Proc. Natl. Acad. Sci. USA Vol. 76, No. 4, pp. 1668-1672, April 1979 Biochemistry Complete amino acid sequence of a histidine-rich proteolytic fragment of human ( structure// binding/genetic polymorphism/evolution) I. BARRY KINGSTON*, BRIONY L. KINGSTONt, AND FRANK W. PUTNAMt Department of Biology, Indiana University, Bloomington, Indiana 47405 Contributed by Frank W. Putnam, January 22, 1979

ABSTRACT The complete amino acid sequence has been a chain of human ceruloplasmin studied by McCombs and determined for a fragment of human ceruloplasmin [ferroxidase; Bowman (6) and probably corresponds to the a subunit pro- iron(II)oxygen , EC 1.16.3.1]. The fragment posed by Simons and Bearn (4) and the chain of (designated Cp F5) contains 159 amino acid residues and has light Freeman a molecular weight of 18,650; it lacks carbohydrate, is rich in and Daniel (5). histidine, and contains one free cysteine that may be part of a The relation of Cp F5 to the structure of the intact cerulo- copper-. This fragment is present in most commer- chain has to be deduced from indirect observations cial preparations of ceruloplasmin, probably owing to proteo- because of the small amount of single-chain ceruloplasmin lytic degradation, but can also be obtained by limited cleavage available to us and the strong interaction of the fragments. From of single-chain ceruloplasmin with plasmin. Cp F5 probably is the kinetics of the proteolytic cleavage of intact ceruloplasmin an intact domain attached to the COOH-terminal end of sin- gle-chain ceruloplasmin via a labile interdomain peptide bond. and the chemical properties of the fragments, we have sug- A model of the secondary structure predicted by empirical gested that Cp F5 is from the COOH terminus of the single methods suggests that almost one-third of the amino acid resi- chain (8). We have obtained a unique amino acid sequence and dues are distributed in a helices, about a third in ,8-sheet conclude that Cp F5 is probably an intact domain; it appears structure, and the remainder in ft turns and unidentified struc- to be attached to the COOH-terminal end of the ceruloplasmin tures. Computer analysis of the amino acid sequence has not chain by a labile interdomain peptide bond. Prediction of the demonstrated a statistically significant relationship between this ceruloplasmin fragment and any other protein, but there secondary structure by the empirical method of Chou and is some evidence for an internal duplication. Fasman (9, 10) leads to a model in which approximately 30% of the residues occur in f3 sheets, 25% in a helices, and 45% in Ceruloplasmin [ferroxidase; iron(II):oxygen oxidoreductase, j turns and other structure. EC 1.16.3.1], is a blue, copper-containing a2- that Because the ceruloplasmin preparation studied is derived is normally present in human plasma at a concentration of from a pool of plasma from more than 10,000 donors from the 20-40 mg/100 ml (1). Although the biological role of cerulo- ethnically mixed American population, the finding of a unique plasmin (Cp) is not entirely clear, at least three functions have amino acid sequence indicates that the frequency of genetic been ascribed: ferroxidase activity, copper transport and de- polymorphism is too low to interfere with sequence determi- toxication, and maintenance of copper homeostasis in the nation of this fragment. tissues. These functions are not mutually exclusive, but the most In a previous report (8) we identified a cysteine-containing important is thought to be the action of plasma ceruloplasmin sequence in human Cp F5 that appeared similar to a series of as a ferroxidase in oxidizing ferrous iron to the ferric form cysteine-containing sequences that are apparently homologous which is then incorporated into (2). The only es- to each other and contain part of the copper-binding sites in tablished biochemical abnormality involving ceruloplasmin is bacterial azurins and plant plastocyanins. Computer analysis its deficiency in Wilson disease (hepatolenticular degeneration). of an intersequence comparison of Cp F5 and azurin from Al- The deficiency is due to a genetic defect in the rate of cerulo- caligenes sp. indeed showed that the most similar segment of plasmin synthesis that leads to abnormal copper sequence was that which we had identified. However, no sta- and deposition of copper in the tissues (3). tistically significant relationship could be demonstrated be- Despite several reports that it has a subunit structure (1, 4-6), tween the ceruloplasmin fragment and any other protein, in- ceruloplasmin has been shown to be a single polypeptide chain cluding azurin, plastocyanin, and dismutase, all of with a molecular weight of about 130,000 that is readily cleaved which are copper-containing . Further computer to large fragments by proteolytic (7, 8). We have iso- analysis indicated a possible internal duplication in the ceru- lated and characterized three fragments of ceruloplasmin that loplasmin. appear to be nonoverlapping and that have approximate mo- lecular weights of 20,000, 53,000, and 67,000 (8). We report MATERIALS AND METHODS here the complete amino acid sequence of the smallest frag- Materials. The purified human ceruloplasmin from normal ment, which we call the histidine-rich fragment and designate pooled sera used for sequence analysis was a preparation made Cp F5. This fragment contains 159 amino acid residues and by the method of Sgouris et al. (11), starting with ethanol lacks carbohydrate. It is present to a varying degree in all the fraction IV-1, and was generously provided by J. J. Hagan commercial preparations we examined and can be prepared from single-chain ceruloplasmin by digestion with plasmin (8). Abbreviation: Cp, ceruloplasmin. Cp F5 is identical in NH2-terminal sequence to the so-called * Present address: National Institute of Agricultural Botany, Hun- tingdon Road, Cambridge, England. The publication costs of this article were defrayed in part by page t Present address: Commonwealth Bureau of Plant Breeding and charge payment. This article must therefore be hereby marked "ad- Genetics, Department of Applied Biology, Pembroke St., Cambridge, vertisement" in accordance with 18 U. S. C. §1734 solely to indicate England. this fact. t To whom reprint requests should be addressed. 1668 Downloaded by guest on September 25, 2021 Biochemistry: Kingston et al. Proc. Natl. Acad. Sci. USA 76 (1979) 1669 (Squibb). This was preparation Cp 1 of our previous paper (8); (which has specificity for peptide bonds having aspartic or several other preparations described there were used for ref- glutamic acid on the carboxyl side) was used to prepare sub- erence purposes, including single-chain ceruloplasmin (Cp 3 peptides of tryptic peptides T3 and T4 and also T6. In addition, and Cp 5) supplied by Yu lee Hao (National Fractionation Cp F5 was citraconylated and digested with trypsin to give a Laboratory of the American National Red Cross). peptide covering the segment from Lys-7 through Arg-58. Purification of Histidine-Rich Fragment Cp F5. Fragment Sequence analysis of these peptides permitted completion of Cp F5 was prepared from Cp 1 after reduction and ami- the missing area. noethylation. Cp 1 (420 mg) was dissolved in a buffer (20 ml) containing 6 M guanidine-HCI, 2 mM EDTA, 0.5 M Tris-HCI, RESULTS AND DISCUSSION at pH 8.0, and reduced under N2 at 50'C with dithiothreitol Amino Acid Sequence. The complete amino acid sequence (360 mg). After cooling to 4°C, three aliquots (80 Al each) of of the human ceruloplasmin fragment Cp F5 derived by the ethylene imine (Pierce) were added at 10-min intervals. The strategy described above is given in Fig. 2. Except for the solution was then dialyzed against deionized water (4 liters, abundance of histidine, there is nothing remarkable about the three times) and lyophilized. The reduced aminoethylated primary structure. The single cysteine residue (Cys-134) must protein was dissolved in 10 ml of 6 M urea/0.2 M formic acid be present in the sulfhydryl form because reduction made no and applied to a column (150 X 4 cm) of Sephadex G-150 or difference in the banding pattern of the original ceruloplasmin G-200 equilibrated with the same solution. The column effluent preparation on electrophoresis in polyacrylamide gel containing was monitored at 280 nm, and good separation of Cp F5 was 0.1% sodium dodecyl sulfate (8). obtained. After volume reduction by ultrafiltration, the frag- The NH2-terminal sequence of Cp F5 is identical to the ment was desalted on Sephadex G-15 in 1 M acetic acid and 21-residue NH2-terminal sequence given by McCombs and Iyophilized. Bowman (6) for a ceruloplasmin polypeptide they designated Methods. The methods for protein characterization, amino a chain. With the exception of this and of three glycopeptides acid analysis, and sequence determination were the same as in from human ceruloplasmin (12), no other sequence data are our previous report (8). published on ceruloplasmin from any species. However, the Strategy for Sequence Determination. In addition to au- amino acid composition of a cysteine-containing peptide iso- tomated sequence analysis of the intact Cp F5 fragment for 18 lated by Egorov et al. (13) from undegraded human cerulo- steps, the strategy for sequence determination involved two plasmin corresponds to the sequence around Cys-134. different routes in order to obtain independent verification of Fragment Cp F5 has previously been referred to by us (8) each position and to secure all necessary overlaps of peptides. as the 20,000 Mr fragment because that was the molecular Because of the presence of seven methionine residues, CNBr weight estimated from polyacrylamide gel electrophoresis. cleavage was chosen for the initial procedure. Nine CNBr Calculation of the exact molecular weight from the sequence peptides were obtained (Fig. 1); these were purified and sub- yields a value of 18,650, which is in good agreement. mitted directly to automatic sequence analysis, except for the Carbohydrate Content. The most difficult part of the se- dipeptide CB5. In some instances (CB1, CB2, CB6, and CB9) quence to determine was the segment from Asp-18 through the CNBr peptides were digested with trypsin, and the tryptic Lys-30. This acidic stretch contains seven dicarboxylic acid subpeptides were purified and analyzed and their amino acid residues, of which five are in the acidic form. In addition to the sequences were determined. In other cases the subpeptides of difficulty of separating subpeptides of this region, nonstoi- the CNBr fragment were digested further with another en- chiometric amounts of glucosamine appeared to be present on zyme-e.g., thermolysin for CB6, pepsin for CB8, or Staphy- amino acid analysis. The sequence determined does not cor- loccocus aureus V8 protease for CB9. The results of sequence respond to any of the three glycopeptides reported by Ryden analysis by this procedure are diagrammed schematically in and Eaker (12). However, the tripeptide Asn-Glu-Ser (positions Fig. 1. 20-22) contains an example of the triplet acceptor sequence The second route for sequence determination consisted of Asn-X-Ser/Thr, to which glucosamine oligosaccharides usually digestion of the Cp F5 fragment with trypsin, followed by are attached (14). For this reason, and because of the possible isolation, analysis, and partial or complete sequence determi- presence of glucosamine in the impure peptides from this nation of the 19 tryptic peptides obtained; these are denoted segment, we gave a sample of Cp F5 to Jacques Baenziger schematically in Fig. 1. Because of difficulty in purifying the (Washington University School of Medicine, St. Louis, MO) for peptides and determining the sequence in the highly acidic complete carbohydrate analysis. In personal communication, region from Asp-18 through Glu-37, staphylococcal protease Dr. Baenziger reported to us that "a preliminary analysis of the

Residue 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 159 Number I I II T l ' l ' l ' l ' l ' l ' l CB7 CB1 ICB2 ICB3 I CB4 151 CB6 CB8 C9 CNB r Peptides

T13 T1l TlT8T1| T3 T5 T4 T9 T10 ITll I T15 IT16 I T17 Tryptic Peptides _1 T4 I- T6 1-12 1 T T18 'B T2 T14 FIG. 1. Diagram for sequence determination of Cp F5, The scale at the top gives the amino acid residue number. The nine CNBr peptides are demarcated and are designated CB1-CB9. The extra line indicates CB7, which contains CB6 and CB8 joined together. In a separate procedure 19 tryptic peptides, designated T1-T19, were obtained; some of these are overlapping. Subpeptides of the CNBr or tryptic peptides are not indicated in the diagram. Shaded areas denote positions for which the amino acid sequence was established in each procedure. Downloaded by guest on September 25, 2021 1670 Biochemistry: Kingston et al. Proc. Natl. Acad. Sci. USA 76 (1979)

10 20 30 H2N-Val -Phe-Asn-Pro-Arg-Arg-Lys -Leu- G1 u-Phe-A1 a-Leu- Leu-Phe- Leu-Val -Phe-Asp-Gl u-Asn-GI u-Ser-Trp-Tyr- Leu-Asp-Asp-Asn- I 1e- Lys- 40 50 60 Thr-Tyr-Ser-Asp-Hi s-Pro-G1 u-Lys-Val -Asn-Lys-Asp-Asp-Gl u-Gl u-Phe- Il e-Gl u-Ser-Asn-Lys-Met-Hi s-Ala-I le-Asn-Gly-Arg-Met-Phe- 70 80 90 Gly-Asn-Leu-Gl n-Gly-Leu-Thr-Met-Hi s-Val -Gly-Asp-Gl u-Val -Asn-Trp-Tyr-Leu-Met-Gly-Met-Gly-Asn-Gl u- Ile-Asp- Leu-Hi s-Thr-Val - 100 110 120 Hi s-Phe-Hi s-Gly-Hi s-Ser-Phe-Gl n-Tyr-Lys-Hi s-Arg-Gly-Val -Tyr-Ser-Ser-Asp-Val -Phe-Asp- Ile-Phe-Pro-Gly-Thr-Tyr-Gl n-Thr-Leu- 130 140 150 Gi u-Met-Phe-Pro-Arg-Thr-Pro-Gly- Ile-Trp-Leu-Leu-Ili s-Cys-Hi s-Val -Thr-Asp-Hi s-Ile-His-Ala-Gly-Met-G1u-Thr-Thr-Tyr-Thr-Val -

Leu-Gln-Asn-Gl u-Asp-Thr-Lys-Ser-Gly-COOH FIG. 2. Amino acid sequence of the histidine-rich fragment of human Cp F5.

ceruloplasmin 20,000 Mr fragment would indicate that less than rosine at position 60. We do not know whether the tyrosine- 10% of the peptide could be glycosylated." Expressed as moles containing peptide was lost during purification or if it was of carbohydrate per 20,000 daltons, the fragment contained: present only in the first preparation of ceruloplasmin. It is 0.05 glucosamine, 0.12 galactose, 0.12 mannose, and 0.00 fu- possible that a tyrosine-phenylalanine exchange at position 60 cose. It is possible that Asn-20 is partially glycosylated and, al- represents a true genetic polymorphism, but it is unlikely that though our calculation of the secondary structure of Cp F5 (see the tyrosyl variant is Cp A because the latter differs in charge later) places Asn-20 at the end of an a helix, residues 20-23 also from Cp B. have a high probability of being in a (3turn. Since it appears Our evidence for proteolytic degradation coupled with the likely that carbohydrate is frequently attached to observation of "gull-wing" patterns in immunoelectrophoresis at (3turns (15, 16), Asn-20 may be in a favorable position for of aged preparations suggests that some genetic variants pre- glycosylation. However, the bulky tryptophan residue just after viously reported may be artifacts. Poulik and Weiss (1) have the triplet (i.e., Asn-Glu-Ser-Trp) may hinder glycosylation. discussed the questionable genetic significance of one reported Genetic Polymorphism. Genetic polymorphism of cerulo- genetic variant that has a band with identical electrophoretic plasmin is rare in whites, but in black Americans a variant mobility to one observed in aged serum that originally had only designated Cp A occurs with an allele frequency of 0.052, the Cp B type. compared to a frequency of 0.939 for the common form Cp B Role of Histidine in Histidine-Rich Fragment. Although and 0.003 and 0.006 for the rare variants Cp C and Cp NH, exceeded in histidine content by the "histidine-rich a2 glyco- respectively (1, 3). In white Americans the frequency of Cp B protein" of human plasma (17), which has a histidine content is 0.994, whereas the frequency of Cp A is only 0.006, though of 9.9%, and by the a globin chains of some species that have Cp A frequency rises to a high of 0.149 in Nigerians (1). Other 10 or more histidines per a chain, the Cp F5 fragment of variants of ceruloplasmin have a very low incidence in all human ceruloplasmin has a higher histidine content than most populations studied (1). Because blacks constitute about 11% other known proteins (8.82%, calculated from the sequence). of the American population, the incidence of Cp A and other Furthermore, 7 of the 12 histidines occur in an alternating se- known variants in pooled plasma collected by the American quence of His-X-His-i.e., His-Phe-His-Gly-His, His-Cys-His, Red Cross must be less than 0.6%. Such a value is too low to be and His-Ile-His (beginning at positions 91, 133, and 139, re- detected by the methodology of amino acid sequence analysis. spectively). This leads to two short segments that, when other We estimate that because of the cumulative errors in the nearby histidines are included, are exceedingly rich in histi- methodology, genetic variants at an individual frequency of dine-i.e., there are four histidines in the eight-residue segment less than 0.05 in pooled plasma would not be detectable by from His-88 to His-95 and four more in the nine-residue seg- amino acid sequence analysis. It is thus not surprising that we ment from His-l33 through His-141, which contains the free obtained a unique amino acid sequence for the fragment Cp cysteine. In view of the role of histidine in the binding of copper F5. Furthermore, Cp F5 represents only a fragment of the by serum (18), the proximity of so many histidines close whole ceruloplasmin molecule and may not be the site of any to Cys-134, which is assumed to be a copper-binding site (8), polymorphic substitution. is probably significant. Because of the alternation, all of the However, one observation suggests to us the possibility that histidines could be exposed to the surface in the ( sheet postu- unidentified genetic variants not detectable by the usual lated for residues 129-137 and the f turn following it. method of electrophoretic screening may exist in the American Prediction of Secondary Structure. The problems of pre- population. In early work with Cp 2 (8), a different preparation dicting protein structure from amino acid sequence have re- of ceruloplasmin than the one used for complete sequence cently been reviewed by Sternberg and Thornton (19). None- analysis (Cp 1), we obtained a CNBr fragment§ corresponding theless, they point out that "the method of Chou and Fasman to the sequence given in Fig. 2 as Phe-Gly-Asn-Leu-Gln-Gly- (9, 10) has attracted most attention as it is simple to understand, Leu-Thr-Met for positions 60-68; however, the composition can be applied without a computer, and has been relatively of the fragment showed 0.67 residue of phenylalanine and 0.33 successful." Hence, we have applied their rules for predicting of tyrosine, and the dansyl method and sequenator analysis secondary structure, with the result depicted in Fig. 3. demonstrated both amino acids at the NH2 terminus. Yet, in The model depicts a mixed a/,3 secondary structure. Almost the complete sequence study neither the CNBr fragment nor one-third of the residues appear to be distributed in five a the tryptic peptide covering this region gave evidence of ty- helices and about a third in ,B sheets. The remainder are in ( turns, coils, or disordered structures, including one segment of § The nonapeptide also had a bluish color and appeared to bind 15 residues (residues 100-115) that has no clear a or (3 initiation copper. sites. The charged residues are concentrated at three sites: the Downloaded by guest on September 25, 2021 Biochemistry: Kingston et al. Proc. Natl. Acad. Sci. USA 76 (1979) 1671 fragments of ceruloplasmin. Probably this is its natural role, for it appears to be a separate domain structure of the intact mol- 6 7 ecule. Intersequence Comparison of Homology. We have pro- posed (8) that the cysteine-containing COOH-terminal se- quence of Cp F5 may be involved in binding one of the two type 1 cupric ions in ceruloplasmin because the sequence from Gly-128 through Val-150 shows similarity to the cysteine- containing COOH-terminal sequence of blue copper-con- taining azurins and plastocyanins, which contain a single type 1 copper ion. The cysteine, histidine, and methionine residues, which are invariant in the latter proteins and which have been implicated to be in the copper-binding site (8), could be matched by Cys-134, His-139, and Met-144 in Cp F5. This 84 proposal has since been strengthened by the publication of the results of x-ray crystallographic studies of plastocyanin (20) and azurin (21). In both cases, the copper atom appears to be bound to the cysteine and methionine residues which were matched (8) with Cys-134 and Met-144 in Cp F5 and by two histidine residues one of which was matched with His-139. In our pre- 108 116 120 123 dicted secondary structure shown in Fig. 3, the spacial ar- His Cys rangement of Cys-134, His-139, and Met-144 would give them 139 134 the potential to act as copper ligands. Furthermore, it is possible to interpret Chou and Fasman's rules for predicting secondary structure in a manner (L.-C. Lin, personal communication) that predicts that Cp F5 will be composed of two a-helical regions 144 146 152 at the NH2 terminus and six f-sheet strands separated into two Met groups of three by an intervening a-helical or random region; these elements could then fold up to produce a structure rather 159 similar to the cylindrical barrel shape formed largely by strands FIG. 3. Diagram of the secondary structure of Cp F5 as predicted of 3 sheet in poplar plastocyanin (20) and Pseudomonas by the method of Chou and Fasman (9, 10). Residues are represented aeruginosa azurin (21). in helical (A.), ,3-sheet (A), and coil (-) conformational states. Chain To search for any additional similarity to other proteins, we reversals denote /-turn tetrapeptides (::. requested the National Biomedical Research Foundation to undertake a computer search comparing the sequence of Cp F5 with all sequence data available, not only for the blue cop- NH2 terminus, the COOH terminus, and the two a helices and turn in the segment from Asp-34 through Lys-51. There are per-containing proteins, but also for all other proteins. With the several hydrophobic stretches, notably in the first helix from program RELATE (22), ceruloplasmin was compared to itself Phe-10 through Phe-17. The alternation of the histidines re- and to plastocyanin, azurin, and , by using lieves the hydrophobic character of the sheet beginning at a segment length of 15. In all cases the comparison of real se- Ile-129. quences gave a score less than 1 SD from the mean score with Noncovalent Interaction of Cp F5 with Other Fragments. 100 randomized sequences. Under the same conditions, azurin The several present in against plastocyanin gives a score of 3.9 SD. The ceruloplasmin fragments commercial preparations was rerun against itself by using a segment length of 25. The interact strongly, and harsh dissociating conditions are required score was 1.8 SD, which is interesting though not statistically to separate them. Thus, the blue color and the copper are lost significant (Winona C. Barker, personal communication). during the purification of Cp F5. The association is not due to However, as we had earlier identified by visual comparison of disulfide bonds because ceruloplasmin preparations that exhibit the segments from 1-53, 54-83, and 108-137, there is an indi- multiple bands in polyacrylamide gel electrophoresis in 0. 1% cation that the chain had duplicated. In the computer search, sodium dodecyl sulfate give similar patterns before and after 19 of the top 20 scores came from displacing the sequence by reduction although they sediment largely as single components 53 residues. with s2o = 6-7 S in analytical ultracentrifugation (8). The reason Two 25-residue segments of ceruloplasmin (Met-79 through for the strong interaction of Cp F5 with the 53,000 Mr and Gly-i1S, which has six histidioes, and Pro-127 through Leu-151, 67,000 Mr fragments is not evident from the primary structure. which contains the cysteine) were searched against the entire However, one clue is offered by the model of the secondary data collection of the Atlas of Protein Sequence and Structure structure given in Fig. 3; this illustrates the concentration of (95,365 comparisons). The highest scoring segments retrieved charged groups in certain segments of the structure, notably were from a variety of proteins. Nothing stood out significantly the NH2 terminus, the COOH terminus, and the two a helices from the rest. The cysteine-containing segment of the azurins and the ,B turn proposed for residues 34-55. Half the residues gave a good score, but the highest scores included a series of in the latter segment are either positively or negatively charged. ferredoxins and the a and.B globins of many species. The glo- If the largely hydrophobic structures from residues 55-144 are bins and Cp F5 have a high histidine content in common. The mainly in the interior of the molecule and the ionized residues ferredoxins are of interest because the highest scoring segment are at the surface, as would be expected from the general rules is at the COOH terminus of both Cp F5 and ferredoxin, and the for protein conformation (19), then the Cp F5 fragment would matching residues are Trp-130, Leu-132, and Cys-134, all close be capable of strong electrostatic interaction with the other to the site proposed for copper binding in Cp F5. Downloaded by guest on September 25, 2021 1672 Biochemistry: Kingston et al. Froc. Natl. Acad. Sci. USA 76 (1979) We thank J. Dwulet, A. Galen, J. Madison, and S. Dorwin for tech- 10. Chou, P. Y. & Fasman, G. D. (1978) Annu. Rev. Biochem. 47, nical assistance, Dr. J. Baenziger for carbohydrate analysis, and Dr. 251-276. W. C. Barker for computer analysis of the . This 11. Sgouris, J. T., Coryell, F. C., Gallick, M., Storey, R. W., McCall, work was supported by National Institutes of Health Grant AM K. B. & Anderson, H. D. (1962) Vox Sang. 7,394-405. 19221. 12. Ryden, L. & Eaker, D. (1974) Eur. J. Biochem. 44, 171-180. 13. Egorov, T. A., Svenson, A., Ryden, L. & Carlsson, J. (1975) Proc. NatI. Acad. Sci. USA 72,3029-3033. 14. Clamp, J. R. (1975) in The Plasma Proteins, ed. Putnam, F. W. 1. Poulik, M. D. & Weiss, M. L. (1975) in The Plasma Proteins, ed. (Academic, New York), 2nd Ed., Vol. 2, pp. 163-211. Putnam, F. W. (Academic, New York), 2nd Ed., Vol. 2, pp. 15. Aubert, J. P., Biserte, G. & Loucheux-Lefebvre, M. H. (1976) 51-108. Arch. Biochem. Biophys. 175,410-418. 2. Frieden, E. & Hsieh, H. S. (1976) Adv. Exp. Med. Biol. 74, 16. Huber, R., Deisenhofer, J., Colman, P. M., Matsushima, M. & 505-529. Palm, W. (1976) Nature (London) 264,415-420. 3. Gitlin, D. & Gitlin, J. D. (1975) in The Plasma Proteins, ed. 17. Heimburger, N., Haupt, H., Kranz, T. & Baudner, S. (1972) Putnam, F. W. (Academic, New York), 2nd Ed., Vol. 2, pp. Hoppe-Seyler's Z. Physiol. Chem. 353, 1133-1140. 321-374. 18. Peters, T., Jr. (1975) in The Plasma Proteins, ed. Putnam, F. W. 4. Simons, K. & Bearn, A. G. (1969) Biochim. Blophys. Acta 175, (Academic, New York), 2nd Ed., Vol. 1, pp. 133-181. 260-270. 19. Sternberg, M. J. E. & Thornton, J. M. (1978) Nature (London) 5. Freeman, S. & Daniel, E. (1973) Biochemistry 12,4806-4810. 271, 15-20. 6. McCombs, M. L. & Bowman, B. H. (1976) Biochim. Biophys. 20. Colman, P. M., Freeman, H. C., Guss, J. M., Murata, M., Norris, Acta 434,452-461. V. A., Ramshaw, J. A. M. & Venkatappa, M. P. (1978) Nature 7. Ryden, L. (1972) Eur. J. Biochem. 26,380-386. (London) 272, 319-324. 8. Kingston, I. B., Kingston, B. L. & Putnam, F. W. (1977) Proc. 21. Adman, E. T., Stenkamp, R. E., Sieker, L. C. & Jensen, L. H. Natl. Acad. Sc$. USA 74,5377-5381. (1978) J. Mol. Biol. 123,35-47. 9. Chou, P. Y. & Fasman, G. D. (1974) Biochemistry 13, 211- 22. Barker, W. C., Ketcham, L. K. & Dayhoff, M. 0. (1978) J. Mol. 245. Evol. 10, 265-281. Downloaded by guest on September 25, 2021