- Email: [email protected]
Crystallization and subunit structure of canine myeloperoxidase
./. Mol. Hid. (1987) 196, 919-925
Crystallization
and Subunit Structure of Canine Myeloperoxidase Roger E. Fenna
Department of Biochemistry University of Miami School of Medicine P.O. Box 016129, Miami, FL 33101, U.S.A. (Received 16 March 1987) Three crystal forms of canine myeloperoxidase are described, An orthorhombic form in space group P2,2,2, has unit cell dimensions: a = 108.3 A (1 J%= 0.1 nm) b = 205.9 A and c = 139.9 8. A trigonal form in space group P3,21 or P3,21 has unit cell dimensions: u = 6 = 138.9 A and c = 145.2 A. A monoclinic form in space group C2 has unit) cell dimensions: a = 117.2 A, b = 96.9 A, c = 131.4 f% and B = 116.3”. Unusual features in the diffraction patterns of the monoclinic form place restrictions on the molecular packing in the crystal. The proposed model for the molecular packing requires that the myeloperoxidase molecule consist of two identical or near-identical halves. In the intact molecule these halves may be related either by a crystallographic dyad axis or by an approximate dyad axis in which one subunit is translated relative to the other by 3.2 w along the symmetry axis. The trigonal crystal form appears most suitable for highresolution X-ray structural analysis.
1. Introduction Myeloperoxidase (donor: H,O, oxidoreductase, EC 1.11.1.7) is localized in the azurophil granules of where it plays an neutrophils, mammalian important role in killing phagocytized bacteria (Klebanoff & Clark, 1978). Under acidic conditions myeloperoxidase catalyzes the oxidation of halide ions by hydrogen peroxide, and in viva, where chloride is the predominant halide, this reaction results in the synthesis of HOCl, an effective antibacterial chlorinating agent (Harrison & Schultz, myeloperoxidase has been 1976). Recent’ly, of the implicated as a possible modulator inflammatory response (Stendahl et al., 1984; Thomas et al., 1983). The mechanism appears to involve the impairment of receptor-mediated cells mechanisms of phagocytic recognition (Stendahl et al., 1‘984). At the molecular level this may be brought about by oxidation of methionine or cyst,eine residues in either the ligands or their receptors. Chloramines, produced by the action of HOC1 on primary amines can cause the oxidation of sulfydryl groups to disulfides. Both canine and human myeloperoxidases have been purified to homogeneity following extraction from white blood cells with the detergent cetyltrimethylammonium bromide (Harrison et al., 1977). Canine myeloperoxidase has a molecular weight of about 140,000, consisting of two large polypeptides (57.500 il&). two small polypeptides
(10,500 M,), two covalently bound heme groups and 3 to 4 y0 by weight carbohydrate (Harrison et al., 1977). There is evidence that the heme groups, and all of the carbohydrate, are associated exclusively with the large polypeptides (Harrison et al., 1977); however, the chemical identities of the heme groups have not been established. In the native enzyme the visible and soret absorption bands are considerably red-shifted compared to those of protoheme and heme c containing peroxidases (Adar. 1978; Ronnberg et al., 1981). A pyridine hemochrome with spectral characteristics resembling those of heme a was isolated from canine meyloperoxidase by Harrison & Schultz (1978). However, recent resonance Raman scattering studies by three different groups have all shown that the chromophore is an iron chlorin (Ikeda-Saito et al., 1984: Sibbett & Hurst, 1984; Babcock et al., 1985). Similar studies also suggest that the chloride ion substrate may bind as a sixth ligand to the iron atom that retains its high-spin configuration (Ikeda-Saito, 1985). Inequivalence of heme binding has been suggested by a variety of studies: Agner (1958) and Harrison & Schultz (1978) reported extraction procedures that labilize only 500/, of the heme. Oxidation of thiocyanate ion by myeloperoxidase follows biphasic kinetics in which each phase contributes approximately equally to the overall reaction (Harrison & Schultz, 1978); and cyanide binding to the heme has been reported to occur with either 1 : 1 (Odajima. 1980) or 2 : 1
920
R. E. Fenna
(Eglington et al., 1982) stoichiometry. In contrast, Andrews & Krinsky (1981) reported that human myeloperoxidase could be cleaved under nondenaturing conditions to yield an enzymatically active hemimyeloperoxidase, consisting of one large and one small polypeptide. The cleavage involves reduction of a single disulfide bridge between the two large polypeptides and does not result in any loss of heme from the enzyme nor produce any detectable changes in the spectral properties of the enzyme in either the reduced or oxidized form. Canine myeloperoxidase has previously been crystallized as either needles or flat plates by dialysis against O-2 M-K,HPO, containing 20 y. (v/v) ethanol at 4°C (Harrison et al., 1977). When transferred to room temperature, t,hese crystals dissolve and have not been used for X-ray diffraction studies. In this paper we report preliminary X-ray diffraction studies on three crystal forms of canine myeloperoxidase obtained by precipitation with polyethylene glycol, diol. ammonium sulfate and methylpentane Unusual features in the diffraction patterns of one crystal form are explained in terms of constraints on the molecular packing in the crystal. The proposed molecular packing leads to the conclusion that two halves of the myeloperoxidase molecule are related by either an exact or an approximate 2-fold axis.
over a period of several weeks to a maximum size of 05 mm in diameter and 2 mm in length. The crystals are stable when transferred to a protein-free substitute mother liquor containing 2.5 M-ammonium sulfate in 50 rnM-Hepes buffer at pH 7.7. Monoclinic crystals were grown by slow evaporation of protein solutions containing 30 to 50 mg ml-’ protein and 30% (v/v) 2-methyl-2,4-pentanediol (MPD) in the presence of either 5 mw-Tris. HCI (pH 8.3) or 5 mMpiperazine-N.N’-bis(2-ethanesulfonic acid) (Pipes) (pH 6.5). In a typical experiment 60 ~1 of this solution was placed in a small glass vial and sealed with Parafilm. A pin-hole was made in the Parafilm and the solution was allowed to evaporate slowly at room temperature until the first appearance of small plate-like crystals. The tube was then re-sealed and the crystals gradually increased in size to about 1.5 mm x I.0 mm x0.4 mm over a period of several weeks. The crystals are insoluble when transferred t)o 4Oo/o (v/v) MPD in 5 mM buffer. However, the best diffraction patterns were obtained from crystals trannferred to 70% (v/v) MPD in 5 mm-Pipes (pH 6.5). (c) X-ray methods Screened precession photographs of the principal projections of all 3 crystal forms were recorded using an Enraf-xonius precession camera and nickel-filtered CuKa radiation from an Elliott GX20 rotating anode X-ray generator operating at. 1.6 kW. Kodak DEFB direct exposure X-ray film, Kodak GBX developer and Kodak rapid fixer were used throughout. (d) Molecular weight determinations
2. Materials and Methods (a) Source of canine myeloperoxidase
The canine myeloperoxidase used in this study was a generous gift from Dr J. Schultz of the Papanicolaou Cancer Research Institute in Miami, Florida. Preparation of the enzyme from canine pus has been described Harrison et al. (1977).
by
The densities of the crystals were measured by the method of Low & Richards (1952) using a bromobenzene/ xylene density gradient. The substitute mother-liquor densities were measured directly with a hydrometer. The molecular weight (M,) of canine myeloperoxidase was calculated from t’he relationship given by Matthews (1974a): M =
(b) C’ryYtaZlizations Orthorhombic crystals of myeloperoxidase were grown by precipitation with polyethylene glycol (PEG?) of average polymer M, 6000 at room temperature (approx. 24°C). 10 ~1 of a 20% (w/v) solution of PEG was added to 40~1 of a 75 mg ml-’ solution of freeze-dried canine myeloperoxidase in 10 mM-Tris 9HCl buffer (pH 8.0). The glass vial was sealed with Parafilm and left undisturbed at room temperature. An amorphous precipitate was observed within 24 h and clusters of rod-shaped crystals grew out of the precipitate over a period of several weeks. The best X-ray diffraction patterns were recorded from crystals mounted directly from their own mother liquor. When crystals were transferred to protein-free substitute mother-liquor solutions containing PEG at concentrations up to 10% (w/v) they became disordered. Trigonal crystals were grown by the addition of ammonium sulfate to a solution of the protein in 50 mMHepes buffer (pH 7.7) at room temperature. The final was 15 mg ml-’ and the concentration protein ammonium sulfate concentration was 1.7 M. The crystals appear as elongated hexagonal prisms and grow slowly t Abbreviations used: PEG, polyethylene MPD, 2-methyl-2,4-pentanediol.
glycol;
N V(d, -d,) n[l--2.‘d,-zu(d,-d,)/d,]
where d,, d, and d, refer to the densities of the crystal. the substitute mother-liyuor and water. V is the unit cell volume n, the number of molecules in the unit cell, Iv, Bvogadro’s number, G, the partial specific volume of the protein and UJ, the weight fraction of bound water to protein.
3. Results (a) Orthorhombic
crystal diffraction
data
The mm symmetry of all three principal projections together with the requirements h = 2n, k = 2n and 1= 2n for reflections along the three principal axes indicate the space group P2,2,2,. The unit cell dimensions are: a = 108.3 ( f 0.3) A (1 A = 0.1 nm),
h = 205.9
(kO.5)
A and
c = 139.9
(10.3) A. Using the value of 142,000 M, the myeloperoxidase of canine weight molecular (Harrison et al., 1977) the volume occupied per dalton would be 549 A3 if there were a single molecule in the asymmetric unit. However, it is more likely that there are two molecules per asymmetric unit, since the corresponding value of
Crystal Forms of Canine Myeloperoxidase
921
L’,,,= 2.75 A3 dalton-’ is closer to the average value of 2.96 given by Matthews (19746) for crystalline proteins of molecular weight greater than 10’. Jn the hO1zone of the diffraction pattern, rows of reflections with 1 index odd are notably absent at low resolution indicating a sub-cell in the 010 projection with n’ = a and c’ = c/S. However, since t,he asymmetric unit consists of two whole molecules of myrloperoxidase, we have been unable to deduce any definitive information concerning possible intramolecular symmetry. (11) Trigonal cry&al diflraction
data
The symmetry of the diffraction patterns together with the systematic absence of reflections along the 001 row indicate either space group P3,21 or P3,21. The unit cell dimensions are: a = h = 138.9 (+0*3) 8: c = 145.2 (kO.3) 8. The unit cell volume (2.43 x IO6 A3) is consistent with the presence of only one molecule of molecular weight 142,000 per asymmetric unit. The volume occupied per dalton ( V,,,) is 2.85 A3. The crystals diffract strongly and reflections corresponding to Bragg spacings of less than 2 A have been observed on still phot’ographs. Jn addition, the crystals appear quite stable in the X-ray ray beam. Following 100 hours of exposure to C,‘uKa X-rays from a rotating anode X-ray generator operating at 1.8 kW, the diffraction patt,ern retained measurable reflections to Bragg spacings of about 3.5 A. This crystal form would appear to be suitable for medium t’o high-resolution X-ray structure analysis of canine myrloprroxidase. ((a) Monorlinic
csystal diffraction
data
Figure 1 showrrj a 1.?C screened-precession photograph of the hO1 zone from a monoclinic c*rysbal of caanine myeloperoxidase grown by precipitation with MJ’l). The systemat’ic absence of rows of rrfle(+ions with h index odtl, together with the general condition that reflections occur onl) when h+k = 2~). indicate that the crystals belong to space’ group (‘2. The unit cell dimensions measured from precession photographs of the principal zones are: a = 117.2kO.3 A, h = 96.9 and /? = 116.3 (kO.3) A. c = 131.4 (kO.3) x (+we) ). In spac’e group C’2 there are four s~mnlrtr?,-rt,lat~d positions in the unit cell. Using thr value of 142.000 for the molecular weight of canine mvrloprroxidase, the volume occupied per dalton ( I’,,,) is cbalculated to be 2.36 A3, assuming a single molecule per asymmetric unit. The presence of more than one molecule in the asymmetric unit would give values of V,,, well outside the range observed for crystalline proteins (Matthews, 1968, 19746). The measured crystal density was 1.224 ( f 0.005) g cm -’ and that of the substitute mother liquor 0.970 (kO.002) g crh3. Using the value of 0.73 cm3 g-l determined for the partial specific volume of tht, protein (Ehrenberg & Agner, 1958) and assuming that the weight fraction of bound
Figure 1. 15” screened-precessionphotograph of the hOZ zone from a C2 crystal of caninr mgeloperoxidase showing weakness of rows of reflections with I index odd in the c* (vertical) direction.
water to protein is 0.25, the molecular weight of canine myeloperoxidase is calculated t.o be 1.71 ( + 0.2) x 105. This value is about 20°& greater than that determined by Harrison et al. (1977) by equilibrium sedimentation in the ultracentrifuge Several unusual features have been noted in t,he diffraction patterns of the C2 crystal form. Jn the h0Z zone (Fig. 1). in addition to the systematic absence of rows with h index odd, the rows of reflections with 1 index odd are extremely weak and essentially absent at low resolution. This implies that in the 010 project,ion, at low t.tAsolution. the crystal lattice consists of a repeating sub-car11with parameters a’ = n/2. c’ = c/2 and fl = I 16.5’ Figure 1 also illustrates some disorder in thrb diffraction pattern at Bragg spacings <+5 AAin thcb c* direction. The diffraction spots become indist;nct and there is some diffuse scattering risible AS fluctuations in the background intensit,v between reflections. This phenomenon is a general feature of’ the diffraction pat,terns from these cryst,als and presumably signifies some disorder in t,he spacing between molecules in the c direction of thr crystal lat,tice. The hk0 zone (Fig. 2) has a very striking striped resulting from a modulation in appearance diffracted intensity with maxima when k = 15 x 2n and minima when k = 15 x (2n-1). X related modulation in intensities is evident in t.he h,Vt diagonal zone (Fig. 3). where for rows with h index even the modulation exhibits sirnilar int)ensity maxima when k = 15 x 2n and minima when I%= 15 x (en- 1). But, for rows with h indes odd the dependence of the positions of itensity maxirna and minima on the value of the k index is the reverse of the above The intensity modulation in the hk0 zone can be explained as resulting from a convolution of the molecular and lattice transforms with the transform of two identical points approximately h/30 (3.2 a)
922
R. E. Fenna
Figure 2. 17.5” screened-precession photograph of the hkO zone from a C2 crystal of canine myeloperoxidase showing an intensity modulation in the 6* (horizontal) direction.
This would occur if the unit cell contained two identical, or almost identical, objects having the same orientation in the (a, c) plane but separated by distance of b/30 when projected down the c axis of the crystal lattice. Modulation of intensities in the hkh zone, presumably, arises as a result of the same disposition of near identical objects when viewed in projection. apart.
(c) Pseudo-symmetry in the monoclinic crystals The asymmetric unit of the C2 cell has been shown to contain one molecule only of myeloperoxidase; and assuming that the canine and human forms of the enzyme are not dissimilar, this is likely to consist of two identical, or near identical, halves possibly linked together by a single disulfide bridge. The halving of the c axis giving rise to the sub-cell in the 010 projection implies that for each molecular feature with co-ordinates X,Z there is a near-identical feature with co-ordinates approximately equal to x,(1/2+2). As a result, additional pseudo 2-fold axes are introduced into the projection at x = 0, z = l/4 and x = l/4, z = l/4. The asymmetric unit of the C2 cell now consists of two near-identical objects related in projection down the b axis by a pseudo 2-fold axis at approximately x = l/4, z = l/4 (Fig. 4). The intensity modulations observed in the hk0 and hkh zones of the diffraction pattern can now readily be explained in terms of a small difference between the y co-ordinates of the near-identical features at X,Z and x,(1/2+2). Consequently, the pseudo 2-fold axis at x = l/4, z = l/4 in projection, becomes a pseudo f-fold screw axis in the full unit cell. The relationship between the proposed pseudo-
Figure 3. 12.5” screened-precession photograph of the hkh diagonal zune from a C2 crystal of canine myeloperoxidase showing an intensity modulation in the 6* (horizontal) direction. Positions of the intensity maxima and minima differ for rows with h index odd and even.
symmetry and the crystallographic symmetry of space group C2 is shown in Figure 5. The asymmetric unit of the C2 cell thus consists of two identical or near-identical features at x,y,.z and where Ab is the small difference x,(y+Ab),(V+z), in the y co-ordinates described above. By considering the contributions of these two near-identical features in the asymmetric unit to the structure factors in the hk0 and hkh zones of the diffraction pattern it can be shown that the proposed packing arrangement is consistent with the two types of intensity modulation observed. For space group C2 the real and imaginary parts of the structure factor are given by: cos Znhx cos 27cky cos 2nhx sin dzky. The first term, in each case, determines the absence of reflections for which systematic h + k = 2n. The contribution, A,, of an atom at x,y,z to the real part of the structure factor is therefore: A 1chkO,= 4 cos 2nhx cos 2nky and the contribution, x,(y+Ab),(1/2+z) is:
A,, of an identical
atom at
A 2chkO,= 4 cos 2nhx cos 2nk( y + Ab). When cos 2nky is equal in magnitude but opposite in sign to cos Snk(y+Ab) contributions from the two atoms will cancel out. However, when the two terms are equal in both magnitude and sign, their contributions will add maximally. The same argument holds true for the terms sin 2zky and sin 2nk(y+ Ab) in the imaginary part of the structure factor. When extended to all pairs of
Crystal Forms of Canine Myeloperoxidase
Figure 4. Locat,ion of pseudo %fold axes (open symbols) in the 010 projection of t,he C2 cell of canine myeloperoxidase. Sear-identical features related by the pseudodyad are indicated as open and filled symbols. The resultant subc*rll has C’= c/P.
pscudos?-mmet,ry-related
in intensities
in the b* direction of the hkO zone. The modulation will have maxima for integer values of kAb and minima for integer values of 2kAb and will be of the h index. independent The intensity modulation observed in the hkO zone shown in Figure 2 has a minimum at k = 15 followed by a maximum at k = 30. The displacement between the two near-identical features along the b axis can thus be calculated to be b/30 = 32 A. In the hkh zone the real and imaginary parts of factor
Figure 5. Relationship between the pseudosymmetry elements (open symbols) and crystallographic syrnmrt,ry elements (filled symbols) in the (‘2 cell of canine myeloperoxidase. Fear-identical feat,ures with y coordinates differing by h/30 are indicated by open and filled symbols. Circles indicate possible moler>ular positions: the filled circle for a molecule with a true dyad axis and the broken circle for a molecule with an approximate dyad axis.
atoms in the asymmetric
unit t
the structure
923
are given
by:
(‘OS27cI$r+ z) cos 2nky (‘OH2zh(x+z)
sin 27cky.
Again. for two atoms at z.y,.z and z,(y+b),(1/2+z) the c~ont~ributions .-I, and A, to the real part of the struc&ture factor arc’ given by:
.-I I(hkh)= 4 (YE 2nh( .r + z) cos 2zky 2nh(.r+ z + I /2) cos 2zk(y + Ab). ;I Z,hkh)= 4 CY>S
Following the reasoning given above. when extended to all pairs of pseudosymmetry,-related atoms in the asymmetric unit there will again be an intensity modulation in the b* direction of the hkh zone. However. in t,his case, since ws &h(.r+~) and cos 2nh(x + t + l/2) are only equal for even values of’ h. having opposite signs for odd va111t~ of’ h. tht~ values of k corresponding to posit,ions of maxima and minima in the int,ensity modulation will br reversed for rows with II index odd compared with rows having h index even. The periodicGt.~ of the modulation will, however. be the same as in the hk0 zone. This is exactly the case in the hkh zone shown in Figure 3. For rows with h index WYW. maxima occur at k = 0 and 30, wit,h minima at A,= I5 and 45, while for rows with h index odd the reverse is true.
(c) Symmetrfy of myeloperoxidns~ The proposed disposition of pseudosymmetry element,s in the C2 unit cell of monoclinic crystals of myeloperoxidase (Fig. 5) permits a number of choices to be made for the location of’ a mofec~ule
924
R. E. Fenna
known to consist of two identical or near-identical halves. It is unlikely that the molecule would consist of halves arranged either head to tail or related by either crystallographic or pseudo 2-fold screw axes since both can give rise to infinite polymers. Two further possibilities exist,: (1) the myeloperoxidase molecule could have a 2-fold axis of symmetry, which in the (‘2 crystal form coincides with a crystallographic a-fold axis. Pairs of molecules located on non-equivalent crystallographic 2-fold axes would then be related by a pseudo 2-fold combined with a relative t’ranslation of 3.2 A in the b direction. (2) Alternatively. the pseudo 2-fold axis could be a molecular feature, in which ca,se t,he myeloperoxidase molecule would consist of two near-identical halves related by an approximate 2-fold axis but separated by a displacement of 3.2 a along this axis. Tdentical ha,lves of two different molecules would then bc related by the crystallographic X-fold axes. From the information presently available it is not possible to distinguish between these two cases.
4. Discussion Other examples of int’ensit’y modulations giving rise to characterist’ically st’riped diffraction patterns have been noted in the literature. (Irvstals of hemerythrin in space group P4 give d%fraction patterns illustrating both of the types of intensity modulation described above (North & Stubbs, 1974). In this case the modulation resulted from the displacement’ of two octameric molecules in approximately the same orientation by 10 A in the c direction of the unit cell. These authors also considered the possibility that the molecule could consist of two identical halves related by a rotation of 180” combined with a translation of 10 A along the 2-fold axis. However. since there was chemical evidence to suggest that the subunits of hemerythrin were equivalent this possibility was excluded. More recently a very similar intensity modulation was reported for crystals of myosin subfragment-l in space group P2,2,21 (Winkelmann et al.. 1985). In this case the asymmetric unit contained two molecules related by a local 2-fold axis. The intensity modulation and halving of the unit cell in projection arose because the local S-fold was parallel with a crystallographic screw axis but displaced from it by 4.3 A. Steitz et al. (1973) reported a crystal form of yeast hexokinase B in which two subunits are related by a 180” rotation plus a relative translation of 3.6 L\ along this axis. The presence of the approximate dyad axis was originally detected from molecular packing considerations based on an intensit;\ modulation observed in the hk0 zone of the diffraction pattern (Steitz, 1971). A different crystal form of yeast hexokinase B, in which heterologous interactions are responsible for dimer formation was later described by Anderson et al. (1974). In this case the subunits were related by a 156” rotation plus a translation of 13.8 A about the rotation axis.
One factor that) distinguishes the subunit interactions in myeloperoxidase from the examples discussed above is t,he possible presence of a disulfide bond bet,ween the two halves of the molecule. Presumably t’his bond plays an important role in determining the spatial disposition of the interacting subunits. The only example of a protein consisting of two halves linked by disulfide bridges where the S-ray cryst’al structure is known is the intact immunoglobulin molecule Kol (Colman rt 01.. 1976). Tn this case the molecule is located on a crystallographic d-fold axis and density cxorresponding to the two disulfide bridges is seen to cross this axis. However. whereas the two Fah portions of the molecule are related by the dyad axis the portions of the polypeptide chains on the (I-terminal side of the disulfides (corresponding to the Fc fragments) are not: and appear disordered in the electron density map. Even though the 2-fold axis is crystallographic, this can occur provided that t,he portions of the molecule that) are not related by t,he 2-fold do not participate in intermolecular contacts essential to the symmetry of the crystal lat,tice. The data we have presented indicate that the canine myeloperoxidase molecule consists of two halves t,hat are either identical or almost ident,ical and this is caonsistent wit’h t,he finding that the human enzyme can be cleaved int>o t.wo apparently identical halves by cleavage of a single disultide bridge between the large subunits (Andrews & Krinsky, 1981). The evidence further suggests that’ the two halves of the molecule are related by either an exact or an approximat~e 2-fold symmetry axis. Tn the lat,ter case t’he two halves of the molecule would not be exa&ly equivalent since environmental differences could occur as a result of heterologous interactions at the interfacse. Such a model could account for previously reported evidrnre that the two heme groups in myeloperoxidase are non-equivalent. Resolution of the question as to whether the molecular d-fold axis in myeloperoxidase is exact or only approximate must await determination of the X-ray cryst,al structure of this enzyme. Attempts to prepare suitable heavy-atom isomorphous derivatives of the trigonal crystal form of canine myeloperoxidase are now underway in this laborat.ory. This work was supported by a grant from the National Tnstit,ut.es of General Medical Sciences (no. GM33384). The author was the recipient of a Research Career Development Award from the National Institute of Diabetes and Digestive Kidne! Diseases and (no. DKOl132) during the time that this work was performed.
References Adar. F. (1978). Porphyrins, 3, 167-209. Agner. K. (1958). Acta Chem. Stand. 12, 89-94. Anderson, W. F., Fletterick, R. ,J. & Steitz. T. A. (1974). .I. Mol. Viol. 86, 261-269.
Crystal Forms of Canine M yeloperoxidasr Andrews, 1’. (‘. & Krinsky. Xv. I. (1981). J. Hiol. (‘hem. 256. 421 l--4%18. I. J.. f2olwhw. H. (:. J. &I. & Wrwr, R. (1985). Riochim. Wiophy.
A ctct 1 828, .‘iti-fiB.
(‘olman. 1’. %I.. Drisrnhofrr. .J.. Huber. R. & Palm. 1%‘. (1!)76). .1. MO/. Biol. loo. 257%482. Eglington. 1). (i.. Barber. l).. Thompson. A. J.. (~rrwwood. (‘. & &gal. A. W. (1982). Biochim. Bioph!/s. d&l. 703. 187-l 95. Ehrrnberg. .\. Xr Agnrr. K. (1958). Acln (‘hem. Scnnd. 12. !)5- 100. Harrison. .I. tS. & Schultz. -1. (1976). ,I. Hiol. Chwr. 251. 1371 1374. Harrison. .I. lC. & Schultz. .J. (1978). Hiochim. Hiophys. Act/r. 536. :HLf-t!l. Harrisorl. .I. E.. I’abalan. S. & Schultz. J. (1977). lliorhinr.
Hiophys.
Actu, 493. 247-259.
lkrda-Saito. M. (1!%5). I. Riol. (‘hrm. 260, 11688-l 1696. lkeda-Saito. M.. l’rinw. R.. Agarde. I’. \‘. & Rousseau. I). I.. (I !M). Pcv/. I’roc. Frd. A mcr. Sot. Il:xp. Hiol. 43. l5AI. Klrhanofl’. S. J. & (‘lark. R’. A. (1978). Tn Thp Neutrophil. II nd (‘lin id Disorders. ~t///CtiW/ Xorth-Holland Edited
HP5
Publishing C‘n.. Amsterdam. Low, I<. \V. & Richards. F. 13. (195”). .1. .-1n/~r. f’hrm. sot. 74. 1660&l 666. Matthews. R. W. (1968). ,J. Mol. Viol. 33. 491 1!#i. Matthews. H. W. (1974a). .I. Mol. Riol. 82. 513-326. Matthews. R. &‘. (19746). In The I’rotrirrs (Xeurath. H. &, Hill. R. I,.. ~1s). Zirtl rdit.. pp. 40:~ ,590. A\~adtwic~ Press. Kew York. Xorth. A. C’. T. & Stubbs, (:. .I. (1974). .I. Jlol. Hid. 88. l&-1:31. Odajima. T. (1980). I. Biorhem. (Tokyo). 87. 379-391. Ronnberg. M.. Araiso. T.. Ellfolk, r\;. & Dunford. H. R. (1981). Arch. Riochrm. Hioyhys. 207. 197 -%N. Sibbrtt, S. S. & Hurst. .I. K. (1984). Riochrn~i.str~y. 23. :Nw)7-30 13. Steitz. T. A. (1971). .I. Mol. Hiol. 61. 695-700. Steitz. T. .I.. Flrtteric*k. R. J. & Hwang. K. J. (l!GR). I. Mol. Hiol. 78, 551~-561. Stendahl. 0.. (‘able. 1%.1.. Dahigren. (‘.. Hrd. J. & Molin. L. (1984). I. (‘/in. Inwst, 73. M&i-373. Thomas. E. T,.. (irishan. $1. 13. & .J&frrson hl. ( IWi). .I. (‘lit,. I,/wst. 72. 341-454. 12’inkrlmann. I). .A.. Mrkrrl. H. R Ra>~rnrnt. I. (IOH.‘). ./. 901. Hiol. 181. 487501.
by A. Kluy
Crystallization
and Subunit Structure of Canine Myeloperoxidase Roger E. Fenna
Department of Biochemistry University of Miami School of Medicine P.O. Box 016129, Miami, FL 33101, U.S.A. (Received 16 March 1987) Three crystal forms of canine myeloperoxidase are described, An orthorhombic form in space group P2,2,2, has unit cell dimensions: a = 108.3 A (1 J%= 0.1 nm) b = 205.9 A and c = 139.9 8. A trigonal form in space group P3,21 or P3,21 has unit cell dimensions: u = 6 = 138.9 A and c = 145.2 A. A monoclinic form in space group C2 has unit) cell dimensions: a = 117.2 A, b = 96.9 A, c = 131.4 f% and B = 116.3”. Unusual features in the diffraction patterns of the monoclinic form place restrictions on the molecular packing in the crystal. The proposed model for the molecular packing requires that the myeloperoxidase molecule consist of two identical or near-identical halves. In the intact molecule these halves may be related either by a crystallographic dyad axis or by an approximate dyad axis in which one subunit is translated relative to the other by 3.2 w along the symmetry axis. The trigonal crystal form appears most suitable for highresolution X-ray structural analysis.
1. Introduction Myeloperoxidase (donor: H,O, oxidoreductase, EC 1.11.1.7) is localized in the azurophil granules of where it plays an neutrophils, mammalian important role in killing phagocytized bacteria (Klebanoff & Clark, 1978). Under acidic conditions myeloperoxidase catalyzes the oxidation of halide ions by hydrogen peroxide, and in viva, where chloride is the predominant halide, this reaction results in the synthesis of HOCl, an effective antibacterial chlorinating agent (Harrison & Schultz, myeloperoxidase has been 1976). Recent’ly, of the implicated as a possible modulator inflammatory response (Stendahl et al., 1984; Thomas et al., 1983). The mechanism appears to involve the impairment of receptor-mediated cells mechanisms of phagocytic recognition (Stendahl et al., 1‘984). At the molecular level this may be brought about by oxidation of methionine or cyst,eine residues in either the ligands or their receptors. Chloramines, produced by the action of HOC1 on primary amines can cause the oxidation of sulfydryl groups to disulfides. Both canine and human myeloperoxidases have been purified to homogeneity following extraction from white blood cells with the detergent cetyltrimethylammonium bromide (Harrison et al., 1977). Canine myeloperoxidase has a molecular weight of about 140,000, consisting of two large polypeptides (57.500 il&). two small polypeptides
(10,500 M,), two covalently bound heme groups and 3 to 4 y0 by weight carbohydrate (Harrison et al., 1977). There is evidence that the heme groups, and all of the carbohydrate, are associated exclusively with the large polypeptides (Harrison et al., 1977); however, the chemical identities of the heme groups have not been established. In the native enzyme the visible and soret absorption bands are considerably red-shifted compared to those of protoheme and heme c containing peroxidases (Adar. 1978; Ronnberg et al., 1981). A pyridine hemochrome with spectral characteristics resembling those of heme a was isolated from canine meyloperoxidase by Harrison & Schultz (1978). However, recent resonance Raman scattering studies by three different groups have all shown that the chromophore is an iron chlorin (Ikeda-Saito et al., 1984: Sibbett & Hurst, 1984; Babcock et al., 1985). Similar studies also suggest that the chloride ion substrate may bind as a sixth ligand to the iron atom that retains its high-spin configuration (Ikeda-Saito, 1985). Inequivalence of heme binding has been suggested by a variety of studies: Agner (1958) and Harrison & Schultz (1978) reported extraction procedures that labilize only 500/, of the heme. Oxidation of thiocyanate ion by myeloperoxidase follows biphasic kinetics in which each phase contributes approximately equally to the overall reaction (Harrison & Schultz, 1978); and cyanide binding to the heme has been reported to occur with either 1 : 1 (Odajima. 1980) or 2 : 1
920
R. E. Fenna
(Eglington et al., 1982) stoichiometry. In contrast, Andrews & Krinsky (1981) reported that human myeloperoxidase could be cleaved under nondenaturing conditions to yield an enzymatically active hemimyeloperoxidase, consisting of one large and one small polypeptide. The cleavage involves reduction of a single disulfide bridge between the two large polypeptides and does not result in any loss of heme from the enzyme nor produce any detectable changes in the spectral properties of the enzyme in either the reduced or oxidized form. Canine myeloperoxidase has previously been crystallized as either needles or flat plates by dialysis against O-2 M-K,HPO, containing 20 y. (v/v) ethanol at 4°C (Harrison et al., 1977). When transferred to room temperature, t,hese crystals dissolve and have not been used for X-ray diffraction studies. In this paper we report preliminary X-ray diffraction studies on three crystal forms of canine myeloperoxidase obtained by precipitation with polyethylene glycol, diol. ammonium sulfate and methylpentane Unusual features in the diffraction patterns of one crystal form are explained in terms of constraints on the molecular packing in the crystal. The proposed molecular packing leads to the conclusion that two halves of the myeloperoxidase molecule are related by either an exact or an approximate 2-fold axis.
over a period of several weeks to a maximum size of 05 mm in diameter and 2 mm in length. The crystals are stable when transferred to a protein-free substitute mother liquor containing 2.5 M-ammonium sulfate in 50 rnM-Hepes buffer at pH 7.7. Monoclinic crystals were grown by slow evaporation of protein solutions containing 30 to 50 mg ml-’ protein and 30% (v/v) 2-methyl-2,4-pentanediol (MPD) in the presence of either 5 mw-Tris. HCI (pH 8.3) or 5 mMpiperazine-N.N’-bis(2-ethanesulfonic acid) (Pipes) (pH 6.5). In a typical experiment 60 ~1 of this solution was placed in a small glass vial and sealed with Parafilm. A pin-hole was made in the Parafilm and the solution was allowed to evaporate slowly at room temperature until the first appearance of small plate-like crystals. The tube was then re-sealed and the crystals gradually increased in size to about 1.5 mm x I.0 mm x0.4 mm over a period of several weeks. The crystals are insoluble when transferred t)o 4Oo/o (v/v) MPD in 5 mM buffer. However, the best diffraction patterns were obtained from crystals trannferred to 70% (v/v) MPD in 5 mm-Pipes (pH 6.5). (c) X-ray methods Screened precession photographs of the principal projections of all 3 crystal forms were recorded using an Enraf-xonius precession camera and nickel-filtered CuKa radiation from an Elliott GX20 rotating anode X-ray generator operating at. 1.6 kW. Kodak DEFB direct exposure X-ray film, Kodak GBX developer and Kodak rapid fixer were used throughout. (d) Molecular weight determinations
2. Materials and Methods (a) Source of canine myeloperoxidase
The canine myeloperoxidase used in this study was a generous gift from Dr J. Schultz of the Papanicolaou Cancer Research Institute in Miami, Florida. Preparation of the enzyme from canine pus has been described Harrison et al. (1977).
by
The densities of the crystals were measured by the method of Low & Richards (1952) using a bromobenzene/ xylene density gradient. The substitute mother-liquor densities were measured directly with a hydrometer. The molecular weight (M,) of canine myeloperoxidase was calculated from t’he relationship given by Matthews (1974a): M =
(b) C’ryYtaZlizations Orthorhombic crystals of myeloperoxidase were grown by precipitation with polyethylene glycol (PEG?) of average polymer M, 6000 at room temperature (approx. 24°C). 10 ~1 of a 20% (w/v) solution of PEG was added to 40~1 of a 75 mg ml-’ solution of freeze-dried canine myeloperoxidase in 10 mM-Tris 9HCl buffer (pH 8.0). The glass vial was sealed with Parafilm and left undisturbed at room temperature. An amorphous precipitate was observed within 24 h and clusters of rod-shaped crystals grew out of the precipitate over a period of several weeks. The best X-ray diffraction patterns were recorded from crystals mounted directly from their own mother liquor. When crystals were transferred to protein-free substitute mother-liquor solutions containing PEG at concentrations up to 10% (w/v) they became disordered. Trigonal crystals were grown by the addition of ammonium sulfate to a solution of the protein in 50 mMHepes buffer (pH 7.7) at room temperature. The final was 15 mg ml-’ and the concentration protein ammonium sulfate concentration was 1.7 M. The crystals appear as elongated hexagonal prisms and grow slowly t Abbreviations used: PEG, polyethylene MPD, 2-methyl-2,4-pentanediol.
glycol;
N V(d, -d,) n[l--2.‘d,-zu(d,-d,)/d,]
where d,, d, and d, refer to the densities of the crystal. the substitute mother-liyuor and water. V is the unit cell volume n, the number of molecules in the unit cell, Iv, Bvogadro’s number, G, the partial specific volume of the protein and UJ, the weight fraction of bound water to protein.
3. Results (a) Orthorhombic
crystal diffraction
data
The mm symmetry of all three principal projections together with the requirements h = 2n, k = 2n and 1= 2n for reflections along the three principal axes indicate the space group P2,2,2,. The unit cell dimensions are: a = 108.3 ( f 0.3) A (1 A = 0.1 nm),
h = 205.9
(kO.5)
A and
c = 139.9
(10.3) A. Using the value of 142,000 M, the myeloperoxidase of canine weight molecular (Harrison et al., 1977) the volume occupied per dalton would be 549 A3 if there were a single molecule in the asymmetric unit. However, it is more likely that there are two molecules per asymmetric unit, since the corresponding value of
Crystal Forms of Canine Myeloperoxidase
921
L’,,,= 2.75 A3 dalton-’ is closer to the average value of 2.96 given by Matthews (19746) for crystalline proteins of molecular weight greater than 10’. Jn the hO1zone of the diffraction pattern, rows of reflections with 1 index odd are notably absent at low resolution indicating a sub-cell in the 010 projection with n’ = a and c’ = c/S. However, since t,he asymmetric unit consists of two whole molecules of myrloperoxidase, we have been unable to deduce any definitive information concerning possible intramolecular symmetry. (11) Trigonal cry&al diflraction
data
The symmetry of the diffraction patterns together with the systematic absence of reflections along the 001 row indicate either space group P3,21 or P3,21. The unit cell dimensions are: a = h = 138.9 (+0*3) 8: c = 145.2 (kO.3) 8. The unit cell volume (2.43 x IO6 A3) is consistent with the presence of only one molecule of molecular weight 142,000 per asymmetric unit. The volume occupied per dalton ( V,,,) is 2.85 A3. The crystals diffract strongly and reflections corresponding to Bragg spacings of less than 2 A have been observed on still phot’ographs. Jn addition, the crystals appear quite stable in the X-ray ray beam. Following 100 hours of exposure to C,‘uKa X-rays from a rotating anode X-ray generator operating at 1.8 kW, the diffraction patt,ern retained measurable reflections to Bragg spacings of about 3.5 A. This crystal form would appear to be suitable for medium t’o high-resolution X-ray structure analysis of canine myrloprroxidase. ((a) Monorlinic
csystal diffraction
data
Figure 1 showrrj a 1.?C screened-precession photograph of the hO1 zone from a monoclinic c*rysbal of caanine myeloperoxidase grown by precipitation with MJ’l). The systemat’ic absence of rows of rrfle(+ions with h index odtl, together with the general condition that reflections occur onl) when h+k = 2~). indicate that the crystals belong to space’ group (‘2. The unit cell dimensions measured from precession photographs of the principal zones are: a = 117.2kO.3 A, h = 96.9 and /? = 116.3 (kO.3) A. c = 131.4 (kO.3) x (+we) ). In spac’e group C’2 there are four s~mnlrtr?,-rt,lat~d positions in the unit cell. Using thr value of 142.000 for the molecular weight of canine mvrloprroxidase, the volume occupied per dalton ( I’,,,) is cbalculated to be 2.36 A3, assuming a single molecule per asymmetric unit. The presence of more than one molecule in the asymmetric unit would give values of V,,, well outside the range observed for crystalline proteins (Matthews, 1968, 19746). The measured crystal density was 1.224 ( f 0.005) g cm -’ and that of the substitute mother liquor 0.970 (kO.002) g crh3. Using the value of 0.73 cm3 g-l determined for the partial specific volume of tht, protein (Ehrenberg & Agner, 1958) and assuming that the weight fraction of bound
Figure 1. 15” screened-precessionphotograph of the hOZ zone from a C2 crystal of caninr mgeloperoxidase showing weakness of rows of reflections with I index odd in the c* (vertical) direction.
water to protein is 0.25, the molecular weight of canine myeloperoxidase is calculated t.o be 1.71 ( + 0.2) x 105. This value is about 20°& greater than that determined by Harrison et al. (1977) by equilibrium sedimentation in the ultracentrifuge Several unusual features have been noted in t,he diffraction patterns of the C2 crystal form. Jn the h0Z zone (Fig. 1). in addition to the systematic absence of rows with h index odd, the rows of reflections with 1 index odd are extremely weak and essentially absent at low resolution. This implies that in the 010 project,ion, at low t.tAsolution. the crystal lattice consists of a repeating sub-car11with parameters a’ = n/2. c’ = c/2 and fl = I 16.5’ Figure 1 also illustrates some disorder in thrb diffraction pattern at Bragg spacings <+5 AAin thcb c* direction. The diffraction spots become indist;nct and there is some diffuse scattering risible AS fluctuations in the background intensit,v between reflections. This phenomenon is a general feature of’ the diffraction pat,terns from these cryst,als and presumably signifies some disorder in t,he spacing between molecules in the c direction of thr crystal lat,tice. The hk0 zone (Fig. 2) has a very striking striped resulting from a modulation in appearance diffracted intensity with maxima when k = 15 x 2n and minima when k = 15 x (2n-1). X related modulation in intensities is evident in t.he h,Vt diagonal zone (Fig. 3). where for rows with h index even the modulation exhibits sirnilar int)ensity maxima when k = 15 x 2n and minima when I%= 15 x (en- 1). But, for rows with h indes odd the dependence of the positions of itensity maxirna and minima on the value of the k index is the reverse of the above The intensity modulation in the hk0 zone can be explained as resulting from a convolution of the molecular and lattice transforms with the transform of two identical points approximately h/30 (3.2 a)
922
R. E. Fenna
Figure 2. 17.5” screened-precession photograph of the hkO zone from a C2 crystal of canine myeloperoxidase showing an intensity modulation in the 6* (horizontal) direction.
This would occur if the unit cell contained two identical, or almost identical, objects having the same orientation in the (a, c) plane but separated by distance of b/30 when projected down the c axis of the crystal lattice. Modulation of intensities in the hkh zone, presumably, arises as a result of the same disposition of near identical objects when viewed in projection. apart.
(c) Pseudo-symmetry in the monoclinic crystals The asymmetric unit of the C2 cell has been shown to contain one molecule only of myeloperoxidase; and assuming that the canine and human forms of the enzyme are not dissimilar, this is likely to consist of two identical, or near identical, halves possibly linked together by a single disulfide bridge. The halving of the c axis giving rise to the sub-cell in the 010 projection implies that for each molecular feature with co-ordinates X,Z there is a near-identical feature with co-ordinates approximately equal to x,(1/2+2). As a result, additional pseudo 2-fold axes are introduced into the projection at x = 0, z = l/4 and x = l/4, z = l/4. The asymmetric unit of the C2 cell now consists of two near-identical objects related in projection down the b axis by a pseudo 2-fold axis at approximately x = l/4, z = l/4 (Fig. 4). The intensity modulations observed in the hk0 and hkh zones of the diffraction pattern can now readily be explained in terms of a small difference between the y co-ordinates of the near-identical features at X,Z and x,(1/2+2). Consequently, the pseudo 2-fold axis at x = l/4, z = l/4 in projection, becomes a pseudo f-fold screw axis in the full unit cell. The relationship between the proposed pseudo-
Figure 3. 12.5” screened-precession photograph of the hkh diagonal zune from a C2 crystal of canine myeloperoxidase showing an intensity modulation in the 6* (horizontal) direction. Positions of the intensity maxima and minima differ for rows with h index odd and even.
symmetry and the crystallographic symmetry of space group C2 is shown in Figure 5. The asymmetric unit of the C2 cell thus consists of two identical or near-identical features at x,y,.z and where Ab is the small difference x,(y+Ab),(V+z), in the y co-ordinates described above. By considering the contributions of these two near-identical features in the asymmetric unit to the structure factors in the hk0 and hkh zones of the diffraction pattern it can be shown that the proposed packing arrangement is consistent with the two types of intensity modulation observed. For space group C2 the real and imaginary parts of the structure factor are given by: cos Znhx cos 27cky cos 2nhx sin dzky. The first term, in each case, determines the absence of reflections for which systematic h + k = 2n. The contribution, A,, of an atom at x,y,z to the real part of the structure factor is therefore: A 1chkO,= 4 cos 2nhx cos 2nky and the contribution, x,(y+Ab),(1/2+z) is:
A,, of an identical
atom at
A 2chkO,= 4 cos 2nhx cos 2nk( y + Ab). When cos 2nky is equal in magnitude but opposite in sign to cos Snk(y+Ab) contributions from the two atoms will cancel out. However, when the two terms are equal in both magnitude and sign, their contributions will add maximally. The same argument holds true for the terms sin 2zky and sin 2nk(y+ Ab) in the imaginary part of the structure factor. When extended to all pairs of
Crystal Forms of Canine Myeloperoxidase
Figure 4. Locat,ion of pseudo %fold axes (open symbols) in the 010 projection of t,he C2 cell of canine myeloperoxidase. Sear-identical features related by the pseudodyad are indicated as open and filled symbols. The resultant subc*rll has C’= c/P.
pscudos?-mmet,ry-related
in intensities
in the b* direction of the hkO zone. The modulation will have maxima for integer values of kAb and minima for integer values of 2kAb and will be of the h index. independent The intensity modulation observed in the hkO zone shown in Figure 2 has a minimum at k = 15 followed by a maximum at k = 30. The displacement between the two near-identical features along the b axis can thus be calculated to be b/30 = 32 A. In the hkh zone the real and imaginary parts of factor
Figure 5. Relationship between the pseudosymmetry elements (open symbols) and crystallographic syrnmrt,ry elements (filled symbols) in the (‘2 cell of canine myeloperoxidase. Fear-identical feat,ures with y coordinates differing by h/30 are indicated by open and filled symbols. Circles indicate possible moler>ular positions: the filled circle for a molecule with a true dyad axis and the broken circle for a molecule with an approximate dyad axis.
atoms in the asymmetric
unit t
the structure
923
are given
by:
(‘OS27cI$r+ z) cos 2nky (‘OH2zh(x+z)
sin 27cky.
Again. for two atoms at z.y,.z and z,(y+b),(1/2+z) the c~ont~ributions .-I, and A, to the real part of the struc&ture factor arc’ given by:
.-I I(hkh)= 4 (YE 2nh( .r + z) cos 2zky 2nh(.r+ z + I /2) cos 2zk(y + Ab). ;I Z,hkh)= 4 CY>S
Following the reasoning given above. when extended to all pairs of pseudosymmetry,-related atoms in the asymmetric unit there will again be an intensity modulation in the b* direction of the hkh zone. However. in t,his case, since ws &h(.r+~) and cos 2nh(x + t + l/2) are only equal for even values of’ h. having opposite signs for odd va111t~ of’ h. tht~ values of k corresponding to posit,ions of maxima and minima in the int,ensity modulation will br reversed for rows with II index odd compared with rows having h index even. The periodicGt.~ of the modulation will, however. be the same as in the hk0 zone. This is exactly the case in the hkh zone shown in Figure 3. For rows with h index WYW. maxima occur at k = 0 and 30, wit,h minima at A,= I5 and 45, while for rows with h index odd the reverse is true.
(c) Symmetrfy of myeloperoxidns~ The proposed disposition of pseudosymmetry element,s in the C2 unit cell of monoclinic crystals of myeloperoxidase (Fig. 5) permits a number of choices to be made for the location of’ a mofec~ule
924
R. E. Fenna
known to consist of two identical or near-identical halves. It is unlikely that the molecule would consist of halves arranged either head to tail or related by either crystallographic or pseudo 2-fold screw axes since both can give rise to infinite polymers. Two further possibilities exist,: (1) the myeloperoxidase molecule could have a 2-fold axis of symmetry, which in the (‘2 crystal form coincides with a crystallographic a-fold axis. Pairs of molecules located on non-equivalent crystallographic 2-fold axes would then be related by a pseudo 2-fold combined with a relative t’ranslation of 3.2 A in the b direction. (2) Alternatively. the pseudo 2-fold axis could be a molecular feature, in which ca,se t,he myeloperoxidase molecule would consist of two near-identical halves related by an approximate 2-fold axis but separated by a displacement of 3.2 a along this axis. Tdentical ha,lves of two different molecules would then bc related by the crystallographic X-fold axes. From the information presently available it is not possible to distinguish between these two cases.
4. Discussion Other examples of int’ensit’y modulations giving rise to characterist’ically st’riped diffraction patterns have been noted in the literature. (Irvstals of hemerythrin in space group P4 give d%fraction patterns illustrating both of the types of intensity modulation described above (North & Stubbs, 1974). In this case the modulation resulted from the displacement’ of two octameric molecules in approximately the same orientation by 10 A in the c direction of the unit cell. These authors also considered the possibility that the molecule could consist of two identical halves related by a rotation of 180” combined with a translation of 10 A along the 2-fold axis. However. since there was chemical evidence to suggest that the subunits of hemerythrin were equivalent this possibility was excluded. More recently a very similar intensity modulation was reported for crystals of myosin subfragment-l in space group P2,2,21 (Winkelmann et al.. 1985). In this case the asymmetric unit contained two molecules related by a local 2-fold axis. The intensity modulation and halving of the unit cell in projection arose because the local S-fold was parallel with a crystallographic screw axis but displaced from it by 4.3 A. Steitz et al. (1973) reported a crystal form of yeast hexokinase B in which two subunits are related by a 180” rotation plus a relative translation of 3.6 L\ along this axis. The presence of the approximate dyad axis was originally detected from molecular packing considerations based on an intensit;\ modulation observed in the hk0 zone of the diffraction pattern (Steitz, 1971). A different crystal form of yeast hexokinase B, in which heterologous interactions are responsible for dimer formation was later described by Anderson et al. (1974). In this case the subunits were related by a 156” rotation plus a translation of 13.8 A about the rotation axis.
One factor that) distinguishes the subunit interactions in myeloperoxidase from the examples discussed above is t,he possible presence of a disulfide bond bet,ween the two halves of the molecule. Presumably t’his bond plays an important role in determining the spatial disposition of the interacting subunits. The only example of a protein consisting of two halves linked by disulfide bridges where the S-ray cryst’al structure is known is the intact immunoglobulin molecule Kol (Colman rt 01.. 1976). Tn this case the molecule is located on a crystallographic d-fold axis and density cxorresponding to the two disulfide bridges is seen to cross this axis. However. whereas the two Fah portions of the molecule are related by the dyad axis the portions of the polypeptide chains on the (I-terminal side of the disulfides (corresponding to the Fc fragments) are not: and appear disordered in the electron density map. Even though the 2-fold axis is crystallographic, this can occur provided that t,he portions of the molecule that) are not related by t,he 2-fold do not participate in intermolecular contacts essential to the symmetry of the crystal lat,tice. The data we have presented indicate that the canine myeloperoxidase molecule consists of two halves t,hat are either identical or almost ident,ical and this is caonsistent wit’h t,he finding that the human enzyme can be cleaved int>o t.wo apparently identical halves by cleavage of a single disultide bridge between the large subunits (Andrews & Krinsky, 1981). The evidence further suggests that’ the two halves of the molecule are related by either an exact or an approximat~e 2-fold symmetry axis. Tn the lat,ter case t’he two halves of the molecule would not be exa&ly equivalent since environmental differences could occur as a result of heterologous interactions at the interfacse. Such a model could account for previously reported evidrnre that the two heme groups in myeloperoxidase are non-equivalent. Resolution of the question as to whether the molecular d-fold axis in myeloperoxidase is exact or only approximate must await determination of the X-ray cryst,al structure of this enzyme. Attempts to prepare suitable heavy-atom isomorphous derivatives of the trigonal crystal form of canine myeloperoxidase are now underway in this laborat.ory. This work was supported by a grant from the National Tnstit,ut.es of General Medical Sciences (no. GM33384). The author was the recipient of a Research Career Development Award from the National Institute of Diabetes and Digestive Kidne! Diseases and (no. DKOl132) during the time that this work was performed.
References Adar. F. (1978). Porphyrins, 3, 167-209. Agner. K. (1958). Acta Chem. Stand. 12, 89-94. Anderson, W. F., Fletterick, R. ,J. & Steitz. T. A. (1974). .I. Mol. Viol. 86, 261-269.
Crystal Forms of Canine M yeloperoxidasr Andrews, 1’. (‘. & Krinsky. Xv. I. (1981). J. Hiol. (‘hem. 256. 421 l--4%18. I
A ctct 1 828, .‘iti-fiB.
(‘olman. 1’. %I.. Drisrnhofrr. .J.. Huber. R. & Palm. 1%‘. (1!)76). .1. MO/. Biol. loo. 257%482. Eglington. 1). (i.. Barber. l).. Thompson. A. J.. (~rrwwood. (‘. & &gal. A. W. (1982). Biochim. Bioph!/s. d&l. 703. 187-l 95. Ehrrnberg. .\. Xr Agnrr. K. (1958). Acln (‘hem. Scnnd. 12. !)5- 100. Harrison. .I. tS. & Schultz. -1. (1976). ,I. Hiol. Chwr. 251. 1371 1374. Harrison. .I. lC. & Schultz. .J. (1978). Hiochim. Hiophys. Act/r. 536. :HLf-t!l. Harrisorl. .I. E.. I’abalan. S. & Schultz. J. (1977). lliorhinr.
Hiophys.
Actu, 493. 247-259.
lkrda-Saito. M. (1!%5). I. Riol. (‘hrm. 260, 11688-l 1696. lkeda-Saito. M.. l’rinw. R.. Agarde. I’. \‘. & Rousseau. I). I.. (I !M). Pcv/. I’roc. Frd. A mcr. Sot. Il:xp. Hiol. 43. l5AI. Klrhanofl’. S. J. & (‘lark. R’. A. (1978). Tn Thp Neutrophil. II nd (‘lin id Disorders. ~t///CtiW/ Xorth-Holland Edited
HP5
Publishing C‘n.. Amsterdam. Low, I<. \V. & Richards. F. 13. (195”). .1. .-1n/~r. f’hrm. sot. 74. 1660&l 666. Matthews. R. W. (1968). ,J. Mol. Viol. 33. 491 1!#i. Matthews. H. W. (1974a). .I. Mol. Riol. 82. 513-326. Matthews. R. &‘. (19746). In The I’rotrirrs (Xeurath. H. &, Hill. R. I,.. ~1s). Zirtl rdit.. pp. 40:~ ,590. A\~adtwic~ Press. Kew York. Xorth. A. C’. T. & Stubbs, (:. .I. (1974). .I. Jlol. Hid. 88. l&-1:31. Odajima. T. (1980). I. Biorhem. (Tokyo). 87. 379-391. Ronnberg. M.. Araiso. T.. Ellfolk, r\;. & Dunford. H. R. (1981). Arch. Riochrm. Hioyhys. 207. 197 -%N. Sibbrtt, S. S. & Hurst. .I. K. (1984). Riochrn~i.str~y. 23. :Nw)7-30 13. Steitz. T. A. (1971). .I. Mol. Hiol. 61. 695-700. Steitz. T. .I.. Flrtteric*k. R. J. & Hwang. K. J. (l!GR). I. Mol. Hiol. 78, 551~-561. Stendahl. 0.. (‘able. 1%.1.. Dahigren. (‘.. Hrd. J. & Molin. L. (1984). I. (‘/in. Inwst, 73. M&i-373. Thomas. E. T,.. (irishan. $1. 13. & .J&frrson hl. ( IWi). .I. (‘lit,. I,/wst. 72. 341-454. 12’inkrlmann. I). .A.. Mrkrrl. H. R Ra>~rnrnt. I. (IOH.‘). ./. 901. Hiol. 181. 487501.
by A. Kluy