Non enzymatic glycation of apolipoprotein A-I. Effects on its self-association and lipid binding properties

Vo1.153, No. 3,1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages ] 060-1067

June 30,1988

NON ENZYMATIC GLYCATION OF APOLIPOPROTEIN A-I. EFFECTS ON ITS SELF-ASSOCIATION AND LIPID BINDING PROPERTIES Carlos Calvo, Corinne Talussot, Gabriel Ponsin and Francois Berth~z~ne INSERM U 197, Laboratoire de Metabolisme des Lipides Hopital de l'Antiquaille 69321 Lyon C&dex 05, Francs Received May 5, 1988

In diabetic patients, hyperglycaemia results in the non enzymatic glycation o f many proteins including apolipoprotein A-I. We purified glycated apo A-I and compared its lipid binding properties to those of normal apo A-I. Analysis of tryptophan fluorescence spectra and of fluorescence quenching in the presence of iodine showed that glycation of apo A-I induces a decrease in the stability of the lipid-apoprotein interaction% and in that of the apoprotein self-association. Repetitive ultracentrifugations of High Density Lipoprotein (HDL) samples containing radioiodinated apo A-I or glycated apo A-I revealed that glycation of the apoprotein facilitates its dissociation from HDL. These results suggest that the non enzymatic glycation of apo A-I may affect the structural cohesion of HDL particles. © 1988 AcademicPress, Inc.

The plasma lipopr~teins are water-soluble complexes that transport lipids in blood (i). A large body of evidences has shown that apollpoproteins interact with lipoprotein surface through amphiphilic helical segments containing opposite polar and non-polar faces (2-6). Therefore, an alteration of these structural domains might induce changes in the lipid-protein interaction. Such protein alterations may result from genetic deficiencies. For example, the apo A-I variant (Lye iO7-~0 ) has a decreased lipid binding affinity (7) and a decreased ability to activate Lecithin : cholesterol acyl transferase (8). Post translational modifications of apolipoproteins may also occur. Exposure to high glucose concentrations results in the non enzymatic glycation of proteins. Condensation of glucose on the E-amino group of lysine residues forms a Shift base which subsequently

Abbreviations used : Apo A-I, Apolipoprotein A-I ; Gluc-Apo" A-I, Glucosyl-apo A-I ; DMPC, i-2, di-myristoyl-sn-glycero-3-phosphorylcholine ; POPC, l-palmitoyl-2-oleoyl-sn-glycero-3-phospborylcholine ; Gdm.Cl, Guanidium Chloride ; HDL, High Density Lipoprotein.

0006-291X/88 $1.50 Copyright © 1988 by Academic Press, Inc. All rights of reproduction in any form reserved.

1060

Vol. 153, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

undergoes a slow Amadori rearrangement leading to more stable ketoamine or hemiketal adducts (9). This process occurs in vivo in hyperglycaemic diabetic patients (i0). We and others have shown that, in these patients, apo A-I may be glycated (ii, 12). This report describes the alterations of the self-association and lipid binding properties of apo A-I resulting from its non enzymatic glycation.

MATERIALS AND METHODS Except when indicated, all experiments were carried out using a buffer containing 150 mM NaCI, i0 mM EDTA, I0 mM NaN 3 and i0 mM Tris-HCl, pH 7.4. Purification of glugosyl-apolipopFotein A-I : Apo A-I was purified to homogeneity from plasma of patients with type 2 diabetes, using well-established procedures (13). Glucosyl-apo A-I (Gluc-apo A-I) was then separated from the bulk of the apoprotein after affinity chromatography on a column (0.8 xlO cm) of aminopherdl boronats (Affi-gei 501, BioRad). The non-retained fraction was eluted with a 0.25 M ammonium acetate buffer, pH 8.5. The retained gluc-apo A-I fraction was then eluted with 0.2 M sorbitol in the same buffer. After extensive dialysis, the glycation degree of gluc-apo A-I wa~ measured after reduction of the glucose-lysine bonds in the presence of NaBVHA, as previously described (12). Tritiated apoprotein was then submittea to acid hydrolysis in 6 M HCI for 18 h at I08°C. After adjustement to pH 9.0, the hydroiysate was chromatoKraphied on amino-phenyl boronate to determine the actual amount of tritium associated ~ith glucosyl-lysine bonds. Calculations based on the specific activity of NaB H. and on apo A-I 4 concentration revealed that the apparent stochioemetry of g/uc-apo A-I was > 0.9 molecule of glucose per molecule of spo A-I (average of 3 ~ndependent experiments). Since the efficency of reduction of glucosyl-iysine bonds by NaBVH4 is less than I00 % (14), one may consider that the actual stochioemetry is liMely close to 1 molecule of glucose per molecule of apo A-I. Fluorescence measurements : The intrinsic fluorescence of tryptophan was monitored by using an SLM-Aminco SPF-500 c spectrofluorimeter. The excitation wavelength was 280 nm. Emission spectra were recorded between 300 and 450 nm. When the quenching of tryptophan fluorescence by potassium iodide (KI) was measured, the emission wavelength was set at 340 nm. Kinetics of DMPC-apo. A-I interaction : Apo A-I (75 ~g) was added to DMPC (225 p~) in a final volume of 1.5 ml, at 23.7°C. The lipid-protein interaction was assessed by monitoring the right ankle light scattering of DMPC. These experiments were carried out with the fluorimeter mentioned above, using 325 nm for both emission and excitation wavelengths. Serum clearance measurements : Following radioiodination by the chloramine T method (15), apo A-I or gluc-apo A-I were injected to rats for in vivo screening. After 1 h, the rats were exsanguinated and the plasma was reinjected to acceptor rats for serum clearance measurements. One jugular vein was used for injecting the samples and the other for periodic blood sampling. The clearance curves were obtained after calculation of the R/Ro ratio, where R is the plasma radioactivity at a given time and Ro is the plasma radioactivity measured immediately after sample injections. HDL apo A-I interaction : HDL (2 mg) from either healthy subjects or diabetic patients were incubated with radioiodinated apo A-I or ~luc-apo A-I (i00 pg), I061

Vol. 153, No. 3, 1 9 8 8

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

in a total volume of 3 ml, for 30 min at 37°C. After adjustment of the density to 1.22 by addition of solid KBr, the samples were ultracentrifuge d for 48 h at 45 000 rpm. The HDL fractions were separated, counted for radioactivity and ultracentrifuEed again in the same conditions. This procedure was repeated 4 times. The apoprotein radioactivity remaining associated with HDL was expressed as percent of the initial value. RESULTS The tryptophan fluorescence relative to that of tryptophan self-associate additional residues

in water,

spectra of apo A-I were blue-shifted indicating that both apoproteins

(Table I). Addition of either DMPC or POPC induced an

shift of fluorescence

to the hydrocarbon

corresponding

region of the lipid. The reversible

shift of the two s~oproteins was measured The mid-points Gdm.C1.

to the transfer of tryptophan

in the presence of Gdm.Cl.

for reversing the fluorescence

This indicated that the

driving force of gluc-apo A-I was weaker than that of apo

A-I. Potassium iodide was able to quench the tryptophan fluorescence two apoproteins Stern-Volmer

(Fig. 1).

shifts were 1.05 M and 0.85 M

for apo A-I and gluc-apo A-I, respectively.

self-association

fluorescence

(Fig. 2). The data were analyzed

equation

of the

according to the

(16), which predicts that I/Io = KSV [KI] + i

where I is the fluorescence

intensity for a given KI concentration,

Io is the

TABLE I Tryptophan fluorescence of Apo A-I and Gluc-Apo A-I Apo A-I

gluc-Apo A-I

Additions )&max

None

L~kmax a

kmax

Akmax a

341

- 12

342

- 11

338

- 15

339

- 14

0.075 mM

337

- 16

338

-

15

0.150

mM

336

-

17

337

-

15

, 0.150 mM

333

- 20

335

- 18

b

c , 0.030 ram DMPC

POPC

c

a Z~max are the shifts relative to the ~max in 3.5 M Gdm.Cl (353 nm for both Apo A-I and Gluc-Apo A-I) In the absence of lipid, the concentration of either Apo A-I was I0 pg/ml c Lipid-Apo A-I complexes (I00 : i, molar ratio) were prepared by the cholate method (5).

b

1062

Vol. 153, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

100

o~ ==

75

o o

o

F4

50

#, a

25 0 0

®

1

0

Cdm. C,

2

Ol 0

3

04

O8

i

I

i,

i.

0.2

0.4

08

0 8

(~o,/,)

KT Cmo,/,)

.Figure i. Self-association properties of apo A-I (0) and Kluc- apo A-I (0). The apoproteins (i0 g/ml) were incubated in the presence of various concentrations of Gdm.Cl. The reversible shifts of the fluorescence spectra ~max) were expressed as relative values. The unshifted ~max obtained in the presence of 3.5 M Gdm.Cl was 353 nm. F i b r e 2. Stern-Volmer plots of apo A-I (@), gluc-apo A-I (0) and Tryptophan The fluorescence intenslties were recorded at 25 C. The apoprotein concentration was 50~g/ml. The Stern-Volmer constants (Ksv) were calculated from the slopes of the straight lines. -

--

.

.

o

.

£iuorescence intensity in the absence of KI and Ksv is the Stern-Volmer constant. The Kay for apo A-I and gluc-apo A-I were 2.0 and 2.6 respectively, demonstrating a better accessibility to water in the latter case. The affinities of the two apo A-I for lipid were measured by monitoring the decrease of the right angle light scattering of D ~ C ,

after

addition of the proteins (Fig. 3). Although the rates of decrease of DNPC turbidity were not linear (18), that obtained with apo A-I was clearly fester ~han that observed in the presence of gluc-apo A-I. In the latter case, a two-time longer incubation was necessary to obtain a 50 % decrease of light scattering. To compare the respective bindings of gluc-apo A-I and apo A-I to HDL, we measured the plasma clearances of both apoproteins in rats (Fig.4). These were noticeably similar, suggesting that the glycation of apo A-I does not alter, at least quantitatively, its ability to bind to HDL. Consistent

1063

Vol. 153, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

100 t

11111

~ 2s e

"o ~

10

m

E

s 2.5

0

10 20 Time {min)

Q

30

0

' 20

1()

Q

S~) T

Time(h)

Figure 3. kinetics of lipid-apoprotein interactions. The riEht anEle liEht scattering of DMPC, at 23.7°C was monitored after addition of either apo A-I (0) or Eluc-apo A-I (0). The concentrations of lipid and epoproteins were 150 and 50 IAE/ml, respectively. Figure 4. Serum clearances of radioiodinated apo A-I (0) or Eluc-apo A-I (0). The values were expressed as percent of those observed immediately after injection of the samples (see Methods). Each point represents the mean of 6 rats.

with this result, electrophoresis

analysis of rat plasma by gradient polyacrylamide

in non denaturing conditions,

Eel

revealed that radioiodinated

apo

A-I or gluc-apo A-I co-migrated with HDL in a similar way (date not shown). The human HDL-apoprotein experiments.

interactions

were studied in an other group of

HDL semples containing either radioiodinated

A-I were submitted to highly disturbing conditions ultracentrifugations

apo A-I or gluc-apo

(i.e., repetitive

; FiE. 5). Although both apoproteins

strongly dissociated

loc .-,

8C

>,

o

~,n,

2c

"" 10

Number

of

ultracentrifugations

Fisure 5. HDL-bindinE properties of apo A-I (@) or Eluc-apo A-I (0). After incubation with radioiodinated apoproteins, HDL obtained from healthy subjects (A) or diabetic patients (B,~ were submitted to several successive ultracentrifuEations (see methods). The radioactivity associated with HDL samples were expressed as percent of the initial values. Each point is the mean of results obtained from 6 individual HDL samples in two independent experiments.

1064

Vol. 153, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

from HDL, the rate of dissociation of gluc-apo A-I was faster than that of apo A-I. When HDL from normal subjects were subtituted for HDL from diabetic patients, the results were comparable.

DISCUSSION In the present report, we have compared the self-association and lipid-bindir~ properties of non enzymatically glycated apo A-I to those, well characterized of normal apo A-I. Analysis of the tryptophan fluorescence revealed that the two spoproteins undergo a strong blue-shift of their spectra both in the absence and in the presence of phospholipid. Therefore, like apo A-I, glyc-apo A-I is able to self-associate and to interact with lipid. Only minor quantitative differences were observed between the two proteins. However the reversibility of the blue shift of fluorescence in denaturing conditions indicated that the transition of self-associated to monomeric status required less denaturinE agent for Eluc-apo A-I than for apo A-I. This suggested that the glycation of apo A-I decreased the stability of its hydrophobocitymediated self-association. Our studies of the quenching of tryptophan fluorescence by iodine were consistent with this concept. The accessibily to water of tryptophan residues may be estimated from the Ksv (apoprotein)/Ksv (tryptophan) ratio, which corresponds to a steric restriction factor, R (17). This factor is a measure of the fractional exposure of the residue to aqueous environnement. Apo A-I and Eluc-apo A-I had R values of 0.19 and 0.25, respectively. This 30 % increase of R indicates that the hydrophobic domain of apo A-I is less restrictive to the diffusion of water, after glycation of the protein. This may explain the decrease of stability of the self-association of gluc-apo A-I relative to that of apo A-I. Such an alteration of the stability of the hydrophobic interactions likely explains also that the rate of clarification of DMPC by gluc-apo A-I was slower than that obtained in the presence of apo A-I. Collective consideration of this data suggest that glycation of apo A-I alters the protein-protein and lipid-protein interactions in a qualitative rather than quantitative way. Experiments performed to study

1065

Vol. 153, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

the HDL-apo A-I interaction led to comparable conclusions.

The plasma

clearance curve of Eluc-apo A-I was superimposable to that of apo A-I, suEEestinE that the capacity of HDL to bind gluc-apo A-I was quantitatively normal. However, when HDL containin E radioiodinated apo A-I or Eluc-apo A-I were submitted to disturbinE conditions, the dissociation rate of Eluc-apo A-I was faster than that of apo A-I. This aEain shows that the stability of the interaction was decreased in the case of gluc-apo A-I. The mechanism by which the Elycation of apo A-I may alter the structural cohesion of HDL particles cannot be answered at the present time. However,

it should be point out that non enzymatic glycation of proteins

occurs at the epsilon amino Eroup of lysine residues. A large body of evidences has established that several lysine residues appear at the polar-non polar interface of the amphipathic helical seEments of apolipoproteins (18,19). Therefore it is tenable that Elycation of lysine residues in an helical domain alter locally its amphiphilicity.

ProEress in this direction

will require, as a first step, the determination of the preferential site of Elycation of apo A-I. Finally, our data do not answer, in any ws~, to the question as to whether the Elycation of apo A-I miEht be of importance reEarding the abnormalities of lipid metabolism in diabetic patients. One may only consider that the hypothesis of fonctionnal deficiencies resultin E from structural alterations of HDL seems to be a reasonable speculation.

REFERENCES

i. Smith, L.C., Pownall, H.J., sad Gotto, A.M., Jr. (1978) Ann. Rev. Biochem. 47, 751-777. 2. Segrest, J.P., Jackson, R.L., Morrisett, J.D., sad Gotto, A.M., Jr. (1974) FEBS Lett. 38, 247-253. 3. Mac, S.J.T., Sparrow, J.T., Gilliam, E.B., Gotto, A.M., Jr., sad Jackson, R.L. (1977) Biochemistry 16, 4150-4156. 4. Fukushima, A., Yokoama, S., Kroon, A.J., Kezdy, F.J., and Kaiser, E.T. (1980) J. Biol. Chem. 255, 10651-10657. 5. Ponsin, G., Strong~ K., Gotto, A.M., Jr., Sparrow, J.T., sad Pownall, H.J. (1984) Biochemistry 23, 5337-5343. 6. Ponsin, G., Hester, L., Gotto, A.M., Jr., Pownall, H.J., sad Sparrow, J.T. (1986) J. Biol. Chem. 261, 9202-9205. 7. Ponsin, G., Gotto, A.M., Jr., Utermann, G., sad Pownall, H.J. (1985) Biochem. Biophys. Res. Co.mum. 133, 856-862.

1066

Vol, 1.53, No. 3, 1988

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

8. Rall, S.C., J r . , Weisgraber, K.H., Nalhey, R.W., Ogawa, Y., Fielding, C.J., Utermann, G., Hass, J . , Steinmetz, A., Menzel, J . , and Assmann, G. 9. i0. II. 12. 13. 14. 15. 16. 17. 18.

19.

(1984) J. Biol. Chem. 259, i0063-i0070. Means, G.E., and Chang, M.K. (1982) Diabetes 31 (Suppl. 3), 1-4. Kennedy, L., andBaynes, J.W. (1984) Diabetologia 26, 93-98. Curtiss, L.K., and Witzum, J.L. (1985) Diabetes 34, 452-461. Calvo, C., Ponsin, G., and Berthezene, F. (1988) Diabete Metab., in press. Pownall, H.J., Massey, J.B., Kusserow, S.K., and Got~o, A.M., Jr. (1978) Biochemistry 17, 1183-1188. Garlick, R.L., Mazer, J.S., Chylack, L.T., Jr., Tung, W.H., and Bunn, H.F. (1984) J. Clin. Invest. 74, 1742-1749. Greewood, F.C., Hunter, W.M., and Glover, J.S. (1963) Biochem. J. 89, 114-123. Stern, 0., and Volmer, M. (1919) Phys. Z. 20, 183-188. Pownall, H.J~, and Smith, L.C. (1974) Biochemistry 13, 2590-2593. Kanellis, P., Romans, A.Y., Johnson, B.J., Kercret, H., Chiovetti, R., Jr., Allen, T.M., and Segrest, J.P. (1980) J. Biol. Chem. 255, 11464-11472. Anantharamaiah, G.M., Jones, J.L., Brouillette, C.G., Schmidt, C.F., Chung, B.H., Hughes, T.A., Bhown, and Segrest, J.P. (1985) J. Biol. Chem. 260, 10248-10255.

1067