Solution structure of the tissue-type plasminogen activator kringle 2 domain complexed to 6-aminohexanoic acid an antifibrinolytic drug

J. Mol. Biol. (1991) 222, 1035-1051

Solution Structure of the Tissue-type Plasminogen Activator Kringle 2 Domain Complexed to 6-Aminohexanoic Acid an Antifibrinolytic Drug In-Ja L. Byeon and Miguel Llinbt Department of Chemistry, Carnegie Mellon University Pittsburgh, PA 15213, U.S.A. (Received 7 May 1991; accepted 21 August 1991) The solution structure of a recombinant tissue-type plasminogen activator kringle 2 domain, complexed with the antillbrinolytic drug 6-aminohexanoic acid (8AHA) was determined via ‘H nuclear m agnetic resonance spectroscopy and dynamical simulated annealing calculations. The structure determination is based on 610 intramolecular kringle 2 and 14 intermolecular kringle 2-6-AHA interproton distance restraints, as well as on 82 torsion angle restraints. Three sets of simulated annealing structures were computed from three different classes of starting structures: (1) random conformations devoid of disulfide bridges; (2) random conformations that contain correct disulfide bonds; and (3) a folded conformation modeled after the homologous prothrombin kringle 1 X-ray crystallographic structure. All three sets of structures are well defined, with averaged atomic root-meansquare deviations between individual structures and mean set structures of 977, 999 and 070 A for backbone atoms, and l-36, 1.55 and 1.41 A for all atoms, respectively. Kringle 2 is an oblate ellipsoid with overall dimensions of approximately 34 A x 30 A x 17 A. It exhibits a compact globular conformation characterized by a number of turns and loop elements as well as by one right-handed a-helix and five (1 extended and 4 rudimentary) antiparallel /I-sheets. The extended b-sheet exhibits a right-handed twist. Close van der Waals’ contacts between the Cys22-Cys63 and Cys51-Cys75 disulfide bridges and the central hydrophobic core composed of the Trp25, Leu46, His48a and Trp62 sidechains are among the distinguishing features of the kringle 2 fold. The binding site for 6-AHA appears as a rather exposed cleft with a negatively charged locus defined by the Asp55 and Asp57 side-chains, and with an aromatic pocket structured by the Tyr36, Trp62, His64 and Trp72 side-chains. The Trp62 and His64 rings line the back surface of the pocket, while the Tyc36 and Trp72 rings confine it from two sides. The Trp62 and Trp72 indole rings conform a V-shaped groove. The methyl groups of Va135 also contribute lipophilic character to the ligand-interacting surface. It is suggested that the positively charged side-chains of Lys34 and, potentially, Arg69 may favor interactions with the carboxylate group of the ligand. The Trp25 and Tyr74 aromatic rings, although conserved elements of the binding site structure, seem not to undergo direct contacts with the ligand.

Keywords: tissue plasminogen activator

kringle; kringle 2 n.m.r. solution structure; kringle 2 binding site

1. Introduction Tissue-type physiological

plasminogen activator (t-PA)$ is a fibrin-selective thrombolytic agent

t Author to whom all correspondence should be

addressed. $ Abbreviations used: t-PA, human tissue-type plasminogen activator; Pgn, plasminogen; 3D, three dimensional; n.m.r., nuclear magnetic resonance; Ptb,

prothrombin; 6-AHA, 6-aminohexanoic acid; BASA, p-benzylaminesulfonic acid; pH*, glass electrode pH reading uncorrected for deuterium isotope; NOESY, two-dimensional NOE correlated spectroscopy; t,,,, NOESY mixing time; COSY, two-dimensional chemiqal

shift correlated spectroscopy; D&F-COSY, double quantum filtered COSY; NOE, nuclear Overhauser effect; p.p.m., parts per million; RMSD, root mean square difference; ABNR,

$03.00/O

Newton-

(urokinase) . 1035

0022-2836/91/241035-l?

adopted-basis

Raphson; u-PA, kidney type plasminogen activator 0

1991 Academic

Press

Limited

1036

I.-J. L. Byeon and M. Llinds

currently exploited for the clinical treatment of blood clotting disorders such as acute myocardial infarction, brain stroke, deep vein thrombosis, etc. (Collen et al., 1989). Its mode of action involves specific hydrolysis of the Arg560-Va1561 peptide bond in plasminogen (Pgn), thereby converting the zymogen into the active protease plasmin, which efficiently degrades polymerized fibrin (Collen, 1980). t-PA is a -57,000 Da serine protease which consists of six distinct structural modules: a finger domain; an epidermal growth factor domain; two kringle domains; and a serine protease two-domain unit with trypsin-like specificity (Pennica et al., 1983; Bltnyai et al., 1983; Ny et al., 1984). Among these, the kringle 2 domain (Fig. 1) has drawn considerable attention because it contains a lysinebinding site which is likely to mediate t-PA interactions with fibrin (van Zonneveld et al., 1986u), as is thought to be the case for kringles 1, 4 and 5 in Pgn. It has also been demonstrated that the lysinebinding site in the t-PA kringle 2 domain stimulates Pgn activation by fibrin (van Zonneveld et al., 19863; Verheijen et al., 1986; Gething et al., 1988). This stimulatory effect results from an alignment of the substrate Pgn and t-PA on the fibrin matrix so that the latter can efficiently activate Pgn. Despite the important role attributed to kringle 2 in the t-PA mechanism of action, its threedimensional (3D) molecular structure has not yet been determined. Only a model for its putative lysine-binding site (Tulinsky et al., 1988a) has been proposed, derived from nuclear magnetic resonance (n.m.r.) evidence for Pgn kringles 1 and 4 (Ramesh et al., 1987; Motta et at., 1987) and the crystallo-

graphic structure of the prothrombin (Ptb) kringle 1 (Park & Tulinsky, 1986; Tulinsky et al., 1988b)t. Elsewhere (Byeon et al., 1989), we have reported a preliminary ‘H n.m.r. spectroscopic characterization of a recombinant t-PA kringle 2 domain (Fig. 1) in the absence and in the presence of antifibrinolytic drugs such as lysine, 6-aminohexanoic acid (6-AHA), and p-benzylamine sulfonic acid (BASA). More recently, we have published t.he complete ‘H n.m.r. sequence-specific assignments for the kringle 2-6-AHA complex (Byeon et al., 1991). In this paper we report a determination of the 3D solution structure of the t-PA kringle 2-6-AHA complex. The conformational analysis is based on ‘H n.m.r. derived distance and dihedral angle estimates, and dynamical simulated annealing calculations (Nilges, 1990). Particular attention is given to defining the structure of the kringle 2 lysine-binding site.

2. Materials and Methods (a) Sample preparation

Recombinant t-PA kringle 2, expressed in Escherichia coZi and purified as described previously (Cleary et al.: 1989), was provided by Dr R. F. Kelley, Genentech, Inc. 6AHA was purchased from Aldrich Chemical Co. BASA belonged to a batch previously described (Hochschwender et al., 1983). ‘H,O was obtained from Merck, Sharp & Dohme Ltd., Montreal. n.m.r. samples contained w 2 mM kringle 2 and N 12 mM-6-AHA in ‘H,O or 9 : 1 (v/v) ‘H,O:‘H,O (henceforth referred to as ‘H,O), pH (or pH*) 49. Kringle 2 samples with BASA ligand were prepared in 2H,0 or ‘H,O containing 014 M-ammonium sulfate (pH 49). For structure calculations, the n.m.r. experimental data were collected from kringle 2-6-AHA samples. Spectra recorded from kringle 2-BASA samples were found useful for clarifying ambiguities in the kringle 2-6-AHA data. (b) n.m.r. spectroscopy ‘n.m.r.

spectra were recorded on a Bruker AM-560 with an Aspect 3ooO computer linked to an Aspect X32 workstation. Two-dimensional nuclear Overhauser effect correlated spectroscopy (NOESY) spectra (Bodenhausen et al., 1984) were acquired with mixing time (t,,,) = 70, 125, 200 and 356 ms. The NOESY experiments involved 512 t, increments (64 scans per t, experiment) of 4666 data points. Prior to Fourier transformation, the time domain data in t, and t, were zero-filled to 8192 and 2948 points, respectively. Digital resolutions along 6, and 8, are 1.95 Hz/point and 7.80 Hz/point, respectively. The NOESY spectra were baseplane-corrected by fitting to 3rd-order polynomial functions.

spectrometer interfaced

Figure 1. Recombinant kringle 2 domain of human t-PA: outline of the primary structure. The sequence corresponds to the Ser174-Thr263 segment of native t-PA. Site numbering, insertions (sites 44a to 44c, 48a and 66a) and deletions (sites 29, 38 and 59) are based on homology with the Pgn kringle 5 (Motta et al., 1987; Thewes et al., 1987). and follow previous convention (Byeon et aE., 1989, 1991).

t A 243 A resolution X-ray crystallographic structure of the ligand-free t-PA kringle 2 has recently been solved by A. M. de Vos, M. H. Ultsch, R. F. Kelley. K. Padmanabhan, A. Tulinsky, M. L. Westbrook & A. Kossiakoff (personal communication). At the time of submission of this paper, the authors have not inspected, nor received, any data related to the crystallographic model.

1037

n.m.r. Solution Structure of t-PA Kringle 2

Table 1 Proton resonance chemical shifts of the 31 stereo-specijcally of 6-AHA

assigned residues of t-PA kringle 2 in presence

Chemical shiftst Residue

NH

CH”

CHB$

Asp-l Cysl TyrP Asn5 Ser7 Tyr9 Serl4 Glu17 Serl8 Leu23 Trp25

820 7.91 864 857 853 7.54 867 R97 7.82 R92 836

4.90 517 537 496 441 441 526 404 4.58 458 408

292*, 262 301, 285* 314*, 2.37 357*, 281 410*, 404 319*, 313 448, 366* 212*, 1.99 4.10*, 390 1.35*, 1.17 378*, 318

Ser27 Va135 Gln40 Asn41 Ser43 Leu46 His48a Asn49 Tyr50 cys51 Asn53 Lys60 Trp62

7.49 1040 6.09 7.49 897 816 786 7.46 1008 905 856 7.44 900

418 369 433 485 496 357 4.53 488 515 461 531 445 560

401, 367* 1.72 2.23, 1.82* 31s*, 264 4.22, 376’ 1.58*, 0.46 2.83*, 1.66 2.52*, 1.81 331, 264* 337*, 260 2.51*, 1.82 1.79*, 692 372*, 341

Cys63 His64 Val65 Arg69 La70 Trp72

894 811 872 7.81 842 897

466 519 461 456 4.70 5.60

349*, 325 2.92, 280* 1.88 1.87*, 1.80 1.87, 1.32* 333*, 317

cys75

932

503

338*, 312

(p.p.m.) Others

CHY$

CH2,6 647, CH3,5 6.98 NH; 7.19, 7.12 CH2,6 686, CH3,5 7.14 229, 222 1.78

CHd, 992,671 CH2 7.45, CH4 7.81, CH5 696 CH6 557, CH7 7.65, NH1 11.57

094, 0.11* 2.41, 231

NH; 7.41, 7.24 1.10

CHd, 650, -695 CH2 7.63, CH4 6.85, NH3 lOSl§ NH; 7.76, 667 CH2,6 628, CH3,5 6.75 NH; CH: CH2 CH6

1.26

679, 1.46, 7.74, 7.07,

672 CH’, 2.97, 283 CH4 7.30, CH5 566 CH7 652, NH1 11.05

CH2 847, CH4 683, NH3 990 @86*, 672 1.67, 1.57 1.62

CHd, 323, NH” 7.17 CHd, 987 CH2 706, CH4 684, CH5 551 CH6 684, CH7 7.21, NH1 973

t Chemical shifts are referred to the internal dioxane signal at 3766 parts per million (p.p.m.) and are accurate to within approx. +OOl p.p.m. Two sets of cross-peaks were observed for Asp- 1. For this residue, the chemical shifts for the major form are listed. Experimental conditions: (kringle 2:6AHA = 1: 6), pH 49, 53°C. $ Stereospecific assignments are denoted as follows: for the /?-methylene protons, the asterisk (*) indicates the CHB2 resonance; for the Val y-methyl protons, the asterisk (*) indicates the CHG’ resonance. d The NH3 resonance was not observed under the exnerimental conditions. The listed chemical shift is in the presence of L-Lys (kringle 2:~-Lys = 1 : 4), pH 60, 37°C (Byeon et al., 1989;

3JaN coupling constants, excluding those for glycyl residues. were estimated from the NH doublet splittings measured on phase-sensitive 2-dimensional chemical shift correlated spectroscopy (COSY) spectra recorded in ‘H,O at 53 “C (Byeon et al., 1991). After appropriate zero-filling, digital resolutions in 6, and 6, were 697 Hz and 7.80 Hz, respectively.

‘Jap coupling

constants

were measured

from

phase-sensitive double quantum filtered COSY (DQFCOSY) spectra recorded in ‘H,O. Digital resolutions in the final spectra were 673 Hz in 6, and 58 Hz in 6,. (c) Torsion angle restraints and stereospecijic assignments Values of 3JzNcoupling constants 2 80 Hz and <6-O Hz were interpreted in terms of c$ backbone torsion angles lying within the ranges - 160” < 4 I - 80” and - 99” I 4 5 -3O”, respectively (De Marco et al., 1978; Pardi et al., 1984). This afforded 46 0 restraints. Stereospecific assignments of the prochiral B-protons were obtained from 3Ja,

splittings and the relative magnitudes of intraresidue CH”-CHflvB’ and NH-CHfl*” nuclear Overhauser effects (NOES) (Wagner et al., 1987). This afforded stereospecific assignments for /&protons of 29 residues (Table 1). Stereospecific assignments for the Val prochiral y-methyl groups were based on the 3JE,,splittings and the relative magnitudes of intraresidue CH”-CHP and NH-CH$f NOES (Zuiderweg et al., 1985). This led to stereospecific assignments for the methyl groups of Va135 and Va165 (Table 1). A total of 31 side-chain x’ torsion angles were derived from the stereospecific assignments. All 5 Pro residues in kringle 2 were characterized as linked via buns peptide bonds (Byeon et al., 1991). Therefore, 175” < w < 185” was assumed for the Pro residues. (d)

Interproton

distance

restraints

About 1996 Overhauser connectivities were assigned from t, 296 ms and 356 ms NOESY spectra of the kringle

’

1038

I.-J.

L. Byeon and M. Llinds

Sequence

Figure 2. Diagonal plot indicating kringle 2 residues related via interproton Overhauser connectivities. Backbonebackbone NOES are shown below the diagonal and connectivities involving side-chain protons are shown above the diagonal.

2-6-AHA complex dissolved in ‘H,O. The cross-peaks were classified as strong, medium, weak or very weak based on their intensities in the t, 70 ms and 125 ms NOESY spectra. The cross-peaks were interpreted as resulting from interproton distances ranging between 18 and 2-7 A (I A = 61 nm; strong), 18 and 3.5 A (medium), 18 and 50 A (weak) and 18 and 60 A (very weak). Upper limits for distances involving equivalent and nonstereospecifically assigned protons were corrected foi center averaging (Wiithrich et al., 1983). An extra 65 A was added to the upper limit in the case of distances involving methyl protons (Clore et al., 1987; Wagner et aE.. 1987). The NOE assignments resulted in a total of 585 kringle 2 intramolecular (117 intraresidue and 468 interresidue; 264 short range (Ii-j1 < 5) and 294 long range (Ii-j1 > 5)) distance restraints, and a total of 14 kringle 2-ligand intermolecular distance restraints. The NOES used to determine the kringle 2 structure are summarized in Fig. 2. In addition, 22 supplementary distance restraints corresponding to 11 backbone NH to CO hydrogen bonds were incorporated. These were predicted from the secondary structure (Byeon et aE., 1991). The NH-CO hydrogen bond paired residues are: (16,20), (25,48a), (62,52), (63,73), (73,63), (65,71) (71,65), (66a,69) and (69,66a) within the p-sheet regions, and (44c,43) and (45,44) within the a-helical region (residue numbering as in Fig. 1). The hydrogen bond constraints were rNHu-OtiJ = l-9( *@3) A and r,u-007 = 2.9( +65) A. Finally, the 3

disulfide bridges linking Cysl to Cys80, Cys22 to Cys63 and CysBI to Cys75 were restrained to ~~~~~~~~~~ = 202( *@lo) A. (e) Structure

calculations

Elsewhere (Byeon et al., 1991), we have reported the complete analysis and sequential resonance assignments of kringle 2 two-dimensional (2D) ‘H n.m.r. spectroscopic data. Only methods related t,o the structure elucidation are described here. Three-dimensional solution structures of kringle 2 were calculated from data measured on kringle 2-6-AHA samples via a dynamical simulated annealing protocol (Clore et al., 1985, 1986a.6; Briinger at aZ., 1986; Nilges et aE., 1988a,b,c; Nilges, 1999) using X-PLOR (Briinger, 1988), a program developed from CHARMM (Brooks et aE., 1983). All computations were carried out on the CRAY Y-MP machine of the Pittsburgh Supercomputing Center. Structures were displayed on either an Evans & Sutherland PS300 color graphics system with the program HYDRA (Hubbard. 1986) or a Silicon Graphics personal Iris 40/25 workstation with the program MOL17 (Sneddon, 1990). The ribbon model (Fig. 9) was generated using the Ribbon Suite Program written by T. Lybrand. The simulated annealing protocol (Nilges et al., 1988a,b: Nilges, 1990) involves solving Newton’s equations of motion to locate the regions of global minimum of a

1039

n.m.r. Solution Structure of t-PA Kringle 2 target function. The target function is made up of covalent, non-bonded and experimental n.m.r. restraint terms: (1)

F,,, = Fcov+ f’rep + FNOE+ F,,, F,,, is the term accounting for idealized metries such as bond lengths, angles, planarity, and is expressed as: Fe,,, = 1 w&&-G,)~+ bO”dS

covalent chirality

c wb(~-&,)Z angles + c qk~(l-Ld2.

geoand

(3) 10 ps dynamics at 1600 K, during which the constants (2)

improper3

High values of the force constants for bond (kb), angle (k,) and improper torsion angles (kc) terms were used to ensure near perfect stereochemistry throughout the calculations, and 600 kcal/mol . A’, 500 kcal/mol’ . rad2 namely, 500 kcal/mol’ . rad’, respectively (1 cal = 4184 J); wr,, w, and wi, are arbitrary weighting factors. The improper torsion angle terms are introduced to maintain planarity of aromatic rings and, when suitable, chirality of carbon atoms, as well as to enforce tram and planar conformations of peptide bonds. Only a soft repulsion energy, instead of the van der Waals’ (Lennard-Jones’) potential and the Coulombic electrostatic energies, was introduced for Frsp in order to account for the non-bonding interactions by simply preventing too close non-bonded contacts. It is expressed by: Frep

=

0

if r 2 armin

krep(82r$n-r2)2

if T < armin,

(3)

where krep is a force constant. The values of rmin (s is a scale factor) are those standard for van der Waals’ radii as represented by the Lennard-Jones’ potential radii used in the CHARMM empirical energy function (Brooks et al., 1983). F NOE and F,,, stand for the nuclear Overhauser effect (NOE) and torsion angle restraints target functions, respectively. For FNoE, either square-well or soft squarewell potential functions were used:

F NOE

ho, (rij - $‘J’ =

0

i kmE(rij-r~j)2

if rij > r; if rij I rij I r: if rij < rij,

(4)

or a+b/(rij-r$) if rij>r;+0.5 +c(rij-r!.) F NOE = if r&
(5)

if $i>$y kor(4i-4Y)2 if &I~ilf$~ 0 (6) if $i<&, i ktor(4i-4t)’ where k,,, and k,,, are force constants, r; and 4: are the upper limits for the distance and torsion angle restraints, respectively, and rij and 4: are the corresponding lower limits (Clore et al., 1987). The lower limits of the interproton distance restraints were set to 18 A. The constants a and b in the soft square-well NOE potential (eqn (5)) are set to (l-25- 1.5~) and (@5c-O-5), respectively, such that the FNOEfunction is continuous and differentiable at rij = rt+@5 (Nilges et al., 19886). The constant c stands for the asymptote slope. The simulated annealing protocol involves a total of 278 ps dynamics. It proceeds in 5 stages (Nilges, 1990): F,or =

(1) 50 cycles of initial Powell minimization to ensure planarity of some planar groups; (2) 15 ps dynamics at 1060 K using a very low value of k,, (0902 kcal/mol . A4), which allows atoms to easily pass each other, incorporating a soft square-well NOE potential with small asymptotic slope c (01 kcal/mol* A), thus imposing a strong force on atoms exhibiting small, but a weak force to those showing large, NOE deviations (see eqn (5)); le,,, and c are gradually increased from O-002 to O-1 keal/ mol.A4 and from @l to 1 kcal/mol. A, respectively; (4) 2.8 ps dynamics while gradually cooling the systems from 1000 K to 300 K; and (5) 200 cycles of Powell minimization. Bond lengths are fixed during dynamics using the SHAKE algorithm (Ryckaert et al., 1977). Each structure calculation required approximately 0.5 h on the CRAY Y-MP computer.

3. Results and Discussion (a) Converged

kringle

2 solution structures

Although the ‘H n.m.r. data were collected from kringle samples containing 6-AHA, the structure calculations were initially carried out without including the ligand, as its location relative to the protein was not definable a priori. Thus, ligandkringle 2 intermolecular NOES were, in a first step, neglected. Three sets of 12 kringle 2 structures each were calculated following three different schemes. The first scheme starts from 12 different randomcoil structures that omit the three disulfide bonds. These random structures were subjected to simulated annealing calculations using all the experimental restraints (except for the disulfide bonds), according to the protocol outlined under Materials and Methods. The obtained kringle 2 structures (labeled SA0) are shown superimposed in Figure 3(a). The individual SA0 structures converged to essentially the same global conformational minimum (average root-mean-square difference (RMSD) = 1.4 A for backbone atoms), and satisfied the experimental distance restraints reasonably well (average violations = 0.061 A). The Cysl-Cys80, Cys22-Cys63 and Cys51-Cys75 disulfide bonds were then incorporated to the SAO structures followed by 500 steps of Powell energy minimization without experimental restraints. The latter was necessary in order to eliminate bad covalent bond geometries caused by the abrupt introduction of disulfide bonds. The resulting structures were then subjected to a second round of simulated annealing calculations that included all the experimental restraints plus the disulfide bond distance restraints. This led to 12 final structures (Fig. 3(b)), which are referred to as SAl. In a second calculation scheme, 12 random structures with the complete set of disulfide bonds were generated first. Such random structures often exhibit poor covalent geometries which could be corrected via 1OOO cycles of Powell unrestrained energy minimization. The simulated annealing protocol was then applied to the corrected structures, enforcing the complete set of experimental

Figure 3. Stereo views showing best fit superposit,ions of the backbone (N, C” and (I) atoms (d). 8A3 structures. The front views show the surface exposing the ligand-binding s&e.

(b) of each of the 12 calculated

kringle

2 structures:

(d> (a). SAB;

(b) SAL: (c). RAY; and

1041

n.m.i. Solution Structure of t-PA Kringle 2

Table 2 Statistics for the SAl, SAZ, SA3 Eringle 2 structures @Al)

(=i)r

@-W

WW,

(SA3)

(SA3),

RMSD values from experimental restraints (A)? All (016) Intro, residue (117) Short range (Ii-j1 5 5) (264). Long range (ii-j1 > 5) (207) Hydrogen bond (22)

9059+0901 0042 It 0903 0037 f om4 0086 f om4 0041+ 9006

0059 0043 0040 0084 0020

0060*0405 0053+6017 0037 f 0605 0083 f 0907 0053f9019

0062 tiO56 9037 0.089 0032

0058fO6O4 0.039 +0002 0038 f om8 0083 f 0004 0043~@010

O-059 0945 6036 0085 0044

RMSD values from idealized geometryt Bonds (A) (1351) Angles (deg) (2415) Impropers (deg) (624)

0010+0000 2.316+0905 1.063 &-0.003

0010 2310 1.019

0010f00OO 2322 kO.020 1.075 + 0035

6010 2.299 1.034

0010rf:0600 2.313&0608 1963 f 0043

6010 2308 1.030

Energies$ . FVdw (kcal/mol) Fclcc (kcal/mol)

-224flO -232*39

-218 - 386

-225i16 -239f33

- 220 - 402

-224+11 -228+38

-227 -396

The structures are denoted as follows: (SAl), the 12 final simulated annealing structures obtained starting from 12 random structures without dieulfide bonds; (m),, the structure obtained by restrained energy minimization of the mean structure (SAl) obtained by averaging the co-ordinates of the 12 SAl structures that were best fitted to each oths(SAQ), the 12 final simulated annealing structures obtained starting from 12 random structures with correct disulfide bonds; (SA2),, the structure obtained by restrained energy minimization of the mean structure (SA2) obtained by averaging the co-ordinates of the 12 SA2 structures that were best fitted to each other; (SA3), the 12 final simulated annealing structures obtained starting from a structure modeled after the Ptb kringle 1 crystallographic structure; (m),, the structure obtained by restrained energy minimization of the mean structure (SA3) obtained by averaging the co-ordinates of the 12 SA3 structures that were best fitted to each other. t The RMSD values from experimental restraints refer to deviations from experimental upper and lower limits of the distance restraints. Numbers in parentheses refer to the number of restraints in each individual category. $ The idealized geometry refers to that used in X-PLOR program. The number of bond, angle and improper terms is given in parentheses. 5 The F values refer to energies of the final structures, calculated using the CHARMM empirical energy functions. Fvdw is the van der Waals’ (Lennard-Jones potential) energy. Felecis the Coulomhic electrostatic energy calculated with a distance-dependent dielectric constant (E = cc r) where &c = 1.

restraints. The resulting kringle 2 structures are referred to as SA2. The 12 SA2 structures are shown in Figure 3(c). In the third calculation scheme, a structure modeled after the Ptb kringle 1 crystallographic structure was used to start simulated annealing calculations. Modeling was initiated by first mutating the Ptb kringle 1 sequence to generate the t-PA kringle 2 sequence, incorporating the appropriate insertions and deletions, using methods formerly applied to Pgn kringle 5 (Thewes et al., 199O), followed by 1OOO cycles of adopted-basis Newton-Raphson (ABNR) unrestrained energy minimization. Twelve t-PA kringle 2 structures (SA3 structures, Fig. 3(d)) were obtained by starting the simulated annealing calculations with different, sets of initial velocities. Comparison of structures clearly shows that those omitting the disulfide bonds (SAB; Fig. 3(a)) dispiay basically the same global fold as those incorporating the disulfide bonds (SAl, SA2 and SA3; Fig. 3(b) to (d)). This indicates that the n.m.r. experimental restraints suffice to determine the overall polypeptide conformation. We note, however, that the SA0 structures are somewhat less well defined and exhibit comparatively expanded conformations, especially near the Cysl-Cys80 disulfide bridge that closes the domain by bringing the N and C-terminal segments together. All final kringle 2 structures (Fig. 3(b) to (d))

satisfy the experimental restraints excellently (Table 2). None of the structures exhibits distance violations >O5 A, except for one or two constraints whose violations are <0.7 A. Among the experimental distance restraints, the long range (Ii-jl) >5) interresidue interproton distances show, as expected, the largest violations. All three sets of final structures exhibit very small deviations from the idealized covalent geometry (Table 2). No structure shows violations larger than Ol A, 18” and 8” from the idealized bond lengths, 8 angles and c (improper) torsion angles, respectively. Moreover, the relatively large negative values of van der Waals’ energy and Coulombic electrostatic energies (Table 2) indicate that the structures are endowed with reasonably good non-bonding contacts. Figure 4 shows the atomic RMSD values for the 12 SAl structures about the mean SAl structure. As discussed above, the final structures exhibit relatively well-defined conformations for the kringle domain proper, i.e. for the Cysl to Cys80 region of the sequence (Fig. 3(b) to (d)). However, poor definition is observed for the unconstrained N and C-terminal peptide “tails” of the protein (Ser-6 to Asp - 1 and SerSl to Thr82), a reflection of insufficient NOE data. Such is often the case when in the presence of segmental flexibility. The SAl and SA3 structures are considerably better converged than the SA2 structures: the average atomic RMSD value of the SAI, SA3 and

I.-J. L. Byeon and M. Llinds

1042

0.0’

0

1

10

20

30

40

50

60

70

60

Residue

Figure 4. Atomic RMSD values of the 12 SAl structures about the mean structure obtained by bestfitting residues within the CyslLCys80 peptide stretch. (m) Profiles for backbone atoms and (0) profiles for all atoms.

SA2 structures

from

mean structures

are 0+77, 070

and @99 8, respectively, for the Cysl to Cys80 atoms (Table 3). backbone polypeptide Furthermore, the SAl and SA3 structures show somewhat smaller distance violations than the SA2 structures. This observation indicates that betterconverged simulated annealing structures can be obtained either by carrying out two rounds of calculations that start from random structures without disulfide bonds followed by with disulfide bonds, or by resorting to reasonably close initial structure, e.g. from a related homolog. The fit within the individual SAl, SA2 and SA3 sets of structures was optimized and their coordinates averaged- to yield mean structures which are referred to as SAl, SA2 and SA3, respectively. The three mean structures exhibit very similar conformations (Table 3). Poor stereochemistry and

Table 3 Atomic RMSD values for the SAl, SA2 and SA3 kringle 2 structures

(SAl) @Al) SAl (SA2) (SA2) sic9 (SA3) (SA3) SA3 SAl SAl SC2 (SAl), (SAl), (SA2),

vereua SAl ver9u8 (SAl), versus (SAl), versu-s SA2 verse (m), ‘ver8uB (SE), versus SA3 vereue (m), versus (m), verse &@ ve.r.~.~ m vwsus m versus (s7iz), vereus (SA3), versus (m),

Backbone atoms (4

All atoms (4

077f@17 0.85&@17 0.35 099*e22 1.06 f O-23 039 070f@ll 077fO.13 0.39 0.40 037 051 063 0.57 0.58

1.36*0-22 1.55kO.23 075 1.59+_@27 1.82 kO.25 083 1.41*@14 1.62+0.19 0430 0.72 074 0.79 1.26 1.23 1-26

I

-180

-60

I

0

I

60

,’

1 0

120

Figure-- 5. RamachandraGot for the energy minimized mean (SAl), , (SA2), and (SA3), structures. The backbone dihedral Cpand Y angles for all non-glycyl residues within the Cysl-Cys80 peptide stretch are indicated by filled circles (e), and those for glycines by open circles (0).

non-bonded contacts found were readily corrected by unrestrained descent (CHARMM), followed by

in the mean structures 1000 cycles of steepest minimization energy 500 cycles of Powell restrained energy minimization (X-PLOR). Tt is noteworthy that the unrestrained energy minimization caused only minor atomic shifts for the mean structures (@38, @43 and @42 A for backbone atoms of SAL, SA2 and SA3, respectively), indicating that even when omitting the experimental restraints, the structures remain close to those originally determined from the n.m.r. data. The energy-minimized mean structures are referred to as (m),, and (SA3),, respectively. All the minimized

(SA2), mean

structures sati&y the experimental restraints quite well, with (SAl), and (SA3), exhibiting smaller RMSD values from experimental restraints than the (SA2), structure (Table 2). Furthermore, the structures

uniformly

exhibit

reasonably

covalent

good

bond geometries and non-bonding energies which are even lower than those for the individual SAl, SA2 and SA3 structures. As__ shown__ by the Ramachandran plot for the (SAl),, (SA2), and (SA3), structures (Fig. 5), more than 95% of the kringle 2 residues lie within the allowed CDand Y backbone dihedral angle regions or within the bridging region (Ramachandran & Sasisekharan , 1968; Richardson, 1981).

(b) Solution structure of the kringle 2-6-AHA complex

Despite collected

Atomic RMSD values are calculated on the atoms within the CyslLCys80 peptide stretch. Structures are labeled &s in Table 2.

-120

the fact that on

solutions

complex, the structures derived without explicitly

the NOESY of

the

kringle

were 2-6-AHA

data

described above were including the ligand in

1043

n.m.r. Solution Structure of t-PA Kringle 2

Figure 6. All-atom best fit superpositions of the 12 SAl structures. Stereo view of binding site region.

the calculations. A total of 16 NOESY cross-peaks (not shown) were assigned to connectivities between 6-AHA and the kringle 2 Tyr36, Trp62, His64 and Trp72 aromatic protons. These NOES afforded 14 intermolecular distance restraints. From inspection of kringle 2 structures as well as from the dipolar contact points, the binding site was unambiguously located to a protein area characterized by an aromatic pocket consisting of the aromatic rings mentioned above and an anionic center consisting of the Asp55 and Asp57 carboxylate groups. Note that the Asp55 and/or Asp57 have previously been proposed to interact with the ligand amino group in plasminogen kringles 1, 4 and 5 (Lerch & Rickli, 1980; Trexler et al., 1982; Thewes et al., 1987, 1999). Figure 6 shows the binding site area, best fit superpositions of 12 SAl structures. It is apparent that the Trp62, His64 and Trp72 side-chains, which conform a central aromatic pocket for the ligand, are rather well defined. By comparison, the positions of the Tyr36, Asp55 and Asp57 side-chains show rather wide variabilities. In addition, neighboring the binding site, the Tyr74 phenol ring is relatively well defined, while the Lys34 and Arg69 side-chains are poorly positioned, which may reflect the latter representing flexible surface residues that yield sparse experimental constraints. Once the binding site was unambiguously located, we were able to further refine the kringle 2 structure by incorporating the 6-AHA molecule in the calculations. In a first step, the ligand was docked in an extended conformation at the binding pocket of @Al),, one of the better-defined kringle 2 structures, by placing the positively charged amino group of 6-AHA close to the two carboxylate groups of Asp55 and Asp57, and positioning the hydrocarbon chain of 6-AHA to contact the exposed, mostly aromatic, kringle surface. Then, 1000 cycles of Powell restrained minimization were carried out including the 14 kringle 2-6-AHA intermolecular distance restraints as well as all the intramolecular

restraints discussed above. During these calculations, instead of the soft repulsion. function (eqn (3)), the Lennard-Jones and the Coulombic electrostatic (with distance-dependent dielectric constant) potentials were turned on to account for the non-bonding interactions. The resulting structure of the kringle 2-6-AHA complex is henceforth referred to as SAA4 (not shown). A statistical analysis and atomic RMSD values for the SAA4 structure are summarized in Tables 4

Table 4 Statistics for the kringle 2-6-AHA complex structures SAA4 RMSD values from experimental All (624) Intra residue (117) Short range (h-j1 55) (264) Long range (h-j1 > 5) (207) Hydrogen bond (22) Intermolecular (14)

restraints (A)? 0949 0.032 o-027 9073 0.017 0.073

SAA5 6048 0033 6027 9070 0016 om6

RMSD values from idealized geometryt Bonds (A) (1351) Angles (deg) (2415) Impropers (deg) (624)

9010 2.272 I.085

9010 2.269 I.035

Energies7 Fvdr (kcal/mol) Feloc (kWm4

- 237 -908

- 239 -921

Structures are labeled as follows: SAAI, a structure obtained by 1000 Powell restrained minimization of the (SAI), structure docked with 6-AHA, based on all experimental restraints including 14 kringle 2-ligand (B-AHA) intermolecular NOE distance constraints; SAA5, a structure obtained by 500 Powell restrained minimization of the (SAl), structure docked with 6-AHA based on 2 postulated NOES favoring close contacts between the 6-AHA carboxylate group and the Lys34 and Arg69 side-chain cationic groups as well as all the restraints for SAA4, and followed by an additional 500 steps of Powell restrained minimization based on the restraints for SAA4. t Values were derived as in Table 2.

1044

I.-J. L. Byeon and M. L&is

Table 5 Atomic RMSD values for the lcringle 2-6-AHA complex structures

SAA4 SAA4 SAA4 SAA5 SAA5 SAA5 SAA4

ver~..~ (SAl) veraus m ve~su.s @ii), weraus (SAl) versus SAl verger (SAl), verauS SAA5

Backbone atoms (4

All atoms (4

@98+@14 @59 0.45 097 &@15 058 047 0.25

1+37+0.22 0.97 0.77 1.69+027 @99 082 @47

Atomic RMSD values are for atoms within the Cysl-Cys80 peptide stretch. Structures are labeled as in Tables 2 and 4.

and 5, respectively. Note that restrained energy minimization with ligand caused only small atomic RMSD shifts in the kringle 2 structure (645 A for backbone atoms and O-77 A for all atoms, Table 5). It is also noteworthy that the SAA4 structure exhibits somewhat smaller experimental distance violations (6049 A, Table 4) than the starting (SAl), structure (6059 A, Table 2). The result reflects the fact that the experimental restraints have been derived from n.m.r. spectra of the kringle 2 in the presence of 6-AHA so that introduction of the latter is not expected to perturb the structure. We also note that SAA4 results in a much more favorable electrostatic energy than (SAl), (-908 kcal/mol and 2). Tables 4 - 386 kcal/mol; versus Furthermore, SAA4 exhibits improved van der Waals’ contacts relative to (SAl), (-237 kcal/mol versus - 218 kcal/mol; Tables 4 and 2). As expected (Tulinsky et al., 1988a), the SAA4 structure shows the Asp55 and Asp57 carboxylate groups forming ion pairs with the 6-AHA amino group, while the kringle 2 aromatic pocket interacts with the ligand hydrocarbon chain. Although the SAA4 structure satisfactorily displays a predesigned ligand fit at the binding site, it does not incorporate suitable electrostatic positively charged loci to interact with the anionic carboxylate group of the 6-AHA dipole. From the structure of the kringle 2-6-AHA complex, it is apparent that the Lys34 and Arg69 side-chains, located on the surface of the protein, are sufficiently close to the binding site to potentially interact with the 6-AHA carboxylate group, without significantly perturbing the kringle structure. Lys34, in particular, is most likely to be involved in ligand binding as, in the spectrum of the complex, its C&H, protons yield broad, nondegenerate resonances suggestive of some degree of immobilization (Byeon et al., 1991). We recall that Lys C”H, protons in unrestricted environments are magnetically equivalent and yield a sharp 2-proton (Bundi et al., 1979; multiplet at ~3.02p.p.m. Chazin et al., 1987). On this basis, we have tentatively incorporated two fictitious distance restraints in order to favor pulling together the side-chain cationic groups of Lys34 and Arg69 to the 6-AHA

carboxylate group. Subsequently, 500 cycles ot Powell restrained minimizations were carried out on the minimized mean structure (SA 1), Additional 500 cycles of restrained Powell minimizations followed, in which the two fictitious distance restraints were omitted. The resulting second kringle 2-6-AHA complex structure is referred to as SAA5 (Fig. 7). Its binding site structure is shown in Figure 8 (a) without and (b) with 6-AHA docked in place. SAA4 (not shown) and SAA5 exhibit very similar backbone structures (atomic backbone atom RMSD = 0.25 A; Table 5). Their binding pockets, although shaped alike, differ in a few distinct features: in SAA5 (Fig. 8), only the Asp55 y-carboxylate group is located close enough (-2 A) to form an ion pair with the 6-AHA amino group, but the corresponding Asp57 group is placed rather far (-3.7 A) from the ligand. Also, in SAA5 the ligand molecule is pulled towards the Lys34 and Arg69 side-chains which, in turn, orient themselves towards the binding site, although not close enough to achieve bona fide ionic interactions with the 6-AHA carboxylate group. Interestingly, rather close contacts between 6-AHA and the Lys34 and Arg69 side-chain polar groups were obtained when a distance-independent dielectric constant was used in the restrained minimization calculations (not shown). SAA5 satisfies the experimental restraints slightly better than SAA4 does: and exhibits improved covalent geometries (Table 4). Furthermore, the non-bonding contacts in SAA5 are somewhat more favorable than in SAA4 (Table 4). suggesting that, indeed, the Lys34 and Arg69 sidechains may orient themselves towards the ligand when the latter docks at the binding site. (c) Binding site structure The ligand 6-AHA hydrocarbon chain rests within a V-shaped aromatic groove conformed by the indole rings of Trp62 and Trp72, which contact, each other. The Trp72 indole group is exposed on does not the protein surface, and ligand-binding seem to significantly affect its solvent-accessibility. In contrast, the Trp62 ring is considerably less solvent-exposed as it contributes to the inner surface of the binding pocket (Fig. 8) and its solvent-accessibility is further hindered upon occupancy of the binding site by the ligand. The relative differences between the Trp62 and Trp72 environments (Fig. 8) are also reflected in the overall broader spectrum that the Trp62 indole CH resonances exhibit when compared to those from Trp72 (Byeon et al., 1989). Also, the sharp Trp72 indole NH1 signal, but not the corresponding one from Trp62, broadens beyond detection upon increasing the temperature from 40°C to 67°C when recording the spectrum of the kringle 2-BASA complex in ‘Hz0 at pH 7.28, 300 MHz (not shown). This may reflect a temperature-activated dynamic process, such as H-exchange with the solvent, indicative of side-chain exposure. The His64 imidazole CH2 proton shows retarded

n.m.r. Solution Structure of t-PA Kringle 2

(a)

(b)

Figure 7. Stereo view of the kringle 2-6-AHA complex structure. The model corresponds to structure SAA5 For kringle 2, the backbone (N, C” and C’) atoms as well as the Cys CB and Sy atoms am shown, while for the ligand all atoms are included. (a) View similar to that shown in Fig. 3: (b) view generated by rotating the molecule, as shown in (a), by -90” about the y-axis.

1046

I.-J. L. Byeon and M. Llincis

la)

Fig1 Ire 8. Stereo view of the binding site in SAA5 (a) without 6-AHA, (b) with 6-AHA. For kringle 2, only 1leavy atoms and polar hydrogen atoms are shown (stick representation), while all atoms are shown for 6-AHA (dot and stick represcsntation). The view corresponds closely to that shown in Fig. 7(a).

n.m.r. Solution Structure of t-PA Kringle 2

1047

(b) Figure 9. Space-filling representation of the kringle 2-6-AHA complex, structure SAAB. (a) Front view similar to that shown in Fig. 7(a); (b) side view which is obtained by rotating the molecule, as shown in (a), by -90” about the y-axis. Kringle 2 atoms are shown green, while the ligand atoms are shown red.

1048

I.-J. L. Byeon and M. Llincis

Figure 10. Stereo view ribbon representation of kringle 2 indicating u-helical and /?-sheet structural elements. The model corresponds to @Al),. The cystine bridges are omitted for clarity. The view is approximately the same as in Fig. 3.

‘H-2H exchang e kinetics in response to addition of ligand (Byeon et al., 1989). This is consistent with the structure illustrated in Figure 8, which shows that the His64 aromatic ring gains protection from solvent upon binding 6-AHA, as discussed above for the Trp62 indole group. The model in Figure 8 also shows that the Tyr36 ring is relatively exposed to solvent, and achieves close contacts with the ligand. The Tyr74 phenol ring, positioned near the Asp57, Trp62 and Trp72 side-chains is, however, rather removed from 6-AHA. The Trp25 ring and the Leu46 methyl groups (not shown) are found near the binding site, in a second, less exposed layer in contact with the Trp62 indole ring but without directly interacting with the ligand. The model places the Va135 and Leu70 methyl groups (not shown) at the hydrophobic binding site surface, although unambiguous intermolecular cross-peaks connecting these protons to 6-AHA were not identified in the NOESY spectra. While the ligand amino end group shows a delinite interaction with the Asp55 and/or Asp57 y-carboxylate groups (Fig. 8), the ligand carboxylate group exhibits rather poorly defined interactions with the positively charged side-chains of Lys34 and/or Arg69. Figure 9(a) and (b) shows the front view and the side view, respectively, of a space-filling representation of the kringle 2 structure with 6-AHA attached. It is seen that while the ligand amino head group is completely embedded in the deep (upper) end of the binding site, the corresponding ligand carboxylate group is rather exposed (Fig. 9(b)).

structure are illustrated as a ribbon drawing in Figure 10. The sole extended antiparallel /?-sheet found in kringle 2 (the Trp62-Tyr74 sequence) is right-hand twisted. Rudiments of p-sheet elements are also identifiable at sites where the strands simply cross each other and interact through one or two hydrogen bonds. A right-handed a-helical structure is formed by the Ser43-Ala44-Gln44a-Ala44bLeu44c-Gly45 sequence, which contains a threeresidue (44a to 44~) insertion characteristic of plasminogen activators’ (t-PA and kidney-type plasminogen activator (u-PA)) kringles. The Cys22Cys63 and Cys51-Cys75 disulfides are located near the center of the kringle 2 globule, revealing close van der Waals’ contacts (Fig. 7). The backbone chain outlines a large pouch adjacent to the main /?-sheet (Figs 7 and lo), in which the ligand binding site is contained. The Trp62 aromatic ring at the binding site interacts with the Leu46 and Trp25 side-chains, defining a centrally placed hydrophobic core. This hydrophobic cluster further establishes contacts with the His48a and Tyr50 aromatic rings, which are found opposite the binding site face. In conjunction with the antiparallel /?-sheet components and the unique insertion that generates an u-helix (Fig. lo), this array of aromatic groups affords another characteristic feature of the kringle 2 structure. It is noteworthy that the Tyr9-ArglO loop region exhibits a relatively poorly defined side-chain conformation, with significant structural variations (Fig. 4). This may relate to the fact that the Tyr9 residue shows a doubled spectrum, indicative of a slow (n.m.r. time scale) equilibrium between two different conformational states (Byeon et al., 1989).

(d) General features of the kringle 2 structure Kringle 2 exhibits an oblate ellipsoidal shape of approximately 34 A x 30 A x 17 A (Fig. 9(a) and (b)). The folding is characterized by a number of turns and loops (Fig. 7). Features of secondary

4. Conclusions The solution complexed

conformation of t-PA kringle 2 with 6-AHA has been determined on the

1049

n.m.r. Solution Structure of t-PA Kringle 2 basis of ‘H n.m.r. spectroscopic experiments. In the process we have characterized the topology of the kringle 2 ligand-binding site. All final 36 structures, calculated by dynamical simulated annealing using three different classes of starting structures (random coil polypeptides with/without disulfides or an approximate kringle structure modeled after the homologous prothrombin kringle 1 domain), exhibit well-defined conformations with relatively small atomic RMSD values (Fig. 3; Table 4). It is encouraging that the 12 SAO structures, calculated omitting the kringle 2 disulfide bridges, show a close resemblance to the 36 (SAl + SA2 + SA3) structures calculated with the disulfide bonds included, indicating that the n.m.r. data, by itself, are sufficient to determine the protein folding. As found by other authors (Driscoll et al., 1989; Kraulis et al., 1989), well-converged structures were obtained, even though estimated distance restraints were used. This precision reflects highly correlated interproton distances combined with a number of long-range NOE constraints which force proximity among residues far apart in the sequence (Fig. 2). Furthermore, the close resemblence of the SA0 structure, computed without assuming S-S bridges, to the SAl, SA2 and SA3 structures which incorporate cystines based on homology to prothrombin kringles (Pennica et al., 1983; Ny et al., 1984), affords first experimental evidence that the disulfide pairing assumed for the t-PA kringle 2 is, indeed, the correct one. The backbone structures clearly outline a groove or channel placed next to the main antiparallel P-sheet that serves to dock the ligand, 6-AHA (Figs 7(a) and 10). The groove is filled with several ionic and aromatic side-chains, defining a binding pocket for w-amino acids. As predicted (Tulinsky et al., 1988a), the binding site is capped by the negatively charged side-chains of the Asp55 and Asp57 residues, and consists of a rather exposed hydrophobic surface lined by the Tyr36, Trp62, His64 and Trp72 aromatic rings. The Va134 and Leu70 methyl groups protrude into the groove and thus contribute lipophilic character to the binding site. We further suggest that opposite the Asp55 and Asp57 anionic centers, the Lys34 and Arg69 side-chain cationic end groups are likely to contribute an electrostatic field configuration that favors interaction with the ligand carboxylate group. It is most satisfying that the t-PA kringle 2 binding pocket was obtained via structure calculations where the ligand was not explicitly taken into account. We thank Mr S. Cleary and Dr R. F. Kelley for the kringle 2 sample and Mr K. Constantine and Dr M. Madrid for advice on computational procedures. We are also grateful to Dr S. F. Sneddon for graphics assistance, to Prof. C. L. Brooks III for generating Fig. 10 and to Prof. A. Tulinsky for the X-ray crystallographic coordinates of Ptb kringle 1. This research was supported by the U.S. Public Health Service, NIH grant HL 29409. Support from the Pittsburgh Supercomputing Center through the NIH Division of Research, Cooperative

Agreement lP41RR06009, is gratefully acknowledged. Supplementary material, listing t-PA kringle 2 n.m.r. coordinates as well as torsion angle and interproton distance constraints, have been deposited at the Brookhaven Protein Data Bank, accession numbers lPK2 (co-ordinate data) and RlPK2MR (constraints file).

References Banyai, L., Varadi, A. t Patthy, L. (1983). Common evolutionary origin of the fibrin-binding structures of fibronectin and tissue-type plasminogen activator. FEBS Letters, 163, 37-41. Bodenhausen, G., Kogler, H. & Ernst, R. R. (1984). Selection of coherence-transfer pathways in NMR pulse experiments. J. Map. Reson. 58, 370-388. Brobks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. & Karplus, M. (1983). CHARMM: a program for macromolecular energy, minimization, and dynamic calculations. J. Comput. Chem. 4, 187-217. Briinger, A. T. (1988). X-PLOR Version 1.5 User Manual, Yale University Press, New Haven, CT. Briinger, A. T., Clore, G. M., Gronenborn, A. M. t Karplus, M. (1986). Three-dimensional structure of proteins determined by molecular dynamics with interproton distance restraints: application to U.S.A. 83, 3801crambin. Proc. Nat. Acad. Sci., 3805. Bundi, A. & Wiithrich, K. (1979). ‘H NMR parameters of the common amino acid residues measured in of the linear tetrapeptides. aqueous solutions B&polymers, 18, 285-297. Byeon, I.-J. L., Kelley, R. F. & Llinas, M. (1989). ‘H NMR structural characterization of recombinant kringle 2 domain from human tissue-type plasminogen activator. Biochemistry, 28, 9350-9360. Byeon, I.-J. L., Kelley, R. F. & Llinas, M. (1991). Kringle-2 domain of the tissue-type plasminogen activator: ‘H NMR assignments and secondary structure. Eur. J. Biochem. 197, 155-165. Chazin, W. J., Rance, M. & Wright, P. E. (1987). Complete assignment of lysine resonances in ‘H NMR spectra of proteins as probes of surface structure and dynamics. FEBS Letters, 222, 109114. Cleary, S., Mulkerrin, M. G. & Kelley, R. F. (1989). Purification and characterization of tissue plasminogen activator kringle-2 domain expressed in Escherichia coli. Biochemistry, 28, 1884-1891. Clore, G. M., Gronenborn, A. M., Briinger, A. T. & Karplus, M. (1985). Solution conformation of a heptsdecapeptide comprising the DNA binding helix F of the cyclic AMP receptor protein of Escherichia coli. J. Mol. Biol. 186, 435-455. Clore, G. M., Nilges, M., Sukumaran, D. K., Briinger, A. T., Karplus, M. 6 Gronenborn, A. M. (1986u). The three-dimensional structure of crl-purothioninin solution: combined use of nuclear magnetic resonance, distance geometry and restrained molecular dynamics. EMBO J. 5, 2729-2735. Clore, G. M., Briinger, A. T., Karplus, M. & Gronenborn, A. M. (19865). Application of molecular dynamics with interproton distance restraints to threedimensional protein structure determination. J. Mol. Biol. 191, 523-551. Clore, G. M., Gronenborn, A. M., Nilges, M. & Ryan, C. A. (1987). Three-dimensional structure of potato carboxypeptidase inhibitor in solution. A study using nuclear magnetic resonance, distance geometry, and

1050

I.-J. L. Byeon and M. Llinh.

restrained molecular dynamics. Biochemistry, 26. 8012-8023. Collen, D. (1980). On regulation and control of fibrinolysis. Thromb. Haemostasis, 43, 77-89. Cohen, D., Lijnen, H. R., Todd, P. A. $ Goa, K. L. (1989). Tissue-type plasminogen activator: a review of its pharmacology and therapeutic use as a thrombolytic agent. Drugs, 38, 346-388. De Marco, A., Llinis, M. & Wiithrich, K. (1978). Analysis of the ‘H NMR spectra of ferrichrome peptides II: the amide resonances. B&polymers, 17, 637-650. Driscoll, P. C., Gronenborn, A. M., Beress, L. & Clore, G. M. (1989). Determination of the three-dimensional solution structure of the antihypertensive and antiviral protein BDS-I from the sea anemone Anemonia suleata: a study using nuclear magnetic resonance and hybrid distance geometry-dynamical simulated annealing. Biochemistry, 28, 2188-2198. Gething, M.-J., Adler, B., Boose, J.-A., Gerard, R. I).. Madison, E. L., McGookey, D., Meidell, R. S., Roman, L. M. & Sambrook, J. (1988). Variants of human tissue-type plasminogen activator that lack specific structural domains of the heavy chain. EMBO J. 7, 2731-2740. Hochschwender, S. M., Laursen, R. A., De Marco, A. $ Llinas, M. (1983). 600 MHz ‘H nuclear magnetic resonance studies of the kringle 4 fragment of human plasminogen. Arch. Biochem. Biophys. 223, 58-67. Hubbard, R. (1986). HYDRA: Harvard York Drawing Program, University of York, Heslington, York. Kraulis, P. J., Clore, G. M.. Nilges, M., Jones, T. A., Pettemson, G., Knowles, J. C Gronenborn, A. M. (1989). Determination of the three-dimensional solution structure of the C-terminal domain of cellobiohydrolase I from Trichoderma reesei. A study using nuclear magnetic resonance and hybrid distance geometry-dynamical simulating annealing. Biochemistry, 28, 7241-7257. Lerch, P. G. & Rickli, E. E. (1980). Studies on the chemical nature of lysine-binding sites and on their localization in human plasminogen. Biochim. Biophys. Acta, 625, 374-378. Motta, A., Laursen, R. A., Llinas, M., Tulinsky, A. & Park, C. H. (1987). Complete assignment of the aromatic proton magnetic resonance spectrum of the kringle 1 domain from human plasminogen: structure of the ligand-binding site. Biochemistry, 26, 38277 3836. Nilges, M. (1990). In X-PLOB Version 2.1 User Manual (Briinger, A. T., author), pp. 257-264, Yale University Press, New Haven. CT. Nilges, M., Clore, G. M. & Gronenborn, A. M. (1988u). Determination of three-dimensional structure of proteins from interproton distance data by hybrid distance geometry-dynamical simulated annealing calculations. FEBS Letters, 229, 317-324. Nilges, M., Gronenborn, A. M.. Briinger, A. T. & Clore. G. M. (1988b). Determination of three-dimensional structure of proteins from interproton distance restraints. Application to crambin, potato carboxypeptidase inhibitor and barley serine proteinase inhibitor 2. Protein Eng. 2, 27-38. Nilges, M., Clore, G. M. & Gronenborn, A. M. (1988c). Determination of three-dimensional structure of proteins from interproton distance data by dynamical simulated annealing from a random array of atoms. FEBS Letters, 239, 129-136. Ny, T., Elgh, F. & Lund, B. (1984). The structure of the human tissue-type plasminogen activator gene: corre-

lation of intron and exon structures to functional and structural domains. Proc. Nat. Acad. Sci., U.S.A. 81, 5355-5359. Pardi, A., Billeter, M. t Wiithrich, K. (1984). Calibrafion of the angular dependence of the amide proton-C” proton coupling constants, ‘Jnn., in a globular protein: use of 3&NX for identification of helical secondary structure. J. Mol. Biol. 180, 741-751. Park, C. H. & Tulinsky, A. (1986). Three-dimensional structure of the kringle sequence: structure of prothrombin fragment 1. Biochemistry, 25, 3977-3982. Pennica, D., Holmes W. E., Kohr, W. J., Harkins, R. N., Vehar, G. A., Ward, C. A., Bennett, W. F., Yelverton, E., Seeberg, P. H., Heyneker, H. L. & Goeddel, D. V. (1983). Cloning and expression of human tissue-type plasminogen activator cDNA in E. co&. Nature (London), 301, 214-221. Ramachandran, G. M. & Sasisekharan, V. (1968). Conformation of polypeptides and proteins. Advan. Protein Chem. 23, 283-437. Ramesh, V., Petros, A. M., Llinas, M., Tulinskg, A. S: Park, C. H. (1987). Proton magnetic resonance study of lysine-binding to the kringle 4 domain of human plasminogen: the structure of the binding site. J. Mol. Biol. 198, 481-498. Richardson, J. S. (1981). The anatomy and taxonomy of protein structure. Advan. Protein Chem. 34, 167-339. Ryckaert, J. P., Ciccotti, G. & Berendsen, H. J. C. (1977). Numerical integration of the Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys. 23. 327-337. Sneddon, S. F. (1990). MolX: A Series of General Purpow Prograw~a (Version 17). Molecular Graphics Thewes, T., Ramesh, V.. Simplaceanu, E. L. & Llinas, M. (1987). Isolation, purification and ‘H NMR characterization of a kringle 5 domain fragment from human plasminogen. Biochim. Biophys. Acta. 912, 254-269. Thewes, T., Constantine, K., Byeon, I.-J. 1,. & Llinas, M. (1990). Ligand interactions with the kringle 5 domain of plasminogen. J. Biol. Ghsm. 265, 3906-391s. Trexler, M., Vali, Z. & Patthy. L. (1982). Structure of the o-aminocarboxylic acid-binding sites of human plasminogen: arginine 70 and aspartic acid 56 are essential for binding of ligand by kringle 4. J. Biol. Chem. 257, 7401-7406. Tulinsky, A., Park, C. H., Mao. B. & Llinas. M. (1988n). Lysine/fibrin binding sites of kringles modeled after the structure of kringle 1 of prothrombin. Proteins, 3. 85-96. Tulinsky, A., Park, C. H. & Skrzypczak-Jankun, E. (1988b). Structure of prothrombin fragment 1 refined at, 2.8 A resolution. J. Mol. Biol. 202, 885-991. van Zonneveld, A.-J.. Veerman, H. & Pannekoek. H. (1986a). On the interaction of the finger and the kringle-2 domain of tissue-type plasminogen a&vator with fibrin. ,I. Biol. Chem. 261, 14214-14218. van Zonneveld. A.-.J., Veerman, H. & Pannekoek. H. (198%). Autonomous funct.ions of structural domains on human tissue-type plasminogen activator. Proc. Nat. Acad. Sci., U.S.A. 83, 4670-4674. Verheijen, J. H., Caspers, M. P. M.. Chang. G. T. G.. de Munk, 0. A. W., Pouwels, P. H. & Enger-Valk, B. E. (1986). Involvement of finger domain and kringle 2 domain of tissue-type plasminogen activator in fibrin binding and stimulation of activity by fibrin. EMBO .J. 5, 3525-3530. Wagner, G.. Braun, W., Havel, T. F.. Schautnann, T.. GO,

n.m.r. Solution Structure of t-PA Kringle 2 N. & Wiithrich, K. (1987). Protein structures in solution by nuclear magnetic resonance and distance geometry: the polypeptide fold of the basic pancreatic trypsin inhibitor determined using two different algorithms, DISGEO and DISMAN. J. Mo2. Biol. l%, 611-639. Wiithrich, K., Billeter, M. & Braun, W. (1983). Pseudostructures for the 20 common amino acids for use in

1051

studies of protein conformations by measurements of intramolecular proton-proton distance constraints with nuclear magnetic resonance. J. Mol. Biol. 169, 949-961. Zuiderweg, E. R. P., Boelens, R. & Kaptein, R. (1985). Stereospecific assignments of ‘H NMR methyl lines and conformation of valyl residues on the lac repressor headpiece. Biopolymers, 24, 60-611.

Edited by P. E. Wright