letters Solution struture of the DNA-inding domin of MfG Hideki Kusunoki 1,2, Hozumi Motohshi 1,2, umiki Ktsuok 1, Akio Morohshi 3,4, Msyuki Ymmoto 1,2 nd Toshiyuki Tnk 2,3 1 Institute of Bsi Medil Siene, University of Tsuku, Tsuku, Irki 305-8575, Jpn. 2 Center for Tsuku Advned Reserh Alline, University of Tsuku, Tsuku, Irki 305-8577, Jpn. 3 Institute of Applied Biohemistry, University of Tsuku, Tsuku, Irki 305-8572, Jpn. 4 Bnyu Tsuku Reserh Institute, Tsuku, Irki 300-2611, Jpn. Pulished online: 4 Mrh 2002, DOI: 10.1038/ns771 The Mf fmily proteins, whih onstitute sugroup of si region-leuine zipper (ZIP) proteins, funtion s trnsriptionl regultors of ellulr differentition. Together with the si region, the Mf extended homology region (EHR), onserved only within the Mf fmily, defines the DNA inding speifi to Mfs. Here we present the first NMR-derived struture of the DNA-inding domin (residues 1 76) of MfG, whih ontins the EHR nd the si region. The struture onsists of three -helies nd resemles the fold of the DNA-inding domin of Skn-1, developmentl trnsription ftor of Cenorhditis elegns. The struturl similrity etween MfG nd Skn-1 enles us to propose possile mehnism y whih Mf fmily proteins reognize their onsensus DNA sequenes. The Mf fmily of trnsription ftors re si regionleuine zipper (ZIP) proteins, hrterized y the presene of speifi mino id sequene, the Mf extended homology region (EHR), loted on the N-terminl side of the si region 1,2. So fr, seven fmily memers hve een reported nd sudivided into two groups: lrge Mfs (26 39 kd), whih hve puttive trnstivtion domin t their N-termini, nd smll Mfs (17 18 kd), whih lk suh domin 1 3 (ig. 1). Mf fmily proteins ply importnt roles in ellulr differentition nd morphogenesis, nd their unique funtions seem to e determined y their seletive dimeriztion nd DNA-inding speifiity 1 3. Mfs n form homodimers or heterodimers with other ZIP trnsription ftors, suh s AP-1 nd Cp-N- Collr (CNC) fmily memers, through their leuine zipper domins. or exmple, smll Mfs re known to dimerize with p45 (the lrge suunit of the nuler ftor-erythroid 2, N-E2) to tivte erythroid ell-speifi trnsription 1 3. As for DNA-inding speifiity, Mf fmily proteins reognize reltively long plindromi DNA sequene, TGC TGACTCAGCA or TGCTGACGTCAGCA, known s the Mf reognition element (MARE). The MARE sequene ontins the well-known AP-1 reognition site, the 12-O-tetrdenoylphorol-13-ette (TPA)-responsive element (TRE; TGACT- CA) or the yli AMP-responsive element (CRE; TGACGTCA) (ig. 1) 1 5. Therefore, Mfs my reognize the TRE or CRE site in similr wy s AP-1 proteins 6. However, the reognition of extended sequene elements (flnking regions) on oth sides of the AP-1 ore site distinguishes Mfs from AP-1 proteins in the DNA-inding mode. The DNA-inding speifiity of Mfs is elieved to e hieved through reognition of the flnking region y the EHR, whih is onserved only within the Mf fmily nd not found in other ZIP proteins 2,6. To understnd the unique DNA-inding mode of Mfs, we determined the solution struture of the MfG ( smll Mf) DNA-inding domin (residues 1 76), whih shres no sequene homology with ny protein whose struture hs een determined. This struture provides the first opportunity to understnd how the EHR reognizes the flnking region. ig. 1 DNA reognition y MfG(1 76)., Sequene lignment of Mf fmily proteins for the region orresponding to residues 1 76 of MfG. α-helies determined in the present study re indited y purple retngle. Conserved mino ids mong Mf fmily proteins re shown in red. The residues involved in the pping ox of MfG re mrked with sterisks. SWISS-PROT ession numers re s follows: O54790 (mouse MfG), Q61827 (mouse MfK), O54791 (mouse Mf), Q92171 (hiken -Mf), P54841 (mouse MfB), P54846 (mouse NRL) nd O42290 (hiken L-Mf)., DNA sequenes reognized y Mf fmily proteins. T-MARE nd C-MARE stnd for TRE-type MARE nd CRE-type MARE, respetively. The TRE nd CRE onsensus sequenes re shown in green nd lue, respetively. lnking nd ore reognition elements of the hlf site re oxed., EMSA of MfG(1 76) using n oligonuleotide ontining the T-MARE-like (left, proe #25 in ref. 5) or its mutnt sequene (right, proe #23 in ref. 5). The onentrtions of purified protein were 0 µm (lnes 1 nd 6), 0.05 µm (lnes 2 nd 7), 0.1 µm (lnes 3 nd 8), 0.2 µm (lnes 4 nd 9) nd 0.4 µm (lnes 5 nd 10). 252 nture struturl iology volume 9 numer 4 pril 2002
DNA-inding ility of MfG MfG(1 76) ontins oth the EHR nd the si region, whih re neessry for DNA reognition nd inding, ut lks the leuine zipper domin required for dimeriztion. irst, we exmined the funtionl properties of MfG(1 76). Eletrophoreti moility shift ssys (EMSA) showed tht this protein inds to the TRE-type MARE (T-MARE)-like sequene TGCTGACTCATCA with K d of 0.3 µm, ut not to CAATGACTCATTG, whih hs muttions introdued in the flnking region (ig. 1). In ddition, sedimenttion equilirium nlysis reveled tht MfG(1 76) exists s monomer in solution, with lulted nd mesured moleulr weights of 9,000 nd 9,700 ± 800, respetively. These results indite tht MfG(1 76) lone funtions s the DNA-inding domin tht retins the DNA-inding speifiity of the nturl protein. The reltively low ffinity of MfG(1 76) for the DNA sequene results from its lk of leuine zipper domin, whih is responsile for stilizing the protein DNA omplex through dimeriztion. Struture desription The 3D struture of MfG(1 76) hs een determined on the sis of totl of 843 NMR-derived restrints. With the exeption of the 23 N-terminl nd 12 C-terminl residues, whih re disordered nd highly moile s indited y the puity of NOEs, the hemil shift index 7 nd 15 N{ 1 H} NOE vlues 8 (dt not shown), the struture is well defined (ig. 2; Tle 1). MfG(1 76) onsists of three α-helies, H1 (26 31), H2 (34 41) nd H3 (46 61) (ig. 2). An N-terminl pping ox of the type SXX(E/Q) or TXXE (X stnds for ny mino id residue) 9, whih is known to funtion s helix stop signl, is found in ll helies (ig. 1). The helies re stilized y hydrophoi intertions mong the following residues: Leu 24 from the N-terminl region; Leu 29 nd Vl 30 from H1; Met 32 d ig. 2 3D struture of MfG(1 76)., Stereo view of est-fit superposition of the kone toms (N, Cα nd C ) of the 20 NMR-derived strutures of MfG(1 76). The min hin toms of the 20 strutures re superimposed ginst the energy-minimized verge struture using residues 24 64. The 23 N-terminl nd 12 C-terminl residues, whih re not well defined euse they lk mny experimentl restrints, re omitted throughout pnels ( d)., Rion digrm of the energyminimized verge struture of MfG(1 76). The α-helies re shown in purple nd leled.,d, Eletrostti potentil surfes of MfG(1 76). Positive nd negtive potentils re in lue nd red, respetively. The orienttion of the imge (d) is the sme s tht in pnel (). The imge in () is relted to tht in (d) y 180 rottion long the vertil xis. from turn etween H1 nd H2; Leu 37 nd Leu 41 from H2; Leu 44 from turn etween H2 nd H3; nd Ile 49, Leu 52 nd Leu 59 from H3. Conservtion of these hydrophoi residues nd the N-terminl pping oxes mong Mfs (ig. 1) indites tht there is similrity in the 3D strutures of their DNAinding domins. The EHR onstitutes the region from helix H1 to the first one-third of helix H3, wheres the si region omprises the rest of helix H3 nd the C-terminl flexile prt. Most of the si mino id residues reside on one side of the protein surfe (ig. 2,d): Arg 35 from H2 nd Lys 46, Lys 53, Arg 56, Arg 57 nd Lys 60 from H3. These si residues, exept for Lys 46, re ompletely onserved within the Mf fmily (ig. 1). In ddition, loss of DNA inding ws reported for the R57E mutnt of MfG nd the orresponding mutnts of MfK, Mf nd MfB 4,5. Therefore, the si luster on the protein surfe is proly involved in the intertion with trget DNA sequenes. We will refer to this surfe s the DNAinding surfe. DNA-inding site of MfG(1 76) To identify the DNA-inding site of MfG(1 76), we performed n NMR titrtion, in whih the 2D 1 H- 15 N HSQC spetr of MfG(1 76) were reorded with suessive dditions of DNA ontining the T-MARE-like sequene. Lrge hemil shift hnges were oserved upon DNA inding, minly in the region from helix H2 to the C-terminus (ig. 3). This region ontins oth the EHR nd the si region, inditing their involvement in the intertion with DNA. The hnges ontinued on the ddition of up to hlf the equivlent of DNA, nd only single set of protein signls ws oserved throughout the titrtion experiment. These results indite tht MfG(1 76) inds to DNA s 2:1 omplex with two-fold symmetry nd tht the exhnge etween free nd ound protein onformtions is fst on the NMR time sle. nture struturl iology volume 9 numer 4 pril 2002 253
ig. 3 Chemil shift hnges of mide protons (upper pnel) nd nitrogens (lower pnel) of MfG(1 76) upon inding to the DNA ontining the T-MARE-like sequene. Asolute hemil shift differenes etween DNA-free nd DNA-ound sttes re plotted versus residue numer. Residues showing lrge hemil shift hnges ( 0.1 p.p.m. for 1 H or 0.5 p.p.m. for 15 N) re mrked with one-letter ode nd residue numer. Unssigned residues in either stte nd Pro residues re indited using sterisks. A representtion of the seondry struture of MfG(1 76) is shown t the top. Comprison of MfG(1 76) with other DNA-inding domins A helix-turn-helix (HTH) motif, one of the most ommon DNA-inding motifs, is found in mny eukryoti nd prokryoti trnsriptionl regultory proteins. The seond helix of the motif usully inds to the mjor groove of DNA nd, therefore, is minly involved in reognition nd intertion with speifi DNA sequenes 10. MfG(1 76) ontins similr HTH motif (H2 nd H3), ut the interhelil ngle etween the two helies (149 ) is onsiderly lrger thn the typil vlue for the ommon HTH motif ( 110 ) 10. Thus, MfG(1 76) my hve protein DNA intertion mode distint from tht of the ordinry HTH-type DNA-inding domins. Reently, the 3D struture of the DNA-inding domin of Skn-1, developmentl trnsription ftor from Cenorhditis elegns, ws determined in oth DNA-free 11 nd DNA-ound sttes 12. This domin shres 25% sequene homology with MfG(1 76), ut their sequene similrity hs not een identified y dtse serh. The strutured region of MfG resemles the fold of the Skn-1 DNA-inding domin (kone root men squre (r.m.s.) devition of 1.2 Å for helies H1, H2 nd H3), lthough MfG lks n N-terminl helix (H0) nd hs shorter C-terminl helix (H3) (ig. 4). Their struturl similrity is supported y the following fetures of the three helies: similr interhelil ngles, pping ox sequene t the N-terminus nd onservtion of the hydrophoi residues involved in their stiliztion (ig. 4,). In ddition, si luster is present on the surfe of Skn-1 in position similr to tht of MfG(1 76) (ig. 4). DNA-inding speifiity of Mf proteins The struturl similrity etween Skn-1 nd MfG provides insight into the mehnism y whih Mf fmily proteins might reognize the onsensus sequenes (ig. 1). Skn-1 hieves seletivity for the onsensus sequene (G/A)TCAT, n AP-1-like hlf site 11,12, y diret intertion through the five residues in the NXXAAXXCR sequene of its si region (ig. 4d) 12. In ddition, mny of the si residues in the si region (losed irles, ig. 4) mke ontt with the phosphte kones of the DNA 12. A similr mino id sequene, NXXYAXXCR, nd most of the si residues re onserved in the si region of Mfs (igs 1, 4) nd form DNA-inding surfe nlogous to tht of Skn-1 (ig. 4). urthermore, hemil shift perturtion experiment showed tht these residues re involved in the intertion with the MARE sequene (ig. 3). Considering the ove, the Mf fmily proly uses DNA-inding mode ommon to Skn-1 in reognition of the AP-1 site. A similr mode of DNAinding is lso proly used y AP-1 ZIP proteins; they possess the NXXAAXXC(/S)R sequene, nd lmost identil intertions etween the protein nd the DNA were found in their omplex strutures 13,14. However, omplete onservtion of Tyr insted of n Al residue t the fourth position in the NXXYAXXCR sequene (ig. 1) indites the possiility tht the si region of Mfs hs slightly different DNA-inding mode. How then does the EHR, essentil for the unique DNAinding speifiity of Mfs, reognize the flnking region? In the struture of the Skn-1 DNA omplex, the hydrophoi side hin of Leu 482 in helix H2 ws found to intert with the riose group of the gunine nuleotide tht resides outside the onsensus sequene (red, ig. 4d) 12. Therefore, the orresponding region on the DNA-inding surfe of MfG my intert with the nuleotide ses in the flnking region tht lie within three se pirs from the AP-1 ore region. Side hins of Vl 34, Arg 35 nd Asn 38 re present in this region (ig. 4), nd the six residues of helix H2 tht ontin them re ompletely onserved within the EHR (ig. 1). In ddition, these residues displyed pertured hemil shifts upon DNA inding (ig. 3). To ssess the roles of these residues in reognition of the flnking region, EMSA ws performed using series of mutnts, in whih one of the three residues ws repled y n Al residue. A signifint redution in DNA-inding ffinity ws oserved for ll three mutnts (V34A, R35A nd N38A) (ig. 4e), inditing tht Vl 34, Arg 35 nd Asn 38 on the DNA-inding surfe re the most prole residues involved in the inding of the flnking region nd the ontriution of helix H2 to DNA reognition is quntittively different etween MfG nd Skn-1. Truntion of the N-terminus of -Mf to Glu 259 (orresponding to Glu 36 of MfG) ws reported to result in loss of DNA-inding speifiity 6. This trunted -Mf ws le to ind to oth the T-MARE sequene nd the sequene whose flnking region ws mutted. This suggests tht Vl 34 nd Arg 35 ply n importnt role in the speifi reognition of the flnking region. Conlusions The present struture determintion of MfG(1 76), whih shows no sequene homology to ny protein of known struture, hs reveled struturl feture relevnt to its DNA reognition mehnism. Within Mf fmily proteins, the struture formed y the three helies of the EHR (the EHR fold) is present just 254 nture struturl iology volume 9 numer 4 pril 2002
d e ig. 4 Comprison of MfG(1 76) with the Skn-1 DNA-inding domin., DNA-inding domins of MfG (left) nd Skn-1 (right) (PDB ession ode 1SKN). The helies of MfG nd the orresponding regions of Skn-1 re in purple; the other helil regions of Skn-1 re in yn., DNA-inding surfes of MfG (left) nd Skn-1 (right). The si residues of MfG (Lys 53, Arg 56, Arg 57 nd Lys 60) nd Skn-1 (Arg 503, Arg 506, Arg 507, Arg 508 nd Lys 510) re in lue. The five residues of Skn-1 tht intert with the DNA ses nd the orresponding residues of MfG (only Asn 61 nd Tyr 64 in this pnel) re in green. Vl 34, Arg 35 nd Asn 38 of MfG nd the orresponding residues of Skn-1 re shown in red., Sequene lignment of the MfG nd Skn-1 DNA-inding domins (SWISS-PROT ession numer P34707). Seondry struture elements of MfG (upper) nd Skn-1 (lower) re shown. The residues tht form the hydrophoi ores of MfG nd Skn-1 re underlined. The residues orresponding to the olored residues in pnel () re highlighted in mthing olor. The pping ox sequenes re indited with sterisks. Open nd losed irles show the residues of Skn-1 tht intert with the ses nd phosphte kones of DNA, respetively. d, DNA sequene used in the struture determintion of the Skn-1 DNA omplex nd its intertion with the NXXAAXXCR sequene of Skn-1. The onsensus reognition site of Skn-1 is oxed. The AP-1 ore region is in green. Hydrogen onds nd vn der Wls ontts re indited y solid nd dotted lines, respetively. e, EMSA of MfG(1 76) nd its mutnts using n oligonuleotide ontining the T-MARE-like sequene (proe #25 in ref. 5). The onentrtions of purified protein were 0 µm (lnes 1, 5, 9 nd 13), 0.2 µm (lnes 2, 6, 10 nd 14), 0.8 µm (lnes 3, 7, 11 nd 15), nd 3 µm (lnes 4, 8, 12 nd 16). efore the si region. This struture enles Mfs to mke roder re of ontt with DNA nd to reognize longer DNA sequenes. In prtiulr, the two residues t the eginning of helix H2 (Vl 34 nd Arg 35) re positioned to reognize the flnking region. Therefore, disruption of the EHR fold will led to loss of DNA-inding ffinity nd/or speifiity of Mfs. This EHR fold is not found in AP-1 fmily memers suh s GCN4, -os nd -Jun. The EHR fold is designed for speifi reognition of the flnking region nd, thus, distinguishes Mfs from other proteins y virtue of its DNA-inding speifiity. Methods Smple preprtion. MfG(1 76) or its mutnt protein ws expressed in Esherihi oli s His 6 -tgged protein. M9 miniml medium ontining 15 NH 4 Cl or 15 NH 4 Cl/ 13 C 6 -D-gluose ws used to prepre 15 N- or 15 N/ 13 C-enrihed protein, respetively. The His 6 - tgged protein ws purified y Ni-NTA superflow olumn (Qigen) nd then leved with thromin. The reominnt protein tht ontined three extr vetor-derived mino ids (Gly-Ser-His) t the N-terminus ws further purified using S2 (BioRd), HiLod 16/60 Superdex 75 pg (Phrmi) nd Benzmidine Sephrose 6B (Phrmi) olumns. NMR smples ontined 1.5 2.0 mm protein in 20 mm sodium phosphte uffer, ph 6.7, 10 mm dithiothreitol (DTT)-d 10 nd 10% (v/v) 2 H 2 O. Comprison of the proton NMR spetr indited tht none of the muttions disrupted the threedimensionl struture of MfG(1 76). NMR spetrosopy. NMR spetr were otined t 25 C using Bruker AVANCE DRX 800 or Vrin UNITY INOVA 500 spetrometer. All dt were proessed using nmrpipe nd nmrdrw 15, nd nlyzed with PIPP 16. Bkone nd side hin 1 H, 13 C nd 15 N resonnes were ssigned using stndrd triple resonne experiments 17. Stereo-speifi ssignments of Vl nd Leu methyl groups were otined y nlyzing 2D 1 H- 13 C CT-HSQC spetrum on 10% 13 C-enrihed smple 18. Three sets of 15 N{ 1 H} NOE spetr with nd without the NOE effet were reorded nd nlyzed s desried 8. Struture lultion. Distne restrints were otined from 2D NOESY nd 3D 15 N-edited nd 13 C-edited NOESY experiments (mixing time of 100 ms) s desried 19. Dihedrl Φ ngle restrints were dedued from 3 J NHα oupling onstnts mesured y 3D HNHA experiment 20 : Φ ws restrined to 60 ± 30 for residues with 3 J NHα < 5.5 Hz. Struture lultions were performed using simulted nneling protool 21 within X-PLOR v.3.851 (ref. 22). rom the 50 lulted strutures, 20 strutures with the lowest totl restrint energy were seleted. Struture figures were generted using MOLMOL 23, MOLSCRIPT 24 nd RASTER3D 25, or GRASP 26. EMSA. Proteins were inuted with 0.25 nm of 32 P-leled DNA proe t 37 C for 30 min in 10 µl of EMSA uffer (20 mm sodium phosphte, ph 6.5, 10 mm DTT, 4 mm MgCl 2, 50 mm NCl, 0.1 mg ml 1 BSA nd 0.1 mg ml 1 poly(di-dc)). The resulting mixture ws sujeted to ntive polyrylmide gel eletrophoresis nd nture struturl iology volume 9 numer 4 pril 2002 255
Tle 1 Struturl sttistis of the 20 strutures of MfG(1 76) R.m.s. devitions 1 Experimentl distne restrints (Å) All (817) 0.031 ± 0.002 Interresidue sequentil ( i j = 1) (237) 0.028 ± 0.011 Interresidue short rnge (1 < i j < 5) (200) 0.036 ± 0.005 Interresidue long rnge ( i j 5) (94) 0.029 ± 0.002 Intrresidue (286) 0.029 ± 0.008 Experimentl dihedrl restrints ( ) (26) 0.27 ± 0.02 Idelized ovlent geometry Bonds (Å) 0.006 ± 0.0001 Angles ( ) 1.73 ± 0.002 Impropers ( ) 1.02 ± 0.01 Energies (kl mol 1 ) 2 NOE 39.1 ± 7.0 2 dih 0.19 ± 0.28 3 repel 8.4 ± 2.1 E 4 L-J 94.5 ± 17.0 Averge r.m.s. differene (Å) 5 Residues 24 64 0.69 ± 0.17 (1.38 ± 0.13) Residues in α-helies 0.56 ± 0.13 (1.34 ± 0.13) Rmhndrn mp nlyses (%) 6 Allowed regions 99.2 ± 1.0 Disllowed regions 0.8 ± 1.0 1 The numer of eh type of restrint used in the struture lultion is given in prentheses. None of the lulted strutures show violtions >0.50 Å for the distne restrints or 5.0 for the dihedrl restrints. 2 NOE nd dih were lulted using fore onstnts of 50 kl mol 1 Å 2 nd 200 kl mol 1 rd 2, respetively. 3 repel ws lulted using finl vlue of 4.0 kl mol 1 Å 4 with the vn der Wls hrd sphere rdii set to 0.75 those in the prmeter set PAR- ALLHSA supplied with X-PLOR 22. 4 E L-J is the Lennrd-Jones vn der Wls energy lulted using the CHARMM 30 empiril energy funtion nd is not inluded in the trget funtion for the simulted nneling lultion. 5 The verge r.m.s. differenes from the energy-minimized verge struture re given for seleted residues. The vlue for kone toms (N, Cα nd C ) is followed y tht for ll hevy toms in prentheses. 6 The vlues were otined y PROCHECK-NMR 31. visulized y utordiogrphy. The K d vlue ws determined s desried 27 on the sis of the results otined using protein onentrtions from 0 to 50 µm. DNA titrtion. The 15-p oligonuleotides 5 -GTGCTGACT- CATCAG-3 nd 5 -CTGATGAGTCAGCAC-3, whih ontin the T-MARE-like sequene, were used. The 2D 1 H- 15 N HSQC spetr of MfG(1 76) with 0, 0.13, 0.26, 0.39, 0.52, 0.65, 0.78, 0.91 nd 1.04 equivlents of DNA were reorded t 37 C. In this experiment, sodium hloride ws dded to finl onentrtion of 50 mm. Sedimenttion equilirium nlysis. Sedimenttion equilirium experiments were rried out t 20 C using Bekmn XL-I nlytil ultrentrifuge. Dt sets were olleted t three rotor speeds of 20,000, 25,000 nd 30,000 rpm nd three protein onentrtions of 150, 220 nd 370 µm. The nine dt sets otined t n sorne of 280 nm were sujeted to glol fitting, nd the moleulr weight ws lulted using nonliner lest-squres tehnique 28. Prtil speifi volume ws lulted from the mino id omposition of the reominnt protein y using SEDNTERP 29. Coordintes. Coordintes hve een deposited in the Protein Dt Bnk (ession ode 1K1V). Aknowledgments We thnk E. Ari nd. Arisk for ultrentrifuge nlysis, T. Med for useful disussion, K. Yp for providing progrm to lulte interhelil ngles nd T. O Connor for ritil reding of the mnusript. This work ws supported y grnts from JSPS nd TARA (T.T.); the Ministry of Edution, Siene, Sports nd Culture of Jpn (H.M. nd M.Y.); JSPS nd CREST (M.Y.); nd PROBRAIN (H.M.). Competing interests sttement The uthors delre tht they hve no ompeting finnil interests. Correspondene should e ddressed to T.T. emil: ttnk@tr.tsuku..jp Reeived 2 Otoer, 2001; epted 24 Jnury, 2002. 1. Motohshi, H., Shvit, J.A., Igrshi, K., Ymmoto, M. & Engel, J.D. Nulei Aids Res. 25, 2953 2959 (1997). 2. Blnk, V. & Andrews, N.C. Trends Biohem. Si. 22, 437 441 (1997). 3. Motohshi, H., Ktsuok,., Shvit, J.A., Engel, J.D. & Ymmoto, M. Cell 103, 865 875 (2000). 4. Ktok, K. et l. Mol. Cell. Biol. 15, 2180 2190 (1995). 5. Ktok, K., ujiwr, K.T., Nod, M. & Nishizw, M. Mol. Cell. Biol. 14, 7581 7591 (1994). 6. Kerppol, T.K. & Currn, T. Onogene 9, 3149 3158 (1994). 7. Wishrt, D.S. & Sykes, B.D. Methods Enzymol. 239, 36 81 (1994). 8. rrow, N.A. et l. Biohemistry 33, 5984 6003 (1994). 9. Hrper, E.T. & Rose, G.D. Biohemistry 32, 7605 7609 (1993). 10. Wintjens, R. & Roomn, M. J. Mol. Biol. 262, 294 313 (1996). 11. Lo, M.C., H, S., Pelzer, I., Pl, S. & Wlker, S. Pro. Ntl. Ad. Si. USA 95, 8455 8460 (1998). 12. Rupert, P.B., Dughdrill, G.W., Bowermn, B. & Mtthews, B.W. Nture Strut. Biol. 5, 484 491 (1998). 13. Ellenerger, T.E., Brndl, C.J., Struhl, K. & Hrrison, S.C. Cell 71, 1223 1237 (1992). 14. Glover, J.N. & Hrrison, S.C. Nture 373, 257 261 (1995). 15. Delglio,. et l. J. Biomol. NMR 6, 277 293 (1995). 16. Grrett, D.S., Powers, R., Gronenorn, A.M. & Clore, G.M. J. Mgn. Reson. 95, 214 220 (1991). 17. Sttler, M., Shleuher, J. & Griesinger, C. Prog. NMR Spet. 34, 93 158 (1999). 18. Neri, D., Szyperski, T., Otting, G., Senn, H. & Wüthrih, K. Biohemistry 28, 7510 7516 (1989). 19. Tomomori, C. et l. Nture Strut. Biol. 6, 729 734 (1999). 20. Vuister, G.W. & Bx, A. J. Am. Chem. So. 115, 7772 7777 (1993). 21. Nilges, M., Clore, G.M. & Gronenorn, A.M. EBS Lett. 229, 317 324 (1988). 22. Brünger, A.T. X-PLOR version 3.1: A system for X-ry rystllogrphy nd NMR (Yle University Press, New Hven; 1993). 23. Kordi, R., Billeter, M. & Wüthrih, K. J. Mol. Grph. 14, 51 55 (1996). 24. Krulis, P.J. J. Appl. Crystllogr. 24, 946 950 (1991). 25. Merritt, E.A. & Bon, D.J. Methods Enzymol. 277, 505 524 (1997). 26. Niholls, A., Shrp, K.A. & Honig, B. Proteins 11, 281 296 (1991). 27. Azm, T.A. & Ishihm, A. J. Biol. Chem. 274, 33105 33113 (1999). 28. Johnson, M.L., Correi, J.J., Yphntis, D.A. & Hlvorson, H.R. Biophys. J. 36, 575 588 (1981). 29. Lue, T.M., Bhirvi, D.S., Ridgewy, T.M. & Pelletier, S.L. In Anlytil ultrentrifugtion in iohemistry nd polymer siene (eds Hrding, S.E., Rowe, A.J. & Horton, J.C) 90 125 (Royl Soiety of Chemistry, London; 1992). 30. Brooks, B.R. et l. J. Comp. Chem. 4, 187 217 (1983). 31. Lskowski, R.A., Rullmnn, J.A.C., MArthur, M.W., Kptein, R. & Thornton, J.M. J. Biomol. NMR 8, 477 486 (1996). 256 nture struturl iology volume 9 numer 4 pril 2002