The nonenzymatic glycosylation of collagen

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 221, No. 2, March, pp. 428-43’7, 1983

The SARAH Department York

10&l,

ROGOZINSKI,**’

of Bidwmistry, and

Nonenzymatic

*Department

Received

OLGA

Albert Einstein of Biophysics,

August

Glycosylation

of Collagen

0. BLUMENFELD, Cdlege

The

of Medicine,

Weizmann

25, 1982, and in revised

AND

1300

Institute

form

SAM SEIFTER

Morris Park Avenue, Bronx, of science, P.O.B. 26, Rehovot,

November

New Israel

9, 1982

Calf skin acid-soluble collagen, containing about 34 residues of lysine plus hydroxylysine per 100,000 dalton polypeptide chain, was treated with [14Clglucose in the presence or absence of NaCNBH,. In 144 h, under the conditions employed, the presence of NaCNBH3 increased the extent of glycosylation from 8 to 15% of the total residues of lysine plus hydroxylysine. The extent of glycosylation was estimated, using acid hydrolysates of the protein, by isolation and determination of reduced adducts (l-lysinohexitols) employing a system of paper chromatography followed by chromatography on an amino acid analyzer. By those means the difficulties of using specific color reactions such as that with thiobarbituric acid were obviated. Identification of the reduced adducts as forms of 1-lysinohexitol was made by comparison of that substance prepared by treatment of polylysine with [‘4Clglucose in the presence of NaCNBHs. Of interest is the fact that treatment of the polymer with glucose for 144 h under conditions similar to those used for the collagen, resulted in an increase of extent of glycosylation from 3 to about 50% of the total lysine residues when NaCNBH3 was present in the incubation medium. The greater degree of glycosylation of lysine residues in polylysine as compared with collagen (15 versus about 50%) may be ascribed to the different orders of macromolecular structure in the protein that could sequester certain of the residues from reaction with glucose. 1-Lysinohexitol was also identified in hydrolysates of neutral salt-soluble guinea pig skin collagen that had been reacted with glucose and then treated with NaB3H4. The glycosylated collagens were fragmented by treatment with CNBr, and modified lysine residues were found to occur along the entire length of the collagen chains. The use of NaCNBH3 in the manner described above permits measurement of both aldimine and ketoamine forms of the adducts made with glucose. The possible physiological significance of the reversibility of the ketoamine form of adduct is discussed briefly.

The nonenzymatic glycosylation of proteins and its chemical consequences have been known for a long time in relation to the so-called Maillard and Browning reactions (1). Much interest has been generated in the physiological concomitants of such reactions between tissue proteins and glucose present in the blood in normal

ambient concentrations (2); presently even more attention is being paid to the possible pathological effects of such reactions when glucose occurs in abnormally high concentrations in tissues of diabetic persons (3). The study received great stimulus from the demonstration (4) that hemoglobin Ale, a normal hemoglobin containing a borohydride-reducible group (5), is identical with the hemoglobin found to be increased in uncontrolled diabetic persons (6), and from the finding that hemoglobin Al, is in fact a hexosyl derivative of he-

1 Author to whom all correspondence should be addressed: Department of Biochemistry, Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, N. Y. 10461. 0003-9861/83/040428-10$03.00/O Copyright All rights

0 1’33.9 by Academic Press, Inc. of reproduction in any form reserved.

428

NONENZYMATIC

GLYCOSYLATION

moglobin A (7). The specific hexose later was shown to be glucose (8). Changes in the oxygenation curve of hemoglobin Al, when compared with hemoglobin A have been demonstrated (9), but no real pathology has been shown to occur with the changes. In recent years measurement of hemoglobin Al, by use of several different methods has proved of considerable value as an indicator of control in diabetic persons (10,ll). Several excellent reviews have been made of the reactions of glucose with hemoglobin and other proteins (12-14). Although hemoglobin is a special case in that it is a protein locked in a cell for about 120 days exposed to intracellular concentrations of glucose approximating those of plasma, the extracellular and pericellular proteins that have relatively slow turnover rates perhaps should be especially susceptible subjects for glycosylation. Even before the description of the nonenzymatic glycosylation of hemoglobin, Bensusan (15) performed experiments to test the nonenzymatic glycosylation of collagen, and indeed postulated all of the currently used schemes of Schiff base (aldimine) formation, Amadori rearrangement, and possible crosslinking between the rearranged protein-ketoamine and another molecule of protein. The basic chemistry of the reaction of glucose with a protein, either through (Yamino groups of terminal residues or Eamino groups of internal lysine and hydroxylysine residues, is as follows. First an aldimine function forms; this is unstable primarily by virtue of its reversal to glucose and the unmodified amino acid residue. That reaction is relatively rapid (16). The aldimine also can participate in an Amadori rearrangement to form a stable protein-bound ketoamine (8). The rearrangement reaction is relatively slow. The method chosen to quantify extent of glycosylation can be critical both in experimental and clinical studies (17). Reduction with sodium borohydride, or its tritium form, can be used to measure both aldimine and Amadori-rearranged products (8). In contrast, sodium cyanoborohydride, at neutral pH values, will reduce the aldimine but not the ketoamine (18); also

OF

429

COLLAGEN

it will not reduce glucose whereas sodium borohydride will (19). The present paper is concerned with a determination of the total glycosylated protein, in both aldimine and ketoamine forms, resulting from in vitro incubation of collagen with glucose. This has been achieved by incubation in the presence of sodium cyanoborohydride, as used previously with albumin (20) and hemoglobin (19). Similar experiments were performed with polylysine; and the large degree of glycosylation obtained with the polymer enabled us to study more readily the nature and rate of modification of the labile adduct. EXPERIMENTAL

PROCEDURES

Materials Acid-soluble calf skin collagen (Lot YOF-8025 NO. C-3511) and NaCNBHa were purchased from Sigma, St. Louis, Missouri. Guinea pig skin salt-soluble collagen was a gift from Dr. S. Takahashi, Department of Biochemistry, Albert Einstein College of Medicine (Bronx, N. Y.). Poly-L-lysine hydrobromide was generously provided by Mr. J. Jacobson, Department of Biophysics, The Weizmann Institute of Science (Rehovoth, Israel). n-[U-“C]Glucose, sp act 269.7 Ci/ mmol, was obtained from Amersham (Arlington Heights, Ill.). NaBaH, (sp act 200 Ci/mol was purchased from New England Nuclear, Boston, Mass.). Dulbecco’s phosphate-buffered saline, without calcium and magnesium, was obtained from Grand Island Biological Company, Grand Island, New York, as was the penicillin-streptomycin mixture. Bio-Gel, P-6, 200-400 mesh, was obtained from Bio-Rad Lab.

Methods Incubation

of Collagen with Glucose

Acid-soluble calf skin collagen was dissolved in 0.5 M acetic acid (1.7 mg/ml). One-milliliter aliquots were distributed into 15-ml sterile disposable borosilicate centrifuge tubes (Falcon) and lyophilized. An amount of [U-‘%Jglucose (0.13 mCi) was added to 12 ml of a 44 mM glucose solution in sterile Dulbecco phosphate-buffered saline, pH 7.4. The purity of the &J-‘4Clglucose was verified by chromatography for 48 h on Whatman No. 1 using the Fisher-Nebel system (21) of ethyl acetate:pyridine:HaO:acetic acid 5:5:3:1, followed by radioscanning using a Packard Model 7201 Radiochromatogram Scanner. A single radioactive symmetrical peak was obtained. The final specific activity of glucose was 0.55 X lo6 dpm/pmol.

430

ROGOZINSKI,

BLIJMENFELD,

One milliliter of this solution was added to the lyophilized collagen followed by either 0.1 ml of a 440 mM solution of NaCNBHs in Dulbecco buffer (reduced samples) or 0.1 ml of Dulbecco buffer (controls). The final concentration of glucose and of NaCNBHs was 40 mbf. The solutions were sterilized by irradiation with ‘“Cs for 40-50 min at 7000 rad and then incubated at 37°C with constant shaking for 4.5, 29, or 144 h. In other experiments, larger amounts of acid-soluble collagen or neutral salt-soluble guinea pig collagen were incubated under sterile conditions with [U-‘“Clglucose as follows: calf skin collagen, 4 mg/ml with 40 mM [U-i4Clglucose of sp act 0.26 X 10s dpm/ nmol for 12 days at 37°C; and guinea pig collagen, 2 mg/ml with 40 mM [U-“Clglucose, sp act 0.35 X lo6 dpm/pmol, for 15 days at 37°C.

Purification

of Glycosyluted

Collagwns

Acid-soluble calfskin collaga. At the end of incubation, 4 ml of water was added to each tube and the solutions centrifuged in a Sorvall GLC centrifuge at 3000 rpm for 10 min at room temperature. The clear supernatant was withdrawn and the insoluble collagen pellet washed in a similar manner with 7 ml of water usually five to six times until the radioactivity in the supernatant was negligible. We have examined only the collagen that remained insoluble under the conditions of incubation and subsequent washing. In most experiments, the collagen pellet was subjected to cleavage by addition of 0.8 ml of a solution of CNBr in formic acid (310 mg/ml of 75% formic acid) (22). The solution was stirred under nitrogen for 35 min at room temperature, and the reaction terminated by addition of 50 ml of water; this was followed by lyophilization. Both intact glycosylated collagen and the CNBrtreated material were solubilized in 0.5 M acetic acid and purified further by gel filtration on Bio-Gel Pe using 0.5 M acetic acid as eluant. The recovery of collagen in the excluded Bio-Gel Ps pool, measured by determination of hydroxyproline content, was approximately 50% of the amount subjected to the reaction and gel filtration sequence. Neutral salt-soluble guinea pig skin wllu~en. At the end of the incubation period, the solution was dialyzed exhaustively against distilled water at 4”C, treated with CNBr, and further freed of unreacted glucose by gel filtration using Bio-Gel Ps as above. Incubation of polylysine with glucose and purification of glycasylated polylysine. One milliliter of polylysine solution (12.4 am01 of lysine equivalents/ml water) was distributed into disposable borosilicate centrifuge tubes and lyophilized. One milliliter of the 44 mM [U-r4Clglucose solution (see above) was then added followed by either NaCNBHs or phosphatebuffered saline, as described above. Sterilization and

AND

SEIFTER

incubation of all samples were carried out as described for the collagens. The polylysine used in this study proved to be polydisperse. This was shown by the fact that about 60% was lost on dialysis. In fact, analyses were performed on retentates after dialysis; in the case of control polylysine, the retentates represented about 40% of the material. When polylysine was incubated with glucose in the presence of NaCNBHs (for 144 h), the retentate after dialysis represented about 74% of the original material. That higher value, as compared to the recovery of control polylysine, may indicate that glycosylation of polylysine causes an increased size of the polymer. One may note that indeed the reduced glycosylated polylysine formed a gel that dissolved only after acidification with 0.5 M acetic acid. In any event, samples were dialyzed exhaustively against water, and retentates were purified by gel filtration using Bio-Gel Ps as described for collagen; 0.5 M acetic acid was used as eluant. Reduction of 4C-glycml/lated wllagens and polylySinefreecE of excess [“‘CJglumse. In some experiments, in which larger quantities of the collagens were incubated with r’C]glucose for 12 or 15 days, the samples were freed of excess [“Clglucose by washing or dialysis, and then reduced with either NaCNBHs or NaBH4 for 2 h at room temperature. The reaction was terminated by acidification with acetic acid, and excess reagent removed by washing (in the case of calf skin collagen) or dialysis (guinea pig skin collagen) as described above. For preparation and identification of 1-[i4C]lysinohexitol, an amount of polylysine * HBr equivalent to 0.5 mmol lysine was incubated with 1 mmol [14Clglucose, having a sp act of 4 X 10’ dpm/pmol glucose, in 2.0 ml of Dulbecco phosphate buffer, pH 7.8, at 37°C for 9 days. Sterilization was achieved by addition of penicillin and streptomycin. Excess glucose was removed by extensive dialysis and gel filtration. Following dialysis, the glycosylated polylysine was reduced with 140 mM NaBH, for 3 h at room temperature. After acidification, borate was removed by evaporation with methanol, and the reduced glycosylated polylysine purified by gel filtration on BioGel Ps as described. Polylysine was also glycosylated with unlabeled glucose and reduced with NaBsHI for 5 min with 10 mCi (200 Ci/mol) followed by treatment with unlabeled NaBH, as described above. Identification of l-lysinohexitol. “C- or SH-labeled glycosylated polylysine was used as a source of llysinohexitol. Following reduction and Bio-Gel Ps chromatography, the sample was hydrolyzed in constant boiling HCl for 24 h at 105°C under vacuum, and subjected to paper chromatography on Whatman No. 1 strips using n-butyl alcohokacetic acidwater (4:1:5) for 100 h at room temperature (23). Paper chromatograms were dipped in ninhydrin-acetone to visualize adducts and amino acids. The radioactivity

NONENZYMATIC

GLYCOSYLATION

in unstained paper chromatograms was determined by cutting l-cm consecutive strips, eluting with Hz0 and counting a portion of the eluate by liquid scintillation. Radioactive materials eluted from the paper chromatograms were pooled and subjected to ionexchange chromatography using a Beckman Model 119 PL amino acid analyzer employing a split-stream device; [%I]leucine was used as internal standard. Estimation of spec$ic activity of glycosylated wl&errs and polylysine. The contents of radioactive fractions (see Fig. 4) from Bio-Gel Ps chromatography were pooled, lyophilized, and reconstituted in a known volume of 0.5 M acetic acid. The radioactivity of an aliquot was determined, another aliquot was subjected to amino acid analysis to determine the contents of either hydroxyproline (in the case of col-

431

OF COLLAGEN

lagen) or lysine (in the case of polylysine). Collagen content was calculated by assuming 100 residues of hydroxyproline per a chain. In the case of polylysine, the sum of lysine and adduct residues was used in calculation of specific activity (Table I).

Other Anal&al

Procedures

For amino acid analysis, samples were hydrolyzed in www by heating in constant boiling HCI for 24 h at 105% The contents of hydroxyproline and lysine were quantified following chromatography of the hydrolysates on a Beckman Model 119PL or a JLC-6AH amino acid analyzer equipped with a Durrum DC-GA column. Norleucine or cysteic acid was used as internal standard. Radioactivity was determined by

TABLE

I

EFFECT OF NaCNBHx ON THE GLYCOSYLATIONOF COLLAGEN AND POLYLYSINE” Acid-soluble

collagen

r*C]Glc

pm01 hyprob

4.5 h Control Reduced

350 2,031

0.324 0.237

1.56

4.6

29h Control Reduced

2,050 6,458

0.460 0.360

0.81 3.26

2.4 9.6

144b Control Reduced

7,500 14,250

0.494 0.515

2.76 5.03

8.1 14.8

dw Time of incubation

mol [“C]Glc/ a chain’

Percentage glycosylatedd (1~s + hylys)

Polylysine Percentage glycosylated (lysine)

dpm [“C]Glc

pm01 lYS

pm01 lys + adduct

mol [‘“C]Glc/ mol lys

29h Control Reduced

52,600 360,000

3.54 4.61

3.64 5.26

0.026 0.12

2.6 12.0

144h Control Reduced

116,509 2.86 X lo6

5.64 4.04

5.84 9.24

0.036 0.56

3.6 56.0

o Specificity activity of r*C]Glc used was 0.55 X lo6 dpm/Mmol. Calculations are based on data in Fig. 4. b One mole of collagen (Ychain is assumed to contain 100 mol of hypro. ‘A mole of collagen contains 3 a chains, each of about 100,000 daltons. This value is for an average 100,000 a chain. d Contents of lys and hylys in acid-soluble calf skin collagen are, respectively, 26.6 and 7.3 residues per lOO,OOO-dalton units (Piez et al. (27)).

432

ROGOZINSKI,

BLUMENFELD,

liquid scintillation counting. Polyacrylamide-SDS2 gel electrophoresis of collagen CNBr-fragments was carried out on lo-15% gradient slab gels as described by Laemmli (24). Proteins were stained with Coomassie brilliant blue. For fluorography, gels were soaked for 1 h in Enhance and 1 h in distilled water, dried on a Bio-Rad gel drier, placed on X-ray plates, and exposed at -70°C for 2 weeks.

AND

SEIFTER

18 1 M-J ‘;

1i

16

’ p !

14

I I I

RESULTS

Glgcos$ation of Collagen and Ident$cation of l-Lysinohexitol Preliminarily to this study, polylysine was used as a model for the reaction of amino groups of peptide-bound lysine with glucose. Following reduction with NaBH4, both the aldimine and the ketoamine adducts result in protein-bound l-lysinohexitol. Paper chromatography of the acid hydrolysate of the material of the Bio-Gel Ps excluded pool resulted in two radioactive peaks of slower mobility than lysine; these stained with ninhydrin (Fig. 1, top). Peaks of identical position were obtained whether the polymer had been labeled with [‘4Clglucose and reduced with NaBH4 or reacted with unlabeled glucose and reduced with NaB3H4 (Fig. 1, top). The recovery of radioactivity in the two peaks was 80%. Two major radioactively labeled peaks emerging in the vicinity of phenylalanine were obtained also when the acid hydrolysate of the modified polylysine peak from Bio-Gel Ps was chromatographed on the Beckman amino acid analyzer (Fig. 2a). The peaks occurred in the region where llysinohexitol and its dehydration product are usually observed (25,26). The two peaks correspond to the radioactive areas of the paper chromatograms (Fig. 1 top), and can be seen in Figs. 2b and 2~. Figure 3 shows the elution pattern obtained with a hydrolysate of polylysine reacted with [14C!]glucose for 144 h in the presence of NaCNBH3 (see below). In addition to lysine, two peaks with the mobility of l-lysinohexitol may be noted. 1-Lysinohexitol was identified in the acid hydrolysates of reduced calf skin or guinea pig skin collagen as evidenced by the presence of radioactive peaks of identical chro* Abbreviation

used: SDS,

sodium

dodecyl

sulfate.

6

ORIGIN

I

2 LYS

FIG. 1. Chromatography on Whatman No. 1 paper of an acid hydrolysate of reduced glycosylated collagen (bottom) and of glycosylated polylysine (top). The solvent system was n-butyl alcohol/acetic acid/ water (4:1:5); the runs were for 100 h at room temperature. The ninhydrin patterns, shown in the strips below the elution patterns, correspond to the radioactivity in eluates of consecutive l-cm portions of parallel chromatograms. Collagen incubated with [U‘“CJglucose and reduced with NaBH, (0); polylysine incubated with [U-“Cjglucose and reduced with NaBH, (A); polylysine incubated with glucose and reduced with NaBaH, (0).

matographic position to those found in the hydrolysates of modified polylysine (Figs. 1 bottom and 2d). However, the distribution of radioactivity between the peaks from the collagens is different from that obtained from reduced glycosylated polylysine, indicating perhaps a different ratio of isomers or dehydration products after

NONENZYMATIC

I,,

3m-a

I,,

Le”

280024002000EQOIZOO-

,

.

,

,

GLYCOSYLATION

,

I,,

PL

-

l

OF

COLLAGEN

433

lation was carried out in the presence of NaCNBHa, the adduct was in the form of 1-lysinohexitol residues; when it was carried out in its absence, the adduct could be either in the form of an unstable aldimine or the corresponding stable ketoamine. Both aldimine and ketoamine, upon reduction, give rise to 1-lysinohexitol residues. GZ~cos&don of Collagen in the Presence or Absence of NaCNBH,

T”“““““‘l 180 160 140

d

GCSC L.3 I

-I

t

I20 100

80 60 40 20

IO20 304050 60 7080 80100110120 I30 FRACTION

NUMBER

FIG. 2. Chromatography of the radioactive adducts of glycosylated and NaBH4-reduced collagen and of polylysine; a Beckman Model 119 BL amino acid analyzer equipped with a split-stream device was used. (a) An acid hydrolysate of glyeosylated, reduced polylysine (PL); (b) eluate of spot 1 from the paper chromatogram of an acid hydrolysate of polylysine as shown in Fig. 1; (c) eluate of spot 2 from paper chromatogram of an acid hydrolysate of polylysine as shown in Fig. 1; (d) acid hydrolysate of reduced glycosylated calf skin collagen (GCSC). Arrow indicates position of migration of standard leucine.

acid hydrolysis. The additional peaks seen in glycosylated calf skin collagen (Fig. 2d) may be due to adducts of glucose with hydroxylysine residues (25). We conclude that the radioactivity in the Bio-Gel Ps excluded pools of glycosylated collagen or polylysine occurs predominantly in adducts formed between glucose and e-NH2 groups of lysine residues. When glycosy-

The presence of NaCNBHa during incubation of collagen with [‘4Clglucose led to an increased incorporation of radioactivity as compared with incubation in its absence. The radioactivity was found in the excluded peak of Bio-Gel P6 chromatography. Results are shown in Figs. 4A and B for samples incubated with [‘4Clglucose for 4.5, 29, and 144 h, respectively. The extent of glycosylation of collagen at the various time points is presented in Table 1. The collagen glycosylated in the presence of NaCNBH3 showed a two- to fourfold increase in the number of incorporated glucose residues. That reflects the stabilization of labile aldimine groups by NaCNBHa (19). At 144 h, nearly 15% of the available c-amino groups of lysine and hydroxylysine were glycosylated and, in the continuous presence of NaCNBH,, represented the cumulative content of aldimine functions that form during the reaction. The glucose incorporated in the control samples probably reflects the content of stable ketoamines. All of the fragments obtained after treatment with CNBr of reduced glycosylated collagen (144-h sample), showed radioactivity on SDS-polyacrylamide gel electrophoresis. This suggests that covalent reaction with glucose occurred in regions along the entire length of (Y chains of collagen. The glycosylation of the model polylysine in the presence of the NaCNBH3 is also shown in Fig. 4 and in Table I. At 144 h, nearly one-half of all lysine residues were glycosylated. Under those conditions, the adduct present after reduction, l-lysinohexitol, can be clearly seen in the profile obtained by amino acid analysis of the

434

ROGOZINSKI,

BLUMENFELD,

AND

SEIFTER

acid hydrolysate of the glycosylated sample (Fig. 3). Incubation in the absence of NaCNBH3 resulted in preparations that were much less glycosylated (approximately 0.03 mol/mol of polylysine, Table I). The differences between the reduced and control samples illustrate the dissociation of the aldimine adducts, and points to their relative lability. Further Characterization of Stable Adducts Obtained bg Reductim of Glgcosylated CoUqwn When the collagens were incubated with [‘4Clglucose for 12 to 15 days, exhaustively dialyzed, or washed, or treated with CNBr and then reduced with either NaBH4 or NaCNBHa, the incorporation of [14C]glucose was not significantly different from that obtained with the nonreduced con-

FIG. 3. Elution pattern obtained on the amino acid analyzer of an acid hydrolysate of polylysine incubated with 40 mM [Ui’CMlucose (sp act 0.55 X lo6 dpm/ pmol) for 144 h at 3’7’C in the presence of 40 mM NaCNBHr. Products were detected by ninhydrin, and the adduct (1-lysinohexitol) emerged close to the elution position of the phenylalanine. The portion of the chromatogram between elution of leucine and ammonia is shown; no other ninhydrin-positive peaks were present.

Vdume, ml

FIG. 4. Gel filtration on Bio-Gel Ps of collagen and polylysine incubated with r’C]glucose in the absence (A-collagen; C-polylysine) and presence (B-collagen; D-polylysine) of NaCNBHr. Incubations were for 4 h (O), 29 h (O), or 144 h (A). The column size was 1.6 X 94 cm, and elution was with 0.5 M acetic acid. Arrows indicate the void volume (V,). Position of elution of free glucose was at 130 ml (not shown).

trols (Table II). That suggests that those aldimine functions that did form became dissociated or had undergone rearrangement during the pretreatment (dialysis or washing) of collagens prior to reduction, and that the adducts measured under these conditions were mostly the stable ketoamines. Even at the longer times of incubation in the absence of reductant, the extent of glycosylation was less (2-3 mol Glc/cr chain) than in its presence (5 mol Glc/a chain) (Tables I and II). As also shown in Table II, the neutral salt-soluble guinea pig and acid-soluble calf skin collagens showed comparable extents of glycosylation in the form of stable adducts; experimentally, the former collagen had been glycosylated in solution and the latter in fibril form. DISCUSSION

When polylysine was reacted with [14Clglucose for 144 h, about 3% of the ly-

NONENZYMATIC

GLYCOSYLATION TABLE

EXTENT Collagen”

Guinea pig skin Guinea pig skin Guinea pig skin Calf skin Calf skin Calf skin

CNBrb

+ + + + +

II

OF GLYCOS~ATION

Reduction

OF COLLAGENS

dpm [“C]Glc

+ + + +

435

OF COLLAGEN

3,600 17,890 21,575 64,800 18,000 28,548

pmoi hypro

mol [i4C]GIc/a chain”

0.28 1.68 2.76 13.5 2.5 4.7

3.57 3.04 2.21 1.84 2.76 2.34

a The guinea pig and calf skin collagens were incubated with 40 mM glucose of sp act 0.35 X lo6 dpm/pmol and 0.26 X lo6 dpm/pmoi, respectively, for 12-15 days and then dialyzed or washed. bCNBr digestion was as described under Methods. When performed, reduction of CNBr-treated collagens was carried out with either NaBH, of NaCNBHa and the collagen processed as under Methods. ‘A mole of collagen contains 3 a! chains, each of about 100,000 daltons. This value is for an average 100,000 (Ychain.

sine residues became glycosylated as determined from incorporated radioactivity. When the same reaction was carried out for the same period of time in the presence of NaCNBH3, approximately 50% of the lysine residues were found to have become glycosylated, and the main products of reduction of the glucose adduct were found to be 1-lysinohexitols in peptide linkage. One may consider that the cyanoborohydride reacted rapidly and irreversibly with all of the aldimine as it formed, driving the reaction to completion and forming the 1-lysinohexitol derivative. The approximate 50% conversion found in the presence of cyanoborohydride probably represents a maximum possible conversion under the conditions used. In the reaction conducted in the absence of cyanoborohydride, the irreversible intramolecular Amadori rearrangement proceeds too slowly to shift the equilibrium away from aldimine formation. Therefore the reverse reaction occurs to reestablish glucose and free amino groups; and the formation of stable glycosylated products is much less. In the case of calf skin collagen, reaction with glucose resulted in the glycosylation of about 8% of potential glycosylation sites (represented by an average of about 34 residues of lysine plus hydroxylysine (27)) available in each (Ychain. In the presence of NaCNBH3, the extent of glycosylation was about 15%, a figure representing the

maximum glycosylation possible in 144 h under the conditions used. That the data obtained with collagen differed markedly from those obtained with polylysine invites some comment. First, in the absence of cyanoborohydride, a larger fraction of potentially susceptible groups becomes glycosylated in collagen as compared with polylysine, so that one may consider that the aldimine groups formed in the protein more readily undergo Amadori rearrangement, perhaps aided by interactions with side chains of neighboring amino acids. Yet in the presence of cyanoborohydride during the course of the reaction, as compared to its absence, the increase in total glycosylation measured was 2-fold for collagen but about 20-fold for polylysine. Second, only 15% of the free amino groups in collagen became glycosylated even in the presence of cyanoborohydride; this suggests that most of the lysyl and hydroxylsyl residues are not available for reaction with glucose. From structural considerations one may anticipate that the equilibria of the reactions in the case of collagen would differ from those of polylysine. Studies of the incorporation of labeled glucose into rat tail tendon collagen had already identified the adduct, after reduction, as deoxysorbitol-lysine (i.e., a specific 1-lysinohexitol) (13). Approximately 1 mol of glucose per collagen peptide chain of

436

ROGOZINSKI,

BLUMENFELD,

100,000 daltons was found using the thiobarbituric acid color reaction to measure the extent of glycosylation. That value was obtained after incubation of collagen and glucose for 8 days. Other studies (20) have shown that the presence of cyanoborohydride during the course of reaction of glucose with plasma albumin accelerated glycosylation; indeed at a given time point the extent of glycosylation was greater in the presence of cyanoborohydride than in its absence. Nevertheless, under no condition studied was the extent of incorporation of glucose more than could be accounted for by reaction with one or two lysine residues of the total of 58 in the albumin molecule. Again, the reactivity of particular lysine residues must be determined both by neighboring group effects and the total organization of the polypeptide chain. From these studies, and from those with hemoglobin (19), one can infer that the use of cyanoborohydride in the way described allows an estimate of the maximum extent to which a particular protein can be glycosylated in a particular time frame. The clinical and physiological significance of these studies, and of those with hemoglobin (19) and with albumin (20), is that the measurement of the actual degree of glycosylation attained in tivo is compromised by the fact that a significant portion of the aldimine (unstable) form of the glycosylated protein residue is in equilibrium with glucose and unglycosylated residues of the protein, and therefore unavailable for estimation by usual procedures. Any removal of glucose from a sample by dialysis or by passage on columns can drive the reaction back, and cause underestimation of the actual extent of glycosylation as it occurs in viva (28). Yet it is the total extent of glycosylation of a protein over a period of time, even if it is in equilibrium with the unstable aldimine forms, that probably has the greater possible pathophysiological significance. This is implicit in the fact that rapid lowering of blood glucose levels can cause more dissociation of aldimine adducts thereby lowering the extent of measured glycosylation (29). This matter is

AND

SEIFTER

considered in some detail in a recent paper in which stable and unstable forms of glycosylated hemoglobin were determined (30). In viva, any collagen that is glycosylated in the aldimine form, even though it is unstable, is transiently different from nonglycosylated collagen. Furthermore, if, as Bensusan long ago postulated, Amadori rearrangement and crosslinking of the collagen depends on the extent of aldimine formation, the in viva content of aldimine can be of considerable significance. For instance, as shown by Bensusan (15), some collagens, when glycosylated, may become more insoluble. ACKNOWLEDGMENTS This research was supported by NIH Grants AG 99374, GM16389, and AM 20541 and the David Opochinsky Foundation. The authors are grateful to Dr. Anthony M. Adamany for aid and advice. REFERENCES 1. GO~TSCHALK, A. (1972) in Glycoproteins (Gottschalk, A., ed.), Vol. 5, pp. 141-157, Elsevier, New York. BUNN, H. F., GABBAY, K. H., AND GALLOP, P. M. (1978) science zoo, 21-27. CERAMI, A., STEVENS, V. J., AND MONNIER, V. M. (1979) Metubolis7n 28,431-437. RAHBAR, S., BLUMENFELD, O., AND RANNEY, H. M. (1969) Biocha. Biophgx Ra Cbmmun 36,838843. 5. HOLMQUIST, W. R., AND SCHROEDER, W. A. (1966) Biochemistry 5.2489-2503. 6. RAHBAR, S. (1968) Clin Chim Actu 22,296-298. 7. BOOKCHIN, R. M., AND GALLOP, P. M. (1968) Bioch.em Biophys Res. Commun 32.86-93. 8. BUNN, H. F., HANEY, D. N., GABBAY, K. H., AND GALLOP, P. M. (1975) Biock Biophys. Ra Cmnmun 67.103-199. 9. BUNN, H. F., AND BRIEHL, R. W. (1970) J. Clin Invest 49,1088-1995. 10. DODD, R. F., AND BOLMEY, C. (1979) Clin. Chem 25,764-768. 11. FRASER, D. M., SMITH, A. F., GRAY, R. S., BORSEY, D. Q., SINCLAIR, M. E., CLARKE, B. F., ANDDUNCAN, L. J. P. (1979) Brit. Med J. 1,979-981. ~~.:RAHBAR, S. (1981) in Human Hemoglobins and Hemoglobinopathies: A Review to 1981 (Schneider, R. G., Charge, S., and Schroeder, W. A., eds.), Vol. 40, pp. 373-385, Texas Reports on Biol. and Med., University of Texas, Galveston.

NONENZYMATIC

A. J. (1981)

13.

BAILEY,

14.

11, 90-94. BUNN, H. F. (1981)

15.

BENSUSAN,

GALLOP, P. MININSOIIN, BORCH,

M.,

19. HIGGINS,

Chem DAY,

SPIRO,

24.

LAEMMLI, 685.

25.

LE

0. O., AND

ADAMANY,

A. M. (1978)

26.

AND

(1979)

SPIRO, R. G., AND SPIRO, 246,4899-4909.

J. BioL

Chem

S. R., AND

J.

254,9394-

Bid Chem

MILLER, (1980)

K. A., WEISS,

WIDNESS,

A. J. (1981)

E., AND

LEWIS,

BIINN,

H. F.

M. S. (1960)

J.

235,1987-1991.

J. A.,

ROGLER-BROWN,

T. L.,

L., PETZOLD, K. S., H. C., AND SCHWARTZ,

NIEDERAU,

C. M., POWHOFF,

REINAUER, 30.

M. J. (1971)

BAILEY,

MCCOR-

SUSA, J. R. (1980)

B., J.

Lab. Clin Med 95,386-394. 29.

THORPE,

J. P., AND

197,405-412. J. A., GRAVALLESE, E., AND J. Clin Invest 65,896~901.

MICK, K. SCHWARTZ,

Amer. Chem Sot 93,289’7-2904. AND BONN, H. F. (1981) J. Biol R. W.,

A., MUH,

Bid Chem

DURST,

P. J., 256,5204-5208.

U.

PAPE,

27’. PIEZ, 28.

D.,

R. G.

Sci. USA 75,2X7-2731. (1969) J. Biol &em 244, 602-612. K. (1970) Nature (Landon) 227,680-

Biochzem J.

J.

J. W.

437

COLLAGEN

BLUMENFELD,

23.

129.203-208.

R., HANNEKEN, A., GABBAY, K. H. (1981)

117, 427-432. BERNSTEIN, M.

J. F., THORNBURG,

BAYNES, 9400. 21.

Biochem J.

FLUCKIGER, M. M., AND

F.,

H. D. (1971)

20.

Amer. J. Med 70,325~330.

Biodwm. R.

22.

OF

Proc Nat. Ad

H. B. F. (1972)

And 18.

Metubol Rea Suppl

Hormone

H. B. (1965) in Structure and Function of Connective and Skeletal Tissue (Jackson, S. F., Harkness, R. D., Partridge, S. M., and Tristram, G. R., eds.) (Proc. Adv. Study inst. NATO, 1964), pp. 42-46, Butterworth, London.

16. DIXON, 17.

GLYCOSYLATION

9-14. HUISMAN,

H. W.,

(1980)

KUIJKEN,

H. L., DUURKOOP, Clin Chim. Acta

S., GRIES,

F. A., AND

L.&n-aturiumsmdizin

4,

J. P. A. A., TAN-TJIONG,

E. P., AND LEIJNSE, 118, 303-309.

B. (1982)