V4alw IAIE, numkr 3, fl I-m I% r#J @ml duntt IWI t" IWl ~~~~~~4~~ ofWuragea44 ~~~~~~~~~~~ ~~~~~4~~ ~1~~~~~~~~~~~~~ rtfmvKi ~1~~~~~~~~~~~

mdined crystal $~~~~~~~~ d ytterIaium-aubrstituecd carp ~~~~~1~~~~~ 4.25 at 1.5 A, and its ~~~~~~~~~~ with he native and

~~d~i~~a~~~~~tit~~~d structures

vinyl B, KWIW~, %rrna Lad onrd Brim F,P. Edwrds~

b ildtrrron,ui~rrrl~, Sslrmrtm Lelenrr~ry. NC~-Rwlrd~k Chrrepr Rrxcardr crstl ikvdtp~~c~~t Cclr,ic*r, A &Sihxk R~WW~I h#~t~l. PC) ks 0, ~iwkrl&, MD 317W JNI, li”+$d, 1 Eky~nrtmvtr u#C%rtdwy ttutl WlurlrtiWryy, tbIllvcvrily ox Whbur, Qlwdtr NJB

1199 wrd ~Drptrrcwtri rfBImAm&w~. Wtytc Strik+ I/hw~it,r S~hssl rf Afrrtkitrc, DWek. Aft &?Ot. W&4

Rwciue~\ 5 Murclr I‘BI: rwixrd vcmiao ree&%d 15 Mttreh ‘1991

711~ crp;tl ~tpwttm a[ cltrp pwvrlbtrn\in 4,Zj containing II 1 :l mulw rnticl c3f ytwbium ehloridc TV pr~rin htir ken rctincd ut 13 i\ rsriulurian by rpnninccl Icax~-n~~urrex m&o& ro 8, erprratlagrnphic R rulue sl Cl.199, The er@Ml wuctur~ ecWkm~ the MMR %‘l\ldi91% which WW~t th&t hw esnfxntrntiunr et+ yltfxbium CIIISC iw extcnrive clixpl~n~~nr crf eulcium rrtm the EF mail binding rite. h rqmprriron af thcrttcrbiu~~~arrb~tl~ tuted mudel with the nqtivc und radmi~lnl.nrbrrtitrr(ttl xt~ucturc rhaw na rignifkant difkrenrcr. extcpt rrr6und tha substitutrtl EF mCtal-biRdin rcgirm. fhc dixpl~tcrment af’~nj&,~m by ytterbium St the EF xitc he* cttuxcd u m~rmcnt in the polyprptidc b~kb~n~! of SW-91 rrlld Asp-91, fhia

mtwcmcnt F@Ai\lCd in un incrwr in the nunrbrr a$crxrgdn lipmdl bound tu ptcrhiurn in the gp xilc rrsm wen IO eight.

CJaleium bindingprutcin; twhwnidc: Yttrrbiumwbrtituted: X~rwdtructurc

1. lNTRODUCTlQN Up Co this point, controversy remains eowxtning ttlc binding affinities of the n~erals for the two metal-

Larrrhanidcs from lanthanum ro lutetium, form binding sites in parvalbumirr. The original X-ray trivalent ions in aqueous soluiion and have ionic radii erystallagraphic studies on the isomarphous rcplncc- [I] which are comparable to the radius OF divalent ment of calcium with terbium [9] showed an increase in calcium (0.99 A); thus they have frequently been used electron density only at the EF site at low molar ratios to mimic the role of calcium in proteins [2,3]. Lan- of rcrbium to patvalbumin, implying a sequential thanides arc of biological interest because their optical replacement of calcium in the CD and EF sites by ter- and paramagnccic properties can be exploited to obtain bium, The terbium fluorescence data [lo- 121 as a func- information about the structure and function of tion of terbium concentration showed a maximum calcium-binding proteins [4]. The replacement of upon the addition of 1,4-l .8 equivnlerits of terbium calcium by lanchanide ions has been observed in a with quenching at higher terbium ratios, which suggests number of proteins, such as a-amylase [S], trypsinogen an equal displacement of calcium from two sites [6] and thermoiysin f7], follbwed by binding of terbium to a’third weaker site.

The 3-D structure of parvalburnin was originally CavC et al. [ 131 used proton relaxation enhancement established by Kretsinger and Nockolds [8]. They methods to study the binding of gadoiinium and described the unique structural configuration of a reported equal affinities. Rhee et al. .[14] used helix-loop-helix calcium-binding site as an EF hand, a europium and terbium luminescence to study the bin- conformation which is a common structural feature for ding of these metals and reported relatively equal all members of a superfamily of calcium-binding pro- displacement for the two sites. Lee and Sykes [15] teins which includes troponin-C and calmodulin. The studied ytterbium-shifted ‘M-NMR resonances as a two such calcium-binding sites in parvalbumin are function of ytterbium concentration and found sequen- termed CD and EF. Lanthanide ions have been used to tial displacement of the calcium from the two sites. study the affinities by which’ each of the Corson et al. [16] who used optical stopped-flow helix-loop-helix structures in this unique class of pro- kinetics and cadmium NMR to study the displacement teins binds metals. of calcium and cadmium by ytterbium also observed se-

quential displacement in the two sites. The nearly iden-

Corresoo,zdence address: V.D. Kutnar, Macromolecular Structure eical binding affinities of the middle-weight lanthanides

Laboratory, NCI-Frederick Cancer Research and Development for parvalbumin was considered reasonable in view of Center, ARL-Basic Research Program, PO Box B, Frederick, MD the crossover in relative CD/EF site affinities across the 21702-1201, LISA lanthanides. Further NMR investigations suggest that

Published by Elsevier Science Publisishers 3, V. 311

(18), (b) evidence of isomerphoua rcplaccmsnr of calcium with ytterbium occurring only at the EF rncrnl binding site, and Cc) a comparison of che: model with native [I$)]. and cadmium-rubstiturcd models 1201,

2. MATERIALS AND METHQBS

WryxW ol ~ttcrbiun~.substit~~te~~ parvtWmh~ wore grown from Q I : I nrelnr solution of ytterbium chloride (114 mM) and pnrvnlbumin (XL8 mll/ml) u~inp the hrnglnp drop method ax prcviguslr described for the native protein [IO]. The crystals were stabilixcd by ttddin( 0.1 valu~~~eof 65b~rrturntcd nmmonium sulfa~c to the protein drop con= tnjning the cryxtnlr. The xpacc group of the crystals was dctcrmined from preci5sion photogrnpha and rlir unit cell dimcnsionr rrem 14 hiyh&@c rcflcctions refined on the diffractometer. The cryauls belong to rpnrc group C2 with one molccutc in the ssymmctric unit and have unit cell dimensions a = 28.5 A, b = 61 .R A trntl c 81 54.5 A and B = 9J*.

Crystals of ytterbium-substituted carp pqtlbumin were found to be quite stnblc in Ways. The complete 1.5 A resolution da1a Wcrc callcctcd from two crystals on a NicoWSyntcx PZI four circle dif. frecromctcr, which hnd b&n modified for pro1cin darn collection with an extcndcd dctcctor arm and a 385nm-long, hclium~fillcd beam tunnel, The first crystal was used to record 7544 reflections with I > b(l) from 00-1.9 A resolution. The second crystal was used IO record 6684 rcflcc1ions with I > 20(l) in the 2.0-1.5 A range. Radiation damage on each crystal was monhurcd by the Ineasure- mcnt of 5 reflections distributed evenly through the two.0 range after every 200 reflections, The intensity measurements were discontinued when the average intcnsitics for the standard reflections had decreas- ed by 10% for the first crystal and 16% for the second crystal. The symmetry residual was 0,049 for 362 symmetry-related observations on crystal I and 0,092 for 302 observations on crystal 2. The absorption.corrccted data from the two crystals was merged with the ROCKS programs [2l] and placed on a common scale using the reflections in the overlapping range between LO-l.9 A range. The merging residual value for the 966 reflections in fhe overlapping range was 0.084. The final data set contained 12646 reflections in the IO-t.5 A range with I > 20 of a possible 14318 reflections.

Electron density maps were interpreted on an MMSX graphics system using the M3 software package (221.

3. RESULTS AND DISCUSSION

3,1. Refinement The parvalbumin coordinate set 1CPY obtained

from the Protein Data Bank [23], was chosen as the starting point for the refinement because it had the lowest residual while still retaining ideal geometry. The calcium atom in the EF site was subsMuted by ytter- bium in the atomic coordinate File before running PROTIN [l%]. Restrained least-squares refinement was initiated with S-3.0 A data with an overall temperature factor, B = 14 8’. Twenty-six cycles of refinement brought the R value from 0.406 to 0.213, At this point

dnta. Fourier maps naing 2F@-F, caefficisnt4 wcrc

calculated and displayed on rhc graphics system- The electron-density map revealed poor density for the residues I-5, and rhe side chain atoms of residues Ser-39, &p-41, AJpm79, I+=%3 and Lyam87. An F,wF, clectron~dcnoicy map calculated after three cyelcs of refinement (64-66) without these atoms showed unam= bigusus dcnlsity for the residues i-5. Four cycles (67-71) of rcfintlmcnt with the correctly positioned No terminal region reduced the R value to 0.259.

AC this stage peaks in the Fe-fc map rhnc had heights greater than 3 times the standard deviation of the map

and were within 2.5-3,s h of a hydrogen-bonding atom of the protein or R previously included water molecule were lacatcd, A total of 86 acceptabic peaks

were found. They were included in the next, round of structure factor and least-square calwlacions, Nineteen cycles (72-90) of refinement reduced the R value: CO 0.206. Because changes were observed in the coordina- tion of the metal ions in the CD and EF sites, three cycles (91-93) of refinement were performed omitting

the side-chain atoms beyond the CP position for ail residues that directly liganded to the metal ions. Analysis of the resulting F,-F, density map showed chat these atoms wcro correctly fit to the observed density.

In addition the N-acctyl group and some of the side chain atoms of residues Ser-39, &p-41, Asp-79, Lys-83, Lys-87 could be located from the map and were fit into density. A check for contacts for the 86 water molecules revealed that eight of them were symmetry- related molecules and subsequently removed. A total of twenty more cycles (94-l 13) of refinement, were per- formed to give an R value of 0.200, In order to deter- mine if ytterbium caused an equal or sequential displacement of the calcium, an F,-F, electron density map was calculated, omitting the two metal ions. The relative peak height for the metal ion at the EF site was found to be two times higher than the corresponding metal position at the CD site. As a last step the oc- cupancies of the 7% water molecules were refined in three cycles to give an A value of 0.199.

The final I? value for the model, including ail residues, and 78 water molecules is 0.199 for 12646 reflections with I> &(I), in the 10.0-l .§ A resolution range. The distribution of the agreement index R, with resolution ranges for the data is shown in Table I. In the final model t!e rms deviation of bond lengths from ideality is 0.026 A, the rrnsdeviation from planarity for 146 planar groups is 0.01 A, and the rms deviation for the 108 peptide bonds in the rnolecuie is 1.9”. Only 88

312

of 2188 bond dircenceg deviw by more than 0.03 a Since the present work provides the first refined

struerural dewlption of EL 1Pnthanictr=substir\Ited par- from ideal values. vulbumin it is instructive to examine the relative

xtrengthri of the metal-ligand bonds insofar as they can 3.2. Dmription of sttweture be inferred from the metal-ligand bond lengths’in both

The refined model has the same o\crnll structural CD and E,F sires. The CD metal Ion in all three strut- features as the native parvalbwi-h structure. The tures ‘is coordinated by 7 oxygens in a distorted oc- observed peak height at the EF site MS twice that of the tahedral arrangement, Fig. 2 shows the superposition CD sire in R FO-Fe map in which the metals’werc omit- of the CD metal-binding sire for the three refined scruc- ted, This increase in ~leccron density at the EF site at cures. Therms difference in bond length was 0.04 A for Low molar ratios of ytterbium to parvalbumin implies ihe metal-oxygen bonds between the ytcerbfum- a sequential replacement of the calcium by ytterbium. substituted and native structure, and 0.122 A between Based on the number of electronS present in ytterbium the ytterbium.substicuted and cadmium-substituted and calcium, the anticipated peak height for the EF site structure. As it is evident from the small rms difference should be 3.S times higher than CD site. Hoticver, the in the atoni posirians that the relative strengths of the discrepancy in the peak heights could be attributed to metal-ligand bonds at the CD site are very similar in all (1) a partial substitution of calcium by ytterbium at the three scructurcs. Table II compares the bond distances CD site, or (2) an incomplete substitution of calcium by for the CD metal-binding sites for the three structures. ytterbium at the EF site, or (3) differences in the The metal-oxygen distances for the three models rahge

Fig. 1. Stereo view of theCa superposition of Carp parvalbumin (solid lines), thecadmium-substituted structure (dashed lines) and the ytterbium- substituted structure (dotted lines),

313

Vtalulpr m,. nu’mkr 2 N%~ ~~~T~~ hilW WI

berwcsn 2.02 k and a-77 A wlrh an ~v~r~~~ v&Ilur for ‘l”#W Ii

rke 7 2.41 A,

#xy#en &tarns ranging bzltwacn 1.31 (ei tend Ctrmfiiarl~ al ~hr rn~~~~~~~~~~ BIw~w\ (Al 01t flrc CR m*M ~~~~~~~ &a

The EF mm1 is rlfa coardinatctl by oxygm ~umso in VJV Ic-“.- _., I”_ g\ distorted ~~t~h~~r~~ poultry. Fig. 3 shows tlrc t.l@WWf ~~~ C#klwm ~~~~~~~1 rlrirblunt superposition of’ the I% rn~ta~,~~~d~~$ site fc%r 8111 three?

W-m

refined stq~uren. The rms diffarrtncc in the h@I eXM 3.Zl 3.18 &?I

mcrsrl-oxygen bond&ngtho wax O,M A and 0.204 A hSP12 c3B1 1.15 ?.Jcx Ili*PSS 614 xiBt J‘fJ :::J:

laf tfic native and ~~~~~u~.~u~~tit~~t~~ psaccina WWsa 0 X.YT 1.,9$ Z.TJ

reopcetivcly, The moat drastic change is in, the relative GhlGs QR I 2.34 1.94 2.43

pmirisns of thr palypoptic!c backbcsne: atoms of Ser-91 dkl42 cm I ,1,BI 3.71 &$I

@&Aq3-92, Ar U resrtl! af this chulqgc the curircaxyl ox- ClIlJ4 OF,4 3.41 Z.98 34

ygcn atom tOD2) of Asp92 is now ctoncr ta rhe ycter- AWrt@! ?.41 2.40 1.38 PS=emasae%l (Ilr 1

bium ion by 1.12 A compared with the ctrrrcaponditlg metal-oxygen dirrttlnee in rhr native structuri?. The IYW metal ion relative to the native structure. 8~ch ahifts in jor difference between the three structures is the rhe metal ion hrzvc been observed in tllerrnelyain when number of’ oxygen iigands of the! metal ion: in a enlciurn ion is replaced by innthnMcsi, and were ytterbium-substituted parvalbumin there are 8 oxygen found to vary with their atomic numbers 171. The shifts ligands as compared to 7 oxygen ligands in both the in position were the smallest for the lanthnnides of native and cadmium-substitutcG proteins, The increase higher atomic number, presumably kct\use of the from 7 to 8 is as a result of botl~ bidenrata carboxylalz closer similaricy of rheir ionic radii to that of calcium oxygens of Asp-92 now forming, ligands with the yttcr- ion. bium. Table III compares the bond distances for the EF The average temperature factor (B) for the main site for all three models. The average metal-otygcn chain atoms (20.28 f 7.0 A’} of the yrterbium- distances for rhe three models are bctwecn 2.35 A and substituted structure is h,igher than for the native or the 2.42 A. Coordination numbers of 3 br 9 are not uncx- cadmium-substituted structure (la.0 and 16.7 ia’), but petted with oxygen ligands and have been observed for they follow the same trend as the native and cadmium- rare earth metals [25,26]. Although the average substituted structures, The average temperature factor metal-oxygen distance (2.37 A) for the EF metal site in for the ligands around the EF metal-binding site in all the~yrterbium~substitutcd model did. not differ from the three models is higher than that for the ligando around other models significantly (2.42 A and 2.35 A for the the CD metal binding sire, suggesting the EF site may native and cadmium-substituted models), the metal be a more flexible environment. In addition, a com- binding pocket appears to be slightly more compact as parison of the temperature factor of the two metal ions a result of the movement of certain oxygen ligands, in each of the three models shows higher B’s for the These shifts are more pronounced iri the X (Asp-90) metal ion in the EF site, suggesting that it is in a less and -X (Water-128) direction. Another interesting constrained environment and thcrcfore subject to observation is a shift of 0.11 A in the position of the greater movement.

Fig. 2. Stereo view of the superposition of atoms around the CD metal-binding site for the native (open bonds), cadmium-substituted (single bonds) and ytterbium-substituted (filled bonds) parvalbumin structures. The metal atoms for all three models are indicated as open circles (Call 10)

in the center, The bonds to the oxygen ligands are shown as dashed lines.

314

The find refined model of ytterbium-subscic\rz@d par- vnlburnirr revealed 78 ordered water mokulcs, of which 43 have full occupancy, end the rest are partially occupied, the Lowest occupancy being 0.74. Com- psrisan of the ordered solvent structures with the native and the cadmium-substituted models show 413 of the 78 ordered water molecules are laenred tit similar positions (i.e. within 1 A), having the same hydrogen-bond partners.

The results from the refinemcnc of the ycccrbium- substituted parvalbumin structure indicate chat at low molar ratios of ytterbium co parvalbumin (1: I) results in an extensive replacement of calcium by ytterbium at the EF site. This finding supports the NMR studies reported earlier by Lee and Sykes [15], The meral displacement results in only minor perturbation of the overall structure, but significant changes around the substituted EF metal binding site. For instance, both carboxylate oxygens of Asp-92 are now bound by the ytterbium, resulting in an increase in the number of ox- ygen ligands to the EF metal from 7 to 8. There is a small displacement in the position of the substituted yc- terbium ion, resulting in shorter metal-oxygen ligands

with some residues, Higher cemperncurc fwwrs at the EF site suggest the EF site may be a more flexible en- vironment chnn the CD ntetal aice, thus making it more accessible for ditiplaeement. These rosulcs support NMR studies which suggest that the EF domain ligands are flexible whereas the CD domain ligands are rigid [27]. The present study on ytterbium-substituted carp parvalbumin shows the replacement of calcium by yt- terbium rebuits in only minor changes in the overall structure, thus supporting the ability of the lanchanidc ions co mimic the biological function of calcium.

nck/ro~c,/erlff~tt~rnrs: WC ttinnk Dr Amy Swain r0r helpful discuri- sions, and providirlg us with the coordinntes ofcndtnium~suhstitutcd parvalbumin structure. This work wns supported in pnrt by NIH Grants GM33192 and RR 02945, a grant from NSERC (L,L.), and n grant-in&t from the WSU center for Molecular Biology to B.E. This work was also sponsor& in part by rhr National Cancer In- stitu~c, DHHS, under contract NOI-CO-74101 with ABL (V.K.), The contents of this publication do not necessarily reflect the views or policies of the Department of Health and Human Services, nor dots mention of trade names, commercial products, or organizations lm- ply endorsement by the US Government.

REFERENCES

Table III

Comparison of the metal-oxygen distances (dr) in the EF metals binding site

Ligand

Asp-90 0D2 Asp-92 ODI Asp-92 0D2 Asp-94 ODI Lys-96 0 Wat-128 0 Glu-101 OEl Glu-101 OE2

Average

Calcium Cadmium

2.24 2.26 2,37 2.39

2,il 2.17 2042 2.30 2,35 2.39 2.381 2.26 269 2.66

2.42 2.35

Ytterbium

2.14 2,29 2.62 2.48 2.28 2.13 2.35 2.69

2.37

[I] Shannon, R.D. (1976) Acta Crystallog. A 32, 751. [2] Darnall, D.W, and Birnbautn, E.R. (1970) J. Biol. Chem. 245,

6484-64X8. [3] Martin, R.B, and Richardson, F.S. (1979) Quant. Rev.

Biophys. 12, 181. [4] Martin, R.B. (1983) in: Calcium in Biology (Spiro, T.G, ed,)

p, 235, Wiley, New York. [S] Smolka, GE., Birnbaum, E,R. and Darnall, D, (1971)

Biochemistry IO, 4556-4561. [6] Darnall, D.W, and Birnbaum, E.R. (1973) Biochemistry 12,

3489-3491. [7] Matthews, B.W. and Weaver, L.H. (1974) Biochemistry 13,

1719-172s. [8] Kretsinger, R,H, and Nockolds, C.E. (1973) J. Biol. Chem.

248, 3313-3326. [9] Sowadski, .I., Cornick, G. and Kretsinger, R.H. (1978) J. Mol.

Binl. 124, 123-132.

Volume ~]tl3, number 2 FK[~S LETTI~I~$ Jun~ 1991

IlOI I)~n~l~, I1., Jr. and M~rdn, l~J4, t19~4~ Hi~¢h~m|~ I L

1111 Nelson. D,J., MilIll, 'r.L, ~nd Martin, R.B. 119~'/1 ~ivlri~rll,

1121 Miller, T.L., CCvk, R.M,, Nd~n, DJ. ~nd Thcvh~dt~. A.D. 1191101 J, MoI , Itiol, 141, ~2}-2~6.

[141 Rh~,~, M.J., $undi~k, D.R., Atkl~. V,.K. ~nd Hor~o~ks. W,D~W., Jr. {19~1) ~iO~h~mislry 20. ~2~) }34 ,

ll~l Lee, L. ~nd Sykes, B.D. (191l)) lllo~h~mlslr): 22, 4J66~4}';I~I, 116) Corson, D,C., Williams, T.C, .nd Sykes, B,D, (191t~1)

Bioch~mtslr)e 22, 3~#2-.~899 liT] Cot~on, D,C,, Willi~ms, T.C, und Syk~L ~.D. (19#))

Biochemistry 22, 5S82-;~IZS9, [l~| H~udri~kson, W,^, m~d KonnerL J,H, (19R0) in: Comp.ll.~ h~

Crysmllollrapl~y (Diumond, R,, k~ma~sh~n, S. ~nd Venk~t~s~ln, K, ¢ds) Chapter 13, pp. 1-23. Indium A¢~tlemy of Sciences, B,~nttMore,

I191 Kmn,r. V,. L++. L. and Bdsvmrd+. PI,P.P./I~+ 141mh+mh.y 29. 14~ I:41~,

Pol ~|n, A,L.. Kr~t~lnll~r, R.H, ~Iid Amm~, I~.L. (I~3m9) J, l]|pl.

1211 Rcwk¢. ¢LN, t19~4) J, Appl, CrX~I~IIolIr. I?, 125. I~'~1 ~i~l~ki, A,R., J~mcs, M.N.¢3, mid Broullh|on, C,~. {lgt~] in:

Cry~ll~r~phi~ Cempultnlt t~#~r~, D,, cd,) pp, 409, O~r~rd Univ~r~ily Pr~s~. L¢,d~n,

12~1 Mo~w~. P.C. and grel,~inltlr, R,H, (197~) J, ,Mol, Bt~i. 91,

1241 Res~rn~nn, M.G, msd Arl~s. P, (I~?;S) J. Btot. Chcm, 250,

12~I H~Imholx, L, (19~19] J. Am, Chum. $e~, 61, I$44~I~0 1261 M~rtln, L,L, and JaCobson. R.^, (19'~2) Inorp, Chum, II,

2"~g9. [2"~1 Williams, T,C,, Corsou, D.C. and Sykes. ~.D, (1984) .I. Am.

316