases that control specificity of ubiquitination constitute the largest group with ~600 members in humans. Due to their large number and importance in controlling cellular fate, E3 ligases are considered important targets for therapeutic intervention.

Multisubunit Cullin RING Ligases (CRLs) constitute the largest family of E3 ligases with ~250 members (Bhowmick et al., 2013) and are characterized by the presence of a Cullin subunit responsible for tethering a substrate receptor and a RING protein that in turn recruits E2-Ub. Seven distinct Cullin subunits (Cul1, Cul2, Cul3, Cul4A, Cul4B, Cul5 and Cul7) define seven CRL subfamilies, which also differ in terms of the type of substrate receptor (Skp1-F-box, EloBC-VHL, BTB, DDB1-DCAF or EloBC-SOCS-box) and RING protein (Rbx1 and Rbx2) that are tethered (Bulatov and Ciulli, 2015). Skp1-Cul1-F-box (SCF) E3 ligases represent the best characterized CRL family, and they have attracted great attention because of their established and inferred roles in regulating diverse biological processes (reviewed in (Zheng et al., 2016)). In humans, 69 F-box proteins act as substrate receptors for SCF E3 ligases and are bound to the Cul1 subunit through the adaptor Skp1. Binding to Skp1 is mediated by a small F-box domain of 50-residues that is similar in all F-box proteins (Zheng et al., 2002). F-box family members are divided into three subfamilies denoted Fbw, Fbl and Fbo, based on the presence of an additional WD40, LRR or “other” domain, respectively, which functions to recruit substrates.

Small molecule inhibitors of individual SCF E3 ligases have been developed to disrupt substrate binding (Nangle et al., 2013, Chen et al., 2013, Wu et al., 2012, Orlicky et al., 2010) or the interaction between F-box and Skp1 (Aghajan et al., 2010, Chan et al., 2013). However, effective inhibitors are lacking for the vast majority of F-box proteins, and thus, the biological effects and therapeutic potential of inhibiting their E3 ligase functions remain unexplored.

Protein-based modulators of enzymes can aid in the development of small molecule therapeutics by assisting in target validation, by serving as probes in displacement screens, and by serving as leads for structure-based inhibitor design. The Ub scaffold is particularly amenable for protein engineering of variants that function as either inhibitors or activators. Ub variants (UbVs) that bind to a variety of target proteins have been generated through several different approaches (Lorey et al., 2014, Hoffmann et al., 2012, Zhang et al., 2013, Leung et al., 2017). Research by our group using a phage display method to generate UbVs has been successful in targeting diverse components of the Ub proteasome system (UPS) including deubiquitinases (Ernst et al., 2013), UIMs (Manczyk et al., 2017), E2 enzymes (Ernst et al., 2013), HECT E3 ligases (Zhang et al., 2016), RING and U-box E3 ligases (Gabrielsen et al., 2017), the Anaphase Promoting Complex/Cyclosome APC/C complex (Brown et al., 2016), and SCF E3 ligases (Gorelik et al., 2016).

Since many UPS enzymes use weak Ub-binding sites for their normal biological functions, it is not surprising that most UbVs are generated by the strengthening of these natural interactions. However, in the case of SCF E3 ligases, we identified a UbV that bound to a composite surface on the Skp1-Fbw7 subcomplex not previously known to interact with Ub. The UbV-binding surface overlapped extensively with the binding surface on Skp1 for Cul1. Consequently, the UbV functioned as an inhibitor of SCFFbw7 activity by disrupting Cul1 binding. This finding revealed a previously uncharacterized inhibitory site in SCF E3 ligases (Gorelik et al., 2016).

One attractive feature of targeting the Cul1 interacting surface of the Skp1-F-box subcomplex is that it affords a potentially general means to inhibit an SCF complex without any knowledge of substrates and substrate binding mechanisms of the F-box subunit, which remains poorly understood for most F-box proteins. However, the issue of specificity remains a potential liability to this strategy because the Cul1 binding surface is largely contained on the common Skp1 subunit. Indeed, in generating UbVs that bound to Skp1-Fbw7, we observed that our model UbV displayed cross-reactivity to several other Skp1-F-box complexes. By analyzing the co-structure of a UbV bound to Skp1-Fbw7, we confirmed that cross-reactivity stemmed from the UbV forming the majority of its contacts with the common Skp1 component while forming only a minority of contacts, notably involving the β1-β2 loop of the UbV, with the variable F-box component (Fbw7). Thus we hypothesized that specificity could be improved by increasing the contacts between the F-box and the β1- β2 loop. To this end, we increased the length of the β1- β2 loop and subjected it to random diversification (Gorelik et al., 2016). In doing so, we isolated a UbV with an extended β1- β2 loop that bound with high specificity to Fbw11, discriminating even against the highly related protein Fbw1.

Here we sought to determine if the β1- β2 loop diversification strategy is generalizable for generating specific UbV binders for a larger set of Skp1-F-box complexes, including members of Fbw, Fbl and Fbo sub-families. We generated UbV binders targeting 21 different Skp1-F-box complexes, and amongst these, showed that UbVs targeting 17 complexes were highly specific. Furthermore, we solved structures of a representative high specificity UbV generated against Skp1-F-boxFbl11 and a less specific UbV generated against Skp1-F-boxFbl10 in complex with their cognate targets. These structures describe binding mechanism employed by UbVs with extended β1- β2 loops and provide insight into the specificity determinants of UbV-F-box interactions. Together, our findings demonstrate that an F-box targeted UbV library can be exploited to generate specific binders against F-box proteins on a large scale and suggest possible modifications to the library design for further improvements.

Results

Binding selections with an F-box targeted phage-displayed UbV library

Towards developing improved F-box specific UbVs, we designed a phage-displayed library based on one that we employed in our previous study. In particular, we left the predicted Skp1 contacting residues in our UbV template fixed to those observed in a previously characterized Skp1-Fbw7 binder, UbV.Fw7.5, and varied the size and sequence of the ß1-ß2 loop. Our strategy for varying the ß1-ß2 loop differed slightly from our previous UbV library (Gorelik et al., 2016) by inserting 6–8 residues (rather than 7–9 residues) and by randomizing 10–12 residues (rather 11–13 residues) (Figure 1A). We found that by slightly decreasing the number of inserted residues in the ß1-ß2 loop, we increased the fraction of UbVs in the naïve library that were displayed at high levels on phage (Figure S1).

Figure 1. Selection of phage-displayed UbVs targeting a panel of human F-box proteins.

Figure 1.

Open in a new tab

(A) F-box targeted UbV library design. The sequence of wt Ub is shown and UbV positions conserved as wt are indicated by dashes. Positions that were fixed in UbV.Fw7.1 or the library but differ from wt are shown. Each position that was diversified in the library is indicated by an “X”, which represents all 20 amino acids encoded by a degenerate NNK codon (N = A/G/C/T, K = G/T). Three sub-libraries were made with insertions of 6, 7 or 8 codons between positions 11 and 12, and these were combined to generate the final library used for selections. (B) Summary of Skp1-F-box protein purification and UbV selection results. The targeted F-box proteins are shown as follows: phage selections generated UbVs against the F-box target, “Selection successful”; generated UbVs demonstrated preference for the target Skp1-F-box complex (>2-fold higher signal to target Skp1-F-box versus Skp1-Fbw7 in phage ELISA), “Target selective”; generated UbVs showed no preference for the target Skp1-F-box complex (<2-fold higher signal to target Skp1-F-box versus Skp1-Fbw7 in phage ELISA), “Target non-selective”; phage selections failed to produce UbVs against the F-box target, “Selection failed”; Skp1-F-box complexes could be purified at sufficient quantities for the use in phage selections and ELISAs, “Purification successful”; Skp1-F-box complexes could not be purified, “Purification failed”. (C) Number of unique binding UbVs generated for each Skp1-F-box complex. Black bars indicated targets for which UbVs exhibited a clear consensus sequence in the diversified ß1-ß2 loop. (D) Sequence logos (generated by WebLogo (Crooks et al., 2004)) for the diversified residues in the ß1-ß2 loop of unique UbVs selected for binding to the indicated Skp1-F-box complexes.

We chose 32 human F-box family members as targets for Ubv selections, including some with well-characterized biological functions and connections to disease (e.g. Fbl1 (Skp2), Fbw1 (β-TrcpA), Fbo5 (Emi1)) and others with completely uncharacterized functions. We expressed each isolated F-box domain (see Table S1 for domain boundaries) in complex with Skp1. Of 32 F-box domains, 23 could be purified in complex with Skp1 for use in 4 rounds of binding selections with the F-box targeted phage-displayed UbV library (Figure 1B). Following selections, the binding of individual UbV-phage clones was tested by phage enzyme-linked immunosorbent assays (ELISAs). Only those UbVs which met the following criteria were further characterized: 1) Confirmed as binders with greater than 2 fold higher signal towards the target Skp1-F-box relative to GST; 2) Showed some measure of specificity with greater than 2-fold higher signal towards the target Skp1-F-box relative to the reference standard Skp1-Fbw7; 3) Binding inhibited by Cul1 with greater than 20% reduction in binding signal in the presence of Cul1. We expected that these criteria would narrow the population of UbV binders to those that conformed to our design strategy and specifically targeted the Cul1 binding surface of a particular Skp1-F-box complex (Gorelik et al., 2016). Two Skp1-F-box complexes (Fbl13 and Fbl18) did not meet the first selection criterion and four complexes (Fbl6, Fbo24, Fbo28 and Fbl6) did not meet the second selection criterion, and thus, 6 of the 23 purified complexes failed to generate UbVs selective for the target Skp1-F-box (Figure 1B).

Using the above criteria we succeeded in generating UbV binders with at least some evidence of specificity for 17 of the 23 purified Skp1-F-box complexes (Figure 1B, Table S1), representing a success rate of over 70%. The number of unique UbV binders obtained for each Skp1-F-box complex varied considerably (Figure 1C, Table S2). Fbl7, Fbl10, Fbo4, Fbo11, Fbo34, Fbo43, and Fbo45 yielded 6–18 unique UbVs each, which allowed us to define consensus motifs for the diversified positions (Figure 1D). Other selections yielded either fewer unique UbVs (<5) or highly variable UbV sequences, which precluded determination of consensus motifs.

Comparison of the unique UbV sequences for each Skp1-F-box complex revealed that only two UbVs (out of 128 total) were selected by more than one (in this case two in total) F-box proteins. In one case, UbV.Fo5/43.1 was selected as one of 3 or one of 7 unique UbVs by Fbo5 (Emi1) or Fbo43 (Emi2), respectively (Table S2). While these two F-boxes exhibit 44% sequence identity (Figure 2A) and thus represent two of the more similar proteins in our study, two even more similar F-boxes, namely Fbl10 (Kdm2b) and Fbl11 (Kdm2a) (67% identity; Figure 2A), selected distinct UbV sequences. Specifically, Fbl10 selected multiple UbVs that shared a motif (Figure 1D), whereas Fbl11 strongly selected (95% of all clones sequenced) a distinct UbV (Table S2). This result suggested that sequence diversity across the F-box family may be sufficient to generate distinct binding preferences, an issue explored further below. In the second case, two of the more dissimilar F-boxes, Fbo4 and Fbo11 (36 % identify; Figure 2A), both selected UbV.Fo11.1 (Table S2) and families of UbVs with similar consensus motifs (Figure 1D).

Figure 2. Binding specificities of phage-displayed UbVs.

Figure 2.

Open in a new tab

(A) Sequence alignment of the 69 human F-box domains. The F-box domains were aligned based on the available F-box structures and conservation calculated at each position (see Materials and Methods). Residues with >75% conservation and >50% conservation are shaded dark grey or light grey, respectively. Only 24 F-box domains included in the UbV binding specificity analysis are shown. Asterisks (*) indicate F-box domain positions involved in UbV interactions based on the structures of UbVs in complex with Skp1-Fbl11 or Skp1-Fbl10. (B) Phage ELISA binding specificities of the 17 most specific UbVs targeting 17 different Skp1-F-box complexes. For the Skp1-F-box targets with more than one characterized UbV, the UbV with the highest specificity is shown. See Figure S2 for the specificity profiles of all 29 characterized UbVs. Binding of phage-displayed UbV to immobilized Skp1-F-box complexes was detected spectrophotometrically and the absorbance at 450 nM is shown normalized to the signal for cognate Skp1-F-box complex. Grey boxes indicate interactions that were not assayed. The UbV.Fw7.5 and UbV.Fw11.2, targeting Fbw7 or Fbw11 respectively, were described previously (Gorelik et al., 2016). UbVs classified as specific (normalized binding to all non-cognate Skp1-F-box complexes <0.1) are shaded green and the remaining UbVs with different degrees of cross-reactive behavior are shaded red.

Characterization of UbV binding specificity

To investigate specificity in greater detail, we assessed binding of the selected UbVs for the various Skp1-F-box complexes by phage ELISAs. For each Skp1-F-box complex, we chose up to three UbVs for characterization based on prioritization of binding strength, clone abundance in the selected pool, and sequence diversity. We measured the binding signals of 29 unique UbV-phage across a panel of 19–24 immobilized Skp1-F-box complexes. Figure 2B shows the most specific UbV for each of the Skp1-F-box complexes, while Figure S2 shows specificity data for all 29 UbVs that were tested. Providing corroboration that the selection process does impart significant specificity, 27 of 29 UbVs bound to their cognate target with the highest binding signal (Figure 2B, Figure S2). The two exceptions were UbVs selected for binding to Fbw1, which exhibited higher binding signals to the closely related Fbw11 (89% identity; Figure 2B, Figure S2).

We classified UbVs as monospecific if they displayed >10-fold higher binding signal towards their target relative to every other Skp1-F-box tested. Based on this definition, we generated monospecific UbVs for 11 of the 17 Skp1-F-box complexes that yielded specific binders (Figure 2B). UbV.Fw11.2, which targets Fbw11, was previously shown to be highly selective when tested against 6 other Skp1-F-box complexes (including the highly related Fbw1, 89% identity)(Gorelik et al., 2016). When assayed here for binding to 21 Skp1-F-box complexes, it exhibited absolute specificity for Fbw11. Monospecificity was also exhibited by UbV.Fl11.1, which was selected for binding to Skp1-Fbl11 and did not bind appreciably even to the closely related Skp1-Fbl10 (67% identity in the F-box domains). UbVs exhibiting monospecificity were also isolated for Skp1 in complex with Fbl1, Fbl7, Fbl20, Fbo6, Fbo10, Fbo34, Fbo36, Fbo38 or Fbo45.

For two pairs of Skp1-F-box complexes, the selections yielded bispecific UbVs. UbVs selected for binding to Skp1-Fbo4 or Skp1-Fbo11 exhibited similar sequences (Figure 1D) and recognized both complexes. In the case of the Fbo5 and Fbo43 complexes, which share high sequence homology (44% identity in the F-box domain) and biological functions, the same UbV.Fo5/43.1 was the most specific and it recognized both complexes but no others. UbV.Fbl10.1, which was selected for binding to Fbl10, bound most strongly to its target Skp1-Fbl10 and to the similar Skp1-Fbl11 (67% identity in the F-box domain) but also bound weakly to several other Skp1-F-box complexes. Finally, the most specific UbVs selected for binding to Skp1-Fbw1 (UbV.Fw1.1), Skp1-Fbw7 (UbV.Fw7.5)(Gorelik et al., 2016) or Skp1-Fbo7 (UbV.Fo7.1) exhibited significant binding to several complexes beyond the expected targets (Figure 2B). Taken together, these results demonstrate that extension and diversification of the β1-β2 loop enabled the selection of absolutely or highly selective UbVs for the majority of Skp1-F-box complexes that we targeted.

Structural Analysis of UbVs bound to Skp1-Fbl11 and Skp1-Fbl10 complexes

We previously elucidated the structure of UbV.Fw7.1 bound to Skp1-Fbw7 (PDB:5IBK). However, we lacked a structural understanding of the mechanism of action of UbVs bearing extended β1-β2 loops designed to favour particular Skp1-F-box complexes. Thus, to better understand the molecular basis for specificity and to confirm our design strategy, we chose representative examples of highly specific (UbV.Fl11.1) and less specific (UbV.Fl10.1) UbVs for structural analysis.

We solved the structure of UbV.Fl11.1 bound to Skp1-Fbl11 to 2.61 Å resolution by molecular replacement (Figure 3A, See methods and Table 1 for data collection and structure refinement statistics). The crystal asymmetric unit consisted of 3 highly similar (average RMSD = 0.28 Å) trimeric complexes, and as such we focused our description on a single complex (chains B,F,H) (Figure S3A). The binding mode of Ubv.Fl11.1 to Skp1-Fbl11 was similar in some respects to the binding mode of UbV.Fw7.1 to Skp1-Fbw7 (Figure 3B). UbV.Fl11.1 targeted a very similar Cul1 binding surface on the Skp1-F-box complex and the total surface areas buried by the UbVs were similar (1913 Å2 for UbV.Fl11.1 and 1738 Å2 for Ubv.Fw7.1). Despite sharing a common binding surface on the Fbl11-Skp1 complex, UbV.Fl11.1 bound in a strikingly different orientation relative to UbV.Fw7.1 (a 70° rotation and a 3 Å translation of the UbVs) (Figure 3B). The change in binding orientation of UbV.Fl11.1 relative to UbV.Fw7.1 may be caused by one or a combination of the following factors (see also Figure S4 for a detailed illustration of differences between the two UbVs originating from 3 successive engineering cycles): 1) the 8 residue insertion within the β1-β2 loop of Ubv.Fl11.1; 2) 9 substitutions outside the β1-β2 loop that UbV.Fl11.1 contains relative to UbV.Fw7.1, which were incorporated to improve Skp1 binding (A12T, I42V, L49R, H62Q, K63R, I72V, R74G, G75R, G76R, substitutions are written as UbV.Fw7.1 to UbV.Fl11.1); 3) the presence of Skp1 acidic loops in the Skp1-Fbl11-UbV.Fl11.1 complex that were shortened in the Skp1-Fbw7-Ubv.Fw7.1 complex (Schulman et al., 2000); 4) the two extra C-terminal residues on UbV.Fw7.1 relative to UbV.Fl11.1. It is not readily apparent which of these changes are responsible for the altered binding modes of the two UbVs.

Figure 3. The crystal structure of UbV.Fl11.1 bound to Skp1-Fbl11.

Figure 3.

Open in a new tab

A) Overall architecture of the UbV.Fl11.1-Skp1-Fbl11 complex. B) The structures of UbV.Fl11.1-Skp1-Fbl11 and UbV.Fw7.1-Skp1-Fbw7 superimposed by their Skp1 subunits. The rotation axis, rotation angle and translation required to superimpose UbV.Fw7.1 onto UbV.Fl11.1 are shown. Translation is occurring outwards from page. C) Open book view of the contact surfaces between Ubv.Fl11.1 and Skp1 and Fbl11. Skp1 and Fbl11 residues engaged by the ß1-ß2 loop of Ubv.L11.1 are colored yellow and those engaged by Ubv.L11.1 residues that were fixed in the library are colored orange. Residues that contact both the ß1-ß2 loop and the fixed residues are colored based on which part of Ubv.L11.1 they form the majority of their contacts with. Unengaged surfaces of Skp1 are colored blue and unengaged surfaces of Fbl11 are colored red. The UbV.L11.1 ß1-ß2 loop and core residues engaged in the interaction with the Skp1-Fbl11 complex are colored yellow or orange, respectively. Unengaged surfaces on Ubv.L11 are colored green. D) Details of the molecular interactions between UbV.L11.1 and Fbl11. Dashed lines indicate hydrogen bonds.

Table 1.

Crystallographic data collection, processing and refinement statistics.

UbV.Fl11.1-
Skp1-Fbl11		 UbV.Fl10.1-
Skp1-Fbl10		 UbV.Fl10.1-
Skp1-Fbl11
Data Collection
Space group P212121 P212121 P21
Wavelength (Å) 1.54 0.979 0.979
Cell dimension
 a, b, c (Å) 60.6, 118.2, 146.3 65.4, 81.8, 128.7 38.2, 119.6, 63.7
 α, β, γ (°) 90, 90, 90 90, 90, 90 90, 98.5, 90
Resolution range (Å) 50 – 2.61 50 – 2.66 19.82 – 3.27
Rsym 0.102 (1.00) 0.174 (1.07) 0.179 (1.07)
CC (1/2) (0.864) (0.781) (0.547)
Total no. of observations 218630 109832 31378
Total no. unique observations 32211 (2735) 20716 (2028) 8718 (1811)
Mean [(I)/σ(l)] 24.7 (1.7) 16.5 (2.15) 5.2 (1.1)
Completeness (%) 98.5 (85.4) 99.8 (99.9) 98.9 (99.3)
Multiplicity 6.8 (5.2) 5.3 (4.6) 3.6 (3.5)
Refinement
Resolution (Å) 43.4 – 2.61 47.5 – 2.66 19.82 – 3.27
No. of reflections 31608 20406 8658
Rwork/Rfree 23.7/29.4 21.6/26.3 26.0/31.7
No. of atoms
 Protein 6166 4071 3529
 Ligand/ion 0 0 0
 Water 70 40 0
Average B-factors
 Protein 66.88 69.54 91.84
 Ligand/ion N/A N/A N/A
 Water 51.17 49.33 N/A
R.m.s.d
 Bond length (Å) 0.006 0.006 0.003
 Bond angles (°) 0.799 0.881 0.615
Ramachandran statistics
 Residues in favoured regions (%) 95.21 96.39 95.06
 Residues in allowed regions (%) 4.66 3.61 4.94
 Residues in disallowed regions (%) 0.13 0 0
Open in a new tab

Statistics in brackets are for highest resolution shell

Consistent with the library design, the extended β1-β2 loop of Ubv.Fl11.1 engaged the F-box domain, and furthermore, the surface area buried between UbV.Fl11.1 and the Fbl11 F-box subunit (Figure 3C) was increased greatly relative to that between UbV.Fw7.1 and the Fbw7 F-box (727 Å2 and 322 Å2, respectively). In particular, the β1-β2 loop of UbV.Fl11.1 formed the majority of the contacts with Fbl11 (~75% of the buried surface area), but other residues that were fixed in the library design also contributed to the interaction. Contacts with the F-box mediated in part by the extended β1-β2 loop included both hydrophobic and hydrogen bond interactions, and provided a likely basis for the specificity of the UbV. Specifically, hydrophobic residues Trp892 and Met893 on Fbl11 interacted with UbV residues Tyr11*, Leu71* and Phe73*, while Met899 on Fbl11 interacted with Leu11d* (UbV residues are indicated by “*” and insertions relative to the Ub sequence are indicated by lower case letters) (Figure 3D, left panel and detailed stereo view in Figure S5, top panel). Hydrogen bond interactions involved the side chain of Trp892 with the side chain of Gln40* and backbone oxygen of Val72* (Figure 3D, left panel), the side chain of Arg895 with the main chain oxygens of Ser10*, Tyr11* and His11b*, the side chain of Arg928 with the main chain oxygen of His11b* (Figure 3D, center panel), and the side chain of Glu896 with the side chain of Ser10* (Figure 3D, right panel).

Notably, amongst all the contact residues on Fbl11, only Trp892 (Val1064 in Fbl10) differed from the corresponding residues in Fbl10, but nonetheless, Fbl10 exhibited only weak binding to UbV.Fl11.1. Trp892 makes contacts with residues both within and outside of the ß1-ß2 loop (Figure 3D left panel and Figure S5 top panel). The smaller side chain of Val1064 in Fbl10 relative to Trp892 in Fbl11 would likely make suboptimal packing contacts with a loss of hydrogen bond interactions (Figure S5 bottom panel). Thus, it appears that Trp892 is a key determinant of the specificity of UbV.Fl11.1 for Skp1-Fbl11 over Skp1-Fbl10. To test this hypothesis we substituted Val for Trp at position 892 in Fbl11 and Trp for Val at position 1064 in Fbl10 and tested each for binding to UbV.Fl11.1 (Table 2) using a competitive protein ELISA. In this assay Skp1-Fbl11 displayed an IC50 of 79 nM against UbV.Fl11.1, similar to the KD of 25 nM measured by ITC (Figure 4, left panel). Consistent with the preference for Fbl11 demonstrated by phage ELISA (Fig 2B), UbV.Fl11.1 bound Fbl11 with ~10-fold greater affinity relative to Fbl10 (Table 2). Confirming our prediction that Trp892 specifies preferential binding of UbV.Fl11.1 to Fbl11, the Fbl11 variant with a W892V substitution displayed reduced binding comparable to that of Fbl10, and conversely, the Fbl10 variant with a V1064W substitution displayed increased binding comparable to that of Fbl11 (Table 2).

Table 2. Mutational analysis of Fbl10 and Fbl11 binding to target UbVs.

See also Figure S6 for isothermal titration calorimetry of UbV.Fbl11.1 binding to Skp1-Fbl11 and UbV.Fbl10.1 binding to Skp1-Fbl10 and Skp1-Fbl11.

IC50
(nM)a
Fbl11 Fbl11
W892V		 Fbl10 Fbl10
V1064W		 Fbl10
V1069A		 Fbl10
A1072Y
Ubv.Fl11.1 79 ± 11 690 ± 110 870 ± 80 71 ± 10
Ubv.Fl10.1 430 ± 30 140 ± 11 220 ± 30 22 ± 1.4
Open in a new tab
a

IC50 values were calculated by competitive ELISA as the concentration of Skp1-F-box in solution that blocked 50% of Ubv binding to immobilized Skp1-F-box complexes.

Figure 4. Isothermal titration calorimetry analysis of UbV.Fbl11.1 binding to Skp1-Fbl11, UbV.Fbl10.1 binding to Skp1-Fbl10 and UbV.Fbl10.1 binding to Skp1-Fbl11.

Figure 4.

Open in a new tab

See methods for experimental details. Experiment performed in triplicate. Values represent mean ± SD.

We also used molecular replacement to solve the structure of UbV.Fl10.1 bound to each of its two strongest binding partners, Skp1-Fbl10 and Skp1-Fbl11, at 2.66 Å and 3.27 Å resolution, respectively (Figure 5A; See methods and Table 1 for data collection and structure refinement statistics). Both crystal asymmetric units consisted of 2 highly similar trimeric complexes (RMSD = 0.35 Å for UbV.Fl10.1-Skp1-Fbl10 and RMSD = 0.95Å for UbV.Fl10.1-Skp1-Fbl11), and as such, we focused our descriptions on a single complex for each structure (chains A,C,F and chains A,C,H respectively; Figure S3B,C). The overall binding modes of UbV.Fl10.1 to Skp1-Fbl10 and to Skp1-Fbl11 were highly similar to each other and to the binding mode of UbV.Fl11.1 to Skp-Fbl11, with only small differences in center of mass positions and rotations (Figure S6). Thus, in all three UbV structures with extended β1-β2 loops, the UbVs bound to Skp1-F-box complexes with very similar orientations that were distinct from that of UbV.Fw7.1 with a short β1-β2 loop.

Figure 5. The crystal structures of UbV.Fl10.1 bound Skp1-Fbl10 or Skp1-Fbl11.

Figure 5.

Open in a new tab

A) Overall architecture of the UbV.Fl10.1-Skp1-Fbl10 and UbV.Fl10.1-Skp1-Fbl11 complexes. B) Open book views of contact surfaces between UbV.L10.1 and Skp1-Fbl10 or Skp1-Fbl11. Coloring is the same as Figure 3B, except for unengaged surfaces of Fbl10, which are colored pink. C) Details of the molecular interactions between Ubv.L10.1 and Skp1-Fbl10 or Skp1-Fbl11. Dashed lines indicate hydrogen bonds.

The surfaces between Ubv.Fbl10.1 and Skp1-Fbl10 or Skp1-Fbl11 were highly overlapping with contact areas of 2078 Å2 and 1724 Å2, respectively (Figure 5B). Consistent with the lower specificity of UbV.Fbl10.1 relative to UbV.Fl11.1 (Figure 2B), UbV.Fbl10.1 had a smaller contact surface with Fbl10 (336 Å2) or Fbl11 (344 Å2) than did UbV.Fbl11.1 with Fbl11 (727 Å2).

Among F-box residues that contact UbV.Fbl10.1 and thus may contribute to binding specificity, Val1069 and Ala1072 in Fbl10 and the corresponding Val897 and Ser900 in Fbl11, differ in other F-box proteins that bind weakly to UbV.Fbl10.1 (Figure 5C, Figure 2A). These side chains in the F-box appear to allow efficient packing against Leu11a* in UbV.Fbl10.1 (Figure 5C). In contrast, the weak binding F-box proteins have bulkier residues at the position corresponding to Ala1072/Ser900 and/or contain a residue other than valine at the position corresponding to Val1069/Val897. To test whether these two positions dictate the preference of UbV.Fl10.1 for Fbl10 and Fbl11 over other F-box proteins, we assessed the effect of V1069A (Ala representing an amino acid different than the conserved Val) and A1072Y (Tyr representing a bulkier sidechain compared to Ala/Ser) substitutions in Fbl10 on Ubv.Fl10.1 binding using the competitive protein ELISA. In this assay, Skp1-Fbl10 exhibited an IC50 of 140 nM against UbV.Fl10.1 and Skp1-Fbl11 exhibited an IC50 of 430 nM against Skp1-Fbl11 (Table 2), similar to the Kd of 91 nM or 421 nM, respectively, measured by ITC (Figure 4, middle and right panels). Contrary to our hypothesis, the Fbl10 V1069A substitution did not decrease binding considerably, while the Fbl10 A1072Y substitution actually increased binding (Table 2). Thus the basis for the specificity of UbV.Fbl10.1 for Skp1-Fbl10 and Skp1-Fbl11 remains an open question.

Effects of UbVs on the function of SCFFbl1 and SCFFbo11 in cells.

To test whether UbVs are able to inhibit SCF E3 ligases in cells, we chose UbV.L1.1 and UbV.O11.1 targeting SCFFbl1 and SCFFbo11, respectively, for testing in HEK293T cells. SCFFbl1 (Skp2) is one of the best characterized SCF E3 ligases and is considered an important oncogene with numerous studies attempting to engineer SCFFbl1 inhibitors for cancer treatment (Skaar et al., 2014). SCFFbo11 is also well characterized and, along with SCFFbl1, has been shown to be among the most abundant SCF ligases in HEK293T cells (Reitsma et al., 2017). In order to determine whether cellular expression of UbV.Fl1.1 or UbV.Fo11.1 could inhibit their cognate ligases, we analyzed the stability of ligase substrates. Consistent with anticipated inhibitory effects, UbV.Fl1.1 stabilized the SCFFbl1 substrate p27, whereas UbV.O11.1 stabilized the SCFFbo11 substrate Snail1 (Figure 6A, B) (Jin et al., 2015, Zheng et al., 2014). The observed effect was specific, as we did not observe any stabilization of substrates targeted by other SCF E3 ligases including Cyclin E (Fbw7 substrate) and Cry2 (Fbxl3 substrate) in the case of UbV.L1.1 and Cyclin E, Cry2 and c-Myc (Fbw7 and other E3 ligase substrate) in the case of UbV.O11.1 (Figure 6A, B). Taken together with the results of our previous study (Gorelik et al., 2016), which demonstrated inhibition of cellular SCFFbw7 and SCFFbw11 by UbVs, these results confirm that UbVs can target Cul1 binding sites on SCF E3 ligases to inhibit their functions in cells.

Fig. 6. Biological activity of Ubvs in HEK293T cells.

Fig. 6.

Open in a new tab

A) Expression of UbV.Fl1.1 stabilizes the SCFFbl1 substrate p27, but has no effect on stability of the SCFFbw7 substrate Cyclin E and the SCFFbxl3 substrate Cry2. Cells were transiently transfected with either empty vector or vector encoding FLAG-UbV.Fl1.1. Cells were treated with 100 ug/ml Cycloheximide (CHX) for the indicated time points and cell lysates were probed with antibodies against endogeneous p27, Cyclin E, Cry2, GADPH (loading control) or FLAG (UbV levels). B) Expression of UbV.O11.1 stabilizes the SCFFbo11 substrate Snai1 but has no effect on the stability of the SCFFbw7 substrate Cyclin E, the SCFFbxl3 substrate Cry and the SCFFbw7,Skp2 substrate c-Myc. Cells were transiently transfected with HA-tagged Snai1 and either empty vector or vector encoding FLAG-UbV.Fo11.1. Cells were treated with 20 μg/ml CHX for the indicated time points and cell lysates were probed with antibodies against Snail1, Cyclin E, Cry2, c-Myc, GADPH (loading control) or FLAG (UbV levels).

Discussion

We previously reported a strategy to develop specific UbV inhibitors of SCF E3 ligases using the well-characterized F-box proteins Fbw7 and Fbw11 as test cases (Gorelik et al., 2016). Here we applied this approach to approximately half of the human F-box family by attempting to target 32 of the 69 members (Figure 1B). We successfully purified 23 of the 32 F-box domains in complex with Skp1, and for 17 of these we generated UbV binders with at least some level of specificity. Our results demonstrate that our strategy can generate fairly specific UbV inhibitors for a large majority of stable Skp1-F-box complexes that can be purified by recombinant expression in bacteria. Further improvements in this success rate should be achievable by optimization of the bacterial Skp1-F-box expression system or by using eukaryotic expression systems that are better able to support the correct folding of Skp1-F-box complexes.

A survey of binding specificities across a large panel of Skp1-F-box proteins (Figure 2B) revealed that many UbVs display high specificity for their cognate targets. Notably, for 11 Skp1-F-box complexes, we identified at least one monospecific UbV, and for 4 others, we identified bispecific UbVs. We designed the F-box targeted UbV library with the expectation that extension and diversification of the β1-β2 loop would yield UbV binders with improved specificity through expanded and optimized interactions with the F-box domain. Consistent with our design, the β1-β2 loop of the highly specific UbV.Fl11.1 forms extensive interactions with the Fbl11 F-box domain (Figure 3C, D). The structures reported in this study of UbVs bound to Skp1-F-box complexes reveal that Gln40*, Leu71*, Val 72* and Phe73*, which were fixed in the library design, make contacts with the F-box domain (Figure 3D, Figure 5C) that were prohibited by distance constraints in the original structure of UbV.Fw7.1 bound to Skp1-Fbw7. We posit that these positions may be fertile ground for additional sequence diversification to further optimize specificity and affinity.

Although we have focused our efforts on developing UbV inhibitors for SCF E3 ligases, targeting other CRL families is also of a great interest. While the approach described here is not directly applicable to Cul4A and Cul4B families, since their substrate receptors do not interact directly with Cullin subunits (Angers et al., 2006), it could likely be applied to the other three CRL subfamilies. Namely EloBCVHL-Cul2, EloBC-SOCs-box-Cul5, and BTB-Cul3 families not only share a common architecture with the SCF E3 ligases, they also share a structurally conserved interaction interface with Cul1 (Stogios et al., 2005) (Figure S7). As in the case of SCF E3 ligases, UbV inhibitors of these CRL families could be generated by targeting a defined and relatively small component that is responsible for the interaction with a Cullin subunit, without any knowledge of the overall structure or function. This strategy may be most successful for the BTB-Cul3 family, which is distinct in that BTB proteins interact with Cul3 directly, limiting the issue of specificity that arises when a common component such as Skp1 contributes to the targeted interaction surface. Most importantly, previous UbV inhibitors of Fbw7 and Fbw11 (Gorelik et al., 2016) and the UbVs described here can serve as useful tools to interrogate functions of SCF E3 ligases. Thus, we are well positioned to apply the UbV strategy to systematically develop specific inhibitors to a significant fraction of the large CRL family of E3 ligases.

STAR Methods

Contact for reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact: Frank Sicheri, (sicheri@lunenfeld.ca)

Experimental Model and Subject details

Protein expression

E. coli BL21 (DE3) cells were grown in LB medium at 37°C and proteins expressed at 16°C.

Cell based assays

For cell based assays human 293T cells (female) were cultured in DMEM (Gibco 11995–065) supplemented with 10% fetal bovine serum (Gibco 12483–020) and 1X Penicillin Streptomycin solution (Corning 30-002-Cl) at 37°C and 5% CO2.

Method details

Protein expression and purification

The Skp1-F-box, UbV, and Cul1 proteins employed in phage binding selections and/or ELISAs were expressed and purified as described (Gorelik et al., 2016). Domain boundaries of F-box constructs are listed in Table S1. The Skp1-F-box-UbV complexes used in crystallization were prepared as described previously for the Skp1-F-boxFbw7-Ubv.Fw7.1 complex(Gorelik et al., 2016).

Library construction and binding selections

The UbV-phage library was constructed as described previously for Library 3 (Gorelik et al., 2016), except that 10–12 residues were inserted in the β1-β2 loop. The diversity of the constructed library was 1.8 × 1010. For binding selections, GST-tagged Skp1-F-box complexes were coated on 96-well MaxiSorp plates (Thermoscientific 12565135) by adding 100 μL of 50 nM protein solution and incubating overnight at 4 °C. Four rounds of binding selections with phage library pools were performed with immobilized proteins as described (Fellouse, 2007). To eliminate UbV-phage that bound nonspecifically, input phage pools were pre-incubated on plates coated with non-target Skp1-F-box proteins (Table S1) and the input phage pools were also mixed with 1 μM of the same non-target Skp1-F-box complex during the positive selection step. Several targets (Fbo24, Fbo28, Fbl6), which failed to generate specific UbV-phage in the initial binding selections, were subjected to a second binding selection using GST as the negative selection target.

For each Skp1-F-box target, 24–192 UbV-phage clones were screened by phage ELISA for binding to target Skp1-F-box, a non-target Skp1-F-box and with competition from Cul1. UbV-phage that exhibited >2-fold higher binding signal to target versus GST or Skp1-Fbw7, and at least 20% reduction in binding in the presence of Cul1, were further characterized. Up to 24 positive clones were sequenced from each selection (Table S2).

ELISAs

For phage ELISAs, target Skp1-F-box complexes were immobilized on 384-well MaxiSorp plates (Thermoscientific 12665347) by adding 30 μL of 50 nM protein solutions and incubating overnight at 4 °C. Phage ELISAs were performed as described (Fellouse, 2007), except that three washes were performed for all wash steps and volumes were scaled down from 100 μL to 30 μL to accommodate the 384-well format. Binding of phage was detected using anti-M13-HRP antibody at 1:5000 dilution. For testing whether the binding of UbV-phage was inhibited by Cul1 protein, either 20 uL of PBS or 20 uL of 200 nM Cul1 were added for 10 min prior to addition of 10 uL UbV-phage. Reduction of phage-displayed Ubv binding by >20 % was considered to signify competition with Cul1 for binding. For protein ELISAs, Skp1-F-box complexes were immobilized by adding 30 μL of 200 nM protein solutions to 384-well MaxiSorp plates (Thermoscientific 12665347) and incubating overnight at 4 °C. Ubv.Fl11.1 or Ubv.Fl101.1 at sub-saturating concentration were mixed with the serial dilutions of the tested Skp1-F-box complexes and applied to the plate containing immobilized Skp1-Fbl11 or Skp1-Fbl10 complexes respectively. Binding of FLAG-tagged UbV was detected using anti-FLAG-HRP antibody at 1:5000 dilution. Competitive ELISA IC50 values were derived from the binding curves and corresponded to the concertation of the tested Skp1-F-box complexes at which 50% of the UbV binding was observed.

Cell-based assays.

On day 0, 6-well plates were seeded with 4 × 105 HEK293T cells. On day 1, cells were transfected with 2 μg of plasmid DNA to express FLAG-UbV using the X-tremeGENE transfection reagent according to manufacturer’s protocol. For Snai1 stability assays, the cells were simultaneously transfected with 0.1 μg of plasmid DNA to express HA-tagged Snai1. On day 3, cycloheximide (100 μg/ml for p27 stability assays or 20 μg/ml for Snai1 stability assays) was added for 0–6 hours. Cells were lysed in lysis buffer (Cell Signalling 9803) and cell lysates were subjected to western blot analysis.

F-box alignment

The alignment of F-box domains was performed using Cn3D (Wang et al., 2000). An initial structure based model of the F-box domain was performed using the F-box structures from Fbl3 (PDB:4IGJ), Fbw1 (PDB:1P22), Skp2 (PDB:2ASS), Fbw7 (PDB:2OVP), Cdc4 (PDB:3MKS), and Fbl10 (PDB:5JH5). All other human F-box sequences were then imported into Cn3D and aligned using the “block align single” function. Further refinement of the alignment was done manually. Conservation was calculated using AL2C0 1.0(Pei and Grishin, 2001) in Chimera (Pettersen et al., 2004).

Crystallization and structure determination

The UbV.Fbl11.1-Skp1-Fbl11 complex was crystallized in sitting drops at 20 °C by mixing 0.2 μL protein complex (17 mg/mL) with 0.2 μL mother liquor (0.1 M PCTP buffer pH 6 and 25% w/v PEG 1500). Crystals were transferred to a cryoprotectant solution (mother liquor with 20% ethylene Glycol) and flash frozen in liquid nitrogen. A single crystal dataset was collected at −180 °C on a home-source consisting of a Rigaku MicroMax-007 HF rotating anode generator coupled to a Rigaku Saturn 944 HG CCD detector. Data was processed by HKL-2000 (Otwinowski et al., 1997). The structure was solved by molecular replacement using Phaser(McCoy et al., 2007) and search models of Ub with the 5 C-terminal residues removed (PDB:1UBQ), Skp1 (PDB:5IBK) and Fbl10 residues 1066–1104 (PDB:5JH5). The structure was refined by PHENIX (Adams et al., 2010) using TLS parameters (Painter and Merritt, 2006) and manual building in Coot (Emsley and Cowtan, 2004).

The Ubv.Fbl10.1-Skp1-Fbl10 complex was crystallized in hanging drops at 20 °C by mixing 1 μL protein complex (19 mg/mL) with 1 μL mother liquor (0.1 M Hepes buffer pH 7.5, 0.2 M CaCl2 and 18% w/v PEG 6000). Crystals were transferred to a cryoprotectant solution (mother liquor with 20% ethylene glycol) and flash frozen in liquid nitrogen. A single crystal dataset was collected at −180 °C on a PILATUS 6MF detector at station 24-ID-C, NE CAT beamline, Advanced Photon Source (APS) and processed using HKL-2000(Otwinowski et al., 1997). The structure was solved by molecular replacement using Phaser (McCoy et al., 2007) and search models of ubiquitin with the 5 C-terminal residues removed (PDB:1UBQ) and Skp1-Fbl11 from our UbV.Fbl11.1-Skp1-Fbl11 structure. The structure was refined by PHENIX(Adams et al., 2010) using TLS parameters (Painter and Merritt, 2006) and manual building in Coot (Emsley and Cowtan, 2004).

The Ubv.Fbl10.1-Skp1-Fbl11 complex was crystallized in hanging drops at 20 °C by mixing 1 μL protein complex (18 mg/mL) with 1 μL mother liquor (0.1 M Malic Acid pH 4.5, 0.15 M NaCl, 27% w/v PEG 3350). Crystals were transferred to a cryoprotectant solution (mother liquor with 20% ethylene glycol) and flash frozen in liquid nitrogen. A single crystal dataset was collected at −180 °C on an EIGER 16M detector at station 24-ID-C, NE CAT beamline, Advanced Photon Source (APS) and processed using XDS (Kabsch, 2010). The structure was solved by molecular replacement using Phaser (McCoy et al., 2007) and search models of Ub with the 5 C-terminal residues removed (PDB:1UBQ) (Vijay-Kumar et al., 1987) and Skp1-Fbl10 from our Ubv.Fbl10.1-Skp1-Fbl10 structure (PDB: 1JH5)(Wong et al., 2016). The structure was refined by PHENIX (Adams et al., 2010) and manual building in Coot (Emsley and Cowtan, 2004).

Interactions for structures were analyzed manually using The PyMOL Molecular Graphics System (Schrodinger, LLC) and using the protein interfaces, surfaces and assemblies (PISA) tool (Krissinel and Henrick, 2007). Alignments were performed using Coot. Rotation and translation of structures were analyzed by the Pymol script ‘RotationAxis’ available on the Pymol wiki webpage. Structure representations were performed using the PyMOL molecular graphics system.

Isothermal Calorimetry

Calorimetric titrations were performed on a Malvern MicroCal Auto-iTC200 (SBC Facility at The Hospital for Sick Children) at 25 °C. Protein samples were dialyzed in 300 mM NaCl, 50 mM Hepes pH 7.5, 5% glycerol, 4 mM β-mercaptoethanol. To measure UbV.L11.1 binding to Skp1-Fbl11, 150 μM UbV.L11.1 in the syringe was titrated into 20 μM Skp1-Fbl11 complex in the lower cell. To measure binding between UbV.L10.1 and Skp1-Fbl10 or Skp1-Fbl11, 167 μM or 150 μM UbV.L10.1 in the syringe was titrated into 20 μM Skp1-Fbl10 or Skp1-Fbl11 in the lower cell. Sixteen 2.49 μL injections were performed with an interval of 120 seconds.

Quantification and Statistical Analysis

Competitive ELISA assays

IC50 values were calculated by fitting the obtained binding curves to four parameter logistic nonlinear regression model using GraphPad Prism software. Data represent mean ± SEM of four binding curves.

Isothermal calorimetry assays

Experiments were carried out in triplicate. Analysis was performed by nonlinear curve-fitting of the corrected data to a model with one site using ORIGIN software. Reported values represent the mean ± SD.

Data and Software availability

Accession Numbers

The atomic coordinates and structure factors for UbV.Fbl11.1-Skp1-Fbl11, UbV.Fbl10.1-Skp1-Fbl10 and UbV.Fbl10.1-Skp1-Fbl11 complexes have been deposited to the RCSB PDB (www.rcsb.org) with the PDB ID codes: 6BYH, 6BVA and 6C16, respectively.

Supplementary Material

1
2

Highlights.

Ubiquitin variant (UbV) library designed to target specific F-box proteins

Extended and varied UbV ß1-ß2 loop to improve specificity

Selective UbV inhibitors generated against 17 F-box proteins using phage display

3 crystal structures show common UbV binding mode and ß1-ß2 loop contacts with F-box

Acknowledgements

This work was supported by Canadian Institutes of Health Research (CIHR) grants FDN 143277 to F.S., MOP-136956 to S.S., a CIHR postdoctoral fellowship to M.G. and an OSOTF award to N.M. This work is based upon research conducted at the Northeastern Collaborative Access Team beamlines, which are funded by the National Institute of General Medical Sciences from the National Institutes of Health (P41 GM103403). The Pilatus 6M detector on 24-ID-C beam line is funded by a NIHORIP HEI grant (S10 RR029205). This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02–06CH11357.

Footnotes

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

Declaration of Interests

The authors declare no competing interests.

References

  1. Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung LW, Kapral GJ, Grosse-Kunstleve RW, et al. (2010). PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta crystallographica. Section D, Biological crystallography 66, 213–221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aghajan M, Jonai N, Flick K, Fu F, Luo M, Cai X, Ouni I, Pierce N, Tang X, Lomenick B, et al. (2010). Chemical genetics screen for enhancers of rapamycin identifies a specific inhibitor of an SCF family E3 ubiquitin ligase. Nat Biotechnol 28, 738–742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Angers S, Li T, Yi X, Maccoss MJ, Moon RT & Zheng N (2006). Molecular architecture and assembly of the DDB1-CUL4A ubiquitin ligase machinery. Nature 443, 590–593. [DOI] [PubMed] [Google Scholar]
  4. Bhowmick P, Pancsa R, Guharoy M & Tompa P (2013). Functional diversity and structural disorder in the human ubiquitination pathway. PLoS One 8, e65443. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brown NG, Vanderlinden R, Watson ER, Weissmann F, Ordureau A, Wu KP, Zhang W, Yu S, Mercredi PY, Harrison JS, et al. (2016). Dual RING E3 Architectures Regulate Multiubiquitination and Ubiquitin Chain Elongation by APC/C. Cell 165, 1440–1453. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bulatov E & Ciulli A (2015). Targeting Cullin-RING E3 ubiquitin ligases for drug discovery: structure, assembly and small-molecule modulation. Biochem J 467, 365–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chan CH, Morrow JK, Li CF, Gao Y, Jin G, Moten A, Stagg LJ, Ladbury JE, Cai Z, Xu D, et al. (2013). Pharmacological inactivation of Skp2 SCF ubiquitin ligase restricts cancer stem cell traits and cancer progression. Cell 154, 556–568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen BB, Coon TA, Glasser JR, Mcverry BJ, Zhao J, Zhao Y, Zou C, Ellis B, Sciurba FC, Zhang Y, et al. (2013). A combinatorial F box protein directed pathway controls TRAF adaptor stability to regulate inflammation. Nat Immunol 14, 470–479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Crooks GE, Hon G, Chandonia JM & Brenner SE (2004). WebLogo: a sequence logo generator. Genome Res 14, 1188–1190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Emsley P & Cowtan K (2004). Coot: model-building tools for molecular graphics. Acta crystallographica. Section D, Biological crystallography 60, 2126–2132. [DOI] [PubMed] [Google Scholar]
  11. Ernst A, Avvakumov G, Tong J, Fan Y, Zhao Y, Alberts P, Persaud A, Walker JR, Neculai AM, Neculai D, et al. (2013). A strategy for modulation of enzymes in the ubiquitin system. Science 339, 590–595. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fellouse FA, Sidhu SS(2007). Making antibodies in bacteria In: HOWARD GC, KASER MS (ed.) Making and Using Antibodies. Boca Raton, FL: CRC Press. [Google Scholar]
  13. Gabrielsen M, Buetow L, Nakasone MA, Ahmed SF, Sibbet GJ, Smith BO, Zhang W, Sidhu SS & Huang DT (2017). A General Strategy for Discovery of Inhibitors and Activators of RING and U-box E3 Ligases with Ubiquitin Variants. Mol Cell 68, 456–470 e410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gorelik M, Orlicky S, Sartori MA, Tang X, Marcon E, Kurinov I, Greenblatt JF, Tyers M, Moffat J, Sicheri F, et al. (2016). Inhibition of SCF ubiquitin ligases by engineered ubiquitin variants that target the Cul1 binding site on the Skp1-F-box interface. Proc Natl Acad Sci U S A 113, 3527–3532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hoffmann A, Kovermann M, Lilie H, Fiedler M, Balbach J, Rudolph R & Pfeifer S (2012). New binding mode to TNF-alpha revealed by ubiquitin-based artificial binding protein. PLoS One 7, e31298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Jin Y, Shenoy AK, Doernberg S, Chen H, Luo H, Shen H, Lin T, Tarrash M, Cai Q, Hu X, et al. (2015). FBXO11 promotes ubiquitination of the Snail family of transcription factors in cancer progression and epidermal development. Cancer Lett 362, 70–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kabsch W (2010). Xds. Acta Crystallogr D Biol Crystallogr 66, 125–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Krissinel E & Henrick K (2007). Inference of macromolecular assemblies from crystalline state. Journal of molecular biology 372, 774–797. [DOI] [PubMed] [Google Scholar]
  19. Leung I, Jarvik N & Sidhu SS (2017). A Highly Diverse and Functional Naive Ubiquitin Variant Library for Generation of Intracellular Affinity Reagents. J Mol Biol 429, 115–127. [DOI] [PubMed] [Google Scholar]
  20. Lorey S, Fiedler E, Kunert A, Nerkamp J, Lange C, Fiedler M, Bosse-Doenecke E, Meysing M, Gloser M, Rundfeldt C, et al. (2014). Novel ubiquitin-derived high affinity binding proteins with tumor targeting properties. J Biol Chem 289, 8493–8507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Manczyk N, Yates BP, Veggiani G, Ernst A, Sicheri F & Sidhu SS (2017). Structural and functional characterization of a ubiquitin variant engineered for tight and specific binding to an alpha-helical ubiquitin interacting motif. Protein Sci 26, 1060–1069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mccoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC & Read RJ (2007). Phaser crystallographic software. Journal of applied crystallography 40, 658–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Nangle S, Xing W & Zheng N (2013). Crystal structure of mammalian cryptochrome in complex with a small molecule competitor of its ubiquitin ligase. Cell Res 23, 1417–1419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Orlicky S, Tang X, Neduva V, Elowe N, Brown ED, Sicheri F & Tyers M (2010). An allosteric inhibitor of substrate recognition by the SCF(Cdc4) ubiquitin ligase. Nat Biotechnol 28, 733–737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Otwinowski Z, Minor W (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods in enzymology 276, 307–326. [DOI] [PubMed] [Google Scholar]
  26. Painter J & Merritt EA (2006). Optimal description of a protein structure in terms of multiple groups undergoing TLS motion. Acta crystallographica. Section D, Biological crystallography 62, 439–450. [DOI] [PubMed] [Google Scholar]
  27. Pei J & Grishin NV (2001). AL2CO: calculation of positional conservation in a protein sequence alignment. Bioinformatics 17, 700–712. [DOI] [PubMed] [Google Scholar]
  28. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC & Ferrin TE (2004). UCSF Chimera--a visualization system for exploratory research and analysis. J Comput Chem 25, 1605–1612. [DOI] [PubMed] [Google Scholar]
  29. Reitsma JM, Liu X, Reichermeier KM, Moradian A, Sweredoski MJ, Hess S & Deshaies RJ (2017). Composition and Regulation of the Cellular Repertoire of SCF Ubiquitin Ligases. Cell 171, 1326–1339 e1314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Schulman BA, Carrano AC, Jeffrey PD, Bowen Z, Kinnucan ER, Finnin MS, Elledge SJ, Harper JW, Pagano M & Pavletich NP (2000). Insights into SCF ubiquitin ligases from the structure of the Skp1-Skp2 complex. Nature 408, 381–386. [DOI] [PubMed] [Google Scholar]
  31. Skaar JR, Pagan JK & Pagano M (2014). SCF ubiquitin ligase-targeted therapies. Nat Rev Drug Discov 13, 889–903. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Stogios PJ, Downs GS, Jauhal JJ, Nandra SK & Prive GG (2005). Sequence and structural analysis of BTB domain proteins. Genome Biol 6, R82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Vijay-Kumar S, Bugg CE & Cook WJ (1987). Structure of ubiquitin refined at 1.8 A resolution. J Mol Biol 194, 531–544. [DOI] [PubMed] [Google Scholar]
  34. Wang Y, Geer LY, Chappey C, Kans JA & Bryant SH (2000). Cn3D: sequence and structure views for Entrez. Trends Biochem Sci 25, 300–302. [DOI] [PubMed] [Google Scholar]
  35. Wong SJ, Gearhart MD, Taylor AB, Nanyes DR, Ha DJ, Robinson AK, Artigas JA, Lee OJ, Demeler B, Hart PJ, et al. (2016). KDM2B Recruitment of the Polycomb Group Complex, PRC1.1, Requires Cooperation between PCGF1 and BCORL1. Structure 24, 1795–1801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Wu L, Grigoryan AV, Li Y, Hao B, Pagano M & Cardozo TJ (2012). Specific small molecule inhibitors of Skp2-mediated p27 degradation. Chem Biol 19, 1515–1524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zhang W, Wu KP, Sartori MA, Kamadurai HB, Ordureau A, Jiang C, Mercredi PY, Murchie R, Hu J, Persaud A, et al. (2016). System-Wide Modulation of HECT E3 Ligases with Selective Ubiquitin Variant Probes. Mol Cell 62, 121–136. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Zhang Y, Zhou L, Rouge L, Phillips AH, Lam C, Liu P, Sandoval W, Helgason E, Murray JM, Wertz IE, et al. (2013). Conformational stabilization of ubiquitin yields potent and selective inhibitors of USP7. Nat Chem Biol 9, 51–58. [DOI] [PubMed] [Google Scholar]
  39. Zheng H, Shen M, Zha YL, Li W, Wei Y, Blanco MA, Ren G, Zhou T, Storz P, Wang HY, et al. (2014). PKD1 phosphorylation-dependent degradation of SNAIL by SCF-FBXO11 regulates epithelial-mesenchymal transition and metastasis. Cancer Cell 26, 358–373. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Zheng N, Schulman BA, Song L, Miller JJ, Jeffrey PD, Wang P, Chu C, Koepp DM, Elledge SJ, Pagano M, et al. (2002). Structure of the Cul1-Rbx1-Skp1-F boxSkp2 SCF ubiquitin ligase complex. Nature 416, 703–709. [DOI] [PubMed] [Google Scholar]
  41. Zheng N, Zhou Q, Wang Z & Wei W (2016). Recent advances in SCF ubiquitin ligase complex: Clinical implications. Biochim Biophys Acta 1866, 12–22. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1015708-supplement-1.pdf (7.8MB, pdf)
2
NIHMS1015708-supplement-2.docx (34.2KB, docx)

Data Availability Statement

Accession Numbers

The atomic coordinates and structure factors for UbV.Fbl11.1-Skp1-Fbl11, UbV.Fbl10.1-Skp1-Fbl10 and UbV.Fbl10.1-Skp1-Fbl11 complexes have been deposited to the RCSB PDB (www.rcsb.org) with the PDB ID codes: 6BYH, 6BVA and 6C16, respectively.

RESOURCES