Structural Studies of βTrCP Reveal Plasticity in Binding Modes of Consensus and Nonconsensus Degrons

Abstract

The E3 ligase βTrCP regulates a significant number of important cytosolic proteins by recognizing and binding to a “DSGXXS” consensus phosphodegron sequence, resulting in the ubiquitination and degradation of target proteins. While many of the substrates of βTrCP have strong disease links, there is high-resolution structural data available for just one of these proteins in complex with βTrCP. Here, we describe the development of a robust crystallographic system for βTrCP and report high-resolution crystal structures for βTrCP in complex with degrons from five new targets, encompassing the important cancer proteins, WEE1, claspin, ATF4, PDCD4, and IκBα. Interestingly, these structures reveal the molecular basis by which βTrCP can recognize and bind both consensus and nonconsensus degron peptides and reveal an overall general plasticity in degron binding mode. We also provide a biochemical assessment of the binding affinities of these peptides for βTrCP, adding further insight into the molecular interactions observed in the crystal structures. Finally, computational analyses of the βTrCP complexes identify opportunities for potential molecular glue approaches.

Introduction

The E3 ligase βTrCP (beta transducin repeat-containing protein, also referred to as F-box/WD repeat-containing protein 1A) regulates a diverse array of important cellular proteins by recognizing and binding to a doubly phosphorylated “D_pSGXX_pS” consensus phosphodegron motif on target proteins (where “_pS” denotes phosphoserine). These substrate proteins are then ubiquitinated and degraded via a proteasome system. Many of the substrate proteins recognized by βTrCP regulate key cellular process such as growth, , survival, , and apoptosis, , and as a result, βTrCP represents a potentially important yet underexplored (cancer) drug target, with both the design of small molecules able to inhibit the interaction of βTrCP with target proteins, or compounds able to stabilize βTrCP–target interactions (i.e., “molecular glues”) representing interesting avenues of therapeutic exploration. , Such efforts would be aided significantly by structural characterization of βTrCP–substrate complexes, yet high-resolution crystallographic data are available for just one such complexβTrCP–β-catenin. − Here, we report the development of a robust crystallographic system able to generate high-resolution structures of βTrCP in complex with target protein degrons. In total, we report new βTrCP-degron crystal structures for five disease-relevant proteins encompassing WEE1 (wee1-like protein kinase), claspin, ATF4 (activating transcription factor 4), PDCD4 (programmed cell death protein 4), and IκBα (NFκB inhibitor α). These structures cover both canonical and noncanonical degrons, providing a molecular view as to how βTrCP recognizes and binds degrons that deviate from the “D_pSGXX_pS” consensus. This structural data is augmented with affinity data generated for these peptides using a biochemical FRET-based probe displacement binding assay. Finally, we provide a computational analysis of the complexes with a view to assessing whether any of the complexes would be theoretically suitable for a therapeutic “molecular glue” approach.

Results

We were interested in structurally characterizing βTrCP in complex with substrate protein degrons beyond the reported βTrCP–β-catenin complex , and were particularly interested in βTrCP substrates with strong disease links. While a number of βTrCP–β-catenin crystal structures are available, as mentioned, , we were unable to translate these structures into a system capable of structurally characterizing degrons beyond β-catenin. We thus set out to develop a βTrCP crystal system suitable for our needs. Since the E3 ligase component of the available βTrCP crystal structures is composed of a heterodimeric complex formed of βTrCP and SKP1, we sought to simplify the system by expressing and crystallizing βTrCP in the absence of SKP1. The substrate-recognition domain of βTrCP is composed of a WD40-repeat containing β-propeller fold encompassing, approximately, residues 260–548, with the 100 or so amino acids N-terminal to this domain responsible for binding to SKP1 (Figure ). We thus designed constructs with varying truncations N-terminal to the β-propeller domain. Constructs covering residue range 255–548, 248–548, and 245–548 of βTrCP, with either N-terminal 6xHis-MBP-TEV-, 8xHis-ZZ-TEV-, or 10xHis-TEV-βTrCP expression and purification modules, or a C-terminal -TEV-10xHis module did not result in the expression of soluble protein. However, less drastic truncation of the SKP1-binding region of βTrCP proved successful, with a construct covering residues 186–548 of βTrCP, with an N-terminal 8xHis-ZZ-TEV- expression and purification module, yielding the expression of soluble protein (Figure ). Following purification, however, the resultant protein had very low solubility, and we thus designed a final construct with two surface leucines (exposed by the absence of the N-terminal region of βTrCP) changed to glutamates. Pleasingly, this construct, composed of 8xHis-ZZ-TEV-βTrCP-186–548^{[L188E,L192E]}, resulted in high-yielding protein expression as well as resultant purified protein with excellent solubility (∼15 mg mL^–1).

1.

Crystal structure of the βTrCP–SKP1–β-catenin complex (PDB entry 1P22,). Protein engineering truncation sites of βTrCP explored in this current work are labeled and shown in blue.

We next attempted to investigate whether our βTrCP protein construct would be suitable for characterizing canonical degron peptides by X-ray crystallography. For this, we chose phosphodegron peptides from IκBα and claspinboth proteins known to play a role in cancer. − Pleasingly, cocrystallization of our βTrCP construct with these peptideseach with both consensus serines phosphorylated (see Table for full details of peptides studied)resulted in crystal structures for both complexes (resolution range: 1.16–1.35 Å) (Figure A,B) with clear electron density for the two phosphodegron peptides (Figure S1). Analysis of the binding mode of these phospho-peptides reveals a number of shared features and in line with that reported for the βTrCP–β-catenin complex (Figure S2), with both phosphoserines forming multiple key polar contacts to the side chains of Y271, R285, S309, and S325 (N-terminal phosphoserines) (Figure C) and to the side chains of S448 and R431, and the backbone NH of G432 (C-terminal phosphoserines) (Figure D). In addition, the side chains of the first aspartic acid of the “D_pSGXX_pS” motifs form shared key ionic and polar contacts to the side chains of R474 and Y488 (Figure A,B), with the side chain of N394 of βTrCP forming a hydrogen bond donor/acceptor pair with the backbone amide of the fifth residue of the degron peptides (Figure E).

1. Details of βTrCP Phosphodegron Peptide Sequences Used in This Work and Summary of Key Data.

βTrCP substrate	Phosphodegron peptide	Degron type	Res. (Å)	PDB entry	K i vs βTrCP (nM)
Claspin	SPSDpSGQGpSYET	Canonical	1.35	9TDZ	97.35 ± 26.75
IκBα	DRHDpSGLDpSMKD	Canonical	1.16	9T9W	26.26 ± 6.24
WEE1	TGEDpSAFQEPDS	Noncanonical	1.68	9TG7	566.53 ± 116.48
PDCD4	SSRDpSGRGD pSVSD	Semicanonical	1.22	9TES	6.10 ± 1.07
ATF4	SDNDpSGICMSPES	Semicanonical	2.00	9TFU	3.62 ± 1.14

Core degron is shown in bold. _pS denotes phosphoserine residue. K _i values were determined using a FRET-based probe displacement assay. See Supporting Information for further details. K _i values are the average of two repeat experiments, with standard deviations shown.

2.

Crystal structures of βTrCP in complex with canonical phosphodegron peptides from claspin (A) and IκBα (B). In all panels, carbons and residue labels are colored brown for claspin, green for IκBα, and gray for βTrCP. (C) Conserved ionic interactions formed from the N-terminal phosphoserine residues of claspin and IκBα to βTrCP. (D) Conserved ionic interactions formed from the C-terminal phosphoserine residues of claspin and IκBα to βTrCP. (E) Conserved hydrogen-bonding interactions from N394 of βTrCP to claspin and IκBα. Numbers in red circles denote the relative position of specific residues of claspin or IκBα within the degron consensus sequence. In panels C–E, hydrogen bonds are colored by complex: brown for claspin, green for IκBα. In panels A and B, all atoms of the phosphodegron peptide present in the atomic model are shown. Several terminal residues of the phosphodegron peptides are not resolved in electron density and have therefore not been included in the final model.

While βTrCP is reported to bind to doubly phosphorylated degrons bearing the “D_pSGXX_pS” consensus sequence, there are also reports that βTrCP can recognize and degrade (via the ubiquitin–proteasome system) degrons that deviate from this consensus, either moderately, as seen for the semicanonical degrons for PDCD4 and ATF4 of “D_pSGRGD_pS” or “D_pSGICM”, respectively, or significantly, as represented by a reported noncanonical degron of WEE1 of “D_pSAFQE”. Pleasingly, cocrystallization attempts with phosphodegron peptides for all three semi- or noncanonical phosphodegron peptides (see Table for full details of peptides studied) resulted in crystal structures for all complexes (resolution range 1.22–2 Å), with clear electron density for all three degron peptides (Figure S1). While all three phosphodegrons share the N-terminal “DpS” motifand indeed these residues are seen to form all interactions present in equivalent canonical degron complexes (Figure A–C)the C-terminal regions of these degrons adopt quite varied binding modes, presumably a consequence of their deviation from the canonical consensus degron (Figures D–F). Notable observations include the fact that the additional methyl group of the alanine at the third position of the WEE1 degron does not seem to perturb the binding mode of the peptide (Figure S3), with the glutamate residing at the sixth position of the WEE1 degron appearing to accurately mimic the interactions of the phosphoserine residue that sits in this position for all of the canonical consensus βTrCP–degron peptides (Figure D). Similarly, the C-terminal phosphoserine of the PDCD4 degron peptide adopts a binding mode near-identical to that of the C-terminal consensus phosphoserine, despite being one residue further from the start of the degron sequence. As a consequence, this results in the largest shift in backbone of all the degrons studied here, with a loss of one of the two hydrogen bonds to the side chain of N394 seen in all other βTrCP–degron structures (Figure E). Interestingly, the ATF4 degron peptide, which, as the PDCD4 degron peptide, differs from the canonical consensus degron sequence by lacking a phosphoserine residue at the sixth position of the degron (bearing a methionine at the sixth position instead), appears to account for the loss of polar contacts from the typical/canonical phosphoserine at this position by burying the methionine side chain at the interface between the degron peptide and βTrCP (Figure F). Interestingly, a comparison of all five complexes reported here shows that, despite sharing conserved interactions, there is a surprising degree of plasticity in the binding modes of the phosphodegron peptides, both with respect to the positions of the peptide backbones (Figure A) and with respect to specific molecular interactions formed (Figure B).

3.

Crystal structures of βTrCP bound by nonconsensus phosphodegron peptides from WEE1 (A), PDCD4 (B), and ATF4 (C). In all panels, carbons and residue labels are colored yellow for WEE1, magenta for PDCD4, cyan for ATF4, and gray for βTrCP. (D) Overlay of βTrCP in complex with IκBα (green carbons) and WEE1 (yellow carbons), highlighting ionic contacts made by E57 of WEE1 to R431, G432, and S448 of βTrCP, which mimics the ionic contacts formed from the C-terminal phosphoserine residue _pS36 (at the 6th relative position within the degron) of IκBα (green carbons) to βTrCP. (E) Overlay of βTrCP in complex with IκBα (green carbons) and PDCD4 (magenta carbons), highlighting the loss of conserved contacts from N394 of βTrCP to the backbone of the 5th degron residue (denoted by number 5 in red circle), caused by a shift of the backbone of the PDCD4 peptide (indicated by black arrow). (F) Overlay of βTrCP in complex with IκBα (green carbons) and ATF4 (cyan carbons), highlighting the absence of ionic contacts from ATF4 to R431, G432, and S448 of βTrCP. Numbers in red circles denote the relative position of specific residues of phosphopeptides within the degron consensus sequence. In panels A–C, all atoms of the phosphodegron peptide present in the atomic model are shown. Several terminal residues of the phosphodegron peptides are not resolved in electron density and have therefore not been included in the final model.

4.

Overlay of all five βTrCP–phosphodegron crystal structures reported in this work, highlighting the plasticity in both the backbone position (A) and side-chain position (B) of the bound phosphodegron peptides. βTrCP is shown in gray and phosphodegrons distinguished by carbon color (see the legend for details). Both panels are in the same orientation.

In order to further understand the molecular interactions between the phosphodegron peptides and βTrCP observed in the crystal structures, we determined K _i values for all five peptides listed in Table , including a doubly phosphorylated β-catenin peptide as a benchmark, as this represents one of the best-characterized substrates of βTrCP (Figure and Table ). The K _i value of ∼1 nM determined for the interaction of β-catenin with βTrCP is in line with literature reports, with the remaining five peptides yielding K _i values ranging from single-digit nM (with K _i values for ATF4 and PDCD4 determined to be 3.62 nM and 6.10 nM, respectively) to over 0.5 μM (the K _i for WEE1 was determined to be 566.53 nM). The K _i for WEE1, being the weakest of the six peptides studied here, is perhaps not surprising, considering that this sequence deviates the most from the consensus sequence. Slightly less expectedly, the two lowest K _i values (excluding that of β-catenin) are observed for the two semicanonical sequences (PDCD4 and ATF4), rather than for the canonical phosphodegron peptides claspin (K _i = 97.35 nM) and IκBα (K _i = 26.26 nM). Further, we noted with interest that the tightest binder of the five peptides of focus in this paper, ATF4, is a monophosphorylated peptide (phosphorylated on the N-terminal degron serine), suggesting that the interactions (and therefore affinity) created by the C-terminal phosphoserine residue can perhaps, in some cases at least, be compensated for with appropriate hydrophobic interactions, such as the buried side chain of M223 of ATF4 (see Figure C,F).

5.

Determination of K _i values for all phosphodegron peptides shown in Table for βTrCP using a biochemical FRET-based probe displacement assay. K _i values and standard deviations derived from these plots are shown in Table , along with the sequences of the peptides. The β-catenin peptide is included as a benchmark, with a K _i value of 0.94 ± 0.20 nM. This is a doubly phosphorylated degron peptide with the following sequence: CDRKAAVSHWQQQSYLD_pSGIH_pSGATTTAPSLSG. Further details of the assay can be found in the Supporting Information.

Finally, we noted with interest a report describing the discovery of a small molecule able to enhance the native interaction between (i.e., “glue”) βTrCP and substrate β-catenin. Due to this, we analyzed all of the βTrCP–substrate complexes reported here with SiteMap, a computational tool for identifying potentially ligandable or druggable pockets within proteins or protein complexes. Interestingly, though perhaps disappointingly, this analysis showed that four of the five degrons studied here almost completely fill the phosphodegron binding site on βTrCP, leaving no potential scope for artificially gluing these substrates to βTrCP for a therapeutic effect. However, the complex between βTrCP and claspin reveals the presence of small hydrophobic pocket at the interface between the two proteins, which seems to be a consequence of the lack of side chain of the glycine at the fifth relative position of claspin’s degron (see Table and Figure ). While no small molecule hit finding was initiated, based on this analysis, it seems that there is potential scope for enhancing the affinity of claspin to βTrCP via a hypothetical molecular glue small molecule. Interestingly, this small pocket is within a region similar to where the above-mentioned βTrCP–β-catenin molecular glues bind (Figure S4), and indeed inspection of the βTrCP–claspin crystal structure clearly shows a molecule of ethylene glycol bound to the small pocket identified by SiteMap (Figure ), indicating that the pocket within the βTrCP–claspin complex is amenable to binding by small organic molecules.

6.

Surface representation of the βTrCP–claspin complex revealing the presence of a small pocket within the degron binding pocket at the βTrCP–degron interface. This small pocket is occupied by an ethylene glycol molecule (magenta carbons). βTrCP is shown as a green surface and claspin as a yellow surface.

Conclusions

We have reported here the development of a crystallographic system suitable for easily characterizing βTrCP in complex with substrate phosphodegron peptides, including both canonical and noncanonical degrons. Although these structures highlight key common interactions between the complexes, they also reveal a general plasticity in the degron binding mode. Overall, these structures enhance our understanding of the structural basis by which βTrCP recognizes and thereby regulates key cellular proteins, with biochemical affinity data adding further insight into the molecular interactions observed structurally. Finally, the structures we report were analyzed for suitability for potential therapeutic small molecule molecular glue approaches, with the βTrCP–claspin complex appearing promising within this context. Indeed, recent reports have highlighted the therapeutic potential of rationally designed molecular glues, both within the context of βTrCP but also more generally. ,

Structure factors and atomic models for all five crystal structures reported in this manuscript have been deposited in the Protein Data Bank and will be released upon publication. Accession codes can be found in Table 1.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.5c01007.

• All materials and methods and additional figures (PDF)

∥.

G.W.C. and H.M. contributed equally to this work. All authors contributed to the preparation of this manuscript. All authors except G.W. are or were employees of AstraZeneca at the time this work was conducted and may hold shares for AstraZeneca.

The authors declare no competing financial interest.