J. Mol. Biol. (1992) 224, 413-426

Structure of a Sarcoplasmic Calcium-binding Protein from Nereis diversicolor Refined at 2.0 A Resolution Senadhi Vijay-Kumar’y 3 and William J. Cook2$3t Departments

of ‘Biochemistry,

2Pathology and 3Center for Macromolecular of Alabama at Birmingham Station, Birmingham, AL 35294, U.S.A.

Crystallography

University

UAB

(Received 23 August

1991; accepted 11 November 1991)

The crystal structure of a sarcoplasmic Ca*+-binding p rotein (SCP) from the sandworm Nereis diver&color has been determined and refined at 2.0 A resolution using restrained leastsquares techniques. The two molecules in the crystallographic asymmetric unit, which are related by a non-crystallographic S-fold axis, were refined independently. The refined model includes all 174 residues and three calcium ions for each molecule, as well as 213 water molecules. The root-mean-square difference in co-ordinates for backbone atoms and calcium ions of the two molecules is 0.51 A. The final crystallographic R-factor, based on 18,959 reflections in the range 2.0 A I d I 7.0 A, with intensities exceeding 2.0 r~, is 0.182. Bond lengths and bond angles in the molecules have root-mean-square deviations from ideal values of 0.013 A and 2.2”, respectively. SCP has four distinct domains with the typical helix-loop-helix (EF-hand) Ca *+-binding motif, although the second Ca*+-binding domain is not functional due to amino acid changes in the loop. The structure shows several unique features compared to other Ca *+-binding proteins with four EF-hand domains. The overall structure is highly compact and globular with a predominant hydrophobic core, unlike the extended dumbbell-shaped structure of calmodulin or troponin C. A hydrophobic tail at the COOH terminus adds to the structural stability by packing against a hydrophobic pocket created by the folding of the NH, and COOH-terminal Ca*+-binding domain pairs. The first and second domains show different helix-packing arrangements from any previously described for Ca *+-binding proteins.

Keywords: sarcoplasmic

calcium-binding

protein;

X-ray

crystal

structure

Although the exact function of SCPs is unknown, they are probably responsible for the maintenance of a buffered intracellular Ca*+ concentration. Their inability to form stable complexes with amphiphilic peptides or to bind proteins when mixed with muscle extracts suggests that SCPs are incapable of protein-protein interaction. The amino acid sequences for SCPs from four different phyla have been reported (Takagi & Konishi, 1984a,b; Takagi et al., 1984, 1986; Collins et al., 1988; Jauregui-Adell et al., 1989). Sequence studies have shown that polymorphism of SCPs is quite common; only two SCPs have been shown to consist of one isoform (Cox & Stein, 1981; Collins et al., 1988). All SCPs contain four putative Ca*+-binding domains with the helix-loop-helix conformation that has been called EF-hand (Kretsinger & Nockolds, 1973), but not all of the Ca*+-binding domains are still functional. Ca*+ binding has been preserved in the first and third

1. Introduction Sarcoplasmic Ca*+-binding proteins (SCPsS) are subfamilies of EF-hand major one’ of the Ca*+-binding p roteins (Moncrief et al., 1990). They Ca*+-binding system in inverteare an important brate muscle (Cox, 1989) and are probably the functional counterparts of parvalbumins, which have been found only in vertebrates. SCPs usually contain a mixture of Ca*+ and Mg*+-binding sites. In general, their Ca*+-binding properties are complex and involve positive co-operativity, with two classes of sites of different affinity and (Wnuk et al., 1982). pronounced Mg*+ antagonism 7 Author to whom all correspondence should be addressed. t; Abbreviations used: SCP, sarcoplasmic Ca’+-binding protein; SCP-1 and SCP-2, molecules 1 and 2 in the asymmetric unit; m.i.r., multiple isomorphous replacement; r.m.s., root-mean-square. 413 0022-2836/92/060413-14

$03.00/O

0

1992 Academic

Press Limited

414

8. Vijay-Kumar

domains of all known SCPs, and most SCPs also have-a third site in either domain II or IV. There are some interesting amino acid changes in the Ca2+-binding domains of SCPs compared to most other EF-hand proteins. For example, in the overwhelming majority of functional EF-hands, the 12th residue in the Ca’+-binding loop is glutamate, whereas in the first domain of all SCPs it is aspartate or asparagine. Also, glycine residues at positions 4 and 6 in the Ca2+-binding loop are often replaced by other amino acid residues in SCPs. Another difference between SCPs and other Ca’+-binding proteins is the highly variable number of residues between Ca2+-binding loops and a long COOH-terminal stretch of 17 to 32 residues. This irregular spacing of the Ca’+-binding domains has not been observed in the other subfamilies of the EF-hand family, such as calmodulins and troponin C, which display symmetrical positions of the Ca2+-binding sites. SCP from the sandworm Nereis diversieolor is a single polypeptide chain that contains 174 amino acid residues and has a molecular mass of 19,485 daltons (Collins et al., 1988). Only one isoform has been identified. There are three Ca’+-MgZf sites that are probably occupied by Mg2+ under physiological conditions, although Ca2+ uptake causes a release of Mg” (Cox & Stein, 1981). The second Ca2+-binding domain is non-functional because of amino acid changes in the Ca2+-binding loop. The high resolution X-ray structure of Nereis SCP was undertaken in order to examine the changes caused by having a non-functional domain paired with a functional

Ca’+-binding

domain

and to see if there

were any differences in Ca’+-binding sites that might explain the affinity for Mgzf as well as Ca’+. In addition, we wanted to see what changes, if any, occurred as a result of the amino acid changes in the

Ca2+-binding

loops mentioned above.

2. Experimental

Procedures

(a) Data collection Nereis SCP was crystallized from solutions containing 60 y0 (w/v) ammonium sulfate and 25 mM-CaCl, at pH 78 (Babu et al., 1987); all crystals used for data collection were obtained by seeding. The crystals, which grow as large rectangular prisms, belong to monoclinic space group P2, with a = 43.6(l) A, b = .56.0(l) A, e = 658(3) A (1 A = @l nm), and /? = 92.6(l)“. There are 2 protein molecules in the asymmetric unit (SCP-1 and SCP-2), and the solvent content is approximately 40%. Intensity data for the native and derivative crystals (KAu(CN),, K,PtCl, and Lu(C,H,O,),) were collected at room temperature with an Enraf-Nonius CAD4 automated diffractometer to a resolution of 3.0 A as described by Cook et al. (1991). For the high-resolution refinement, intensity data for native crystals were collected at room temperature with a Nicolet X- 1OOA area detector at 22 “C using CuKol radiation from a Rigaku RU-300 rotating anode generator operating at 40 kV and 100 mA. Two crystals were used for data collection. The detector 20 value was 20”, and oscillation frames covered 025”. For the 1st crystal, the detector-to-crystal distance was

and W. J. Cook 10 cm, and 720 frames of data were measured for 189 s each. For t,he 2nd crystal, the detector-to-crystal distance was 12 cm and a total of 1440 frames were measured for 200 s each (720 frames were collected, and then the crystal was rotated 90” about the phi axis and another 720 frames were collected). Indexing and integration of intensity data were ca.rried out using the XENGEN processing programs (Howard et al., 1987). Native data to 20 A resolution were collected. A total of 73,192 reflections were merged into 20,117 unique reflections; this represents about 92% of the possible reflections at this resolution. The R,,, value (based on I) for the data to 2.0 A was 0088. (b) Development of the initial

model

Initial phase estimates for the structure of Nereis SCP were derived from the 3 isomorphous derivatives. Following completion of the structure refinement at 20 A resolution, cross-difference Fourier maps were calculated for each derivative, using the native phases. The 6 Lu ions substitute for Ca’+ in the 3 Ca’+-binding loops in each SCP molecule, although the Lu3+ positions obtained from the final difference Fourier map differ from the refined Ca’+ positions by @4 to 1.3 A, with an average deviation of 0.9 A, The highest substitution of Lu3+ for Ca*+ occurs in the first Ca*+-binding domain of each SCP molecule (Lust sites 1 and 4). The binding of the heavy-atom derivatives is summarized in Table 1. Except for one of the minor Au sites, the degree of heavy-atom binding was higher in SCP-1 than in SCP-2. Phase angles for the native data were calculated from the 3 isomorphons derivatives including anomalous dispersion effects. The overall figure-of-merit for phases based on the 3 derivatives was 0:69 for the complete data set to 3.0 A resolution (6269 reflections). An electron density map for the protein was calculated including the native structure factors with d > 3.0 A and using centroid phase angles with figure-of-merit weights. Although this map showed the molecular boundary and 8 a-helices, the polypeptide chain could not be followed completely without ambiguities. A modified electron density map was obtained by solvent-flattening techniques using the ISE%/ ISAS package of programs developed by Wang (1985). The initial molecular envelope was determined by assuming a solvent fraction of 30%. A vector for the noncrystallographic 2-fold axis was obtained from the heavyatom positions, and the electron density of the solventflattened map was averaged based on the non-crystallographic symmetry. This map resolved most of the ambiguities in chain tracing, and a partial model for residues 90 to 158 was built and regularized using the computer graphics package FRODO (Jones, 1978). Phases were calculated from this model and combined with the original multiple isomorphous replacement (m.i.r.) phases in the resolution range 5.0 A to 30 A. The resultant map showed the protein density more clearly, allowing all of the model, except residues 61 to 70 and 160 to 174, to be built. The phase combination procedure was repeated, and the remaining residues were then identified. At each stage of model fitting, the corresponding portions of the 2nd molecule in the asymmetric unit were generated using the non-crystallographic symmetry. The entire model then underwent 2 cycles of refitting using OMIT maps (Bhat & Cohen, 1984), followed by refinement using the program PROLSQ (Hendrickson, 1985). The crystallographic R-factor for the final graphics model, based on the 50 A to 3.0 L%data and using an overall isotropic temperature factor of 15 A.‘; was 0.22.

Structure

t3fNereis

SCP at 2-U A &.solution

415

Table 1 Analysis

of heavy-atom

&ding

Lu((:,H,O,),

6

1 05 : 0.7 : 0% : 0.3 : 0.3

KAu(CN),

4

I : 0.2 : 0.7 : 03

K zl’t(:l,

3

1 .03 : 0.2

to ?;ereis SCP

Sites 1 and 4 Substitute for (:a” in domain I Sites 2 and 5 Substitute for (la’+ in domain III Sites d and 6 Substitute for Ca” in domain IV Sites I rend3 Argl13. Lysl27 Sites ‘2 and 4 PheW, M&32, ValOl Sites 1 and 3 Met50 Site 2 Met26 (SCP molecule 2)

The r(:laf ive occupancies are on an arbitrary scale and were obtained from cross-difference:Fourier maI’s calrulsted wing the final rcfinctd native phases. The site with the highest occupancy in each derivative is given the value 1.0.

was refined b3 using PROISQ ‘l-h: model (Hendrickson, 19X5), alternating with manual rebuilding using computer graphics. ltcfinement proceed& in 13 rounds, resulting in significant improvement, in the geometry of the model. A round of refinement consisted of a session of model-building from electron density maps, followed by cycles of least-squares refinement until convergence was achieved. The starting point for refinement was the initial model fitted to a 3.0 .8, m.i.r. electron density map. The model was fit to sum-Fourier maps in the first 2 rounds, followed by 2 rounds of model rebuilding using OMIT maps. In round 5. the data were added in 2 steps to 2.8 A resolution, and subsequent model building was done to fit, the electron density calculated with m.i.r. phases for 12 to 50 A data and calculated phases beyond the 5.0 .& data. Data above 5 .%were not included in the refinement. since they would be especially influenced by disordered solvent. At this point. the features of the sum-Fourier and OMIT maps were very similar. indicating no large changes in the model. In the next 4 rounds, the resolution was gradually extended to 2.2 -4, and the model was rebuilt from 2F,--E; maps. The H-factor at. this rrsnlut.ion was 0.25. In round IO, the examination of difference: Fourier electron density mtlps indicated the prcscncc of a number of solvent molecules. I’caks were identified LS wa.ter molac:‘ult:s when t)hey had well-defined electron densities greater than 3 times i,hc standard deviation of the map and wcrc less than 35 ,h from a polar protein group or another water molecule. Using these criteria, 62 peaks were assigned as oxygec atoms from water molecules and given isot.ropic temperature factors of 20 a2 and occupancies of 1.0. In round I I, individual thermal parameters were introduced with restraints on the differences of tcmpcraturc factors between connected at.oms. Based on a difference Fourier map calculated after this rcfincmcnt. an additional 48 water molecules \vere identified. In round 12. the resolution was extended in 2 steps to 24 A. In subsequent. rounds of refinement, t.he refined tenlperature factors of 84 of the total 213 water molecules were in the range 40 tn 52 A2. Inspection of thcsct positions in the 2F,-FC and F0 - FC maps (*Iearly showed well-defined electron density

and reasonable distances from a polar group of the protein or from other water molecules. Therefore, itssuming that t,his water structure was disordered. the temperature factors and occupancies were refined together. but shifts were applied separately in alternate cycles. Por the last round of refinement, the lower resolution limit wa.~ increased from 5.0 x to 7.0 .A. .4t the end of the refinement. the occupancies of these 84 disordered water molecules wcrc found to bt: in the range: 0.4 to ti7 and the B-factors were 17 to 37 AZ. 0uring the refinement, 3 of 24 amide side-chains were noted to have large differences between the temperature factors of t.he nitrogen and oxygen atoms. For all amide side-chains, the averagr dikerence was 1.7 A*, whereas fnr t,hese 3 side-chains it was 4.9 A’. After interchanging the atomic positions for nitrogen and oxygen atoms in these 3 amide groups, the average differences in temperature factors were 1.0 A2 for all amide side-chains and 2.5 A2 for the 3 that were interchanged. The co-ordinates for +‘ereis SW have been deposited with the Protein Data Bank: Hronkhaven .Vational Laboratory. Upton. SY I1 973. I:.S.A.; accession number PSCP.

3. Results and Discussion (a) Quality

of the structure

The final model includes 2736 protein atoms and six Ca2+ ions from two SW molecules and 213 solvent molecules modeled as oxygen atoms. The R-index, based on 18,959 reflections in the range 2.0 A I d I 7.0 A with intensities exceeding 2.0 rr, is 0.182. The R-index for all reflections in this ra.nge with no sigma cutoff is (kl84. Tahk 2 gives t.he R-index tabulated as a funct.ion of resolution. Superposition of the R-factor curve with theoretical curves for different mean positional errors places an upper limit of 0.15 A on the root-mean-square (r.m.s.) error in the atomic co-ordinates (I,uzatti, 1952). Comparison of the co-ordinates of the initial SW model at 3.0 A resolution wit,h those of the

416

S. Bijay-Kumar

Summary

Table 2 of the reJinement statistics Number of reflections

Resolution 7.00-5.50 550-3.80 3.80-2.95 2.95-2.65 265-2.40 2.40-2.20

290-200

(A)

Total

I > 20(Z)

542 2171 3617 2525 3104 3420 4200

538 2155 3559 2462 2991 3313

3941

R 0301 0.156 0.161 0.188

0198 0.189 0196

R is defined as ZllF..l- IFJl)/ZIFOl

refined model at 2.0 A shows a r.m.s. difference of l-2 a for main-chain atoms and 2.8 .& for all atoms. Although each molecule in the asymmetric unit was refined independently, they show very similar geometry. The overall r.m.s. differences between main-chain atoms and between all atoms are @5 a and 1.2 8, respectively. Figure 1 shows a stereo drawing of the two molecules in the asymmetric unit with their C” atoms superimposed.

and W. J. Cook

As listed in Table 3, the r.m.s. variation in distances from ideal values falls close to or within targeted variances. The final co-ordinates of the SCP model deviate from ideal bond lengths and angles by 0013 ,& and 2.2”, respectively. Dihedral angles of the main-chain conform well to their expected values (Fig. 2). All the phi-psi pairs for non-glycine residues fall in allowed regions of the Ramachandran plot (Ramachandran et al., 1963 j, except for four residues in each molecule that occupy special positions in the Ca2+-binding loops. Three of these four residues (Lysl9, GM07 and Asn141) occur in the fourth position of the 12-residue Ca2+-binding loops in domains I, III and IV; this position is usually occupied by a glycine residue. The average phi, psi angles for this position are 58” and 36”, respectively, which are more easily satisfied by a glycine residue (Strynadka & James, 1989). The fourth residue is AsnlO9, which occupies the sixth position of the Ca”-binding loop in domain III. This position is almost always occupied by a glycine residue, since the average phi, psi angles for this position are -90” and -3” (Strynadka & James, 1989). The variances within categories of temperature

Figure 1. Stereo drawing of the 2 SCP molecules in the asymmetric a least-squares procedure. The first and last residues in the molecules region between helices B and C.

unit. The C” backbones have been superimposed by are labeled. The only large difference is in the linker

Structure of Nereis SCP at 2.0 A Resolution

417

Table 3 Summary ,of the least-squares rejinement parameters r.m.s. delta Distances (A) Bonds Angles Dihedral Planes (A) Chiral volumes (A) Contacts (A) Single Multiple Hydrogen-bonded Torsion angles (“) Planar Staggered Orthonormal Thermal (A B, A’) Main-chain bond Main-chain angle Side-chain bond Side-chain angle

Target 0

0013 0.036 0.043 0011 0.183

0.015 0027 0035 0.015 @150

0350 @228 0.273

0400 0.400 0400

- 25 % G 15 5 0

20

40

60

20

40

60

80 100 120 Residue number (a)

140

160

140

160

35 -

25

30 @Of 0.07

2.1 22.0 25.5

1.000

0934 1.510 1.403 2.291

1.500 1.200 1.800

2G

5 15 I I/ 0

80

100

120

Residue number (b)

t In the first 12 rounds of refinement, these target (r values were set at 150” and 2@0”, respectively. For the final round of refinement, the constraints were completely removed.

Figure 3. Mean temperature factors for (a) main-chain atoms and (b) side-chain atoms of SCP-1.

factors fall near their intended values. Figure 3 shows the temperature factors averaged over atoms for each residue of the main-chain and each sidechain. In general, there is a good correlation between B-factor and solvent accessibility of the chain. The residues in the Ca’+-binding loops tend to have the lowest temperature factors, whereas the more flexible regions of the molecule are the linker regions that connect helix-loop-helix domains. The highest temperature factors in the molecule occur in residues 62 to 67, which are part of the non-functional Ca2 + -binding loop. One other region of high thermal motion is found in residues 161 to 164, which form part of a large loop after helix H that

changes the direction of the polypeptide backbone and enables the hydrophobic tail to pack against the hydrophobic pocket between the two Ca2 + -binding domain pairs. Overall, the temperature factors for SCP-2 are higher than those for SCP-1. The final 2F,- FG electron density map was of good quality, as shown in Figure 4. There are only two regions where the electron density is poorly defined. One of these is the loop in domain II, where the sequence is Gly66-Gly67-Lys68-Gly69. The other region includes residues 126 to 129, which form a type I tight turn. These residues in SCP-1 are well ordered, but in SCP-2 they appear to be disordered. (b) Conformation

180”

PSI

x \

-180”

Phi

180”

Figure 2. A Ramachandran plot of the phi and psi angles for each residue in SCP-1. ( x ) Glycyl residues; (0) remaining residues.

of the molecule

SCP is compact and has overall dimensions of approximately 25 a x 35 a x 40 8. There are four domains that form two pairs of helix-loop-helix motifs (EF-hands). The two domains in each pair are related to each other by an approximate 2-fold rotation. The loops of the four domains are located on the surface at opposite ends of the molecule. For the two SCP molecules, the average distance between Ca2+ ions in the COOH-terminal half is 12.0 d; the average distances between the Ca2+ ion in the first domain and the Ca2+ ions in the third and fourth domains are 27.7 A and 32.1 A, respectively. The eight a-helices, designated A to H, consist of residues 1 to 15, 25 to 38, 43 to 62, 72 to 84, 89 to 103, 113 to 122, 127 to 137 and 147 to 159. These eight a-helices comprise approximately 58% of the structure, which is similar to the percentage found protei ns (Babu et al., 1988; in other Ca ‘+-binding Herzberg & James, 1988; Satyshur et al., 1988;

418

S. Pijay-Kumar

and W. J. Cook

Fig. 4.

Szebenyi & Moffat, 1986; Ahmed et al., 1990). The non-helical portions of the molecule include the three linker regions 38 to 42, 85 to 88 and 123 to 126, and the long COOH-terminal stretch from Asnl60 to Va1174. The lengths of these linker regions between domains are similar to those of calmodulin and troponin C. The linker between domains II and III contains a tight turn at residues 85 to 88 (Fig. 4(a)), which is responsible for the much more compact shape of Nereis SCP compared to the four-domain proteins calmodulin and troponin C. The COOH-terminal extension is quite contorted, with three tight turns in an eight-residue stretch from Ser164 to Gly171. Residues 160 to 167 are hydrophilic and form a large loop on the surface of the molecule. Residues 168 to 174 are hydrophobic and are tucked into a hydrophobic pocket formed by residues from helices A, E, G and H. The two independent molecules of SCP in the crystallographic asymmetric unit are very similar. The r.m.s. deviation between main-chain atoms of the two molecules is only 651 A. From the refined atomic co-ordinates, the two independent molecules in the asymmetric unit are related to each other by a rotation of 181.7” and translation of 25.6 A. There are only 11 contacts of less than 3.5 A between polar

atoms of the two molecules in the asymmetric unit (Table 4). Although there is no extensive contact between the two independent molecules, there is a prominent water structure at the interface, with 12 water molecules forming hydrogen bonds to both molecules. Of the four EF-hand domains in Nereis SCP, only

Ta Hydrogen bonding and favorable dipolar interactions between the two molecules in the asymmetric unit Molecule I

Molecule 2

Residue

Atom

Residue

Atom

Distance iA)

Ser52

OG

Asp58 Asp58 Am59

0 0 ND2 ND2 OG OEl NE NH1 OEl OEl

Am59 ArglOl ArglOl Ser52 Glu148 Glu93 Ser90 Asp58 Asp58 As1185 Am85

ND2 NE NH1 OG OEl OEl OG 0 0 ODl ND2

230 249 3.24 303 2.85 3.30 2.71 353 271 295 318

Am85 Ser90

Glu93 ArglOl ArglOl Glu148 Glu148

Structure

qf Nereis

SCP at 50 A Resolution

(b) Figure 4. Representatiw: stereo drawings elcct,ron density is from a 2F,,- EL map that to 88. This type 1 turn bctwecn the 2nd and compared to calmodulin or troponin C. (b)

of .Ver& WI’ was prepared 3rd EP-hands Residues I I5

with t.he superimposed elwt~ron density cont,our surfaces. The using phases calculated from the refined model. (a) Residues 85 is responsible for the much more globular shape of Xereia Xl to 125. This is t.he exiting helix of the 3rd EF-hand.

the first., third and fourth bind C:a’ ’ ions. The second domain is incapable of binding Ca’+. although it still maintains the helix-loop -helix 2-fold rclat,ionconformat.ion and a.n approximate

calciun-oxygen distances in the three .loops are 2.39, 2.31 and 2.37 A. respec:t.ively (Table 5). These values are in general agreement. with the dist.anccs found for 7-fold calcium-co-ordinatiorl polyhedra in crysta.1 structures (Einspahr & Bugg. 1981). The first. &“-binding l&and in each loop is furnished by an Asp side-chain. The next five ligands, from three side-chains and one carbonyl oxygen atom, arc approximately coplanar with the Ca ion. The rcsiduos at the first, third and fifth positions in each loop contribute a carboxylatc oxygen atom. The residue at the seventh position pont.ributes its carbonyl oxygen atom. The 12t.h residue in the Ca2+ -binding loop is either aspartate (1st loop) or glutamate (3rd and 4th loops); this residue binds to the Ca ion through bot,h carboxylate oxygen atoms. The seventh ligand in each loop is a water molecule, cxccpt for t,he third domain of SCP-2. which has only six ligands with no co-ordinating wa.t.cr molecule. However, there is a wat.er molcculo a.t a distance of 3-6 ;\ from the Ca ion in this loop. 1t, appears that the water molecule from

ship to the first domain. The overall conformation of domains I and II in the NH,-terminal half is markedly different, from t.hat described for other pairs of helix-loop -helix domains (Fig. 5). On the ot.her hand, domains III and IV have the typical helix-loop--helix conformation described in other Ca2--binding proteins with pairs of (la’+-binding doma.ins (Rabu et al.,. l!J88: Herzberg 8r. ,Ja.mes, 1988; Satyshur ef al.. 1988: Ahmed et al.: 1990).

(c) C:nlcium, co-ordinati(m sites exhibit 7-fold coAll t.hree (:a’+-binding ordination (Fig. 6). The seven ligands are arranged around the Ca .ion in a distort.ed pentagonal bipyramidal arrangement. Calcium-oxygen distances were not restrained

during

the refinement.

The average

S. Vijay-Kumar

and

W. S. Cook

Figure 5. Superimposed stereo drawing of each half of Nereis SCP and the NH,-terminal haif of troponin C. The Iview is approximately down the pseudo-Z-fold axis between the 2 EF-hands in each pair. The COOH-terminal half of Nereis SCP is in green, the KHz-terminal half is in yellow, and the NH,-terminal half of troponin C is in red.

The geometrical parameters for all four domains are given in Tables 5 and 6. The Ca2+-binding segment in the first domain, which consists of residues 16 t*o 27, is smaller and more compact than the usual Ca2+-binding loop, which is somewhat elongated (Strynadka & James, 1989; Fig. 8). This probably reflects the presence of an aspartate rather t,han a glutamate residue at the 12th position in the

the co-ordinating polyhedron has been drawn into a more energetically favored position where it makes hydrogen bonding contacts with the third and fifth residues of the loop (AsnlO6 and Aspl08) and the earbonyl oxygen atom of Leu173 of a symmetryrelated SCP-2 (Fig. 7). Overall, the three functional loops show the typical features of Ca2+ -binding loops in EF-hands.

Table GalciuwL-oxygen

ligands

Loop 1 Residue

Atom

Asp16

ODl

Asp18

ODl

Asp20

OD2

Ala22

0

Asp27

ODl

Water375 Water391

OD2

Ca2+

polyhedra

Loop IV

Loop I14

Distance (A)

Residue

Atom

232 2.30

Asp104

ODl

2.42

Am106

ODl

Asp108

ODl

AsnllO

0

2-16 2-54 240 2.21 2.08 250

Distance (A)

Residue

Atom

Distance (A)

2.26

Asp138

OD2

2.26

$sn140

OD1

Asp142

OD2

Leu144

0

Glu149

OEI

228

I.91 2.34 2.34 2-33 254 2.24

211 GM15

OEl

2-49 Asp27

5

in the three

Glu115

OE2

2.53 2.31 254 2.51

248 2% 2-45

Water381

209

251

Water396-i

361

Glu149 Water387 Water402

The corresponding distances for SCPS are given below those for SCP- 1. t The water molecule in this CaZf polyhedron occupies a unique position related SCP-2 molecules. See the text for details.

OE2

2.34 2.32 2-55 2.42 247 2-35

2.57 233 2.43 2.43 F45

1.96

between 2 symmetry.

421

Figure 6. Storco dra\ving of the (I ions and their 7 ligantls from domains I and III. ‘I’hr 2 domains have been superitnposed. (:.A109 is thr r-carbon of =\snlW. which is tile 6th resitluc in the loop. The ll’th residucl in each loop IS on the loft at the nillc: o‘clock position. Sotr the much sharper turn hctwcet~ residues Thr24 and Arg2.i that. rnovcs helix B int,o position to acc~ommoda.tc: the shorter side-chaitl of Asi>27.

-hincling loop. The shorter s&-chain of aspartate requires major vhanpes in the usual conforma1ional angles of the loop &sidues. The loop in domain IT. which dots not bind Ca*-. contains nine non-helical residues (fi3 to il). so it is t Ilr same length as the norl-helical portion of a funct.ional Ca2 + -binding loop. Thr helix preceding this loop ends wit.h a ‘short stretch of 3,0 helix: rather than the standard r-helix. As seen in other st.ruct.ures of helix-loop helix Ca.’ . -binding protcins, there is a short, antiparallel P-sheet bct,wccn domains 1 and It that involves r&dues 22 to 24 and 69 to 7 I. IleiO corrcspontls to the eighth residue in a typicaat 12-residue Ca2 + binding domain. and in fact isolvucine is t.he most common residue in this position (Marsden et cl/.. I!)!)()). The last three residues of this non-binding loop (rcsiducs 70 to 72) form the beginning of the second helix in the helix loot)helix domain, which is analogous to the functional Cla2+ -binding loops in domains 111 and IV. O~~erallz r.his loop is larger than I he usual (;‘a’ ‘-binding loop (Fig. 8). segments The Ca ‘+-binding in the third and fourth domains consist, of residues 104 t.o 1 I.’ and 138 to 149. Thr phi, psi angles are consistent wit,h

CTi12 ’

those dcscribcd in ot.her structures of (‘a”-binding prot.eins, and the overall caonforrnation is not significantl\r different. Thrrc are several unusual fcatnres of the Ca.2 ’ -binding loops in A’or~is SCI’. For (:xample. the most. favorable amino acid residue in the fourth position of il (la’+ -binding loop is glycinc (Marsden et al.: l!)!N). because the avcrag(: phi. psi angles for the rcsidur are 58’ and RG” (Str\:nadka & James. 1989). In SC’P t,he residues in tllis position of the t.hree C;L’ + -binding loops are LyslR, Glu107 and .4sn141: the average phi, psi allgl(:s are 49” and 41”. The residue in t.he sixth posiiion of the loop plays an important role by introducing a right-angle bend (phi = W, psi = 0”) in the pol~:pepii&: hackbone to enable t.hc remaining ligands to co-ordinate to the (.‘a ion. By far the most common residue at this position is glycine (h’larsdcn ef ~1.. I!)!N). However. in *Vrr& WI’ t.he third Ca’T-bindirlg loop contains An109 in this position, and the phi. psi angles arc ii” and 13” for SW-1 and 67’ and I I” for SCP-2. Kesidiics in positions 7, F( a.nd 9 of t.hc loops form a short ar~tiparallel p-sheet. The residue in t,he nint,h position of the loop is us11a1ly scrinc or threonine, and it typically forms a. hydrogen bond wit.h the

422

8. Vijay-Kuwm

ad

W. J. Cook

Figure 7. Stereo drawing of the environment around the Ca ion in domain III of SCP-2 with the superimposed eieetron density contour surfaces. Residues from the 3rd EF-hand of SCP-2 are on the left (in green), and residues from the COOH terminus of a symmetry-related SCP-2 are on the right (in pink). The crystallographic symmetry operation required t,o generate the other molecule from SCP-2 is (1 -x, y - %, 1 -z). Water 396 in the center is hydrogen-bonded to the carbonyl oxygen atom of Leu173 from the symmetry-related SCP-2. The 2F,-F, electron density map used phases that were calculated from the refined model excluding this water molecule.

Table 6 Torsion

angles jar the 12-residue

Loop 1 Residue D16 F17 D18 K19 D20 G21 A22 I23 T24 R25 M26 D27

Loop II

phi

psi

-83 -81 -45 -56 -81 -78 55 78 -92 -98 87 85 -139 -137 -84 -93 -141 -140 -63 -56 -57 -54 -58 -57

67 74 -53 -44 -1 -24 32 40 12 1 -3 3 138 143 130 124 170 165 -28 -36 -36 -33 -47 -49

The corresponding

loops I, If,

Residue A63 V64 A65 G66 G67 K68 G69 170 D71 E72 T73 T74

phi -72 -69 -76 -78 -124 -138 48 46 104 116 - 120 -90 -36 -60 -113 -99 -82 -89 -60 -52 -71 -63 -60 -66

III

and IV Loop IV

Loop III psi

Residue

-10 D104 -17 -49 T105 -46 30 N106 52 61 El07 20 -27 D108 -13 178 N109 168 136 NllO 136 120 1111 129 165 5112 160 -36 R113 -51 -47 D114 -39 -40 El15 -41

phi

psi

-78 -80 -66 -59 -97 -98 44 31 -98 -91 77 67 -138 -136 - 106 -107 -90 -94 -56 -64 -61 -59 -62 -54

72 77 -22 -29 4 26 46 36 5 8 13 11 152 150 121 124 169 172 -46 -35 -46 -59 -41 -49

Residue D138 T139 N140 N151 D142 G143 L144 L145 5146 L147 El48 El49

phi

psi

-75 -78 -66 -69 -98 -109 57 29 -88 -115 75 59 -138 -139 - 106 -95 -85 -85 -60 -59 -59 -60 -59 -61

73 75 -23 -16 2 18 30 60 4 16 15 20 155 138 120 116 164 169 -42 -43 -54 -47 -39 -45

angles for SCP-2 are given below those for SCP-1. Loop II does not bind Ca’*

Structure of Nereis SCP at 2.0 ,d Resolution

423

Figure 8. Stereo drawing of the C” backbones of the 4 U-residue loops in Nereis SCP superimposed. Residues 16 to 27 are in red, residues 63 to 74 are in green, residues 104 to 115 are in dark blue, and residues 138 to 149 are in pink. The 1st residue of each loop is at the lower right, and the 12th residue is at the upper right.

water molecule in the Ca’+-co-ordination polyhedron. In Nereis SCP, the residues in the four loops are Thr24, Asp71, Serl12 and Ser146. Only the sidechain of Thr24 of the first loop is hydrogen-bonded to the co-ordinating water molecule, while the other side-chains form hydrogen bonds with the mainchain NH atoms of the residue in the 12th position of the loop. In domain II, the Asp71 side-chain atom ODl hydrogen bonds to NH atoms of the residues at the 1lth and 12th positions. The residues in position 10, 11 and 12 are in a-helical conformation. The main-chain NH groups of residues in position 10 and 11 are hydrogen-bonded to two water molecules that are conserved in all four loops. The r.m.s. deviations between co-ordinates for main-chain atoms, Ca ions, and six conserved water molecules of loops 3 and 4 of SCP-1 are 927 A. The r.m.s. deviations between atomic co-ordinates of loop 1 and loop 3 or 4 are in the range of 1.27 A to 1.39A. This large deviation is primarily due to the presence of an aspartate residue in the 12th position of loop 1, rather than the glutamate residue found in loops 3 and 4. Hydrogen bonding within the Ca2+-binding loops is similar to that described for other Ca’+-binding proteins (Babu et al., 1988;

Herzberg & James, 1988; Satyshur et al., 1988; Szebenyi & Moffat, 1986; Ahmed et al., 1999). The first four residues at the beginning of each Ca’+-binding loop form a type I reverse turn. All three of these turns have On-NsfJ hydrogen bonds. The first four residues in the non-functional loop also form a reverse turn, but there is no 1 + 4 hydrogen bond. As observed in other Ca2+-binding proteins that utilize the EF-hand motif, there are multiple Asx turns (Rees et al., 1983) within the Ca2+-binding lo op s. The only Asx turn conserved in all three loops involves the Asx residues at the fifth position in each loop. The two hydrogen bonds in the two three-residue /?-sheet segments involve the main-chain N and 0 atoms of the middle residue in each strand; these residues are hydrophobic amino acids that occur at the eighth position in each of the Ca’+-binding loops. (d) Packing of he&es The four helices of each pair of helix-loop-helix domains pack against one another in a highly conserved fashion. Table 7 shows a comparison of the interhelical angles between the helices of SCP

424

Interhelical

8. Vijay-Kumar

angles (“)

Table 7 in four-domain

Ca2 ’ -binding

and W. 4. Cook

@lassijkation

Table 8 of water ir&raction,s

proteins Helix pair

Nereis SCP-1 64 166

80 128

102 124 100 128

Troponin

C

Calmodulin 95

138 127 146

114 85

118 106 127 110 118

110 103 116 91 118

The co-ordinates for troponin C and calmodulin are taken from the Brookhaven Protein Data Bank. The C/D pair of helices in SCP flank a non-functional Ca” -binding site. The A/B and C/D pairs of helices in troponin C flank functional Ca’+-binding sites that do not contain Cat+.

(the NH,-terminal domain pair has only 1 Ca ion), troponin C (both of the NH,-terminal domains are Ca*‘-free) and calmodulin (all 4 domains bind Ca’+). Crystal structures of other Ca’+-binding proteins have shown that when both Ca*+-binding sites in a pair of EF-hands are occupied, the helices of each EF-hand are approximately perpendicular to each other. In SCP, the interhelical angles E-F and G-H in the COOM-terminal Ca’+-binding domain are respectively. In the Ca2+ -free 102” and loo”, EF-hands of troponin 6, the helices are almost antiparallel (138” and 146”) to each other. In Nereis

Total number of water molecules per asymmetric unit Sumber of H-bonds from SCP molecule I to water molecules Number of H-bonds from SCP molecule II to water molecules Water molecules with 1 H-bond to protein Water molecules wit,h 1 H-bond to protein and 1 or more to other water molecules Water molecules with 2 or more H-bonds to pro&in Water molecules with 2 or more H-bonds to protein and 1 or more to other water molecules Water molecules wit.h H-bond to other solvent molecules only Water molecules with no contacts less than 3.5 A

213 187 143

50 43 5.5 29 31 4

SCP, where the NH,-terminal domain pa,ir has only one bound Ca2 + , the angle between heliees A a,nd B of the first EF-hand (Ca’+-bound) is 64”, and the angle’between helices C and D (Cazf-free) is 80”. Nevertheless, the first and last helices in this pair of domains pack in the same highly conserved fashion as helices in pairs of EF-hands with two or no Ca ions (Fig. 5). The angles between helices A and D and E and H are both 128”, while t,he corresponding angles in troponin C and calmodulin range from 110” to 118”. On the other hand, the interhelical angles between helices of adjacent domains K and C and F and G are not conserved in all these prot,eins, to whether both of the but vary according Ca2+ -binding loops of a domain-pair are Ca2’-free

Figure 9. Stereo drawing of the Xereis SCP molecule to demonstrate the positions of the aromit$ic residues green). The C” of Serl is labeled at the lower left. The 4th Ca2+ -binding domain is at the lower right.

(shown

:li

Structure

of Nereis

SCP at 2.0 A Resolution

425

Figure 10. Superimposed stereo drawing of the backbone atoms of the 3 functional Ca2+ -binding loops in Nereis SCP. Residues 16 to 27 are in red, residues 104 to 115 are in dark blue, and residues 138 to 149 are in pink. The 1st residue of each loop is at the lower right, and the 12th residue is at the upper right. The Ca ions and the conserved water molecules in each loop are represented by x

or Ca”-bound. These interhelical angles range from 114” to 127” in troponin C and calmodulin (Table 7). For the NH,-terminal domain-pair in Nereis SCP, which contains only one Ca ion, the interhelical angle B-C is 166”. This may be due to the fact that when Ca2+ is bound in only one of the two domains, the exiting helix of that domain moves to a position almost perpendicular to the incoming helix. The movement in the exiting helix of the first domain is by a similar movement of the accompanied incoming helix of the second Ca’+-binding domain along with the five-residue linker region. Since the second domain is not functional, the absence of Ca2+ from this domain, where both the helices of the domain are aproximately perpendicular, is energetically unfavorable. To overcome this situation, the two adjacent helices of the domain pair undergo internal movements to have greater interactions between these helices. (e) Aromatic

residues

There is a strong hydrophobic core with 20 aromatic residues. Of the 20 aromatic residues in the

molecule, there are nine in each pair of domains and two at the end of the COOH extension. The sidearomatic residues in the chains of the COOH-terminal half are confined to a rather small volume, whereas the side-chains in the NH-terminal half are spread widely (Fig. 9). The two aromatic residues at the COOH terminus (Phe169 and Trp170) help to stabilize the structure by fitting between aromatic residues in helix H and aromatic residues from helices E and F. (f) Solvent structure A total of 213 solvent molecules was located in the crystal structure of Nereis SCP. At the end of refinement, no solvent molecule had an occupancy less than 64 or a temperature factor greater than 37 8’. The temperature factors for most of the water molecules are in the range 20 A2 to 35 A’. Of the 213 water molecules, 178 (84%) are directly hydrogen-bonded to protein within a cutoff distance of 3.5 A (Table 8). The water molecules are distributed uniformly over the surfaces of the two molecules. Thirtyone (15%) of the water molecules

8. Vgjay-Kumar

426

are part of a second hydration shell, with contacts only to other water molecules. Only four water molecules show no hydrogen bonds at all. Each of the three functional Cazf -binding loops has hydrogen bonding contacts with six conserved water molecules (Fig. 10). These six include the water molecule that is co-ordinated to the Ca ion. These conserved water molecules follow the pattern seen in other EF-hands as discussed by Strynadka & James (1989). This research was supported by grant NAGW813 from the NationaI Aeronautics and Space Administration.

References Ahmed, F. R., Przybylska, M., Rose, D. R., Birnbaum, G. I., Pippy, M. E. & MacManus, J. P. (1999). Structure of oncomodulin relined at I85 A resolution: an example of extensive molecular aggregation via Ca’+ J. Mol. Biol. 216, 127-140. Babu, Y. S., Cox, J. A. & Cook, W. J. (1987). Crystallization and preliminary X-ray investigation of a sarcoplasmic calcium-binding protein from Nereis diversicolor. J. Biol. Chem. 262, 11884-11885. Babu, Y. S., Bugg, C. E. & Cook, W. J. (1988). Structure of calmodulin refined at 2-2 A resolution. J. Mol. Biol. 204, 191-204. Bhat, T. N. & Cohen, 6. H. (1984). OMITMAP: an electron density map suitable for the examination of errors in a macromolecular model. J. Appl. Crystallogr. 17, 244-248 Collins, J. H., Cox, J. A. & Theibert, J. L. (1988). Amino acid sequence of a sarcoplasmie calcium-binding protein from the sandworm Nereis dive&color. J. Biol. Chem. 263, 15378-15385. Cook, W. J., Ealick, S. E., Babu, Y. S., Cox, J. A. & Vijay-Kumar, S. (1991). Three-dimensional structure of a sarcoplasmic calcium-binding protein from Nereis diversicolor. J. Biol. Chem. 266, 652-656. Cox, J. A. (1989). Calcium vector protein and sarcoplasmic calcium binding proteins from invertebrate muscle. In Stimulus-response Coupling: The Role of Intracellular Calcium (Dedman, J. R. t Smith, V. L., eds), pp. 85-110, CRC Press, Inc., Boca Raton, Florida, U.S.A. Cox, J. A. & Stein, E. A. (1981). Characterization of a new sarcoplasmic calcium-binding protein with magnesium-induced co-operativity in the binding of calcium. Biochemistry, 29, 5430-5436. Einspahr, H. & Bugg, C. E. (1977). Crystal structures of calcium complexes of amino acids, peptides and related model systems. In Calcium Binding Proteins and Calcium pun&ion (Wasserman, R. H. et al., eds), pp. 13-26, Elsevier/North-Holland, New York. Einspahr, H. & Bugg, C. E. (1981). The geometry of calcium-carboxylate interactions in crystalline complexes. Aeta Crystallqr. see& B, 37, 10441052. Hendrickson, W. A. (1985). Stereochemically restrained re6nement of macromolecular structures. Methods Enzymol. 115, 2522270. Herzberg, 0. & James, M. N. G. (1988). Refined crystal structure of tropoinin C from turkey skeletal muscle at 2.OA resolution. J. Mol. Biol. 203, 761-779. Howard, A. J., Gilliland, G. L., Finzel, B. C., Poulos, T. L., Ohlendorf, D. H. & Salemme, F. R. (1987). The use of an imaging proportional counter in macromolecular crystallography. J. Appl. Crystallogr. 20, 383-387.

Edited

and W. J. Cook Jauregui-Adell, J., Wnuk, W. 8E @ox, J. A. (1989). Complete amino acid sequence of the sarcoplasmic calcium-binding protein (SCP-I) from crayfish (Astacus leptodactilus). FEBL? Letters, 243, 209-212. Jones, T. A. (1978). A graphics model building and refinement system for macromolecules. J. Appt. Crystallop. 1 I, 268-272. Kretsinger, R. II. & Nockolds, C. E. (1973). Carp muscle calcium-binding protein. II. Structure determination and general description. J. Biol. Chem. 248, 3313-3326. Luzzati, V. (1952). Traitement statistique des erreurs dans la determination des structures cristailines. Acta Crystallogr.

Marsden, B. Calcium tions to variants

5, 802-810.

J., Shaw, G. S. & Sykes, B. D. (1996). binding proteins. Elucidating the contribucalcium affinity from an analysis of species and peptide fragments. Biochem. Cell Biol.

68, 587-601.

Monerief, N. D., K&singer, R. II. & Goodman, M. (1996). Evolution of EF-hand calcium-modulated proteins. I. Relationships based on amino acid sequences. J. Mol. Evol. 39, 522-562. Ramaehandran, G. N., Ramakrishnan, C. & Sasisekharan, V. (1963). Stereochemistry of polypeptide chain configurations. J. Mol. Biol. 7, 95-99. Rees, D. C., Lewis, M. & Lipscomb, W. N. (1983). Refined crystal structure of earboxypeptidase A at 1.54 A resolution. J. Mol. Biol. 168, 367-387. Sa,tyshur, K. A., Rao, S. T., Pyzalska, D., Drendel, W., Greaser, M. & Sundaralingam, M. (1988). Refined structure of chicken skeletal muscle troponin C in the two-calcium state at 2 A resolution. J. Biol. Chem. 263, 1628-1647.

Strynadka, N. 6. J. $ James, M. N. 6. (1989). Crystal structures of the helix-loop-helix calcium-binding proteins. Annu. Rev. Biochem. 58, 951-998. Szebenyi, D. M. E. & Moffat, K. (1986). The refined structure of vitamin D-dependent calcium-binding protein from bovine intestine. J. Biol. Chem. 261, 8761-8777.

Takagi, T. & Cox, J. A. (1990). Amino acid sequences of four isoforms of amphioxus sarcoplasmic calciumbinding proteins. Eur. J. Biochem. 192, 387-399. Takagi, T. & Konishi, K. (1984s). Amino acid sequence of a-chain of sarcoplasmic calcium binding protein obtained from shrimp tail muscle. J. Biochem. 95, 1603-1615.

Takagi, T. & Konishi, K. (19846). Amino acid sequence of the ,&chain of sarcoplasmic calcium binding protein obtained from shrimp tail muscle. J. Biochem. 96, 59-67. Takagi, T., Kobayashi, T. & Konishi, K. (1984). Amino-acid sequence of sarcoplasmic calciumbinding protein from scallop (Patinopeeten yessoensisj adductor striated muscle. Biochim. Biophys. Acta, 787, 252-257.

Takagi, T., Konishi, K. & Cox, J. A. (1986). Ammo acid sequence of two sarcoplasmic calcium-binding proteins from the protochordate amphioxus. Biochemistry, 25, 3585-3592. Wang, B. C. (1985). Resolution of phase ambiguity in macromolecular crystallography. Methods Enzymol. 115,99-112. Wnuk, W., Cox, J. A. & Stein, E. A. (1982). Parvalbumins and other soluble high-affinity calcium-binding proteins from muscle. In Calcium and Cell Function (Cheung: W. Y., ed.), vol. 2, pp. 243-278, Academic Press, New York.

by B. W. Matthews