Portal de Periódicos da CAPES

Solution structure of the AT‐rich interaction domain of Jumonji/JARID2

2009; Wiley; Volume: 76; Issue: 4 Linguagem: Inglês

10.1002/prot.22449

1097-0134

Hideki Kusunoki, Takashi Takeuchi, Toshiyuki Kohno,

Genomics and Chromatin Dynamics

Jumonji/JARID2 (Jmj) is a transcriptional repressor protein, which plays important roles in development, cell growth and gene expression.1-3 For example, recent studies have revealed that Jmj represses the expression of cyclin D1, a key component of the cell cycle machinery, by inducing histone H3 lysine 9 (H3-K9) methylation in the complex with G9a and GLP, which are H3-K9 methyltransferase proteins.4, 5 In addition, Jmj has been shown to repress atrial natriuretic factor (ANF) gene expression, by interacting with the transcriptional activities of the cardiac transcription factors, Nkx2.5 and GATA4.6 Therefore, it is believed that Jmj may use different regulatory mechanisms with multiple target genes. The Jmj protein consists of at least six structural and functional domains [Fig. 1(A)]: the nuclear localization signal (NLS) domain, the transcriptional repressor domain, the JmjN domain, the AT-rich interaction domain (ARID), the JmjC domain and the Zinc finger-like C5HC2 domain.3 (A) Domain structure of Jumonji/JARID2. The Jumonji/JARID2 protein contains six structural and functional domains, which are the nuclear localization signal domain (NLS, shown in purple), the transcriptional repressor domain (TR, orange), the JmjN domain (blue), the AT-rich interaction domain (ARID, green), the JmjC domain (cyan) and the zinc finger-like C5HC2 domain (Zf-C5HC2, yellow). The NLS (residues 1-130) and TR (residues 131-222) domains were identified from structural and functional analyses,28 and the boundaries of the JmjN (residues 555-600), ARID (residues 616-726), JmjC (residues 914-1029) and Zf-C5HC2 (residues 1137-1191) domains were obtained from the results of Pfam,29 (B) Stereo view of a best-fit superposition of the backbone atoms (N, Cα, and C′) of the 20 NMR-derived structures of the Jmj-ARID (615-730), and (C) Stereo view of a ribbon diagram of the lowest energy structure of Jmj-ARID (615-730). The secondary structure elements (α-helices, H1 to H7, and β-strands, B1 and B2) and the two loops (L1 and L2) are labeled. The noncanonical HTH motif and the cruciform are labeled and circled, respectively. The ARID domain7, 8 is a distinct DNA-binding module and binds to DNA through a long loop that connects the helices of a helix-turn-helix (HTH) motif, and therefore this HTH motif is called as a noncanonical HTH motif.9, 10 ARID domains have both functional and structural diversities.7, 8 For example, the Dri-ARID9, 10 and Mrf2-ARID11 domains interact with their specific AT-rich DNA sequences, while the RBP2-ARID domain binds to a GC-rich CCGCC DNA sequence.12 Furthermore, the SWI1-ARID domain reportedly interacts with nonspecific DNA sequences.13 So far, three structural classes have been represented: (i) the minimal ARID domain (six α-helices), (ii) both the N- and C-terminal extended ARID domain and (iii) the N-terminal extended ARID domain. However, the structure of the C-terminal extended ARID domain has not been determined yet. We previously performed NMR analyses of the Jmj-ARID domain consisting of residues Leu615-Lys730.14 The secondary structure prediction from the NMR data revealed that the Jmj-ARID domain may possess an additional helix at the C-terminus. Thus, the Jmj-ARID domain seems to be a good candidate to represent the C-terminal ARID domain. Here we report the first NMR-derived structure of the ARID of Jumonji/JARID2. The ARID of the mouse Jumonji/JARID2 [residues 615-730, hereafter referred as to Jmj-ARID (615-730)] was expressed in either the Rosetta or BL21 strain of Escherichia coli as a GST fusion protein and purified as previously described14 (see Supporting Information). The resulting recombinant protein contains five additional vector-derived amino acid residues (Gly-Pro-Leu-Gly-Ser) at the N-terminus. These five residues were numbered from 610 to 614, in this study, for the structure determination. The NMR samples contained 0.25 to 1.6 mM protein in 22.5 mM sodium phosphate (pH 6.4), 90 mM NaCl, 1.8 mM DTT, 0.67 mM DSS and either 90% H2O/10% D2O or 100% D2O. All NMR experiments were acquired on Bruker AVANCE 500 and AVANCE II 700 spectrometers equipped with a 1H/13C/15N cryogenic probe at 15°C. The 1H, 13C, and 15N resonances of Jmj-ARID (615-730) were assigned using triple resonance NMR experiments.15, 16 The stereospecific assignments of the methyl groups of leucine and valine were obtained from a two-dimensional 1H-13C CT-HSQC spectrum on a 10% 13C/100% 15N-labeled protein sample.17 For the residual dipolar coupling (RDC) measurement, NMR samples with 300 mM sodium chloride in the absence and presence of 10 mg/mL Pf1 phage18 (Asla Labs) were prepared, respectively. The one bond NH couplings in the isotropic and anisotropic media were measured using two-dimensional IPAP 1H-15N HSQC experiments.19 The amide resonances were successfully assigned, based on those of the NMR sample with 90 mM NaCl. Two sets of {1H}-15N NOE spectra with and without the NOE effect were recorded and analyzed, as described.20 CYANA version 2.1 was used for the collection of the distance restraints.21 Dihedral ϕ and ψ angle restraints were determined from the chemical shift data, using the program TALOS.22 Hydrogen bond restraints were generated from the secondary structure of the protein, based on the backbone chemical shift data. Final structures were calculated by CNS version 1.1, with 1889 distance restraints, 166 dihedral angle restraints and 60 1DNH values, by a simulated annealing protocol.23 The 60 1DNH RDC values were obtained as the difference in couplings between the isotropic and anisotropic samples, which were selected after excluding the residues representing {1H}-15N NOE 0.25 ppm for 15N nuclei or >0.05 ppm for 1H nuclei) are shown in purple. Almost all of these residues are located in the N- and C-terminal regions of Jmj(615-730) (E) Close-up view of the interactions that form the cruciform between helices H1 and H7 of Jmj-ARID(615-730) The image is related to that in panel D by an approximately −15° rotation along the vertical axis. Two unique features are present in the three-dimensional structure of Jmj-ARID (615-730): (i) an additional helix H7 is located at the C-terminus, and (ii) a cruciform is formed with helices H1 and H7 [Fig. 1(C)]. The cruciform is probably stabilized by the following residue contacts: Leu717 in helix H7 makes van der Waals contacts with Lys631 and His632 in helix H1, Tyr706 in helix H6, and Leu709 in the loop between helices H6 and H7. His632 in helix H1 makes van der Waals contacts with Tyr706 in helix H6, and with Leu709 in the loop between helices H6 and H7. Val721 in helix H7 makes van der Waals contacts with Ala627 and Lys631 in helix H1 and Trp619 [Fig. 2(E)]. It is interesting to note that both the direction and length of the C-terminal α-helix of Jmj-ARID (615-730) are quite different from those of the Dri-ARID domain [Fig. 2(D)]. The C-terminal helix of the Dri-ARID domain is located on the outside of the core fold (helices H1 to H6) and is parallel and adjacent to the fourth helix [corresponding to helix H3 of Jmj-ARID (615-730)], due to the hydrophobic interactions between the C-terminal and fourth helices of the Dri-ARID domain. So far, three structural variants of ARID domains have been represented: (i) minimal ARID domains, such as the RBP2-ARID12 [Fig. 2(B)] and Mrf2-ARID11 [Fig. 2(C)] domains, (ii) both the N- and C-terminal extended ARID domains, such as the Dri-ARID9, 10 domain [Fig. 2(D)], and (iii) the N-terminal extended ARID domains, such as the SWI1-ARID13 domain [Fig. 2(A)]. In this study, the three-dimensional structure of Jmj-ARID (615-730) revealed that it has a C-terminal extended ARID domain, and thus represents a novel fourth ARID fold. In addition, the differences in the C-terminal α-helix between Jmj-ARID (615-730) and the Dri-ARID domain suggest the possibility that ARID domains extended at their N- and C-termini have further structural variety. In fact, a structural difference exists between the Dri- and SWI1-ARID domains: the length of the N-terminal α-helix of the Dri-ARID domain [Fig. 2(D)] is longer than that of the SWI1-ARID domain [Fig. 2(A)]. To examine whether Jmj-ARID (615-730) can bind to DNA via the residues in the HTH motif, we performed NMR chemical shift perturbation experiments, using two oligonucleotides with the Dri-ARID recognition sequence and the ANF (−110) promoter-enhancer sequence. These two oligonucleotides contain either a Jmj (529-798) (a region including residues 529-798 of Jumonji/JARID2) or Jmj (529-792) binding sequence (5′-TATT-3′ or 5′-(A/T)4–6-3′ sequence), as reported previously.6, 28 In our NMR perturbation experiments using both oligonucleotides, the patterns of the chemical shift changes, between the absence and presence of DNA, were almost identical: the chemical shift changes were quite small and the interactions were on the fast exchange time scale, in 106 assigned residues out of 112 nonproline residues within Jmj-ARID (615-730). It should be noted that all of the residues in the noncanonical HTH motif, except for Leu682, showed small chemical shift changes in both NMR perturbation experiments (data not shown). In the Dri-ARID domain/DNA complexes,9, 10 the residues in the noncanonical HTH motif, in the two β-strands of L1 and at the C-terminus, showed large chemical shift changes (>1.0 ppm for 15N nuclei and >0.35 ppm for 1H nuclei).9 The complex structure demonstrated that the residues exhibiting large chemical shift changes contribute to the interaction with DNA [Fig. 2(D), shown in red in the Dri-ARID domain].10 Therefore, our NMR results indicate that Jmj-ARID (615-730) may bind to DNA with relatively low affinity, and the residues in the HTH motif may not be primarily involved in binding DNA. This is presumably because these residues in Jmj-ARID (615-730) are poorly conserved with those in the Dri-ARID domain (Supporting Information Fig. S1). In contrast, the residues showing relatively large chemical shift changes (> 0.25 ppm for 15N nuclei or >0.05 ppm for 1H nuclei) were Arg618, Trp619, Gly620, Asn622, Val623, Leu682, Leu728, and Glu729, and these residues reside in the N- and C-terminal regions [Fig. 2(D), shown in purple in the Jmj-ARID domain]. To clarify whether these residues are involved in the interaction with DNA and to characterize the function of the Jmj-ARID domain, further functional and structural analyses are needed. For example, it seems possible that this domain with either the N-terminal (residues 529-614) or C-terminal (residues 731-798) region, or both, may contribute to the interaction with DNA, based on experimental evidence that Jmj (529-798) and Jmj (529-792) can bind to DNA.6 In addition, another possibility is that the Jmj-ARID domain may contribute to the protein–protein interaction, based on experimental evidence that Jmj (529-792) reportedly interacts with Nkx2.5 and GATA4, respectively.6 In this study, we determined the first NMR-derived structure of the ARID of Jumonji/JARID2. The solution structure clearly showed that the Jmj-ARID domain forms the C-terminal extended ARID fold, and the additional α-helix at the C-terminus forms a cruciform with the first helix on top of the core fold. This structure is a unique feature among the three-dimensional structures of the ARID domains. In addition, our NMR chemical perturbation experiments implied that the residues in the HTH motif may not be primarily involved in the interaction with DNA. The authors gratefully acknowledge the contributions of Ms. Tsukasa Hasegawa and Ms. Chieko Komatsu to the initial stages of this project. Additional Supporting Information may be found in the online version of this article. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.