post-translational heterogeneity of the human vitamin d-binding protein (group-specific component)

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 226, No. 1, October 1, pp. 218-223, 1983

Post-Translational Heterogeneity of the Human Vitamin D-Binding Protein (Group-Specific Component)

DORIAN H. COPPENHAVER,* NICHOLAS P. SOLLENNE,* AND BARBARA H. BOWMAN?’

*University of Texas Medical Branch, Galveston, Texas 77550, and ~University of Texas Health Science Center at San Antonio, San Antonio, Texas 7828.4

Received March 9, 1983, and in revised form June 6, 1983

The vitamin D-binding protein in human serum (the group-specific component) is an az-globulin which is genetically polymorphic in all populations studied. Previous work (J. Svasti and B. H. Bowman (1978) J. Biol. Chem. 253, 5188-5194, and J. Svasti, A. Kurosky, A. Bennett, and B. H. Bowman (1979) Biochemistry l&1611-1617) has shown that the electrophoretic variations of the proteins controlled by two allelic genes, Gc’ and Gc2, are due to at least three amino acid substitutions between Gel and Gc2 (Svasti et al. (1979)) and to heterogeneity in the Gel phenotype arising from carbohydrate dissimilarities. Gel migrates electrophoretically as two protein bands, while Gc2 migrates cathodally as a single band. This study demonstrates a post-translational glycosylation difference occurring in a single area of the Gel sequence which accounts for the heterogeneity observed previously. The glycosylation site, a threonine residue, appears to be in a sequence which differs between Gel and Gc2. The 0-glycosidic bond, which is typical of mucins, is rare in plasma proteins. The cyanogen bromide fragment containing the galactosamine-containing carbohydrate in Gel was partially sequenced through 20 residues from the amino terminus. No detectable galactosamine could be found in the homologous cyanogen bromide fragment in Gc2. A new purification procedure for the vitamin D-binding protein in human plasma has been developed. Three chromatographic steps provide purified protein.

Vitamin D and its metabolites are transported in human blood by a genetically polymorphic plasma protein, the group-specific component (Cc).’ The common allele, Gel, produces two distinct protein electromorphs, Gclanoda’ and Gclcathoda’, after alkaline electrophoresis, while the rarer GC2 allele produces a single, less anodal band under the same conditions (3). Subtypes of the Gc’ allele have been de-

1 To whom correspondence should be addressed. The University of Texas Health Science Center at San Antonio, 7’703 Floyd Curl Drive, San Antonio, Tex. 78284.

’ Abbreviations used: Gc, group-specific component; VDBP, vitamin D-binding protein; TFA, trifluoroac- etic acid.

scribed, the products of which are only dis- tinguishable after prolonged isoelectric focusing (4). Our laboratory has studied the molecular basis for this polymorphism (1,2) and the distribution of Gc phenotypes in normal children and cystic fibrosis pa- tients (5). In a previous report, Svasti and Bowman (1) established that the electrophoretic difference between the “fast” and “slow” gel electromorphs is due to the presence of sialic acid in the more anodal, Gel a”oda’ molecules but missing in the Gel cathoda’ molecules. Svasti et aZ. (2) also reported that, in addition to differences in primary structure, there appeared to be differences in carbohydrate content between the Gel and Gc2 forms. On the basis

0003-9861/83 $3.00 218 Copyright 0 1983 by Academic Press, Inc. All rights of reproduction in any form reserved.

CARBOHYDRATE DIFFERENCES IN GENETIC TYPES OF HUMAN PROTEINS 219

of indirect evidence it was concluded that an 0-glycosidically linked oligosaccharide was present in Gel which was absent in Gc2 (2). The present work confirms the presence of N-acetylgalactosamine in one cyanogen bromide fragment of Gel.

MATERIALS AND METHODS

Materials. Reagents used in automatic sequence analysis were purchased from Pierce Chemical Com- pany. Dithiothreitol, iodoacetic acid, cyanogen bromide, and sodium borohydride were from Sigma Chemical Company. Anti-Gc globulin was a product of Atlantic Antibodies. Sepharose 6B was from Phar- macia Fine Chemicals; Affi-Gel blue and Bio-Gel were from Bio-Rad Laboratories. Guanidine-hydrochloride was purchased as ultrapure grade from Schwarz/ Mann Biochemicals. Iminodiacetic acid disodium salt, 1,4-butanediol diglycidyl ether, and CUSO~ * 5HzO (Gold Label) were from Aldrich Chemical Company. Prepared isoelectric focusing plates (pH 4-6.5) were from LKB. Tetrahydrofuran, acetonitrile, and all high- performance liquid chromatography grade solvents were from Burdick and Jackson. All other chemicals used were reagent grade or better.

Plasma fractions. Single-donor plasmapheresis samples were kindly provided by Dr. D. R. Barnett, University of Texas Health Science Center at San Antonio. Plasma from one Gc? homozygote and one Gc’ (subtype Gcis’is) homozygote were used through- out this investigation. Plasma samples were typed as previously described (4,5). A 35-60% ammonium sulfate precipitate was prepared from whole plasma before further processing.

Cibacron Blue FSGA-Agarose chromatography. Ammonium sulfate fractions (21 to 28 g) were dialyzed against the chromatography buffer (0.01 M KzHPO, 0.15 M NaCl, pH 7.3) prior to chromatography on Ci- bacron Blue FBGA coupled to an agarose support (Affi- Gel blue). The Gc-containing pool (3.5 to 4 g) from 100 ml of plasma was applied to a 2.5 X 60-cm column of Affi-Gel blue. The column was eluted with the same buffer at 4°C; the flow rate was 40 ml/h. After elution of the unbound material the buffer was changed to include 1.5 M NaCl to remove the strongly bound serum albumin from the column (6). The Gc-containing peak was located immunologically, concentrated by ultra- filtration, and dialyzed against 0.05 M Tris-HCI, 0.15 M NaCl, pH 8.0.

Metal-chelate chromatography. Metal-chelate affinity resin was prepared by epoxy-activating Sepharose 6B as described by Porath and colleagues (7,8). A 2.5 X 90-cm column of the biscarboxymethyl amino agarose was charged with Cu2+ by passing a solution of CuSOd*5Hs0 (2 mg/ml) through the column. After equilibration with 0.05 M Tris-HCl, 0.15 M NaCl, pH 8.0, the Gc-containing pools (1.5 to 2.0 g) from five to

six Affi-Gel blue columns were applied to the column. The column was developed at room temperature with a stepwise gradient of (a) equilibration buffer, (b) 0.1 M sodium acetate, 0.8 M NaCl, pH 4.5, and (c) 0.05 M EDTA, 0.5 M NaCl, pH 7.0 (8). The Gc-containing peak was concentrated and dialzyed against equilibration buffer before application to a second copper chelate column. This 1.5 X 30-cm column was developed with the same equilibration buffer.

Preparation of peptide fragments. Purified Gc preparations (30 to 50 mg) were carbamylmethylated as previously described (2). Cyanogen bromide peptides were prepared (9) and fractionated on a 2.5 X 120- cm gel-filtration column (Bio-Gel P-10) developed at room temperature in 22% HCOOH. Peptides of interest were rechromatographed on a second gel-filtration column (1.5 X 60 cm, Bio-Gel P-6) in the same developer. Final purification of the peptides of interest was verified by reverse-phase HPLC. Samples were applied to a 4.1 X 25-cm large-pore Cl8 reverse-phase column (RP-P, Synchrom, Inc.) and eluted with a linear gradient of 0.1% trifluoracetic acid (TFA) in water to 60% 1-propanol, 0.1% TFA in water (10, 11). The purified cyanogen bromide peptides eluted as a single symmetrical peaks under these conditions.

Sequence analysis. Amino acid analysis of purified peptides was determined using single-column meth- odologies employing Durrum DC-GA resin on Beckman 121M and 119 analysis. Automated sequence analysis was performed on an updated Beckman 890B sequenator using the dilute Quadrol program (Beckman No. 011576) (12). Details of sequenator operation have been given previously (13). Sequenator products were converted to phenylthiohydantoins and identified by HPLC and/or by back-hydrolysis followed by amino acid analysis (11, 12, 14).

RESULTS

Purification of vitamin LMinding protein. Vitamin D-binding protein (VDBP) was purified after ammonium sulfate pre- cipitation of whole plasma using a com- bination of dye-mediated affinity and metal-chelate chromatography. Chroma- tography of the ammonium sulfate precipitate on Cibacron Blue F3GA-agarose (6) resolved the VDBP from albumin. Fig- ure 1 is an elution pattern of a typical Affi- Gel blue column showing the elutions of albumin (fractions 135 to 270) and Vitamin D-binding protein (fractions 54 to 65).

Further purification by a copper chelate column is shown in Fig. 2. The VDBP eluted in fractions 38 to 75. Upon rechromatog- raphy on a second copper chelate column

COPPENHAVER, SOLLENNE, AND BOWMAN

3.0

2.0

8

2

1.0

L

P

1.5MNaCI

I - 50 100 150

FRACTION NO.

FIG. 1. AR-Gel blue chromatograpy. The 3560% ammonium sulfate cut from 100 ml of plasma from a Gel-l donor was dialyzed against PBS and applied to a 2.5 X 60-cm column of Affi-Gel blue. The column was developed with phosphate-buffered saline at 4’C, the flow rate was 40 ml/h, Fractions of 8 ml were collected, the buffer was changed at the arrow to include 1.5 M NaCI. The area under the bar was pooled for further purification.

(Fig. 3), the vitamin D-binding protein was purified to homogeneity (fractions 40 to 60). The purity of this fraction was established by immunoelectrophoresis, isoelectric focusing, and HPLC. The two genetic types of vitamin D-binding proteins, Gel and

I f-w 50 100 200 300

FRACTION NO.

FIG. 2. Metal-chelate chromatography. The Gc-containing peaks from five AI&Gel blue columns were concentrated and dialyzed against 0.05 M Tris-HCl, 0.15 M NaCl, pH 8.0, and applied to a 2.5 X %cm column of epoxy-activated Sepharose 4B charged with Cu*‘, equilibrated in the same buffer. The column was developed at room temperature with a stepwise gradient of (A) equilibration buffer, (9) 0.1 M Na acetate, 0.8 M NaCl, pH 4.5, and (C) 0.05 M EDTA, 0.5 M NaCl, pH 7.0. The Gc-containing peak is shown by the bar.

0.5

0.4

8 0.3

q

0.2

0.1

L I;

J L 60

-

&TION NO.

FIG. 3. Second metal-chelate chromatography. The Gc-containing fractions from the copper chelate column were concentrated and applied to a second column containing the same resin. This 1.5 X 30-cm column was developed with 0.05 M Tris-HCl, pH 8.0, in normal saline. The second peak to emerge from this column contained pure Gel as determined by isoelectric focusing.

Gc2, gave similar elution results on the columns used for purification.

Cyanogen bromide fragments. After treatment with cyanogen bromide, the fragments from Gel and Gc2 preparations were separated by gel filtration on Bio-Gel P-10. The elution of approximately six to seven cyanogen bromide fragments can be seen in Fig. 4. This is in agreement with the predicted number of fragments based on the presence of five methionine residues in the amino acid composition (15). Amino acid composition of each cyanogen bromide fragment demonstrated that only one fragment of Gel (shown by bar in Fig. 4) contained N-acetylgalactosamine. Previous studies (2) had indicated that peptide (lsN5) of Gel contained three amino acid substitutions when compared to the ho-

i0 lb0 l&O

FRACTION NO.

FIG. 4. Cyanogen bromide peptide isolation. The cyanogen bromide digest of 30 mg of Scarboxy- methylated Gel was applied to a 2.5 X 120-cm Bio- Gel P-10 gel-filtration column and developed with 22% HCOOH at room temperature. The peptide designated CB III is marked by the dark bar.


mologous Gc2 peptide. This peptide also contained sialic acid which was missing in the homologous region of Gc2 (2). The only candidate for glycosylation in this peptide was a threonine residue. Therefore, an O- glycosidic linkage between the carbohydrate and threonine was suspected. The Gel cyanogen bromide fragment containing N-acetylgalactosamine (Gel-CBIII) was further purified by gel filtration on Bio-Gel P-6 (Fig. 5). The purification of CBIII was confirmed after analysis by reverse-phase HPLC (Fig. 6). The analogous cyanogen bromide fragment from Gc2 (Gc2-CBIII) was identified by its amino acid composition. The presence of galactosamine and glucosamine was determined by amino acid analysis.

Amino acid compositicm and sequence. The cyanogen bromide fragments from Gel (Gel-CBIII) and Gc2 (Gc2CBIII) were proved homologous by amino acid composition and sequence analysis. Amino acid analysis demonstrated that the amino acid composition of fragments from Gel and Gc2 could not be distinguished, although Gel-CBIII contained four residues of glucosamine and three residues of galactosamine, while Gc2CBIII did not contain any hexosamine. Table I is the amino acid composition of Gel-CBIII and Gc2CBIII. Al- though at least three amino acid differences between Gel and Gc2 have been characterized by previous studies (2), these are not evident in the amino acid composition of CBIII fragments. This was not surprising since two of the amino acid differences are reciprocals, Thr (Gel) - Lys (Gc2) and Lys (Gel) - Thr (Gc2). The

03

oo2 Ei

q 01

t 50 100 150

FRACTION NO.

FIG. 5. Gel filtration on Bio-Gel P-6. Peptide CB III was rechromatographed on a 1.5 X 60-cm column of Bio-Gel P-6 under the same conditions. The major peak was lyophilized prior to HPLC purification.

FIG. 6. Reverse-phase HPLC. The final purification of CB III was verified by reverse-phase HPLC on a 4.1 X 25-cm RP-P column. The column was developed with a linear gradient of 0.1% TFA in Hz0 (pump A) to 60% 1-propanol, 0.1% TFA in Hz0 (pump B). The sharply rising baseline is characteristic of propanol gradients at this wavelength.

amino terminal sequences of Gel-CBIII and Gc2CBIII were identical in the first 20 residues, attesting to their homology (Table II).

DISCUSSION

We previously reported that human vitamin D-binding protein could be purified

TABLE I

AMINO ACID COMPOSITION OF CB III FROM Gel AND Gc2

Gel Gc2

Asp 3.1 3.0 Thr 3.7 3.7 Ser 4.1 4.2 GlU 4.2 3.9 Pro 2.7 3.4 GUY 2.8 2.8 Ala 2.5 1.9 % -cys 1.1 + Val 1.9 2.4 Met 1.0 1.0 Ile 0.9 1.2 Leu 3.3 3.6 ‘M 2.2 2.5 Phe 1.0 0.7 His 0.5 + LYS 1.6 1.3 Arg 1.2 1.1 Glu * NHz 3.9 -o- Gal * NH2 2.8 -o-

Note. Compositions standardized relative to methionine (as homoserine + methionine) = 1.0 mol/ mol.

222 COPPENHAVER, SOLLENNE, AND BOWMAN

TABLE II

QUANTITATION-SEQUENCE ANALYSIS

Round No.

Gel-CBIII Gc2-CBIII

Residue nmol Residue nmol

0 (50) 1 Trp (39) Tw 2 Glu (42) Glu 3 5r (36) 5r 4 Ser (14) Ser 5 Thr (14) Thr 6 Asn (16) Asn 7 Tyr (17) Tv 8 GUY (14) GUY 9 Gln cw Gln

10 Ala (1% Ala 11 Pro cw Pro 12 Leu (16) Leu 13 Ser (6) Ser 14 Leu (9) Leu 15 Leu (10) Leu 16 Val (81 Val 17 Ser (3) Ser 18 Tyr (5) Tyr 19 ND ND 20 Asn (2) Asn

Note. Round 0 = nmol in cup.

(40) (30) (32) (26) (32) (18) (31) (2% (23) WY cw (6)

vv (4)

W-3 (20) (1‘3 (4) (3)

(8)

using a series of gel-filtration and ion-ex- change chromatographic steps (1, 2). The procedure used in the present report rep- resents a considerable improvement upon our former methods. Chromatography on Cibacron Blue FBG-A-agarose is known to be an efficient means of removing serum albumin from mixtures of proteins (6). Re- sults shown in Fig. 1 indicate that human VDBP also shows a modest affinity for the dye ligand, since it was slightly retained under the chromatographic conditions em- ployed here. The nature of this interaction is not known, although it presumably in- volves a slight ionic or hydrophobic interaction with the immobilized triazinyl dye. With repeated use columns of Al&Gel blue markedly lose binding capacity, due to the build up of tightly bound lipids from the plasma.

Since its introduction by Porath and colleagues (‘7, 8) metal-chelate affinity chromatography has successfully been used in

the purification of a number of proteins, inducing az-SH glycoprotein (16), fibroblast interferon (17), and al-antitrypsin (18). Of the metal chelates tested, Cu2+ is reported to be by far the strongest adsorbent (8), with irreversible binding reported for some proteins (19). Copper chelates are thought to act primarily through interaction with the available side chains of histidine and cysteine residues (8). Thus it is of considerable interest that human VDBP is not retained by copper or zinc chelate columns (data not shown). The only other plasma protein for which this lack of interaction has been reported is serum albumin (8). Therefore, the use of copper chelate affinity chromatography allows a rapid and efficient purification of VDBP from serum albumin depleted plasma, with an overall yield of 25-30%.

The molecular basis for the genetic (Gel/ Gc2) polymorphism of the VDBP has been studied and shown to involve at least three amino acid substitutions in the primary structure (2). The heterogeneity in Gel, however, was found to be due to the presence of more sialic acid in Gclanodal than in GclCathodal (1). The presence of sialic acid in the anodal Gel electromorph accounts for the increased migration of this species relative to the cathodal product of the Gc’ allele. This difference resides in a single tryptic peptide when digests of the anodal and cathodal electromorphs of Gel are compared, with the sialic acid detectable only in the peptide isolated from the anodal electromorph (1,2). It was of interest that this peptide (IfN6, in Ref. (2)) contained no asparagine, however. In the absence of asparagine the N-acetylglucosaminyl-as- paraginyl linkage, which predominates in plasma glycoproteins, could not occur. Thus, the carbohydrate linkage in the sialic acid-containing Gclanoda’ peptide was predicted to be N-acetylgalactosaminyl- threonyl. This study substantiates the presence of N-acetylgalactosamine in a cyanogen bromide fragment of Gel but absence in the homologous fragment of Gc2. Galactosamine could not be identified in any other cyanogen bromide fragment in Gel nor in any fragment in Gc2. It appears, therefore, that one genetic alteration in


Gel compared to Gc2 (Lys - Thr) has led to the heterogeneity in Gel and the ap- pearance of two electrophoretic bands, one of which had lost some of its sialic acid complement.

We have reported sialic acid concentra- tions in Gclanoda’ (1.16%) which were greater than in Gc2 (0.57%) (1). In this study we found glucosamine present in some Gel and Gc2 cyanogen bromide fragments (not shown), indicating that Gc2 has some carbohydrate associated with it, although less sialic acid than Gel. The presence or absence of sialic acid does not ap- pear to interfere with vitamin D binding, since digestion with neuraminidase did not alter the capacity of the group-specific component to bind vitamin D (1). Galac- tosamine, however, was found only in the single cyanogen bromide peptide (CBIII) isolated from Gel. Our evidence thus in- dicates that post-translational modifications contribute to the structural heterogeneity in the VDBP system, and that these modifications are in part consequent to genetically determined differences in the primary structures of the gene products of the polymorphic alleles.

In conclusion, there are three common genetic alleles at the Gc locus, Gcl’, GoiF, and Gs. GclS and GclF have similar products which each migrate electrophoreti- tally as two bands. The two bands are com- posed of identical polypeptides, but the fast anodal band has retained its full complement of sialic acid, while the slow band has lost some or all of its sialic acid. The product of Gc2 differs from both GclF and Gc” in at least three amino acids. One of the amino acid substitutions is a change of Thr to Lys in the G2 gene product elim- inating the glycosylation site. Since the Gc2 gene product has no sialic acid at this site, it migrates as a single band in electrophoresis. The site of the genetic and bio- chemical differences appears to lie in a tryptic peptide lfN6 (2) described earlier. This peptide is located within CBIII and

apparently occurs C-terminal to the sequence given in Table I.

ACKNOWLEDGMENTS

We thank Billy Touchstone and Peter van Bragt for technical help and Betty Russell for help with the manuscript. We are grateful to Dr. Don R. Barnett for the source of Gel and Gc2. This work was sup- ported in part by NIH Grants HD16584, CA17701, and AM3132.

1.

2.

3. 4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15. 16. 17.

18.

19.

REFERENCES

SVASTI, J., AND BOWMAN, B. H. (1978) .Z. Bid Chem. 253,5188-5194.

SVASTI, J., KUROSKY, A., BENNETT, A., AND Bow- MAN, B. H. (1979) Biochemists l&1611-1617.

CLEVE, H. (1973) Israel J. Afed Sci 9.1X3-1146. CONSTANS, J., AND VIAU, M. (1977) Science 198.

1070-1071. COPPENHAVER, D., KUEPPERS, F., SCHIDLOW, D.,

BEE, D., ISENBERG, J. N., BARNEY, D. R., AND BOWMAN, B. H. (1981) Hum Genet. 57,399-403.

TRAVIS, J., BOWEN, J., TEWKSBURY, D., JOHNSON, D., AND PANNELL, R. (1976) Biochem .I 157, 301.

LUNDBERG, L., AND PORATH, J. (1975) J. Chrvmatog. 90.87-98.

PORATH, J., CARLSSON, J., OLSSON, I., AND BEL- FRAGE, G. (1975) Nature @ondon) 253,598-599.

NUTE, P. E., AND SULLIVAN, B. (1971) Camp. Biochem PhysioL B 39, 797-814.

MAHONEY, W. C., AND HERMODSON, M. A. (1980) .Z. BioL Chem 255, 11199-11203.

COPPENHAVER, D. H., DIXON, J. D., AND DUFFY, L. K. (1933) Hemq$obin 7, 1-14.

DUFFY, L. K., PETERSON, J. W., ANDKUROSKY, A. (1981) .Z. BioL Chem 256,12252-12256.

KUROSKY, A., KIM, H.-H., AND TOUCHSTONE, B. (1976) Camp. Biochem PhysioL B S&453-459.

LAI, C.-Y. (1977) in Methods in Enzymology (Hirs, C. H. W., and Timasheff, S. N., eds.), Vol. 47, pp. 236-243, Academic Press, New York.

BOWMAN, B. H. (1969) Biochemistry 6,4327-4335. LEBRETON, J. P. (1977) FEBS Z.&t. 80,351-354. EDY, V. G., BILLIAU, A., AND DESOMER, P. (1977)

J. Biol Chem 252, 5934-5935. KURECKI, T., KRESS, L. F., AND LASKOWSKI, M.,

SR. (1979) AnaL Biochem 99.415-420. CHADA, K. C., GROB, P. M., MIKULSKI, A. J., DAVIS,

L. R., AND SULKOWSKI, E. (1979) .Z. Gen ViroL 43, 701-706.

post-translational heterogeneity of the human vitamin d-binding protein (group-specific component)

Documents