Proc. Nadl. Acad. Sci. USA Vol. 87, pp. 1496-1500, February 1990 Biochemistry Arrangement of subunits and domains within the Octopus dofleini hemocyanin molecule (protein assembly/subunits/octopus) KAREN I. MILLER*t, ERIC SCHABTACHt, AND K. E. VAN HOLDE* *Department of Biochemistry and Biophysics, Oregon State University, Corvallis, OR 97331-6503; and tBiology Department, University of Oregon, Eugene, OR 97403 Contributed by K. E. van Holde, December 4, 1989

ABSTRACT Native Octopus dofleini hemocyanin appears graphs of the native molecule are shown in Fig. la] and lbl. as a hollow cylinder in the electron microscope. It is composed The molecule is a hollow circular cylinder; the top view (Fig. of 10 polypeptide subunits, each folded into seven globular la]) exhibits a fivefold symmetry with a highly reproducible oxygen-binding domains. The native structure reassociates pattern of five small projections into the central cavity. spontaneously from subunits in the presence of Mg2+ ions. We Diameter is about 320 A. The side view (Fig. lb]) shows a have selectively removed the C-terminal domain and purified three-tiered structure, with no evidence of axial asymmetry. the resulting six-domain subunits. Although these six-domain The decameric whole molecule requires divalent ions for subunits do not associate efficiently at pH 7.2, they undergo stability and can be dissociated into subunits by dialysis nearly complete reassociation at pH 8.0. The resulting molecule against EDTA. This dissociation has been shown to be wholly looks like the native cylindrical whole molecule but lacks the reversible upon restoration of divalent cations to the solution usual fivefold protrusions into the central cavity. Partially (8, 9). reassociated mixtures show dimers of the subunit that have a The subunits obtained by dissociation exhibit the structure characteristic parallelogram shape when lying flat on the shown in Fig. le], each showing seven globular domains. electron microscope grid, and a "boat" form in side view. Specific cleavage between these domains can be achieved by Removal of the C-terminal domain from monomers results in limited proteolysis; through such studies the domains have the removal of two characteristically placed domains in the been shown to be immunologically distinct (10). Each con- dimers. These observations allow the development of a model tains one oxygen-binding site. The domains have earlier been for the arrangement of the subunits within the whole molecule. designated by labels Odl-0d7, on the basis ofimmunological The model predicts exactly the views seen in the electron purification. Since the sequence of domains in the polypep- microscope of both whole molecule and dimeric intermediates. tide chain has now been established (4) to be 7-4-3-6-5-2-1, we now wish to relabel them as domains a-g (Fig. le2). Hemocyanins are copper proteins that serve to transport Sequencing of cDNA clones has to date provided complete oxygen in many species of arthropods and molluscs (1, 2). sequences for the three domains at the C terminus (e, f, g) (4, Although arthropod and molluscan hemocyanins have long 11). been assumed to be closely related proteins, recent sequenc- A key to understanding the structure of the functional ing studies indicate that these two classes of hemocyanins hemocyanin molecule lies in determining the roles played by have probably evolved independently from copper proteins different domains in the assembled structure. For example, such as tyrosinase (3, 4). Pronounced differences in both we may ask: Which are in the cylinder wall, and which are primary and quaternary structure differentiate the two involved in the inner projections? What is the relative ori- classes. The hemocyanins of arthropods are built up from entation of the 10 subunits-do they array in a parallel or subunits of moderate size (ca. 75 kDa), each containing one antiparallel manner? How is the three-tiered wall con- binuclear copper oxygen-binding site. As a consequence of structed? In recent studies we have been examining by x-ray diffraction studies, the structures of these proteins are electron microscopy the assembly of Octopus hemocyanin, becoming well understood (5, 6). using both native subunits and subunits from which a specific Much less is known concerning the structures of molluscan domain has been removed. We believe that these studies now hemocyanins. These proteins exist in the hemolymph as very allow unambiguous answers to the above questions. large molecules, in most cases assembled as 1O-mers or 20-mers of polypeptide chains. Each chain is immense, AND METHODS having a molecular mass 350-450 kDa (1, 2). Each of these MATERIALS subunits contains in turn a number of folded domains (or Hemocyanin was purified from whole blood by gel filtration functional units), each of mass ca. 50 kDa and carrying one on Bio-Gel A-Sm (Bio-Rad) in 0.1 ,I (ionic strength) Tris, pH binuclear copper site. No x-ray diffraction analyses on any of 7.65/50 mM MgCl2/10 mM CaCI2. The C-terminal domain these molluscan hemocyanins or their subunits have been was removed by limited proteolysis with protease from reported to date, although crystals and diffraction patterns Staphylococcus aureus strain V8 (EC 3.4.21.19; Sigma) in have been obtained for the C-terminal domain of Octopus 0.05 M ammonium carbonate at pH 8.0 (enzyme/substrate hemocyanin (7). ratio, 10-4; 370C for 24 hr). A series of trial digestions was For a number of years, our laboratory has been investi- carried out for various times and with several enzyme con- gating the structure ofthe hemocyanin ofthe Pacific octopus, centrations to determine the optimal conditions for removal Octopus dofleini. We have shown that the functional mole- of the C-terminal domain with minimal further cleavage. The cule in the hemolymph is a decamer, of mass 3600 kDa resultant digest was subjected to gel filtration on Bio-Gel composed of 10 chains of mass 360 kDa (8). Electron micro- A-0.5m at 40C in 0.1 ji Tris (pH 7.2) in order to separate the 6D fragment from undigested (7D) subunits and smaller The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviation: nD, n-domain. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 1496 Downloaded by guest on September 29, 2021 Biochemistry: Miller et al. Proc. Natl. Acad. Sci. USA 87 (1990) 1497

al

.2 ; '- Sa bi

cl *:

,vs.v A t > > > s _ s sW - a .4 Xrtr . w

Jktr.-.t. _

IL~~~~ el 1SG ~~ I'4f3.N*S;-iW\~t *>5;F t ;; -

FIG. 1. Electron micrographs of0. dofleini hemocyanin and the proposed model ofthe three-dimensional structure. Column 1, seven-domain (7D) native hemocyanin; column 2, model structure; column 3, 6D modified hemocyanin; row a, top view of whole molecule; row b, side view of whole molecule; row c, top view of dimer; row d, side view of dimer; row e, subunit. (x345,OO0.)

fragments. Reassociation of the 6D fragment was examined mM Mg2+ to concentrations suitable for staining. For reas- by rapid dialysis against 0.1 ,u Tris buffer containing 50 mM sociation studies of 7D or 6D subunits, small amounts of a 4 MgCl2 and was monitored by analytical ultracentrifugation in M solution of MgCl2 were added to give final concentrations a Beckman model E equipped with a photoelectric scanner. between 30 mM and 100 mM to solutions of the subunits in Electron microscopy on both the intact (51S) molecule and 0.1 M or 0.05 M Hepes buffer. The reassociating mixture was the reassociating 7D and 6D subunits utilized negative stain- sampled at timed intervals thereafter. Reassociations were ing with 0.5% or 0.8% uranyl acetate on thin carbon films carried out either at a protein concentration suitable for supported on nets. Grids with support film were rendered staining without further dilution or at higher concentrations hydrophilic by glow discharge shortly before use. Just prior with subsequent dilution with the appropriate buffer imme- to application to grids, preparations of the intact (SiS) diately before staining. All reassociations were done at molecule were diluted with 0.1 M Tris buffer containing 50 22-24°C. Microscopy employed a Philips CM-12 electron Downloaded by guest on September 29, 2021 1498 Biochemistry: Miller et al. Proc. Natl. Acad. Sci. USA 87 (1990) microscope operated at 80 or 100 kV. Low-dose procedures Calibration of the gel confirmed that this was a 6-unit were used. Magnification was calibrated using crystalline residuum (data not shown). Further confirmation came from catalase. electron microscopic analysis ofthe material in peak 6, which showed predominantly 6D rather than 7D particles (i.e., Fig. RESULTS AND DISCUSSION le3). As in the case of the intact 7D subunit, we see here no consistent spatial arrangement of the domains in the isolated In previous studies, we examined the Mg2+-activated reas- subunits. That the single unit first removed by V8 protease is sociation of Octopus hemocyanin subunits into functional in fact domain g has been demonstrated in earlier studies (10). decamers, using light scattering, fluorescence, and sedimen- Purification of the 6D fragment allowed us to ask the tation techniques (12). The sedimentation studies showed following questions. (i) Is domain g required for reassociation that the process was slow, and indicated the major compo- into decamers? (ii) If 6D subunits can reassociate, can we nents present during reassociation to be completely assem- deduce from the structure of the reconstituted decamer the bled molecules and unreacted monomer. Light scattering and position of the g units in the molecule? fluorescence experiments showed that the rate-limiting step Reassociation experiments were performed by rapid dial- is second order in monomer. We now find, by electron ysis of the purified 6D fragment against buffer containing 50 microscopy of samples taken during reassociation, evidence mM MgCl2. Under these conditions, reassociation has been for a number of intermediate species. The smaller of these shown to go to completion with 7D fragments over the pH take the form ofcrescents with two beads on the concave side range 7.0-8.0 (9). At pH 7.2, some reassociation of 6D (Fig. lcJ); others appear to be small parallelograms (Fig. fragments could be detected in the analytical ultracentrifuge, ldJ). We believe these to be two views of a dimer of the but it did not proceed much beyond the =20S dimer. At pH subunit, for very similar structures can be isolated when 8.0, however, at least 75% of the sedimenting protein had Helix pomatia hemocyanin is taken to high pH; these have reassembled into a very homogeneous population of 46S been demonstrated to be dimers by molecular weight mea- particles. Fig. 3 shows the distribution of the sedimentation surement (13). The sedimentation coefficient of such dimers coefficient at each pH, calculated according to the method of is found to be approximately 20 S. van Holde and Weischet (14). These reassociated mixtures It is not the aim of this paper, however, to describe the were then examined in the electron microscope. At pH 8.0 reassociation process in detail; this will be the topic offuture (Fig. 4a) we saw predominantly structures that appeared very publication. Rather, we wish to focus here on the role played similar to the native 51S whole molecule in side view (Fig. by the C-terminal functional unit, domain g, in forming the 1b3), but in the top view the cylinder lacked the five distinct decameric structure. The results will in turn allow us to projections into the central space (Fig. 1a3). Occasionally, a propose a model for the packing of subunits in the whole molecule was observed in which one central projection (or molecule. Earlier studies have shown that domain g can be portion thereof) could be observed. We interpret these as removed from the whole subunit by mild digestion with either molecules in which contaminating 7D fragments have been trypsin or S. aureus V8 protease (10). After exploratory incorporated. As we have already observed for native hemo- studies with both enzymes, we found that the V8 protease cyanin (Fig. 1 al and a2), the diameter ofthis molecule when was the more selective in cleavage, cutting first between measured in face view is about 320 A; in side view the domains f and g and later between domains c and d. Limited diameter is about 380 A and the thickness about 190 A. The digestion followed immediately by gel filtration on Bio-Gel larger diameter seen in side view is almost certainly an A-0.5m produced the elution pattern shown in Fig. 2. It has artifact due to the flattening and consequent elongation ofthe distinct peaks for the 6D, 3D, and 1D fragments. SDS/ molecules when they are oriented "on edge" on the support polyacrylamide gel electrophoresis (Fig. 2 Inset) showed that film. At pH 7.2 (Fig. 4b) a few whole molecules could be seen peak 6 (lane 3 from left) was highly enriched in a fragment that but the major portion of the sample consisted of dimers and migrated slightly more rapidly than the 7D native subunit. monomers. The dimer formed from 6D fragments was also similar in 2 side view (Fig. 1d3) to the 7D version, but in top view (Fig.

1.0 -

0.8

0.6 .> pH 8 a) Z- 0.4

0.2'

0 Fraction 0 30 s value, S FIG. 2. Elution profile of 0. dofleini subunits following digestion with S. aureus V8 protease and gel filtration on Bio-Gel A-0.5m, in FIG. 3. Distribution of sedimentation coefficient (s) calculated 0.1 ,u Tris (pH 7.2). Peaks 6, 3, and 1 contain primary 6D, 3D, and according to the method of van Holde and Weischet (14) for 6D 0. 1D proteins. (Inset) SDS/7.5% polyacrylamide gel electrophoresis dofleini hemocyanin subunits reassociated in 0.1 ,u Tris/50 mM of, from left to right, native 7D subunit, total digested mixture, peak MgCl2 at pH 7.2 or 8.0. Reassociation is much more homogeneous 6, peak 3, and peak 1. and complete at pH 8.0. Downloaded by guest on September 29, 2021 Biochemistry: Miller et al. Proc. Nati. Acad. Sci. USA 87 (1990) 1499

FIG. 6. Two fivefold rotational integration photographs ofthe top view of 0. dofleini hemocyanin. (x 1,075,000.) The question now remaining is: How are the other six domains arranged to form the three-tiered side wall? For evidence on this question we turn to the various intermedi- ates found in reassociation kinetics experiments (15). The side views of the dimer have a characteristic parallelogram shape. If one considers the angles of this parallelogram and the dimensions of the cylinder, the diagram shown in Fig. 5 presents itself as a probable arrangement of the subunits within the side wall. It requires that the subunits be arranged diagonally with two domains from the subunit in each of the three tiers. Ten subunits folded in this way would give a cylinder 3 domains high and 20 domains in circumference, which is in good agreement with the molecular dimensions. A model was constructed on this basis and photographs (Fig. 1 a2-d2) predict exactly the molecular shapes and dimensions seen in the electron microscope (Fig. 1 al-di). It should be noted that such a model, with each domain g projecting into the central space, gives just the kind of fivefold image seen in Fig. lal. Fig. 6 shows fivefold rotational integration FIG. 4. Electron micrographs of partially reassociated 6D sub- ("Markham rotation") (16) oftwo different face views of 51S units of 0. dofleini hemocyanin, showing many dimers at pH 7.2 (a) molecules. These rotations give further support to the model and many complete molecules at pH 8.0 (b). (x 104,400.) structure. In the model (Fig. 1a2) domain g appears as five inner of the cylinder. These are it lacked the two small domains on the inner side of the pairs of beads on the face 1c3) offset from each other rotationally by approximately 1 do- crescent. The unassociated monomer (Fig. 1e3) had only six arrangement domains. main. The rotation photograph shows this offset side wall ofthe cylindrical clearly. The dimer in top view shows a crescent with two The three-tiered structure ofthe beads on the concave side (Fig. 1c2) and in side view a whole molecule, and the fact that this view is unaltered in the parallelogram with 3- and 4-domain-wide sides (Fig. 1d2). reconstitute from 6D subunits, establishes that the first six of Occasionally one sees a trimeric parallelogram (not shown), the seven domains in the Octopus subunit form the side wall. but never a larger side view. Trimeric and larger top views are The seventh domain (g) projects into the interior space; these common. Perhaps side views of partially reassembled mol- must be evenly distributed top and bottom to account for the ecules larger than trimers are structurally unstable due to symmetrical side views. Thus we conclude that the 10 their curvature and break apart when flattened against the subunits must be arranged in antiparallel pairs to form the grid. cylindrical structure, a conclusion consistent with the appar- There are several ways in which the subunit could be ent fivefold axis of symmetry in this decameric molecule. The folded to fit six domains into the side wall in a parallelogram effect produced on electron microscopic images of the dimer array. Fig. 5 depicts one of the simplest. Our model bears by removal of the two g domains gives further confirmation some resemblance to that proposed by van Breeman et al. of the position of domain g. Also, the fact that two domains (17) and by Berger et al. (18) for H. pomatia hemocyanin. are removed from this image when 6D monomers are used However, the evidence that H. pomatia hemocyanin forms gives additional proof that it does indeed represent a dimer. an asymmetric decamer from 8D subunits leads to a quite different structure. The Berger et al. model could, of course, give no information about the location of specific domains. Confirmation of specific details of the model we propose must await further studies using antibodies to individual domains. We especially wish to thank Dr. Fumio Arisaka, who originally suggested using the electron microscope to follow association of Octopus hemocyanin subunits. This work was supported by National Science Foundation Grant DMB-8818697 (K.E.v.H. and K.I.M.). K.E.v.H. is the recipient of an American Cancer Society research professorship. The electron microscope was obtained with funds FIG. 5. Model of a dimer of two subunits of 0. dofleini hemo- from National Institutes of Health Biomedical Research Support cyanin, used to build the molecular models of Fig. 1 a2-d2. Shared Instrumentation Grant 1-S10-RR02756-01 (E.S.). Downloaded by guest on September 29, 2021 1500 Biochemistry: Miller et al. Proc. Nail. Acad. Sci. USA 87 (1990)

1. van Holde, K. E. & Miller, K. 1.(1982) Q. Rev. Biophys. 15, 10. Lamy, J., Lederc, M., Sizaret, P.-Y., Lamy, J., Miller, K. I., 1-129. McParland, R. & van Holde, K. E. (1987) Biochemistry 26, 2. Ellerton, H. D., Ellerton, N. F. & Robinson, H. A. (1983) 3509-3518. Prog. Biophys. Mol. Biol. 41, 143-248. 11. Lang, W. H. (1988) Biochemistry 27, 7276-7282. 3. Lontie, R., Witters, R., Gielens, G. & Prdaux, G., in Inverte- 12. van Holde, K. E. & Miller, K. I. (1986) in Invertebrate Oxygen brate Dioxygen Carriers, eds. Prdaux, G. & Lontie, R. (Univ. Carriers, ed. Linzen, B. (Springer, Berlin), pp. 245-248. of Leuven Press, Louvain, Belgium), in press. 13. van Bruggen, E. F. J., Schutter, W. G., van Breemen, 4. Lang, W. H. & van Holde, K. E., in Invertebrate Dioxygen J. F. L., Bijlholt, M. M. C. & Wichertjes, T. (1981) in Electron Carriers, eds. Prdaux, G. & Lontie, R. (Univ. of Leuven Press, Microscopy of Proteins, ed. Harris, J. R. (Academic, New Louvain, Belgium), in press. 5. Gaykema, W. P. J., Hol, W. G. J., Vereijken, J. M., Soeter, York), pp. 1-38. N. N., Bak, H. J. & Bientema, J. J. (1984) Nature (London) 14. van Holde, K. E. & Weischet, W. (1978) Biopolymers 17, 309, 230-239. 1387-1403. 6. Volbeda, A. & Hol, W. G. J. (1989) J. Mol. Biol. 209, 249-279. 15. Miller, K. I., Schabtach, E. & van Holde, K. E., in Inverte- 7. Cuff, M. E., Hendrickson, W. A., Lamy, J., Lamy, J., Miller, brate Dioxygen Carriers, eds. Prdaux, G. & Lontie, R. (Univ. K. I. & van Holde, K. E., in Invertebrate Dioxygen Carriers, of Leuven Press, Louvain, Belgium), in press. eds. Preaux, G. & Lontie, R. (Univ. of Leuven Press, Louvain, 16. Markham, R., Fry, S. & Hills, G. J. (1963) Virology 20,88-102. Belgium), in press. 17. van Breeman, J. F. L., Schuerhuis, G. J. & van Bruggen, 8. Miller, K. I. & van Holde, K. E. (1982) Comp. Biochem. E. F. J. (1977) in Structure andFunction ofHaemocyanins, ed. Physiol. B 73, 1013-1018. Bannister, J. V. (Springer, Berlin), pp. 122-127. 9. van Holde, K. E. & Miller, K. I. (1985) Biochemistry 24, 18. Berger, J., Pilz, I., Witters, R. & Lontie, R. (1977) Eur. J. 4577-4582. Biochem. 80, 79-82. Downloaded by guest on September 29, 2021