Oncogene (2007) 26, 4941–4950 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Structure of the C-terminal MA-3 domain of the tumour suppressor protein Pdcd4and characterization of its interaction with eIF4A LC Waters1, V Veverka1,MBo¨ hm2, T Schmedt2, PT Choong1, FW Muskett1, K-HKlempnauer 2 and MD Carr1 1Department of Biochemistry, University of Leicester, Leicester, UK and 2Institut fu¨r Biochemie, Westfa¨lische-Wilhelms-Universita¨t Mu¨nster, Mu¨nster, Germany Programmed cell death protein 4(Pdcd4)is a novel 2004; Bitomsky et al., 2004). The protein was initially tumour suppressor protein, which is involved in the control discovered in a screen for genes activated during of eukaryotic transcription and translation. The regula- apoptosis (Shibahara et al., 1995) and then subsequently tion of translation involves specific interactions with identified as a tumour suppressor in studies of a mouse eukaryotic initiation factor (eIF)4A and eIF4G, which keratinocyte model of tumour promotion, in which high are mediated via the two tandem MA-3 domains. We have levels of Pdcd4 were found to render cells resistant to determined the structure of the C-terminal MA-3 domain transformation by the tumour promoter 12-O-tetrade- of Pdcd4(Pdcd4MA-3 C), characterized its interaction canoyl-phorbol-13-acetate (TPA) (Cmarik et al., 1999). with eIF4A and compared the features of nuclear Pdcd4 has been shown to inhibit the activation of AP1- magnetic resonance (NMR) spectra obtained from the responsive promoters by c-Jun, providing a possible single domain and tandem MA-3 region. Pdcd4MA-3 C is explanation for its ability to suppress TPA-induced composed of three layers of helix–turn–helix hairpins transformation (Yang et al., 2001, 2003b). Recently, loss capped by a single helix and shows close structural of Pdcd4 expression has been strongly implicated in the homology to the atypical HEAT repeats found in many development and progression of both human lung eIFs. The sequence conservation and NMR data strongly cancer and aggressive malignant breast cancer (Chen suggest that the tandem MA-3 region is composed of two et al., 2003; Afonja et al., 2004). In addition, the equivalent domains connected by a somewhat flexible expression of Pdcd4 has been shown to be reduced in linker. Pdcd4MA-3 C was found to interact with the N- many renal-, lung- and glia-derived tumours (Jansen terminal domain of eIF4A through a conserved surface et al., 2004). region encompassing the loop connecting a5 and a6 and Pdcd4 has been shown to shuttle between the nucleus the turn linking a3anda4. This site is strongly conserved and the cytoplasm but under normal cell growth in other MA-3 domains known to interact with eIF4A, conditions it appears to be predominantly localized to including the preceding domain of Pdcd4, suggesting a the nucleus (Bohm et al., 2003; Palamarchuk et al., common mode of binding. 2005). Recent work suggests that phosphorylation of Oncogene (2007) 26, 4941–4950; doi:10.1038/sj.onc.1210305; Pdcd4 by Akt plays a key role in the nuclear published online 19 February 2007 translocation of Pdcd4, as well as significantly reducing the ability of both nuclear and cytoplasmic Pdcd4 to Keywords: Pdcd4; MA-3; eIF4A; HEAT; NMR; inhibit AP-1-dependent activation of gene expression translation (Palamarchuk et al., 2005). Human and mouse Pdcd4 are 469 residue proteins, which contain at least three domains, an N-terminal RNA-binding region (residues 1–157) and two MA3 domains (residues 157–284 and 319–449) (Ponting, 2000; Introduction Bohm et al., 2003). To date, MA3 domains have been found in over 200 eukaryotic proteins and appear to be Programmed cell death protein 4 (Pdcd4) is a novel, involved in mediating specific protein–protein interac- highly conserved, eukaryotic protein, which has recently tions (http://www.sanger.ac.uk/Software/Pfam/) (Finn been shown to play critical roles in the regulation of et al., 2006). Characterization of the cellular functions of both transcription and translation (Yang et al., 2003a, Pdcd4 is the focus of many ongoing investigations, however, recent work indicates essential roles in the control of both transcription and translation, mediated Correspondence: Dr M Carr, Department of Biochemistry, University via specific protein–protein and protein–RNA interac- of Leicester, Henry Wellcome Building, Lancaster Road, Leicester, tions (Yang et al., 2003a, 2004; Bitomsky et al., 2004). Leicestershire, LE1 9HN, UK. The best characterized functional interaction is between E-mail: [email protected] Received 13 October 2006; revised 19 December 2006; accepted 29 the MA-3 domains of Pdcd4 and the eukaryotic December 2006; published online 19 February 2007 initiation factors eIF4A and eIF4G, with complex Structure and interactions of the Pdcd4 MA-3C domain LC Waters et al 4942 formation resulting in the inhibition of cap-dependent hydrogen bond constraints in regions of regular helical translation (Yang et al., 2003a, 2004). eIF4A catalyses structure (Supplementary Table S1). Following the final the unwinding of stable secondary structure in the 50 round of CYANA calculations, 49 satisfactorily con- untranslated region (UTR) of mRNAs, allowing the verged structures were obtained from 100 random recruitment of the 40S ribosomal subunit to the 50 cap of starting structures. The converged structures contain mRNA (Hershey and Merrick, 2000). The inherent no distance or van der Waals violation higher than 0.5 A˚ helicase activity of eIF4A is strongly stimulated by and no dihedral angle violations higher than 51,with binding to the scaffold protein eIF4G to form part of an average value for the CYANA target function of the eIF4F complex, which also includes eIF4E (Rozen 1.4670.25 A˚ 2. The NMR constraints and structural et al., 1990). statistics for Pdcd4 MA-3C are summarized in Supple- Pdcd4 has also been shown to bind to c-Jun and Jun mentary Table S1. The family of converged Pdcd4-MA3 N-terminal kinase (JNK), thereby preventing JNK- domain structures, together with the NMR constraints, dependent phosphorylation of the c-jun transactivation have been deposited in the Protein Data Bank (accession domain, as well as the interaction of c-Jun with the code 2HM8). coactivator p300 (Bitomsky et al., 2004). This suggests a The solution structure of Pdcd4 MA-3C is determined mechanism for the suppression of c-Jun activity by to high precision, which is clearly evident from the Pdcd4 and highlights its importance in transcriptional as superposition of the protein backbone shown for the well as translational control. family of converged structures in Figure 1a (best fit for In this communication, we report the high-resolution residues 327–356, 359–412 and 421–448) and is reflected solution structure of the C-terminal MA-3 domain of in low root-mean squared deviation (RMSD) values to mouse Pdcd4 (Pdcd4 MA-3C), characterization of its the mean structure for both the backbone and all heavy interaction with eIF4A and propose a model for the atoms of 0.4970.07 and 0.9270.07 A˚ , respectively. entire tandem MA-3 region based on sequence con- servation and nuclear magnetic resonance (NMR) data. Mapping of binding sites The positions of signals from backbone amide groups in Results proteins are highly sensitive to changes in their local environment and have been used to localize the eIF4A- 15 1 Structural calculations binding site on Pdcd4 MA-3C. Typical N/ Hhetero- The combined automated nuclear overhauser effect nuclear single-quantum coherence (HSQC) spectra 15 (NOE) assignment and structure determination protocol obtained from samples of N-labelled Pdcd4 MA-3C (CANDID) (Herrmann et al., 2002) proved very (100 mM), in the presence and absence of an excess of effective at determining unique assignments for the full-length eIF4A (150 mM), are shown in Figure 2a. NOEs identified in the two- and three-dimensional These spectra allowed backbone amide minimal shift nuclear overhauser effect spectroscopy-based spectra, values to be determined for nearly all the residues of with unique assignments obtained for 94% (3999/4254) bound Pdcd4 MA-3C, which are summarized in the of the NOE peaks, which produced 2545 non-redundant histogram shown in Figure 2b. Examination of the 1Hto 1Hupper distance limits. The final family of Pdcd4 histogram clearly identifies 11 residues (H361, E362, MA-3C structures was determined using a total of 2787 E373, S374, T375, G376, N410, D414, I415, L416 and NMR-derived structural constraints (an average of 21.3 R428) in Pdcd4 MA-3C whose backbone amide signals per residue), including 2545 NOE-based upper distance undergo large chemical shift changes on binding of limits, 174 backbone torsion angle constraints and 68 eIF4A. These residues are highlighted in red on the b a C 7 C 7 6 6 5 5 3 4 3 4 2 2 1 1 N N Figure 1 Solution structure of the Pdcd4 MA-3C domain. (a) Shows a best-fit superposition of the protein backbone for the family of 49 converged structures obtained. (b) Contains a stereo view of the ribbon representation of the backbone topology of Pdcd4 MA-3C based on the structure closest to the mean and is shown in the same orientation as in (a). The seven helical regions of Pdcd4 MA-3C, a1, a2, a3, a4, a5, a6 and a7 are labelled, as are the N and C termini of the domain. Oncogene Structure and interactions of the Pdcd4 MA-3C domain LC Waters et al 4943 Pdcd4 MA-3C sequence shown in Supplementary above helices a5anda6. One notable feature of the Figure S1a. protein backbone is the loop between a5 and a6. In Analogous chemical shift mapping experiments were contrast to the short well-defined turns between a1and also carried out for Pdcd4 MA-3C (100 mM) in the a2 (two residues) and between a3 and a4 (four residues), presence of a 5- and 10-fold molar excess of the isolated the long loop between a5anda6 (13 residues) appears to N-terminal domain of eIF4A.