GH (growth hormone) binding to the GHR (GH receptor) triggers essential signalling pathways that promote growth and metabolic regulation. The sensitivity of the cells to GH is mainly controlled by the endocytosis of the receptor via βTrCP (β-transducin repeat-containing protein). In the present study, we show that βTrCP interacts directly via its WD40 domain with the UbE (ubiquitin-dependent endocytosis) motif in GHR, promoting GHR ubiquitination in vitro. NMR experiments demonstrated that the UbE motif is essentially unstructured, and, together with functional mapping of the UbE and βTrCP WD40 residues necessary for binding, led to a unique interaction model of βTrCP with GHR–UbE. This interaction is different from the conventional βTrCP–substrate interactions described to date. This interaction therefore represents a promising specific target to develop drugs that inhibit GHR endocytosis and increase GH sensitivity in cachexia patients.
- cancer cachexia
- growth hormone receptor (GHR)
- signal transducer and activator of transcription 5 (STAT5)
- β-transducin repeat-containing protein (βTrCP)
- ubiquitin-dependent endocytosis motif
GHR (growth hormone receptor) is a member of the class 1 cytokine receptors superfamily that exists as a dimer at the cell surface of all cells in the body . Upon binding of GH (growth hormone) to the GHR, Jak2 (Janus kinase 2) molecules associated with the cytoplasmic tails of the receptor  are activated, and signalling cascades are triggered. The Jak/STAT (signal transducer and activator of transcription) pathway is the major effector of GHR signalling by inducing IGF-1 (insulin-like growth factor 1) expression. Additionally, the MAPK (mitogen-activated protein kinase) and PI3K (phosphoinositide 3-kinase) pathways are activated . Apart from promoting longitudinal body growth, GHR signalling regulates metabolism and cellular functions such as proliferation and differentiation . Excessive GH signalling has been connected to cancer [4,5], whereas GH insensitivity has been reported in cachexia patients [6,7], which demonstrates the importance of a proper GH sensitivity control. Cachexia is a complex metabolic disorder, occurring in cancer and AIDS patients, characterized by extensive loss of muscle mass and decreased quality of life [8,9]. Cachectic patients would benefit from a treatment increasing the responsiveness of the cells to GH.
The availability of GHRs in cells is mainly determined by the rate of GHR endocytosis/degradation . GHR endocytosis requires an intact ubiquitination system  and depends on the UbE (ubiquitin-dependent endocytosis) motif present in its cytoplasmic tail . Protein ubiquitination involves the sequential action of three classes of enzymes: E1 or activating enzyme, E2 or conjugase, and the E3 or ligase that introduces substrate specificity. The F-box protein βTrCP (β-transducin repeat-containing protein), the substrate recognition subunit of the E3 ligase SCFβTrCP, is necessary for GHR endocytosis . SCFβTrCP is a heterotetrameric complex belonging to the family of RING-type ubiquitin ligases . The scaffold protein Cullin1 interacts via its N-terminal domain with Skp1, and via its C-terminus with Rbx1, a RING finger protein. Skp1 binds to βTrCP, and Rbx1 recruits transiently the E2 that adds the ubiquitin molecules to the substrate. Mammalian SCF ligases can collaborate with Cdc34 or E2s of the Ubc4/5 family in vitro, but it is unclear whether this is also true in vivo .
SCFβTrCP is involved in the ubiquitin-dependent endocytosis of receptors homologous with GHR, e.g. interferon-α receptor , erythropoietin receptor  and prolactin receptor . After ligand binding to these receptors, the serine residues of the DSGXXS motifs become phosphorylated and SCFβTrCP is recruited, mediating their ubiquitin-dependent endocytosis. Also GHR contains a DSGXXS motif. However, we demonstrated recently that the GHR-DSGXXS motif is not involved in GH-mediated, but rather in GH-independent, basal GHR endocytosis and degradation . The GHR UbE motif regulates GHR endocytosis and degradation, in both the presence and the absence of GH stimulation.
βTrCP interacts with its substrates through its WD40 domain. The crystal structure of a complex of βTrCP in complex with β-catenin was solved and revealed the characteristics of this interaction in molecular detail . Additionally, NMR studies have characterized the interaction of βTrCP with classical phosphorylated degrons DSG(X)2+nS in several substrates [21–26]. For the binding of the GHR-DSGXXS to βTrCP, the same residues in the WD40 domain are involved. On the other hand, the GHR UbE motif sequence (DDSWVEFIELD) differs substantially from the classical DSG(X)2+nS motif . We therefore predict that the GHR–UbE–βTrCP interaction is unconventional.
In the present study we found that βTrCP binds directly to the UbE motif in vitro in a manner different from that of the classical DSG(X)2+nS motif. Furthermore, SCFβTrCP mediates GHR ubiquitination in vitro. Mutations in specific residues in the WD40 domain decreased the GHR–βTrCP interaction, GHR ubiquitination in vitro and GHR endocytosis. Functional mapping of amino acid residues of the UbE motif and the WD40 domain was combined with structural NMR data, and a computational model of the interaction was constructed. This model shows a unique interaction of βTrCP with the GHR UbE motif, different from the one with canonical motifs present in all other βTrCP substrates described.
Anti-poly-His monoclonal antibody was purchased from Sigma. Rabbit anti-[GHR (B)] against amino acids 327–493 and anti-GST (glutathione transferase) antibodies have been described previously . Monoclonal anti-ubiquitin (clone FK2) antibody was purchased from Biomol and anti-FLAG (M2) antibody was from Sigma. Alexa Fluor® 680-conjugated goat anti-(mouse IgG) was from Molecular Probes/Invitrogen and IRDye800-conjugated goat anti-(rabbit IgG) was from Rockland Immunochemicals. Streptavidin beads were purchased from Pierce and streptactin beads were from GE Healthcare. Anti-FLAG (M2)–agarose beads and 3×FLAG peptide were purchased from Sigma. Human GH was a gift from Eli Lilly Research Laboratories. Glutathione (GSH) beads were purchased from GE Healthcare. Biotin–GH was prepared using a biotinylation kit purchased from Pierce. GH(his-TEV-STREP3-his)C was purified by U-Protein Express. Peptides were chemically synthesized by Pepscan: UbE (GHR amino acids 314–335), acetyl-YKPEFHSDDSWVEFIELDIDEP-amide; UbE-mutant (F327A/D331A), acetyl-YKPEFHSDDSWVEAIELAIDEP-amide, and β-catenin peptide (amino acids 17–48), with phosphorylated Ser33 and Ser37, DRKAAVSHWQQQpSYLDpSGIHSGATTTAPSLSG.
Full-length rabbit GHR cDNA in pcDNA3 has been described previously . GHR F327A mutant in pcDNA3 was generated as described previously . GHR DAGXXA (S366A/S370A) was generated with the oligonucleotides 5′-GGATGACGACGCTGGACGAACCGCCTGTTACGAACC-3′ and 5′-CCTACTGCTGCGACCTGCTTGGCGGACAATGCTTGG-3′. GHR 326EFIXXD (GHR-EFID mutant) was generated as described previously . βTrCP2 cDNA N-terminally fused with a FLAG tag cloned in pcDNA3 (FLAG–βTrCP) was a gift from Professor Tomoki Chiba (University of Tsukuba, Tsukuba-shi, Japan). The constructs Myc–Cullin, FLAG–Skp1 and T7-Rbx in pcDNA3.1 were a gift from Professor Kazuhiro Iwai (Osaka University, Osaka, Japan). Structural and biochemical studies on the interaction of βTrCP1 with classical substrates such as β-catenin and IκB (inhibitor of nuclear factor κB) identified residues in the WD40 domain of βTrCP1 involved in their interaction (Tyr271, Arg431, Arg474 and Tyr488) .
Cell culture and transfections
Culture media, fetal bovine serum and antibiotics for tissue culture were purchased from Gibco/Invitrogen. HEK (human embryonic kidney)-293 cells, expressing the tetracycline repressor (HEK-293-TR), were a gift from Dr Madelon Maurice (University Medical Centre Utrecht, Utrecht, The Netherlands), and were transfected as described previously . At 24 h after transfection, the cells were used for pull-down experiments, immunoprecipitation, Western blotting and fluorescence microscopy experiments.
Protein expression and purification
Details can be found in the Supplementary Online Data at http://www.biochemj.org/bj/453/bj4530291add.htm.
In vitro binding assays
GST–GHR-(270–318) [GST–GHR(−UbE)] or GST–GHR-(270–334) [GST–GHR(+UbE)], representing short forms of GHR C-terminal intracellular tails, were dissolved in PBS supplemented with 1% BSA and protease inhibitors (1 mM PMSF, 10 μg/ml aprotinin and 10 μg/ml leupeptin) and incubated with GSH beads for 2 h at 4°C (10 μg of GST fusion proteins were used for saturation of 25 μl of GSH beads). After extensive washing, the GSH beads coupled to GST–GHR fusion proteins were incubated with the His–SF-βTrCP2–His–Skp1ΔΔ complex for 2 h in the binding buffer (50 mM Tris/HCl, pH 8, 100 mM NaCl, 0.1% Triton X-100, 0.1% BSA, 1 mM PMSF, 10 μg/ml aprotinin and 10 μg/ml leupeptin) at 4°C. In case of competition with peptides, the beads were washed and incubated for an additional 2 h with increasing concentrations of the indicated GHR or phosphorylated β-catenin peptides (0.7, 7 and 70 μM), diluted in binding buffer. The beads were then washed and boiled in sample buffer, and the proteins were separated by SDS/PAGE and immunoblotted with anti-GST and anti-His antibodies.
In vitro ubiquitination assays
The purified Uba1 and UbcH5 were a gift from Professor Titia Sixma [NKI (Nederlands Kanker Instituut), Amsterdam, The Netherlands]. GHR ubiquitination reactions were performed at 37°C, for 1 h, in a total volume of 60 μl. The reactions contained 50 mM Tris/HCl (pH 7.4), 75 mM NaCl, 0.6 mM DTT (dithiothreitol), 5 mM MgCl2, 2 mM ATP, 2 mM NaF, 1 μg of ubiquitin, 0.15 pmol of Uba1 (E1) and 90 pmol of UbcH5b (E2). As an E3 ligase source, SCFβTrCP complexes purified from HEK-293 cells were used. Purified GST–GHR tails (50 pmol per reaction) or full-length GHR (one-seventh of the total amount of GHR purified from a 10-cm-diameter dish per reaction) were used as substrates.
Biotin–GH pull downs, immunoprecipitation, Western blotting and confocal microscopy
At 24 h after co-transfection of the indicated GHR and βTrCP constructs, biotin–GH (180 ng/ml) was added to HEK-293-TR cells for 10 min at 37°C. Subsequently, cells were washed three times with PBS, and lysed with cold lysis buffer (1% Triton X-100, 1 mM EDTA, 100 mM NaF, 1 mM Na3VO4, 1 mM PMSF, 10 μg/ml aprotinin and 10 μg/ml leupeptin, in PBS). After centrifugation at 16000 g for 5 min at 4°C, the supernatant was incubated for 1 h with streptavidin beads at 4°C. Beads were boiled and subjected to SDS/PAGE and Western blotting. Western blotting and confocal microscopy were performed as described previously .
All NMR experiments were carried out at 278 K on a Bruker Avance 600 MHz spectrometer equipped with a 5-mm triple-resonance z-shielded cryogenic probe. Spectra were processed using Bruker Topspin (Bruker BioSpin) and analysed using CcpNmr Analysis version 2.1.5 . The secondary structure of the UbE peptide was determined from a consensus of chemical shift indices of 1Hα, 13Cα and 13Cβ nuclei using the CSI (chemical shift index) software . Restraints were derived and structures were calculated in torsion angle space using CYANA 2.1 software . NMR samples contained the UbE motif peptide, acetyl-YKPEFHSDDSWVEFIELDIDEP-amide, with or without the purified βTrCP2 (His–SF-βTrCP–His–SkpIΔΔ complex). Both UbE peptide and βTrCP samples were prepared in 50 mM phosphate buffer (pH 7.2), 200 mM NaCl and 5% 2H2O. 1H chemical shifts were directly referenced to the resonance of 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt.
The following 2D (two-dimensional) experiments were performed: 2D-TOCSY  (4096×600 complex points), 2D-TRNOESY (transferred nuclear Overhauser enhancement spectroscopy) [32,33] (1024×1024 complex points), 2D heteronuclear 1H-13C HMQC (heteronuclear multiple-quantum coherence)  (centred on aliphatic signals) and 1H-13C CT-HSQC (constant time heteronuclear single-quantum coherence)  (centred on aromatic signals).
The fast exchange regime between the bound and free states required for TRNOESY NMR experiments was investigated using STD (saturation transfer difference) across a range of protein/peptide concentration ratios from 1:600 to 1:10 . Saturation transfer was achieved with on-resonance selective low power irradiation (60 db attenuation) at −2 p.p.m., and off-resonance at −10 p.p.m. for the reference spectrum. The optimal protein/peptide ratio was set at 1:350, corresponding to protein and peptide concentrations of 4.5 μM and 1.6 mM respectively. 1H-1H NOE (nuclear Overhauser effect) peaks of the UbE peptide were collected from TRNOESY spectra with 100 ms mixing time in both the absence and the presence of βTrCP (at a protein/peptide ratio of 1:350). Peak volumes were converted into distance restraints using the CYANA 2.1 internal calibration protocol. Of the 200 structures minimized in CYANA 2.1, 20 with the lowest target function values were selected as the representative ensemble and used for the molecular docking stage to βTrCP.
Modelling of βTrCP2 structure and molecular docking of UbE peptide
Ten models of βTrCP2 were derived from the homologous structure of βTrCP1 (PDB code 1P22) using the Modeller (9v1) software [37,38]. All ten models were used in the subsequent docking stage to the TRNOESY-derived structures of the UbE peptide. Molecular docking was generated using the HADDOCK server  (see the Supplementary Online Data for details). On the basis of the mutagenesis studies, the following AIRs (Ambiguous Interaction Restraints) were incorporated as active restraints in the docking protocol: Glu326, Phe327, Asp331 from the UbE peptide, and Tyr244, Arg258, Ser282, Leu284, Leu324, Lys338, Ala364, Ala365, Asn367, Arg383, Gly405, Ala407, Ser421, Glu444 and Arg447 from βTrCP2. After water-refinement, the cluster with the lowest HADDOCK score (−137.7±10.6) and smallest RMSD (root mean square deviation) from the lowest-energy structure (1.4±0.8 Å; 1 Å=0.1 nm) was chosen as the representative ensemble of the complex. Consistent with the HADDOCK score, this cluster displayed the largest buried surface area (1675.1±115.7 Å2) of all clusters. The lowest-energy structure within this cluster was chosen for the comparison with previously published βTrCP1–ligand complexes [21–26].
The UbE motif of GHR interacts directly with βTrCP
Previously, we found that the ubiquitin ligase SCFβTrCP regulates the endocytosis and degradation of the GHR . The UbE motif of GHR and the WD40 domain of βTrCP appeared to be crucial for this interaction. However, it was not formally proven whether this interaction was direct. In order to show the direct interaction, we purified GST–GHR tails from Escherichia coli, and βTrCP from Sf9 cells. βTrCP was purified in complex with Skp1, since Skp1 is required for βTrCP solubility and stability . The optimized purification procedure resulted in a stoichiometric complex His–SF-βTrCP–His–Skp1ΔΔ (Figure 1A). The purified βTrCP–Skp1 complex bound specifically to the GST–GHR tails containing the UbE motif [GST–GHR(+UbE)]. No binding was observed if the UbE motif was absent [GST–GHR(−UbE)] (Figure 1B). The specificity of the interaction was confirmed by a competition assay (Figure 1C). The binding of the purified His–SF-TrCP–His–Skp1ΔΔ to the GST–GHR(+UbE) tails was competed off by increasing concentrations of peptides containing the UbE motif sequence (Figure 1C, lanes 1–3; all lane numbering is from left to right). A peptide with a mutated UbE motif sequence did not interfere with the GHR–βTrCP interaction (Figure 1C, lanes 7–9). A peptide containing the phosphorylated DSGXXS motif sequence competed more efficiently than the UbE motif peptide (Figure 1C, lanes 4–6). These results show that the UbE motif interacts directly and specifically with βTrCP, with a lower affinity than the phosphorylated DSGXXS motif.
SCFβTrCP ubiquitinates GHR upon interaction with the UbE motif in vitro
Interaction of SCFβTrCP with the phosphorylated DSG(X)2+nS motif on its substrates results in their ubiquitination . To investigate whether SCFβTrCP is able to mediate the ubiquitination of the GHR by interacting with its UbE motif, the multi-subunit SCFβTrCP complex (E3) was immunoprecipitated from HEK-293-TR cells (Figure 1D) and, together with purified Uba1 (E1) and UbcH5b (E2), used in the in vitro ubiquitination of GST–GHR tails (±UbE motif). GST–GHR(+UbE) was clearly ubiquitinated as can be seen by several discrete bands and a smear up in the gel, detected with both anti-GST (Figure 1D, upper panel, lane 2) and anti-ubiquitin (Figure 1D, lower panel, lane 2) antibodies. When GST–GHR(−UbE) was used as substrate, or either E2 or E3 were absent from the reaction, no ubiquitination pattern was detectable (Figure 1D, upper panel, lanes 5, 3 and 4 respectively). This shows that ubiquitination of the GST–GHR tails depends on the direct interaction of SCFβTrCP with the UbE motif, confirming the binding results of Figures 1(B) and 1(C). When detected with anti-ubiquitin antibody, a smear up in the gel was detected that occurred independently of substrate addition (Figure 1D, lower panel, lane 1), most likely to be due to ubiquitination of other components of the reaction. We also performed the assay with full-length receptors pulled from HEK-293-TR cells, transiently transfected with WT (wild-type) GHR or the UbE mutant (EFID) . GHR WT was very efficiently ubiquitinated, as seen by a complete shift of GHR signal up in the gel, provided that SCFβTrCP and ATP were present (Figure 1E, compare lane 2 with lanes 4 and 5). The GHR-EFID mutant was hardly ubiquitinated (Figure 1E, lane 3). Taken together, we conclude that SCFβTrCP mediates GHR ubiquitination directly, in an UbE motif-dependent manner.
The GHR–βTrCP interaction has unconventional characteristics
We recently published that βTrCP can bind to two motifs in GHR: DSGRTS and the UbE motif . However, the βTrCP binding mode to these two motifs differed. Whereas binding to the classical DSGRTS motif required the WD40 residues Tyr244, Arg404, Arg447 and Tyr461, similarly to other βTrCP substrates , the binding to the UbE motif did not seem to require Arg404 and Tyr461. We evaluated further the differences by analysing the effect of the same βTrCP mutations in the in vitro ubiquitination of the truncated GST–GHR tails. SCFβTrCP complexes containing mutant βTrCP were purified and used in the ubiquitination reactions. Whereas the R404A and Y461A mutations did not affect the ubiquitination of GST–GHR tails (Figure 2A, lanes 4 and 6), the Y244A and R447A mutations reduced it substantially (Figure 2A, lanes 3 and 5), indicating that, of these four residues, only Arg447 and Tyr244 are involved in βTrCP interaction with the GHR UbE motif. As an extra parameter, we analysed the role of the same residues in the endocytosis of Cy3 (indocarbocyanine)–GH in cells expressing GHR together with WT and mutant FLAG–βTrCP constructs (Figure 2B). Expression of the endocytosis mutant GHR F327A with WT βTrCP was performed as a control, and resulted in inhibition of Cy3–GH uptake. In WT βTrCP, Cy3–GH was efficiently internalized, as all of the Cy3–GH appeared as small punctae inside the cells with no detectable surface labelling. The same phenotype was seen with βTrCP Y244A, R404A and Y461A mutants. In contrast, overexpressing the βTrCP R447A mutant resulted in inhibition of Cy3–GH uptake, comparable with the GHR F327A mutant. This can be explained by a dominant-negative effect where the overexpressed FLAG–βTrCP mutants compete with endogenous βTrCP for the other components of the SCF complex, causing inhibition of GHR endocytosis and degradation.
We conclude that, of the four WD40 residues needed for β-catenin and Iκβ binding  tested, only Arg447 and, to a lesser extent, Tyr244 are involved in GHR binding, GHR ubiquitination and GHR endocytosis. Interestingly, Arg404 and Tyr461 seem to not be involved. These characteristics reveal that the GHR–βTrCP interaction is unconventional. Since GHR is the only protein described to date that uses the UbE motif for βTrCP binding, we can classify the GHR–βTrCP interaction as unique.
Mapping more residues in the UbE and WD40 domain of βTrCP2 involved in the interaction
To characterize further the UbE–βTrCP interaction, we evaluated the relative contributions of each UbE amino acid residue in an in vitro binding assay by alanine scanning. The UbE mutant tails were allowed to bind to purified His–SF-βTrCP–His–Skp1ΔΔ complex. The amount of βTrCP/Skp1 bound per each GHR tail is shown and quantified in Figure 2(C). The most critical UbE residues for the interaction were Glu326, Phe327 and Asp331. These results are in agreement with previous functional studies analysing the effect of the mutations in the UbE motif in the internalization of 125I-GH  (Figure 2C, lower histogram).
To map the residues involved in the GHR–βTrCP interaction, we extended our analysis of Figure 2(A) to the complete WD40 repeat domain of βTrCP. Wu et al.  have shown that the residues contacting β-catenin are distributed along all the blades and are located mainly in three positions: the second residue of the A strand, the residue immediately preceding the A strand (A−1), and the residue immediately after the B strand (B+1). To identify UbE motif-interacting residues in the WD40 domain, we performed extensive site-directed mutagenesis of residues from the A strands and from the loops connecting strands B and C, of all seven WD40 repeats. Corresponding FLAG-tagged mutants were transfected in HEK-293-TR cells together with GHR mutated in the DSGXXS motif (DAGXXA mutant) to prevent conventional βTrCP binding to this motif . FLAG–βTrCP WT or βTrCP mutants were immunoprecipitated and the amounts of GHR were analysed. The results from this analysis are shown in Figure 2(D) and in Supplementary Figure S1 at http://www.biochemj.org/bj/453/bj4530291add.htm. In agreement with Figures 2(A) and 2(B), the data confirm the importance of Arg447 and, to a lesser extent, of Tyr244 for the interaction with the GHR UbE motif. Additional residues were identified as important for the interaction: Arg258 and Ser282 (Figure 2D), Leu284, Leu324, Lys338, Ala364, Ala365, Asn367, Arg383, Gly405, Ala407, Ser421 and Glu444 (Supplementary Figure S1). Some βTrCP mutations resulted in increased binding to GHR, which probably reflects unspecific steric effects on the interaction. The overexpression of βTrCP mutants defective in GHR binding resulted in an increase in total GHR levels: overexpressed FLAG–βTrCP mutants compete with endogenous βTrCP for the other components of the SCF complex, causing inhibition of GHR endocytosis (Figure 2D, total cell lysates). Taken together, we conclude that the UbE–βTrCP interaction depends on the UbE motif residues Glu326, Phe327 and Asp331, and involves residues in every WD40 domain blade, namely Tyr244, Arg258, Ser282, Leu284, Leu324, Lys338, Ala364, Ala365, Asn367, Arg383, Gly405, Ala407, Ser421, Glu444 and Arg447.
Transfer-NMR studies of the UbE–βTrCP interaction
NMR-based methods have been used extensively in the last few years for characterizing interactions of βTrCP with the DSG(X)2+nS motif in its substrates [21–26]. We used this technique to obtain a detailed understanding of how the unconventional UbE motif fits into the WD40 pocket. A 22-amino-acid-long UbE motif peptide was used with or without addition of purified βTrCP in STD and TRNOESY experiments. The UbE peptide is in fast exchange between the free and βTrCP-bound form as suggested by STD experiments: the selective magnetization of βTrCP resulted in saturation transfer from βTrCP to the UbE peptide at a ratio of 1:350 (protein/peptide) (results not shown). Although all of the STD peaks were attributable to the UbE peptide, no quantitative analysis of their intensities could be performed, owing to the low signal to noise ratio.
Initially, the analysis of the CSI obtained from the 13Cα, 13Cβ and 1Hα resonance assignments for the UbE peptide alone supported a random-coil secondary structure (Figure 3A). As the UbE–βTrCP interaction is a fast exchange system, we could incorporate the TRNOEs (transferred NOEs) for a meaningful structure calculation of the bound peptide. A small number of medium- and long-range NOEs, observed in both the bound and free UbE peptide forms, conferred a partially bent structure to the peptide, accessed by a bend formed between residues 11 and 16 (Trp324–Glu329 in full-length GHR). This bend is more prominent in the bound form of the peptide (Figure 3B). In agreement with the CSI analysis, the majority of the NOE restraints, for both unbound and βTrCP-bound UbE peptide, were intra-residual (Figure 3C). The random-coil secondary structure is also supported by the low percentage of dihedral angles occupying the most favoured region in the Ramachandran plot (Figure 3C). This limited the ability to find a unique fold from the structure calculation. Analysis of both structural ensembles reveals an average radius of gyration of 14.2 Å for the unbound ensemble compared with 10.1 Å for the bound ensemble. We conclude that, although the UbE motif has the tendency to be unstructured, it adopts a more compact structure when bound to βTrCP.
Computational modelling of the UbE–βTrCP interaction
On the basis of the mutagenesis data (Figure 2) and NMR experiments (Figure 3), we sought to obtain a model of the structure of the UbE–βTrCP interaction. The WD40 repeat domain of βTrCP2 is completely homologous with βTrCP1, which has been crystallized in complex with β-catenin . The ten lowest-energy structures of the bound form of the UbE peptide, obtained by TRNOESY, were then used for cross-docking against the ten best models of βTrCP2 using the HADDOCK server. Ten representative structures of the βTrCP2–UbE complex from the best-scoring HADDOCK cluster are shown in Figure 4(A). The ten selected have the lowest RMSD to the complex with the best HADDOCK score (Figure 4B). In these models, a succession of turns form a bend in the N-terminal region of the peptide between Tyr314 and Phe327, whereas the C-terminal region (residues 328–335) remains extended. The UbE–βTrCP interaction involves hydrogen bonds, hydrophobic and electrostatic (ionic) interactions, as well as aromatic interactions, as predicted with the webserver-based Protein Interaction Calculator software . A summary of the interactions formed in the ten representative complexes is shown in Figure 5 and Supplementary Table S1 at http://www.biochemj.org/bj/453/bj4530291add.htm. From the interaction models it became apparent that Asp331, Glu326, Phe327 and Trp324 are the main UbE residues stabilizing the interaction with βTrCP2. The importance of Glu326, Phe327 and Asp331 is in agreement with the results from the binding assays (Figure 2). This is expected since these residues were used as restraints during the docking procedure. The additional residue Trp324 plays a more subtle role in the binding to βTrCP2 (Figure 2).
As mentioned previously, the sequence of the UbE and DSG(X)2+nS motifs differ substantially. We therefore predicted the UbE–βTrCP2 interaction to be unique. To investigate this, we compared the βTrCP2–UbE interaction with the interactions of βTrCP1 with the classical DSG(X)2+nS motif present in other substrates (Figure 5). NMR structures of β-catenin, Vpu (viral protein unique), IκB and ATF4 (activating transcription factor 4) peptides docked on βTrCP1, available in the literature, were used for comparison . All of the ligand–βTrCP interactions are stabilized through a network of electrostatic interactions and hydrogen bonds (Figure 5, grey squares) involving all of the WD40 repeats of βTrCP. Additionally, presumed hydrophobic interactions may increase the affinity of the binding region (Figure 5, black squares). Importantly, the peptides bind βTrCP in such a way that one residue is centred into a hydrophobic binding pocket in βTrCP. In the case of β-catenin and ATF4, these residues are Gly34 and Ile221 respectively (Figure 5, black squares). In Vpu and IκB, these residues are not precisely described. In the case of UbE, Phe327 seems to play this role. In all of the examples analysed, ionic interactions and hydrogen bonds are established by phosphorylated serine residues or by negatively charged residues of βTrCP-interacting peptides. Interestingly, the GHR UbE–βTrCP interaction seems to depend on aromatic interactions formed by Phe327 and Trp324. In all ten representative structures, Phe327 is oriented towards the pore of the WD40 repeat domain. In all but two models, Trp324 is stacked between His510 and Arg258 of βTrCP and Phe327 makes hydrophobic contacts with Leu324 within 5 Å. The importance of aromatic interactions in substrate–βTrCP binding has not been documented previously. We conclude that UbE binds to βTrCP by principles similar to those used by the other βTrCP substrates, involving an extensive network of hydrogen bonds, electrostatic and hydrophobic interactions, but with additional involvement of aromatic interactions.
The interaction of the UbE motif of GHR with βTrCP is essential for its endocytosis and degradation . The present study adds crucial information for the understanding of this interaction. First, we prove formally that βTrCP binds to the UbE motif directly. Furthermore, we show that SCFβTrCP mediates GHR ubiquitination in vitro in an UbE motif-dependent manner. Our results suggest that the GHR UbE–βTrCP interaction differs from that of the classical DSG(X)2+nS motif present in β-catenin or IκB [42,43]. We therefore characterized the interaction between the UbE motif and βTrCP in molecular detail. NMR experiments revealed that a peptide containing the UbE motif is essentially unstructured, but adopts a less extended structure when bound to purified βTrCP. On the basis of the NMR and mutagenesis data, a computational model was constructed which shows that the interaction of the WD40 domain of βTrCP with the GHR UbE motif is unconventional.
Recently, we showed that GHR contains two motifs able to bind βTrCP, i.e. the UbE motif and DSGRTS sequence, which corresponds to a classical βTrCP-binding motif . In that study, we concluded that βTrCP binds to the DSGRTS sequence in a classical way, whereas the binding to the UbE motif was different. Therefore the focus of the present study was to characterize the unconventional features of the UbE–βTrCP interaction. For this purpose, all of the experiments of the present study were performed under conditions where the contribution of the DSGRTS sequence could be disregarded. We expected the interaction of βTrCP with the UbE motif to be unconventional since the sequence of the UbE motif, DDSWVEFIELD, has little resemblance to the classical motif. The UbE motif is 11 amino acids long, whereas the classical motif contains six to eight amino acids. Whereas in the classical motif, two phosphorylated serine residues are necessary for the interaction with βTrCP, the UbE motif interacts with βTrCP regardless of a serine phosphorylation event. The role of phosphorylated serine residues in the UbE motif may have been taken over by acidic residues. Also, the characteristic glycine residue of the classical motif is lacking in the UbE motif. Furthermore, except for the prolactin receptor, the UbE motif is unique since it has not been described in any other receptor or other protein so far.
In the present study, we used a 22-amino-acid-long peptide, but we cannot exclude that the length of the peptide may have an influence on its structure. In the context of the full-length GHR protein, the UbE motif is predicted to be unstructured using several secondary structure prediction tools (results not shown), which would suggest that our NMR results might be extrapolated to the full-length receptor. Physiologically, βTrCP binds a GHR dimer at the cell surface . It is not known to what extent GHR dimerization influences the structure of the UbE motif. The UbE motif is also necessary for the interaction of GHR with CHIP (C-terminus of heat-shock cognate 70-interacting protein) . It is tempting to speculate that the unstructured character of the UbE motif confers flexibility for the dynamic regulation of its interaction with different proteins.
Previous NMR studies have revealed that the structures of βTrCP-bound peptides containing DSG(X)2+nS motifs are characterized by an N-terminal loop followed by a large bend centred on the double-phosphorylated motif . Importantly, the TRNOESY data revealed that the bound form of the UbE peptide is rather flexible, or partially unstructured. Nevertheless, we obtained a model of the interaction by docking of the ensemble of TRNOESY-derived UbE ‘structures’ on the βTrCP2 structure. We focused our interaction studies on the βTrCP2 isoform since this seems to be the isoform physiologically active in GHR endocytosis . β-Catenin, Vpu, IκB and ATF4 have been studied in complex with βTrCP1 . In the best-scoring model of the UbE–βTrCP2 interaction, the bound UbE motif peptide showed a succession of turns in the N-terminus that formed a bend in the region before Phe327, and a C-terminal part that remained extended.
The bound structures of all of the peptides, including that of the UbE peptide, expose residues that will form contacts within the βTrCP-binding channel (Figure 5). In the case of the UbE motif, the main residue assuming this role is Phe327 which forms hydrophobic interactions with Tyr244, Leu324, Ala434, Ala364 and Ala365 of βTrCP (Figure 5). Leu324 and Ala434 also interact with Gly34 and Ile221 of β-catenin and ATF4 respectively, which are also inserted in the βTrCP central channel (Figure 5). Additional hydrophobic interactions stabilize the binding to βTrCP, which differ between the UbE motif and the classical substrates β-catenin and ATF4 (Figure 5). Additionally, it is well documented that the majority of the stabilizing contacts in βTrCP–substrate interactions involves phosphorylated serine residues in the substrate-interacting motif . Remarkably, in the case of the UbE motif, this role is likely to be played by the negatively charged residues Glu326 and Asp331 (Figure 5).
A unique characteristic of the UbE–βTrCP binding is the importance of aromatic interactions formed by the UbE motif residues Trp324 and Phe327, while being placed in a βTrCP hydrophobic pocket (Figure 5). Interestingly, the βTrCP2 residue Arg447, which is essential for the binding to the UbE motif, seems to be exclusively involved in a cation–π interaction with Phe327 on the UbE motif. In contrast, the equivalent residue in βTrCP1, Arg474, establishes hydrogen bonds, and ionic and hydrophobic interactions with its substrates. Additional βTrCP2 residues, Tyr244 and Arg258, are predicted to be involved in cation–π interactions, and, when mutated, also substantially impair binding to the UbE motif. These highlight the importance of cation–π interactions in the stabilization of UbE–βTrCP2 binding. In fact, cation–π interactions are now considered to be as important as hydrogen bonding, and ionic and hydrophobic interactions in stabilizing protein structures .
In the present study, we confirm that βTrCP is able to accommodate variable motifs into its WD40 domain through similar principles, involving an extensive network of hydrogen bonds and electrostatic interactions, where certain residues are placed into a hydrophobic pocket in βTrCP. Nevertheless, we found some unconventional characteristics of the UbE–βTrCP interaction when comparing with other βTrCP–substrate interactions, namely the importance of cation–π interactions. These aspects make the UbE–βTrCP interaction a promising target for drug design. Our results show a direct correlation between βTrCP binding to the UbE motif and the efficiency of GHR internalization (Figure 2). Therefore specific inhibition of this interaction would result in a block of GHR endocytosis, increase the GHR levels and, ultimately, improve the GH sensitivity of cells. This would have therapeutic value for patients with GH-insensitivity caused by reduced levels of GHR, such as cancer patients suffering from cachexia. In the present study, we have used a cellular system based on GHR overexpression that does not allow us to verify the expected increased signalling output resulting from interfering with GHR UbE–βTrCP interaction. In these cells, downstream signalling molecules (such as Jak2, STAT5 and others) are now the limiting factors for the GH signalling output. Future modelling studies will evaluate the possibility of designing compounds specifically targeting the interaction of βTrCP with UbE without affecting the interaction of βTrCP with other substrates. One possibility would be to interfere with the aromatic interactions that seem to be essential for the maintenance of the UbE–βTrCP interaction. Aromatic interactions (π–π and cation–π) are considered to be important factors in rational drug design , and have been targeted successfully [48,49]. Drugs acting through the same mechanism could inhibit UbE–βTrCP interaction by competing with/destabilizing the essential cation–π interactions.
GH signalling is involved in a plethora of physiological processes. The characterization of the UbE–βTrCP interaction, reported in the present paper, opens new possibilities for designing strategies that allow the modulation of GH sensitivity of cells.
Ana da Silva Almeida, Henry Hocking, Rolf Boelens, Agnes van Rossum and Ger Strous designed the experiments, Ana da Silva Almeida and Henry Hocking performed the experiments, and Ger Strous and Ana da Silva Almeida wrote the paper.
This research was supported by the European Network of Excellence Rubicon ‘Role of ubiquitin and ubiquitin-like modifiers in cellular regulation’ [grant number LSHG-CT-2005-018683] and the Marie Curie network ‘UbiRegulators’ [grant number MRTN-CT-2006-034555].
We thank all members of the GHR group. We thank Alex Faesen and Professor Titia Sixma for the help with the purification of βTrCP–Skp1 complex and Adrien Melquiond for critical advice on the docking section.
Abbreviations: AIR, Ambiguous Interaction Restraint; ATF4, activating transcription factor 4; CSI, chemical shift index; Cy3, indocarbocyanine; 2D, two-dimensional; GH, growth hormone; GHR, growth hormone receptor; GST, glutathione transferase; HEK, human embryonic kidney; IκB, inhibitor of nuclear factor κB; Jak, Janus kinase; NOE, nuclear Overhauser effect; RMSD, root mean square deviation; STAT, signal transducer and activator of transcription; STD, saturation transfer difference; βTrCP, β-transducin repeat-containing protein; TRNOESY, transferred nuclear Overhauser enhancement spectroscopy; UbE, ubiquitin-dependent endocytosis; Vpu, viral protein unique; WT, wild-type
- © The Authors Journal compilation © 2013 Biochemical Society