The Native Strains in the Hydrophobic Core and Flexible Reactive Loop of A

View metadata, citation and similar papers at core.ac.uk brought to you by CORE Researchprovided Article by Elsevier1181 - Publisher Connector

The native strains in the hydrophobic core and flexible reactive loop of a serine protease inhibitor: crystal structure of an uncleaved ␣1-antitrypsin at 2.7 Å Seong-Eon Ryu*, Hee-Jeong Choi, Ki-Sun Kwon, Kee Nyung Lee and Myeong-Hee Yu

Background: The protein ␣1-antitrypsin is a prototype member of the serpin Address: Protein Engineering Research Division, (serine protease inhibitor) family and is known to inhibit the activity of neutrophil Korea Research Institute of Bioscience and elastase in the lower respiratory tract. Members of this family undergo a large Biotechnology, KIST, P.O. Box 115, Yusong, Taejon 305-600, South Korea. structural rearrangement upon binding to a target protease, involving cleavage of the reactive-site loop. This loop is then inserted into the main body of the enzyme *Corresponding author. following the opening of a central ␤ sheet, leading to stabilization of the E-mail: [email protected]

structure. Random mutageneses of ␣1-antitrypsin identified various mutations Key words: ␣1-antitrypsin, conformational that stabilize the native structure and retard the insertion of the reactive-site loop. transition, loop flexibility, metastability, stabilizing Structural studies of these mutations may reveal the mechanism of the mutations conformational change. Received: 25 June 1996 Revisions requested: 19 July 1996 Results: We have determined the three-dimensional structure of an uncleaved Revisions received: 8 August 1996 ␣1-antitrypsin with seven such stabilizing mutations (hepta ␣1-antitrypsin) at 2.7 Å Accepted: 29 August 1996 resolution. From the comparison of the structure with other serpin structures, we Structure 15 October 1996, 4:1181–1192 found that hepta ␣1-antitrypsin is stabilized due to the release of various strains that exist in native wild type ␣1-antitrypsin, including unfavorable hydrophobic © Current Biology Ltd ISSN 0969-2126 interactions in the central hydrophobic core. The reactive-site loop of hepta

␣1-antitrypsin is an extended strand, different from that of the previously determined structure of another uncleaved ␣1-antitrypsin, and indicates the inherent flexibility of the loop.

Conclusions: The present structural study suggests that the uncleaved

␣1-antitrypsin has many folding defects which can be improved by mutations. These folding defects seem to be utilized in a coordinated fashion in the

regulation of the conformational switch of ␣1-antitrypsin. Some of the defects, represented by the Phe51 region and possibly the Met374 and the Thr59 regions, are part of the sheet-opening mechanism.

Introduction and ovalbumin (a non-inhibitory serpin) the loop forms a Proteins in the serpin (serine protease inhibitor) family distorted or regular ␣ helix, respectively [5,10]. In have a common structural motif of about 370–400 amino antithrombin the loop has an extended conformation and acids comprising two central ␤ sheets surrounded by a the starting region of the loop is partially inserted into small ␤ sheet and several helices [1,2]. The functions of sheet A [11,12]. It is proposed that the helical conformation serpin proteins range from blood coagulation and fibrinoly- is the native form, and the extended conformation is the sis to tumor suppression [3]. Many of these functions are inhibitory form which is used for the target protease mediated by the protease inhibitor activity of serpins. Two binding [10]. However, the role of the mobility and confor- characteristic features of the inhibitory serpins are the mation of the loop in the serpin function is yet to be char- metastability of the native state [4] and the mobility of the acterized in detail. The stability of the inhibitory serpins is reactive loop [4,5]. The inhibitory serpins have a unique greatly increased upon the cleavage and insertion of the loop of about 20 residues that binds tightly to the active loop [4]. The difference in stability between the native and site of the target proteases to inhibit the proteolytic activ- loop-inserted states has an important role in the regulation ity. This reactive loop is cleaved at a specific site by a of the inhibitory-serpin function [2,8,13–15]. The non- target protease [6] and the cleaved loop is incorporated into inhibitory serpins including ovalbumin and angiotensino- ␤-sheet A, leading to a large rearrangement of the strands gen, do not show the structural transition [15,16]. in sheet A with concomitant stabilization of the structure

[5,7–9]. Two kinds of loop conformations were found in The protein ␣1-antitrypsin, which is a member of the serpin different uncleaved serpin structures. In antichymotrypsin family, inhibits the activity of neutrophil elastase in the 1182 Structure 1996, Vol 4 No 10

lower respiratory tract [17,18]. The deficiency of a extends to the region of the P20 (20th residue in the functional ␣1-antitrypsin due to a variety of hereditary N-terminal direction from the protease cleavage site, P1) mutations leads to emphysema and the aggregation of even though some of the hydrogen bonds are not tight mutant ␣1-antitrypsin in liver cells is often associated with (Fig. 3). This is in contrast to the gap between strand 5 liver diseases such as infant hepatitis [19,20]. The and strand 3 of sheet A in the same region shown in the mutations stabilizing the uncleaved ␣1-antitrypsin were uncleaved antithrombin [11,12] and ovalbumin structures found by random mutageneses [21,22]. These mutations do [5]. In the uncleaved antithrombin, the gap is filled with not affect the inhibitory activity of ␣1-antitrypsin. One of the preinserted loop residues (P14–P17). In ovalbumin, these stabilizing mutation (Phe51→Leu) can suppress the the gap, which is relatively small, is occupied with water folding defect of the Z-type variant [23], presumably by molecules. Hepta ␣1-antitrypsin does not show the loop inhibiting the loop-sheet polymerization [24]. Recently, a preinsertion observed in the uncleaved antithrombin. low-resolution structure of an uncleaved ␣1-antitrypsin was This contradicts the proposal based on limited proteolytic determined at 3.45Å [25]. digestion [26] and peptide annealing experiments [27]. A possible method by which stabilization of sheet A occurs We report here the three-dimensional (3D) structure of an in the absence of the preinsertion will be discussed in the uncleaved ␣1-antitrypsin with seven such stabilizing muta- following sections. tions (hepta ␣1-antitrypsin) at 2.7Å resolution. The structure of hepta ␣1-antitrypsin shows differences in loop Before superposition with the cleaved ␣1-antitrypsin, conformation from the previously determined uncleaved using C␣ positions, we divided hepta ␣1-antitrypsin into ␣1-antitrypsin structure and suggests an inherent plasticity two fragments (proposed earlier [9]) to test whether each of the reactive loop. There is no preinsertion of the reac- fragment moves as a unit during the structural transition. tive-site loop into ␤-sheet A, and strands 3 and 5 of the Fragment 1 contained helices A, B, C, G, H and I, sheets sheet have a better hydrogen-bonding network than other B and C and strands 5 and 6 of sheet A (229 residues), and uncleaved serpin structures [5,11,12]. A structural compar- the smaller of the two fragments (fragment 2) comprised ison of hepta ␣1-antitrypsin with other serpin proteins pro- helix F and strands 1, 2 and 3 of sheet A (63 residues). vides an explanation for the mechanism of the stabilizing The flexible joint regions (helices D and E) and the reac- effect of each mutation site and insights into the function tive-site loop were excluded in the superposition. The of the serpin proteins. root mean square (rms) deviation for the superposition when both fragments were included (292 residues) was Results and discussion 1.80Å. When each fragment was superposed separately,

The crystal structure of hepta ␣1-antitrypsin was deter- the rms deviation was reduced to 1.46Å and 1.28Å for mined by molecular replacement using the structure of fragments 1 and 2, respectively. Figure 4a shows that cleaved ␣1-antitrypsin (PDB code 7api) as a search model. residues of fragment 2 of hepta ␣1-antitrypsin are dis- The rotation and translation searches were performed with placed in one direction from those of the cleaved ␣1-antit- the initial low resolution data at 4.0Å. Later, data at higher rypsin when the two proteins were superposed using resolutions of up to 2.7Å were obtained and used for the residues of fragment 1. The observation that individual refinement. The refinement resulted in a model with a fragments superpose more closely than the overall protein crystallographic R value of 19.5% and excellent geome- and the concerted displacement of residues in fragment 2 tries for 3597 protein atoms of the residues 21–394. The by the superposition of fragment 1 support the proposal model agrees well with electron-density maps in most that fragment 2 slides relative to fragment 1 during the regions of the protein including the reactive-site-loop conformational transition [9]. Ovalbumin and the region and the sheet-A region which shows a large uncleaved antithrombin can be superposed with hepta rearrangement of the ␤ strands with respect to the struc- ␣1-antitrypsin with rms deviations of 1.97Å and 1.98Å ture of the cleaved ␣1-antitrypsin (Fig. 1). respectively with both fragments included (Fig. 4b). The relative orientation of the reactive-site loop to the main

Structure description body of hepta ␣1-antitrypsin is similar to that of ovalbumin The structure of hepta ␣1-antitrypsin has a similar topol- even though there are differences in local conformation. ogy to the structures of uncleaved antithrombin [11,12] and the uncleaved antichymotrypsin [10], but with several In the crystal used for the structure determination, hepta differences. The entire reactive-site loop residues from ␣1-antitrypsin contacts another molecule related by a crys- residue 343 to residue 358 (strand 4 of sheet A in the tallographic dyad around the reactive-site loop. This cleaved ␣1-antitrypsin) form an extended loop above the crystal contact generates a dimer of molecules interacting molecule in the view used in Figure 2. Consequently, with each other in a head-to-head fashion. The interface sheet A consists of only five strands. These five strands between two molecules in this dimer does not have have a good hydrogen-bonding network. The hydrogen- enough space to allow a protruding helix as observed in bonding network between strands 3 and 5 of sheet A ovalbumin and antichymotrypsin. The reactive-site loop Research Article Uncleaved ␣1-antitrypsin Ryu et al. 1183

Figure 1

Electron density of sheet A and P1 regions.

The current model of hepta ␣1-antitrypsin is superposed, with 2Fo–Fc electron density maps drawn in stereo at a contour level of 1.5 ␴: (a) sheet A region showing strands 2, 3, 5 and 6 (from right). There is no preinsertion of the reactive-site loop; (b) P1 (Met358) region of the reactive-site loop. A crystallographic twofold related molecule is displayed as broken lines.

in hepta ␣1-antitrypsin forms an extended strand which tions were combined, the stabilizing effect was additive can be accommodated in the intermolecular space. The (KNL and M-HY, unpublished results) and the change in ∆ structure of an uncleaved ␣1-antitrypsin with stabilizing G of hepta ␣1-antitrypsin was 9.0 kcal. Figure 5 shows mutations at three places (triple ␣1-antitrypsin) (Table 1) the position of the mutations in a schematic drawing of was recently determined at 3.45Å resolution [25]. Triple the structure. The mutations are divided into two cate-

␣1-antitrypsin has similar crystal contacts to those of hepta gories depending on the environment of the residues. ␣1-antitrypsin but the loop forms a distorted helix that is squeezed down into the main body of the molecule to be Phe51→Leu, Met374→Ile and Thr59→Ala mutations are accommodated in the limited intermolecular space. surrounded by residues that undergo major conformational changes during the structural transition. Residue 51 Location of each mutation in the structure in strand 5 of sheet B is approximately at the center of the

The uncleaved ␣1-antitrypsin used for the structure deter- hydrophobic core made by sheets A and B. This region mination has mutations at seven different places. Each was termed the shutter region [1,28] due to its critical role mutation stabilizes the protein with a change in Gibbs in the opening of sheet A during the transition. Residue free energy (∆G) of 1.0–2.3 kcal (Table 1) [22]. Among 374 is in the middle of strand 4 of sheet B. The side chain these, the Phe51→Leu and Met374→Ile mutations have of Met374 contacts the side chain of Thr345 in the relatively high stabilizing effects. When all seven muta- cleaved ␣1-antitrypsin. In hepta ␣1-antitrypsin the reac- 1184 Structure 1996, Vol 4 No 10

Figure 2

C␣ trace of hepta ␣1-antitrypsin with every 20th residue indicated as a black ball and 360 360 numbered. The N (residue 21) and C (residue 200 200 394) termini are represented by larger balls 220 220 and labeled. 280 280 240 240 340 340

260 260 C 380 C 380

160 100 160 100 N N 40 40 80 60 80 60 300 140 300 140

120 120 180 320 180 320

tive loop including Thr345 (P14) is not incorporated into sheet A. The significance of these differences in the envi- sheet A. Consequently, the environment for residue 374 ronment will be discussed later. is substantially different between the cleaved and the uncleaved proteins. The environment for the side chain Other mutation sites, including residues 381, 387, 68, and of residue 59 in helix B is also affected by the structural 70, reside in regions where structural changes during the transition due to the movement of strands 2 and 3 of transition are not significant. Residues 381 and 387 are in

Figure 3

Hydrogen-bonding network between strands Thr345 Thr345 3 and 5 of sheet A. The main-chain atoms of strands 3 (s3A) and 5 (s5A) in the P20 (Thr339) region are drawn. Carbon, nitrogen Lys343 Lys343 and oxygen atoms are colored as black, blue, and red, respectively. The putative hydrogen bonds are drawn as broken lines, with the 3.8 3.8 distances indicated. The viewpoint is the same as in Figure 2. (The diagram was drawn using MOLSCRIPT [43].)

4.0 Gly192 4.0 Gly192

Thr339 Thr339 3.9 3.9

Phe190 Phe190 3.7 3.7

Val337 Val337 3.9 3.9

Ile188 Ile188 4.7 4.7

Lys335 Lys335

s5A s3A s5A s3A Research Article Uncleaved ␣1-antitrypsin Ryu et al. 1185

Figure 4

C␣ superposition of the regions with the major structural transition. (a) Hepta (a) ␣1-antitrypsin (thick line) and the cleaved ␣1-antitrypsin (thin line) are compared using the C␣ alignment of fragment 1 (see text). Strand 4 of sheet A (s4A) of the cleaved ␣1-antitrypsin becomes a loop in the hepta s5A* s5A* ␣1-antitrypsin above the molecule. Secondary s4A* s4A* structural elements of the cleaved s6A* s6A* s3A* s3A* ␣1-antitrypsin are labeled with stars. A concerted displacement of strands 1, 2 and 3 s2A* s2A* of sheet A, and helix F (fragment 2) indicates that the fragment slides relative to the main body of the protein (fragment 1) during the conformational transition. (b) The uncleaved structures of three different serpins are compared. The structures of hepta hF* hF* ␣1-antitrypsin, ovalbumin and antithrombin are represented as thick, thin, and broken lines, s1A* s1A* respectively. The antithrombin structure has a gap in the P12–P16 residues due to lack of a s6A s6A model for this region.

s3A s3A s5A s1A s5A s1A

hF hF s2A s2A

(b)

s6A s6A

s1A s1A

s5A s5A s3A s3A

s2A s2A

strand 5 of sheet B. The side chains of these residues are on Conformational strains in the shutter region the opposite side from sheet A and do not make any signifi- Mutations of Phe51 into a smaller amino acid significantly cant contacts with residues having conformational differ- increases the stability of uncleaved ␣1-antitrypsin [21]. ences between the two states. The environments of residues Phe51 is located in the center of the hydrophobic core 68 and 70, in helix B and the following loop, respectively, made by ␤ sheets A and B, and helix B. The critical role of also seem to be unchanged during the transition. the residues in this region suggests the existence of a 1186 Structure 1996, Vol 4 No 10

Table 1 Figure 5

Stabilizing effect of the mutations.

Mutation Location in ∆(∆G)* Hepta Triple the structure (kcal mol–1) s4C Phe51→Leu end of s6B 2.1 o Met374→Ile middle of s4B 2.3 o Ser381→Ala start of s5B 1.0 o s2C Lys387→Arg end of s5B 1.0 o s3C s1B Thr59→Ala start of hB 1.0 o o s1C s2B Thr68→Ala end of hB 1.0 o o Ala70→Gly loop between 1.6 o o hB and hC s6A s5A s3A Hepta 9.0 s3B Triple 3.4 s2A Arg387 s4B Ile374 *The ∆(∆G) value of each mutation is adopted from [22]. The stabilization effect of these mutations is additive (KNL & MHY, s6B s5B unpublished results). The mutations included in the hepta and the triple Leu51 hG hH Ala381 ␣1-antitrypsin proteins are specified as circles. hF hD ‘shutter’ that controls the opening of sheet A. In hepta hA ␣ -antitrypsin, with the Phe51→Leu mutation, the side Ala59 1 hB chain of Leu51 makes extensive hydrophobic interactions with the side chains of Phe372, Phe384, and Ile293. hI s1A Phe372 and Phe384 are in strands 4 and 5 of sheet B, and Ile293 is in strand 6 of sheet A. The side chain of Phe384 hE Ala68 extends the hydrophobic core by interactions with the hC side chains of Ile188 and Phe190 in strand 3 of sheet A. In plasminogen activator inhibitor-1 (PAI-1), the mutations of Phe372 (Phe384 in the ␣ -antitrypsin frame) into 1 Gly70 smaller amino acids, stabilize the active form of the uncleaved PAI-1, as do the mutations in residue 51 of ␣1- antitrypsin [29]. Figure 6 presents the hydrophobic core, Location of stabilizing mutations in the structure. The secondary consisting of residues 51, 384, 188 and 190, in the hepta structure elements and the residues with stabilizing mutations are and cleaved ␣1-antitrypsin structures. specified on the drawing. Helices A–H (hA–hH) are drawn in green and strands of sheets A (s1A–s5A), B (s1B–s6B) and C (s1C–s3C) are drawn in purple. The side-chain atoms for the residues with The structure of hepta ␣1-antitrypsin with the → stabilizing mutations are represented as blue balls. The mutations can Phe51 Leu mutation shows that the side-chain atoms of be divided into two categories depending on the environment of the Leu51 make good van der Waal (VDW) contacts with residues. Phe51→Leu, Met374→Ile and Thr59→Ala mutations are neighboring residues and there is no noticeable hole in the surrounded by residues that undergo major conformational changes region despite the reduction in the size of the side chain. during the native-to-cleaved transition. Other mutated residues including residues 381(Ser381→Ala), 387(Lys387→Arg), The structure of hepta ␣1-antitrypsin suggests that pheny- 68(Thr68→Ala), and 70(Ala70→Gly) reside in the region where the lalanine in position 51 in the wild-type uncleaved ␣1-antit- structural changes during the transition are not significant. (The figure rypsin may cause unfavorably close contacts, mainly with was drawn by MOLSCRIPT [43].) Phe384. The elimination of these contacts in the region of residue 51 may be part of the reason for the stabilization of → the uncleaved ␣1-antitrypsin by the Phe51 Leu muta- If the unfavorably close VDW interactions exist in the tion. A similar phenomenon seems to exist in antithrom- residue 51 region, the sum of other favorable interaction bin. In the cleaved antithrombin [30], the side chain of has to hold the repulsion due to the high energy contacts.

Phe78 (residue 51 in the ␣1-antitrypsin frame) makes good The repulsive force in the ‘squeezed’ structure may con- VDW interactions with neighboring residues, and there is tribute to the opening of sheet A between strand 3 and no high-energy contact between Phe78 and Phe423 strand 5 when the force holding the structure is disturbed

(residue 384 in the ␣1-antitrypsin frame). On the other upon the protease binding. Especially, the repulsive hand, the side chain of Phe77 in the uncleaved antithrom- energy between Phe51 and Phe384 may be used to push bin [11,12] (residue 51 in the ␣1-antitrypsin frame) has the residues of strand 3 of sheet A (Ile188 and Phe190) several high-energy contacts with Phe422 (residue 384 in and the residue of strand 6 (Ile293) in opposite directions the ␣1-antitrypsin frame) (Table 2). for the sheet A opening. Research Article Uncleaved ␣1-antitrypsin Ryu et al. 1187

Figure 6 changes during the native-to-cleaved transition. In the → hepta ␣1-antitrypsin structure with the Met374 Ile mutation, the side chain of Ile374 in strand 4 of sheet B contacts neighboring hydrophobic residues (Phe384, Phe372, Leu338, Phe190, Tyr244 and Ala250) to form a strong hydrophobic core. The hydrophobic core of the native protein with a methionine at this position may not be very strong as methionine only has medium-range hydropho- bicity [32]. The Met374→Ile mutation may increase the stability of the hydrophobic core due to the contribution of the highly hydrophobic side chain of isoleucine. In the structure of cleaved ␣ -antitrypsin, the reactive loop incor- Interaction of residues in the central hydrophobic core of hepta and 1 porated into sheet A runs over the Met374 and the side cleaved ␣1-antitrypsin: (a) hepta ␣1-antitrypsin; (b) cleaved ␣1-antitrypsin. In the drawings, residues 51, 384, 188, and 190 are chain of Met374 contacts that of Thr345 (Fig. 7). An left, center, down and up, respectively, and the carbon, nitrogen and isoleucine in the 374 position results in unfavorable inter- oxygen atoms are represented in white, blue and red, respectively. actions with the hydrophilic side chain of Thr345. The rel- (The figure was drawn with QUANTA [Molecular Simulations Inc.].) ative destabilization of the loop-inserted structure would stabilize the native structure by shifting the equilibrium Phe51 and other residues in the hydrophobic core (Phe384, towards the native conformation in the equilibrium Ile188 and Phe190) are conserved in ovalbumin which is a between this and the partial loop-inserted conformation. non-inhibitory serpin. However, the side chain of Phe51 in ovalbumin does not show any significantly high energy Residue 59 is in the middle of helix B and is close to contacts with neighboring residues (Table 2). This is con- strands 2 and 3 of sheet A. In hepta ␣1-antitrypsin with the sistent with the inability of ovalbumin to have a ␤-sheet Thr59→Ala mutation, the side chain of Ala59 contacts the rearrangement even in the cleaved derivative, plakalbumin side chains of Leu100, Leu99, Val55 and Tyr138 to form a [31]. Despite the limited resolution of the structures used strong hydrophobic core. The threonine in the wild-type in the comparison, the possibility that this repulsion only protein would not fit into this hydrophobic core. In the exists in the inhibitory serpins suggests that it may have an structure of the cleaved ␣1-antitrypsin, the OG1 atom of important role in the function of this class of serpins. Thr59 forms a hydrogen bond with the OD1 atom of Asn186. The loss of a hydrogen bond in the Thr59→Ala Unfavorable hydrophobic interactions mutant would destabilize the cleaved structure. However, The side chains of the residues 374 and 59 are close to the the uncleaved structure is not affected by the loss of the side chains of sheet A that undergo the major structural OH group as this hydrogen bond is not likely to be

Table 2

High-energy contacts of residues in the central hydrophobic core region.*

Protein Contact Distance VDW energy (Å) Pair Total

Hepta api L51(CB):I50(O) 2.99 1.45 -6.70 Cleaved api F51(CD2):A336(CB) 3.38 1.23 -3.53 Cleaved att F78(CE2):F373(O) 2.84 1.97 -4.81 Uncleaved att F77(CZ):F422(CB) 3.27 2.20 -0.70 F77(CZ):F422(CG) 3.33 1.00 F77(CZ):F422(CD1) 3.29 1.23 Ova F51(CZ):A338(CB) 3.37 1.26 -5.02

*The close contacts made by the side-chain atoms of residue 51 residue 51 (or the equivalents) with neighboring atoms. The total (residue 78 in the cleaved antithrombin and residue 77 in the energy difference between the cleaved and uncleaved forms of uncleaved antithrombin) with neighboring atoms are presented. antithrombin indicates a conformational strain in the region. Residue 422 in the uncleaved antithrombin corresponds to residue Phenylalanine in position 51 of the wild-type uncleaved ␣1-antitrypsin 384 in the ␣1-antitrypsin frame. The contacts with van der Waals would increase the interaction energy significantly and also create a (VDW) energy greater than 1.0 kcal mol–1 are listed. The VDW energy conformational strain (see text). Ovalbumin, which is not an inhibitory was calculated from PDB coordinates of each protein by using serpin, does not have the conformational strain. Cleaved api is the X-PLOR [40]. The structures used in the calculation were refined with cleaved ␣1-antitrypsin (PDB code 7api); hepta api, hepta ␣1- similar geometric restraints (rms bond idealities of the structures are antitrypsin; cleaved att, the cleaved antithrombin (PDB code 1att); 0.012–0.020 Å) allowing the comparison made in the table. The total uncleaved att, the uncleaved antithrombin inhibitory form (PDB code VDW energy is the sum of every contact of the side chain atoms of 1ant); ova, ovalbumin (PDB code 1ova). 1188 Structure 1996, Vol 4 No 10

Figure 7

Interaction of residue 374 with neighboring (a) residues. Residues interacting with residue

374 in hepta ␣1-antitrypsin and the cleaved ␣1-antitrypsin structures are drawn in stereo Ala250 Tyr244 Ala250 Tyr244 using MOLSCRIPT [43]; (a) hepta ␣1-antitrypsin; (b) the cleaved ␣1-antitrypsin (PDB code 7api).

Ile374 Ile374 Phe190 Phe190

Phe384 Phe384 Leu338 Leu338

(b) Tyr244 Tyr244 Ala250 Ala250

Thr345 Thr345 Met374 Met374

Phe190 Phe190

Phe384 Phe384 Leu338 Leu338

present due to the displacement of strand 3. As in the the beginning and end regions of the strand 5 of sheet B, Met374→Ile mutation, the Thr59→Ala mutation stabi- respectively. Even though many residues in strand 5 of lizes the uncleaved structure and destabilizes the loop- sheet B contact with sheet A , side chains of both Ala381 inserted structure. However, in this case, the contribution and Arg387 are distant from the residues of sheet A of the destabilization of the loop-inserted structure to the because they are found in the other face of sheet B. The stability of the uncleaved structure is not clear because the side chain of Ala381 is close to Val55 (helix B), Leu30 structural change in the residue 59 region occurs only (helix A) and Leu99 (helix D), which do not have much when the reactive site is cleaved and the loop is com- movement during the structural transition. A serine pletely inserted. Equilibrium between the uncleaved form residue at position 381 in the wild-type protein would and the cleaved form is not possible due to a large spatial have unfavorable interactions with the hydrophobic side separation (70Å) of the cleaved bond. chains of Val55, Leu30, and Leu99. Better hydrophobic interactions of the side chain of Ala381 with the side Mutations away from the major structural change chains of Val55, Leu30 and Leu99 seem to contribute to

Four out of seven mutations in hepta ␣1-antitrypsin reside the stabilization of the protein (Fig. 8). The Lys387 in the in regions where the structural changes between the wild-type ␣1-antitrypsin makes a salt bridge with Asp264. uncleaved and the cleaved forms are not significant. Although the Lys387→Arg mutation preserves the salt These are the Ser381→Ala, Thr68→Ala, Lys387→Arg bridge, the side chain of arginine has more polar atoms and Ala70→Gly mutations. Residues 381 and 387 are at than that of lysine, and these extra polar atoms can be Research Article Uncleaved ␣1-antitrypsin Ryu et al. 1189

used for hydrogen bonding with neighboring residues. there is no preinsertion of the loop. The orientation of the

The atoms of the guanidinium group of Arg387 in hepta loop relative to the main body of hepta ␣1-antitrypsin is ␣1-antitrypsin are within hydrogen-bonding distance of similar to that of ovalbumin even though the loop has an the main-chain atoms of residues 46–48, and the side helical conformation in ovalbumin. The P17–P14 residues chain atoms of Thr48 and Tyr38. These additional hydro- (342–345) are extended without the loop preinsertion. gen bonds would stabilize hepta ␣1-antitrypsin. The P14–P10 residues (345–349) form a unique bulge that accommodates extra residues made by uncoiling of the Residues 68 and 70 reside at the end of the helix B and loop. The lest of the loop (P1–P9, residues 350–358) runs the following loop, respectively. As these residues are in over the molecule in the direction of the helical loop of the loop region, which has a certain flexibility, it is not ovalbumin, but with closer interactions with the main easy to compare their conformation in the uncleaved and body of the molecule than in ovalbumin. The majority of cleaved ␣1-antitrypsin structures. A clear explanation for the P14–P10 residues have short side chains, and the the mechanism of the stabilization effects by these muta- sequence is conserved throughout the serpin family [33]. tions should await higher resolution structures of various The mutations in the residues of this so-called hinge forms of ␣1-antitrypsin, including the wild-type uncleaved region considerably affect the inhibitory activity of serpins protein and cleaved proteins with the mutations. [34,35]. The plasticity of the P14–P10 may allow the uncoiling of the loop without the loop preinsertion. Reactive-loop conformation

The uncleaved reactive loop of serpins appears to be In the crystals of triple ␣1-antitrypsin [25], the reactive mobile. Crystal structures of the uncleaved antithrombin loop forms a distorted helical conformation. In this case, and antichymotrypsin show that the reactive loop can take there is no bulge in the P14–P10 region. In the crystals of either an extended (antithrombin) or a distorted helical both ␣1-antitrypsin mutants, the reactive loop makes (antichymotrypsin) conformation. In antithrombin, with crystal contacts in a head-to-head fashion mediated by a an extended loop conformation, the loop is partially crystallographic symmetry related twofold axis. However, inserted in sheet A [11,12]. This preinsertion, occurring in a subtle difference in crystal contacts in two different crys- the P14–P17 region may exist as an inhibitory conforma- tals may have induced two different reactive loop confor- tion that binds to target proteases. Antichymotrypsin with mations. When the reactive loop helix is uncoiled, serpin a distorted helical loop conformation does not show the protease inhibitors are thought to form an inhibitory loop preinsertion. The reactive loop in hepta ␣1-antit- conformation with preinsertion of the P14–P17 region, as rypsin is in an extended and uncoiled conformation, but observed in the dimeric antithrombin structure. However,

Figure 8

Interaction of residue 381 with neighboring residues in hepta ␣1-antitrypsin and cleaved ␣1-antitrypsin: (a) hepta ␣1-antitrypsin; (b) the cleaved ␣1-antitrypsin (PDB code 7api). (The figure was drawn using MOLSCRIPT [43].) 1190 Structure 1996, Vol 4 No 10

hepta ␣1-antitrypsin does not show the preinsertion even region of ␣1-antitrypsin and we propose that the strain though the reactive loop is uncoiled. The surplus residues acts as a spring in the A sheet opening process that generated by the uncoiling form a bulge in the P10–P14 involves the separation of ␤ strands in the sheet and the region. The nature of these particular mutants of ␣1-antit- incorporation of the reactive-site loop. This is reminis- rypsin may lead to the different conformations. Hepta cent of the spring-loaded mechanism for the conforma- → ␣1-antitrypsin has the Met374 Ile mutation that is not tional transition of the influenza hemagglutinin during → present in triple ␣1-antitrypsin. Met374 contacts with the infection of the virus [37]. The Met374 Ile and → Thr345 (P14) in the cleaved ␣1-antitrypsin when the loop Thr59 Ala mutations stabilize the uncleaved native is inserted into sheet A. The Met374→Ile mutation abro- protein by introducing better hydrophobic interactions. gates the favorable interaction between Met374 and The destabilization of the loop-inserted structure by Thr345 in the presumptive preinserted structure, and sta- these mutations, if such open forms exist in uncleaved bilizes sheet A of the native structure by introducing ␣1-antitrypsin, may also contribute to the stability of the favorable hydrophobic contacts between the side chains of native closed state. Other mutations (Ser381→Ala, Ile374 and neighboring hydrophobic residues. The stabi- Thr68→Ala, Lys387→Arg and Ala70→Gly) are in the lization of sheet A may prohibit the preinsertion of the regions distant from the major structural changes in the P14–P17, even when the reactive loop helix is uncoiled structural transition. due to the crystal contact. The possibility of strain due to the unfavorably close Biological implications van der Waals contacts in the Phe51 region is unique in

The protein ␣1-antitrypsin inhibits the proteolytic activity that the strain occurs in the hydrophobic core of a of neutrophil elastase in the plasma. Failure of native folded protein. It is not clear how this strain

␣1-antitrypsin function by various hereditary mutations could be created in a native folded protein. In the leads to emphysema. ␣1-Antitrypsin is a prototype hydrophobic collapse model for protein folding, fast member of the inhibitory serpin family and undergoes a hydrophobic collapse is followed by slower annealing characteristic conformational transition upon cleavage of through secondary structure formation [38]. It is possi- the reactive-site loop. The native form of ␣1-antitrypsin is ble that the central hydrophobic core residues in the relatively unstable and the stability is substantially inhibitory serpins were squeezed to optimize the inter- increased by the conformational transition. An actions in the overall structure in the annealing stage of understanding of the molecular mechanism for the the folding process. The important role of this strain in conformational transition and the nature of the stability the native conformational switch of the inhibitory difference between the two states would provide serpins may have preserved the strain during evolution. important information on the engineering of serpin proteins with improved functions as well as on the Materials and methods protein folding and stability. Crystallization Hepta ␣1-antitrypsin, overexpressed and purified as previously We report the crystal structure of an uncleaved ␣ -antit- described [39], was crystallized using the vapor diffusion technique. 1 Two µl of the protein solution in 50 mM sodium phosphate pH 6.0–7.5, rypsin with stabilizing mutations at seven different 100 mM NaCl, 1 mM EDTA, 1 mM ␤-mercaptoethanol at a concentra- –1 places (hepta ␣1-antitrypsin). These mutations, which tion of 20 mg ml , was mixed with an equal volume of a mother liquor are generated by random mutageneses, stabilize containing 17–23 % PEG 4 K, 10 % isopropyl alcohol at pH 6.5. This uncleaved ␣ -antitrypsin without affecting its inhibitory mixture was equilibrated with the mother liquor at 20° C. Crystals 1 appear in 2–4 days of incubation with the typical dimension of activity. The reactive loop of hepta ␣1-antitrypsin forms 0.5 mm × 0.2 mm × 0.05 mm. SDS–PAGE and an activity assay of the an extended strand that is different from that of another dissolved crystals confirmed that the uncleaved hepta ␣1-antitrypsin was intact. The fresh crystals diffract X-rays to 2.5 Å. uncleaved ␣1-antitrypsin with a different subset of mutations. This conformational difference suggests that the loop is inherently flexible and its conformation is gov- Data collection and processing The crystals of hepta ␣1-antitrypsin belong to space group C2 with unit erned mainly by the crystal contacts [36] and the nature cell parameters a = 115.78 Å, b = 39.37 Å, c = 90.43 Å, ␤= 103.80 Å. of different mutations. A structural comparison of hepta Judged from the unit cell and the partial specific volume of the protein, ␣1-antitrypsin with other serpin proteins elucidates the the asymmetric unit contains one molecule of hepta ␣1-antitrypsin and mechanism of the stabilization and its relationship with the solvent content is 50 %. Due to limitations of the size of crystals the control of -antitrypsin activity. and radiation sensitivity, data were collected from several crystals and ␣1 merged before use. All data were collected with a R-Axis IIc image plate detector using CuK␣ X-ray radiations from a Rigaku RU-300 X-ray Among these mutations, the Phe51→Leu mutation generator operating at 40 kV × 100 mA. The R-Axis IIc data processing seems to stabilize the protein by releasing a novel con- software was used for data processing. The initial low resolution data formational strain. This strain may result from unfa- set, which was used for the molecular replacement searches, was collected from three different crystals. The data set had 93 % of full data vorably close van der Waals contacts in the region. Σ| | Σ with Rmerge ( –I / I) of 10.5 % to 4.0 Å. Later, a higher resolu- Residue 51 is located in the central hydrophobic core tion data set was obtained and used for refinement of the model. This Research Article Uncleaved ␣1-antitrypsin Ryu et al. 1191

data set was 83 % complete to 2.7 Å and the Rmerge was 9.8 % at the Accession numbers resolution. A summary of crystallographic data is shown in Table 3. Coordinates of hepta ␣1-antitrypsin have been deposited with the Brookhaven Protein Data Bank. Structure determination and refinement The structure was determined by molecular replacement using X-PLOR Acknowledgements [40]. As a target structure for the molecular replacement the crystal We thank Wayne Hendrickson for comments and Se-Won Suh for the coordinates of triple ␣ -antitrypsin. This work was supported by grants from the structure of cleaved ␣1-antitrypsin (PDB code 7api) was used after 1 removal of the reactive loop inserted in sheet A (residues 343–358). Ministry of Science and Technology, Korea. The search model was placed in a P1 box with dimensions of 150 × 150 × 150 Å. The rotation search followed by Patterson correla- References tion (PC) refinement yielded an unambiguous solution at θ1 = 354.01, 1. Stein, P.E. & Carrell, R.W. (1995). What do dysfunctional Serpins tell θ2 = 73.22, θ3 = 159.75 with a correlation value of 0.204. All other us about molecules mobility and disease? Nat. Struct. Biol. 2, peaks had correlation values below 0.100. This solution was used for 96–113. 2. Schulze, A.J., Huber, R., Bode, W. & Engh, R.A. (1994). Structural the subsequent translation search which gave a solution of 6.7 ␴ above aspects of Serpin inhibitions. FEBS Letts. 344, 117–124. the mean. The solution was a = 0.164, b = 0.000, c = 0.382 in a frac- 3. Potempa, J., Korzus, E. & Travis, J. (1994). The Serpin superfamily of tional coordinates system. The next highest peak in the translation protease inhibitors: structure, functions, and regulation. J. Biol. Chem. search was 4.6 ␴ above the mean. The rigid-body refinement using this 269, 15957–15960. molecular replacement solution with 8.0–5.0 Å data gave a crystallo- 4. Carrell, R.W., Rvans, D.L.I. & Stein, P.E. (1991). Mobile reactive graphic R value of 49.0 %. The model was divided into nine rigid center of serpins and the control of thrombosis. Nature 353, groups treating each ␤ strand of sheet A as a different rigid group. The 576–578. subsequent rigid-body refinement with these nine rigid groups gave a 5. Stein, P.E., Leslie, A.G.W., Finch, J.T., Turnell, W.G., McLaughlin, P.J. & Carrell, R.W. (1990). Crystal structure of ovalbumin as a model for crystallographic R value of 39.3 %. the reactive center of Serpins. Nature 347, 99–102. 6. Morii, M. & Travis, J. (1983). Amino acid sequence at the reactive site The 2F –F map after the rigid-body refinement clearly showed a o c of human ␣1-antitrypsin. J. Biol. Chem. 258, 12749–12752. rearrangement of ␤ strands in sheet A. The model was rebuilt by using 7. Loebermann, H., Tokuoka, R., Deisenhofer, J. & Huber, R. (1984). the program TOM FRODO [41] to fit the map and conventional refine- Human ␣1-proteinase inhibitor. J. Mol. Biol. 177, 531–556. ment was performed. Mutated residues were replaced at this stage. 8. Mottonen, J., et al., & Goldsmith, E.J. (1992). Structural basis of Several cycles of rebuilding and conventional refinement yielded a latency in plasminogen activator inhibitor-1. Nature 355, 270–273. crystallographic R value of 28.1 % with 8.0–3.0 Å data. The 2F –F 9. Stein, P.E. & Chothia, C. (1991). Serpin tertiary structure o c 221 map at this stage started to show indicative densities for the reactive transformation. J. Mol. Biol. , 615–621. 10. Wei, A., Rubin, H., Cooperman, B.S. & Christianson, D.W. (1994). loop. The reactive loop residues 343–358 were built into these indica- Crystal structure of an uncleaved Serpin reveals the conformation of tive densities gradually and the model was refined by conventional an inhibitory reactive loop. Nat. Struct. Biol. 1, 251–258. refinement. The R value was improved to 26.3 %. The resolution of 11. Carrell, R.W., Stein, P.E., Fermi, G. & Wardell, M.R. (1994). Biological data was extended to 2.7 Å and subsequent conventional refinement implications of a 3 Å structure of dimeric antithrombin. Structure 2, gave a crystallographic R value of 26.5 %. Subsequent refinement 257–270. cycles including several rounds of simulated-annealing refinement and 12. Schreuder, H.A., et al., & Hol, W.G.J. (1994). The intact and cleaved human antithrombin III complex as a model for serpin–proteinase rebuilding by using both 2Fo–Fc and Fo–Fc maps resulted in a model with a crystallographic R value of 22.3 %. The individual B-factor refine- interactions. Nat. Struct. Biol. 1, 48–54. 13. Schulze, A.J., Baumann, U., Knof, S., Jaeger, E., Huber, R. & Laurell, ment further decreased the R value to 19.5 %. The current model C.-B. (1990). Evidence for the extent of insertion of the active site including 3597 protein atoms (residues 21–394) has a crystallo- loop of intact ␣ proteinase inhibitor in beta sheet A. Eur. J. Biochem. Σ|| | || | | 1 graphic R value ( Fo –|Fc /( Fo ) of 19.5 % for the data (F > 4␴) of 194, 51–56. 8.0–2.7 Å. Rms deviations from bond and angle idealities are 0.013 Å 14. Pemberton, P., Stein, P.E., Pepys, M.B., Potter, J.M. & Carrel, R.W. and 1.935°, respectively. The Ramachandran plot drawn by program (1988). Hormone binding globulins undergo serpin conformational PROCHECK [42] shows that only 2 out of 374 residues are in disal- change in inflammation. Nature 336, 257–258. lowed regions and 14 are in generously allowed regions. 15. Stein, P.E., Tewkesbury, C. & Carrell, R.W. (1989). Ovalbumin and angiotensinogen lack serpin S-R conformational change. Biochem. J. 262, 103–107. 16. Gettins, P. (1989). Absence of large scale conformational change Table 3 upon limited proteolysis of ovalbumin, the prototypic serpin. J. Biol. Chem. 264, 3781–3785. Summary of the crystallographic data. 17. Brantley, M., Nukiwa, T. & Crystal, R.G. (1988). Molecular basis of alpha-1-antitrypsin deficiency. Am. J. Med. 84, 13–31. Space group C2 18. Carrell, R.W. (1986). ␣1-antitrypsin: molecular pathology, leukocytes, Unit cell and tissue damage. J. Clin. Invest. 78, 1427–1431. a,b,c (Å) 115.78, 39.37, 90.43 19. Kalsheker, N.A. (1994). Molecular pathology of ␣1-antitrypsin 87 ␤ (°) 103.80 deficiency and its significance to clinical medicine. Quart. J. Med. , 653–658. Resolution range (Å) 8.0–2.7 20. Psacharopoulous, H.T., et al., & Rodeck, C.H. (1983). Outcome of Number of unique reflections 9154 liver disease associated with ␣1-antitrypsin deficiency (PiZ). Arch. Dis. Number of atoms 3597 Child 58, 882–887. Overall completeness (8.0 Å–2.7 Å)(%) 83.5 21. Kwon, K.-S., Kim, J., Shin, H.S. & Yu, M.-H. (1994). Single amino acid Completeness (3.0 Å–2.7 Å) (%) 70.6 substitutions of ␣1-antitrypsin that confer enhancement in thermal stability. J. Biol. Chem. 269, 9627–9631. Rmerge (%)* 9.8 † 22. Lee, K.N., Park, S.D. & Yu, M.-H. (1996). The native strain of proteins Rvalue (%), F > 4␴(F) 19.5 Rms deviations probed by various stable mutations of ␣1-antitrypsin located at the hydrophobic core. Nat. Struct. Biol. 3, 497–500. Bonds (Å) 0.013 23. Kim, J., Lee, K.N., Yi, G.-S. & Yu, M.-H. (1995). A thermostable mutation Angles (°) 1.94 located at the hydrophobic core of ␣1-antitrypsin suppresses the Dihedrals (°) 26.12 folding defect of the Z-type variant. J. Biol. Chem. 270, 8697–8601. Impropers (°) 1.62 24. Lomas, D.A., Evans, D.L., Stone, S.R., Chang, W.-S.W., & Carrell, R.W. (1993). Effect of Z mutation of the physical and inhibitory Σ| | Σ † Σ|| | | || Σ| | *Rmerge = –I / I; Rvalue = Fo – Fc / Fo properties of ␣1-antitrypsin. Biochemistry 32, 500–508. 1192 Structure 1996, Vol 4 No 10

25. Song, H.K., Lee, K.N., Kwon, K.-S., Yu, M.-H. & Suh, S.W. (1995). Crystal structure of an uncleaved ␣1-antitrypsin reveals the conformation of its inhibitory reactive loop. FEBS Letts. 377, 150–154. 26. Mast, A.E., Enghild, J.J. & Salvesen, G. (1992). Conformation of the reactive site loop of ␣1-proteinase inhibitor probed by limited proteolysis. Biochemistry 31, 2720–2728. 27. Schulze, A.J., Frohnert, P.W., Engh, R.A. & Huber, R. (1992). Evidence for the extent of insertion of the active site loop of intact a proteinase inhibitor in ␤ sheet A. Biochemistry 31, 7560–7565. 28. Carrell, R.W. & Stein, P.E. (1996). The biostructural pathology of the serpins: critical function of sheet opening mechanism. Biol. Chem. Hoppe-Seyler 377, 1–17. 29. Berkenpas, M.B., Lawrence, D.A. & Ginsburg, D. (1995). Molecular evolution of plasminogen activator inhibitor-1 functional stability. EMBO J. 14, 2969–2977. 30. Mourey, L., Samama, J.-P., Delarue, M., Petitou, M., Choay, J. & Moras, D. (1993). Crystal structure of cleaved bovine antithrombin III at 3.2 Å resolution. J. Mol. Biol. 232, 223–241. 31. Wright, H.T., Quan, H.X. & Huber, R. (1990). Crystal structure of plakalbumin, a proteolytically nicked form of ovalbumin. Its relationship to the structure of cleaved ␣1-proteinase inhibitor. J. Mol. Biol. 213, 513–528. 32. Wolfenden, R.V, Cullis, P.M. & Southgate, C.C.F. (1979). Water, protein folding, and the genetic code. Science 206, 576–577. 33. Huber, R. & Carrell, R.W. (1989). Implications of the three- dimensional structure of ␣1-antitrypsin for structure and function of serpins. Biochemistry 28, 8951–8966. 34. Hopkins, P.C.R., Carrell, R.W. & Stone, S.R. (1993). Effects of mutations in the hinge region of Serpins. Biochemistry 32, 7650–7657. 35. Skriver, K., et al., & Bock, S.C. (1991). Substrate properties of C1 inhibitor Ma (alanine 434-glutamic acid). Genetic and structural evidence suggesting that the P12-region contains critical determinants of serine protease inhibitor/substrate status. J. Biol. Chem. 266, 9216–9221. 36. Engh, R.A., Huber, R., Bode, W. & Schulze, A.J. (1995). Divining the serpin inhibition mechanism: a suicide substrate ‘springe’? Trend Biotech. 13, 503–510. 37. Carr, C.M. & Kim, P.S. (1993). A spring-loaded mechanism for the conformational change of influenza hemagglutinin. Cell 73, 823–832. 38. Dill, A.K., et al., & Chan, H.S. (1995). Principles of protein folding – a perspective from simple exact models. Prot. Sci. 4, 561–602. 39. Kwon, K.-S., Lee, S. & Yu, M.-H. (1995). Refolding of ␣1-antitrypsin expressed as inclusion bodies in Escherichia coli: characterization of aggregation. Biochem. Biophys. Acta. 1247, 179–184. 40. Brünger, A.T., Kuriyan, J. & Karplus, M. (1987). Crystallographic R- factor refinement by molecular dynamics. Science 235, 458–460. 41. Jones, T.A. (1978). A graphic model building and refinement system for macromolecules. J. Appl. Crystallogr. 11, 268–272. 42. Morris, A.L., MacArther, M.W. & Thornton, J.M. (1992). Stereochemical quality of protein structure coordinates. Proteins 12, 345–364. 43. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Cryst. 24, 946–950.