485
Clinica Chimica Acta, 71 (1976) 485494 0 Elsevier Scientific Publishing Company,
Amsterdam
-Printed
in The Netherlands
CCA 8043
PHOTODYNAMIC MODIFICATION OF PROTEINS IN HUMAN RED BLOOD CELL MEMBRANES, INDUCED BY PROTOPORPHYRIN
A.F.P.M.
DE GOEIJ,
Sylvius Laboratories, (Received
R.J.C.
VAN STRAALEN
Wasseuarseweg
and J. VAN STEVENINCK
72, Leiden
*
(The Ne~herlo~ds~
May 6th, 1976)
Illumination of erythrocytes or erythrocyte membranes with visible light in the presence of protoporphyrin causes photodynamic damage of the cell membrane. This process is reflected a.o. by a mutilated ultrastructure and changes of the physical properties of the membrane proteins. I~umination in the presence of pro~porphyr~ causes association of membrane proteins, leading to blurring of the protein bands in electropheretograms, disappearance of bands and the appearance of protein aggregates on top of the gels. The formation of large protein aggregates is also indicated by Sephadex gel filtration of the solubilized membrane proteins. In kinetic studies it appeared that spectrin and the bands 2.1, 2.2,2.3 and 6 are most susceptible and that band 3 is least susceptible to this cross-linking reaction. Experimental results indicate that this cross-linking is caused by direct photooxidation of membrane proteins. Peroxidation of unsaturated fatty acids is not involved in the process. The significance of this process for studies on membrane structure and on photodynamic membrane damage is discussed.
Introduction Oxidation of cellular constituents during illumination with visible light in the presence of protoporphyrin as photosensitizer produces presumably the skin lesions in erythropoietic protoporphyria (EPP) and certainly photohemolysis of EPP red blood cells [lf. It was shown previously that protoporphyrininduced photohemolys~ can be attributed to photooxidation of membrane con~ituents [l-4]. Experiment evidence indicates, that medication of * To whom correspondence should be addressed.
486
membrane proteins is an essential link in the disturbance of membrane integrity, leading to increased kation permeability and colloid osmotic hemolysis [1,41. Moreover it seemed likely that the membrane proteins are also the primary target of protoporphyrin-induced photooxidation. Experimental results contradict both cholesterol peroxidation [5] and peroxidation of unsaturated fatty acid side chains [l] as primary events during photooxidation, although a possible role of unsaturated fatty acids could not yet be fully excluded. The protein modification is reflected by profound changes of the architecture of the erythrocyte membranes and of the physical properties of the membrane proteins [ 41. A more detailed study of membrane protein cross-linking seemed to be of crucial importance to elucidate further the molecular background of the photodynamic effects of protoporphyrin. Methods In all experiments heparinized blood was utilized. The red blood cells were spun down shortly after collection of the blood and washed three times in buffered isotonic NaCl solution [ 11. Hemoglobin-free ghosts were prepared according to Weed et al. [6]. Photooxidation was studied with 40% red blood cell suspensions or with 80% ghost suspensions in the presence of 20 pg/ml respectively 100 E.cg/ml protoporphyrin. Illumination was carried out as described previously [ 41. K’ leakage from intact cells was measured with a flame-photometer. The rate of photooxidation was measured manometrically, with differential respirometer units with air as the gas phase. The effects of malonaldehyde and glutaraldehyde on membrane proteins was studied by incubation of ghosts with these agents in a buffered solution, pH 7.2, containing Tris, 4.8 mM; EDTA, 0.84 mM; and NaCl, 10.2 mM (buffer A), prior to SDS gel electrophoresis. Malonaldehyde was prepared by acid hydrolysis of malonaldehyde bis(dimethy1 acetal). The effect of HzOz was studied on 30% red blood cell suspensions in isotonic NaCl, containing 1.2 mM sodium azide and varying concentrations of HzOz. Sodium azide was added to inhibit catalase [ 7 $1. Malonaldehyde was measured with thiobarbituric acid, according to Ottolenghi 191. Lipid peroxides were measured with the KCNS method described by Robey and Wiese [lo]. The influence of linoleic acid oxidation products on membrane proteins was studied by incubating ghosts with pretreated linoleic acid at varying concentrations. Pretreatment of linoleic acid was accomplished by illumination of a 0.2 M solution of the fatty acid in ethanol, in the presence of 150 pg protoporphyrin/ml, for one hour. Lipid extraction from ghosts was performed with acetone/water (9 : 1, v/v) or with n-butanol, according to Zwaal and Van Deenen [ 111. Lipids were measured as described by Hooghwinkel and Van Niekerk [12]. Sephadex chromatography was performed on Sephadex G-200, with membranes solved in 1% sodium dodecyl sulphate in buffer A. Sodium dodecyl sulphate gel electrophoresis of membrane proteins was performed with the meth-
487
od described by Fairbanks et al. [X3], after ~lubilization of the stroma in 10 mM Tris/HCl, containing 1% sodium dodecyl sulphate, 40 mM dithiothreitol and 1 mM EDTA. Gel densitometric scans were recorded on a Zeiss PMQ II spectrophotometer with scanning device. Results The effect of illumination of ghosts in the presence of protoporphyrin on membrane proteins is shown in Figs. 1 and 2. The protein bands obtained after SDS electrophoresis became progressively blurred and increasing amounts of protein failed to enter the gels. With 3% polya~l~ide gels (instead of the normally used 5.6%) the results were similar. In kinetic studies with g~du~ly increasing illum~ation times it appeared that the spectrin bands (1 and 2), band 2.1, 2.2, 2.3 and 6 were more susceptible to this cross-linking mechanism than the other proteins, whereas band 3 was relatively resistant, disappearing only after longer illumination periods. The glycoproteins disappeared with about the same velocity as band 3 (Fig. 3), with a concomitant accumulation of Schiff-positive material on top of the gels.
Fig. 1. Photodynamic effect of pmtoporphyrin on electropheretograms of membrane proteins, Ghost eoncentration: 80%. Protoporphyrin concentration: 100 &g/ml. Iiiumination time in the presence of protoporphyrin: a, 0 min: b. 6 min: c. 15 min: d. 30 min: e, 45 min: f: control, illuminated in the absence of protoporphyrin during 45 min.
b
II
0
G--
Relaiive moblllty
1.0
1.0
0.5 Relatwe
mobMy
Fig. 2. Photodynamic effect of protoporphyrin on densitometric Experimental conditions as described in the legend of Fig. 1.
w?ans of Coomassie Blue stained gels.
This phenomenon of protoporphyr~-induced cross-linking of membrane proteins was not restricted to ghost preparations. When normal cells, with protoporphyrin added to the medium, or EPP red blood cells were illuminated till about 60% of the intracellular K’ had leaked out of the cells, with subsequent ghost preparation and SDS electrophoresis, a similar cross-linking pattern was observed. To investigate the nature of the formed protein complex the illuminated ghosts were treated with several agents, prior to electrophoresis. Treatment during 30 min at 37°C with 40 mM dithiothreitol, prior to addition of SDS had no effect on cross-linking. Treatment with 8 M urea and incubation during 3 min at 100°C in the presence of SDS also failed to abolish cross-linking. Further, pre~~ubation of the ghosts with 40 mM dithiothreitol, ~-ethylm~e~ide or iodoacetate before photoo~dation had no effect on the ultimate results. only illumination in the presence of 1% SDS protected strongly against cross-linking.
489
Fig. 3. Photodynamic effect of pmtoparphyrin eonditicms as described in ttre legend of Fig. 1.
on densitametrie swns of PAS-stained gels. Experimental
In control experiments, however, an SO% inhibition of photooxidation of ghosts was found under these circumstances. The results of Sephadex G-200 filtration of membranes, solved in buffer A containing 1% SDS, are shown in Fig. 4. With normal ghosts two protein peaks were obtained: one in the void volume and a retarded peak, with an apparent protein concentration ratio of about 0.5. After illumination in the presence of protoporphyrin the protein peak in the void volume increased at the expense
Fjg. 4. Saphadex 0 200 fiitratbn of erythrowte membranes w&d in buffer A with 1% SDS. a, control. not illuminated; b, ilEumhated derring 30 tin in the presence of protoporpbyrin.
490
of the retarded peak, with a shift of the apparent concentration ratio to 1.5. Moreover, there appeared a third peak at an intermediate position. Further experiments were designed to discriminate between primary photooxidation of membrane proteins or primary oxidation of other membrane constituents as cause of the observed protein cross-linking. In model experiments it appeared that glucose, galactose and neuraminic acid were not susceptible to photooxidation by protoporphyrin. This virtually excludes the possibility that carbohydrate residues could be involved in protein cross-linking. During illumination of ghosts in the presence of protoporphyrin a very slow lipid peroxidation could be detected [l], leading to the formation of trace amounts of malonaldehyde, up to concentrations of 0.001 mM. The possible role of lipid peroxidation in this context was studied in further experiments. As a first approach the influence of exogenous m~onaldehyde and glutaraldehyde on membrane proteins was studied. At a malonaldehyde concentration of 10 mM a slow cross-linking of membrane proteins was observed, with small amounts of protein unable to enter the gels and a preferential disappearance of spectrin and band 4 (Fig. 5). Even after an incubation period of 2 h, however, cross-linking was limited as compared to the cross-linking caused by protoporphyrin-induced photooxidation. At much lower malonaldehyde concentrations (1 mM) cross-linking was not noticeable. Glutaraldehyde on the other hand caused a very pronounced cross-linking
1.0
io
1.0 d
b 1.0;
__-A
L._-_.A
0.5
0
Relative
I.0
mobility
0
Relatwe
mobiltty
Fig. 5. Effect of m~o~dehyde and ~u~r~dehyde on membrane proteins Densitometric scans of Coomask Blue stained gels a, control. untreated; b, ghosts incubated during 30 min with 10 mM glutat!aldehyde; c, ghosts incubated during 2 h with 10 mM malonaldehyde: d, ghosts incubated during 2 h with 10 mM giutaraldehyde.
491
of the membrane proteins at a concentration of 10 mM (Fig. 5). Again spectrin and band 4 were more susceptible than any of the other proteins. The effect of malonaldehyde generated inside the membrane of intact cells was studied by evoking lipid peroxidation by H,O,, as described by Stocks and Dormandy [7,8]. With this procedure malonaldehyde concentrations up to 0.05 mM were reached in the cell suspensions. No cross-linking of membrane proteins was observed in these experiments, however. In this context it is relevant that in suspensions of intact cells, illuminated in the presence of protoporphyrin, the malonaldehyde concentration never exceeded 0.0001 mM (one order of magnitude less than in illuminated ghost suspensions), despite the high degree of protein cuss-linking [1,5]. Peroxidation of un~t~at~ fatty acids yields primarily fatty acid hydroperoxides and some secondary products, a.o. m~onaldehyde. As pointed out by Matsushita and coworkers, not only malonaldehyde but also the hydroperoxide itself and .other secondary products could be involved in protein modification [ 14-161. To study this possibility linoleic acid peroxidation products were prepared as described in the method section, with subsequent incubation of ghosts with these oxidation products during 30 min at 37°C. The malonaldehyde concentration in the final incubation mixture varied from 0.01 to 0.1 mM, whereas the lipid hydroperoxide concentration was at least two orders of magnitude higher than the lipid peroxide concentration found in ghosts, illuminated in the presence of protoporph~in [ 5 ]. Again, no protein cross-linking could be detected. Finally the effect of lipid extraction prior to photooxidation was investigated. In several experiments it appeared that four extractions of ghosts with acetone/water (9 : 1, v/v) or with n-butanol according to Zwaal and Van Deenen removed at least 90% of the membrane lipids. Illumination of these pretreated ghost preparations in the presence of protoporphyrin yielded cross-linking patterns, undistinguishable from the patterns obtained with fresh ghosts. Discussion The failure of increasing amounts of protein to enter into 5.6% and 3% polyacrylamide gels during SDS electrophoresis indicates the formation of large protein aggregates. This is supposed by the results with Sephadex G-200, Separation of membrane proteins in only two major fractions with this technique is in agreement with the results of gel filtration described in literature ]17-191. With normal, untreated ghosts spectrin is excluded from the gel, appearing in the void volume [18]. Most of the other proteins appear as a rather broad, retarded fraction, Fig. 4 agrees with this pattern for untreated ghosts. After photooxidation the amount of protein excluded from the gel is greatly enhanced, indicating the formation of large protein aggregates. As described by Bhakdi et al. and Carraway et al. some non-covalent bonds in proteins are not disrupted by SDS (refs. 20 and 21). The failure of 8 M urea and of SDS at 100°C to disrupt the bonds makes it quite conceivable, however, that the protein ~~egation observed after photooxidation may be caused by covalent cross-linking of polypeptide chains. Reduction of the illuminated ghost preparation with dithiothreitol prior to
492
addition of SDS and electrophoresis, and treatment of the ghosts with sulfhydry1 reagents before photooxidation had no effect on the final protein pattern on the electrophorograms. Thus formation of disulfide bridges apparently does not play an essential role in the cross-linking reaction. In previous papers it was shown that peroxidation of membrane lipids is not involved in the effectuation of photohemolysis. Thus, if lipid peroxidation would be the cause of protein cross-linking, this would imply that also this profound disturbance of the normal protein pattern is not an essential link in the chain of events leading from photooxidation of membrane constituents via increased kation permeability to photohemolysis. Although this seemed very unlikely, several experiments, designed to elucidate a possible role of lipid peroxidation in the process of protein cross-linking, were conducted. Toxic effects of lipid peroxidation are usually ascribed to malonaldehyde, a major secondary product of lipid peroxidation. Exogenous malonaldehyde, in a concentration of four orders of magnitude higher than the concentration found during illumination in the presence of protoporphyrin, caused only a limited cross-linking. Much lower concentrations (1 mM) had no significant effect. Malonaldehyde, generated inside the membrane structure by Hz02 treatment, caused no cross-linking at a concentration of 0.05 mM. These figures should be compared to the maximal malonaldehyde concentration of 0.001 mM in photooxidized ghosts and of about 0.0001 mM in photooxidized intact cells, with extensive cross-linking in both cases. These observations virtually exclude the possibility that malonaldehyde, formed as a secondary product of lipid peroxidation, could be involved in in the aldehyde-induced crossmembrane protein cross-linking. Moreover, linking spectrin and band 4 are stronger affected than the other proteins, whereas during, photooxidation spectrin and band 6 are more susceptible to cross-linking. Studies of Matsushita et al. have shown that protein modification can also be caused by fatty acid hydroperoxides and possibly secondary products other than malonaldehyde [ 14-161. Therefore in some experiments ghosts were incubated with oxidation products of linoleic acid, as described by these authors. Again no cross-linking was found. Finally the observation that extraction of at least 90% of the membrane lipids prior to photooxidation did not inhibit cross-linking, strongly supports the conclusion that peroxidation of fatty acid side chains is not involved in this process. It seems likely that the cross-linking of membrane proteins and the concomitant change of membrane architecture as described in a previous paper [4] are of crucial importance in protoporphyrin-induced photodynamic membrane damage. Cross-linking experiments can be considered as a possible approach to nearest neighbor analysis of membrane proteins. Considering studies in recent literature, the interpretation of the experimental results is rather difficult however. As discussed above, with malonaldehyde and glutaraldehyde as cross-linking reagents, spectrin and band 4 were more sensitive than the other membrane proteins. This is in accordance with previous studies of Steck [22]. With dimethyl dithiobispropionimidate on the other hand, spectrin and band 3 were most easily cross-linked [ 231. Finally, in protoporphyrininduced photodynamic cross-
493
linking spectrin and band 6 were most susceptible, whereas band 3 was less susceptible than any of the other proteins. Therefore the cross-linking pattern depends on the cross-linking reagent. The only common feature in all these studies is extensive involvement of spectrin in the reaction. This would support the conclusion that spectrin is in close association with many of the other membrane proteins. In a previous paper we have shown that the first photodynamic effect on the architecture of the membrane is particle aggregation, as observed in freeze-etch electron microscopy [4]. Poste et al. [ 251 presented experimental evidence, indicating that a similar redistribution of lectin receptors can be evoked in membranes of mammalian cells by disruption of microtubules and microfilaments at the cytoplasmic face of the membrane. An “anchor” function with respect to intramembrane proteins has been attributed to spectrin [26,27]. Therefore it is tempting to speculate that oxidation and cross-linking of spectrin is a primary effect during photooxidation, leading to impairment of this anchor function and thus to particle aggregation. In a recent paper Girotti described similar experimental results with human erythrocyte membranes, illuminated in the presence of bilirubin as sensitizer [24]. In this system Girotti found the same kind of protein cross-linking as described in this paper. This indicates that this protein modification, not reported before in this kind of studies, may be of crucial importance to understand photodynamic membrane damage in general. Acknowledgement The authors assistance.
wish to thank
Miss Karmi Christianse
for her expert
technical
References 1 Schothorst, A.A.. Van Steveninck, J.. Went, L.N. and Suurmond, D. (1972) Clin. Chim. Acta 3Y, 161-170 2 Schothorst, A.A., Van Steveninck, J., Went, L.N. and Suurmond, D. (1970) Clin. Chirn. Acta 28, 4149 3 Schothorst. A.A.. Van S .veninck, J.. Went, L.N. and Suurmond, D. (1971) Clin. Chim. Acta 33, 207-213 4 De Goeij, AF.P.M., Ververgaert, P.H.J.Th. and Van Steveninck. J. (1975) Clin. Chim. Acta 62. 287292 5 De Goeij, A.F.P.M. and Van Steveninck, J. (1976) Clin Chim. Acta 68, 115-122 6 Weed, R.L. Reed, C.F. and Berg, G. (1963) J. Clin. Invest. 42, 581-588 7 Stocks, J. and Dormandy. T.L. (1971) Br. J. Haematol. 20,95-111 8 Dormandy, T.L. (1971) Br. J. Haematol. 20. 457-461 9 Ottolenghi, A. (1959) Arch. Biochem. Biophys. 79, 355-363 10 Robey, R.F. and Wiese.H.K. (1945) Ind. Eng. Chem. Anal. Ed. 17, 425-426 11 Zwaal. R.F.A. and Van Deenen, L.L.M. (1968) Biochim. Biophys. Acta 163,44-49 12 Hooghwinkel, G.J.M. and Van Niekerk, H.P.G.A. (1960) Proc. Acad. Sci Amsterdam, Ser. B 63, 475-483 13 Fairbanks, G., Steck. T.L. and Wallach, D.F.H. (1971) Biochemistry 10, 26062617 14 Matsushita, S. and Kobayaahi. M. (1970) Agr. Biol. Chem. 34. 825-829 15 Gamage, P.T., Mori. T. and Matsushita, S. (1971) Agr. Biol. Chem. 35, 33-39 16 Garnage, P.T. and Matsushita. S. (1973) Agr. Biol. Chem. 37, l-8 17 Lenard, J. (1970) Biochemistry 9. 1129-1132 18 Marchesi, S.L.. Steers, E., Marchesi. V.T. and Tillack. T.W. (1970) Biochemistry 9, 50-57 19 Carraway, K.L and Shin, B.C. (1972) J. Biol. Chem. 247. 2102-2108
494 20 Bhakdi, S.. Kniifermann. H.. Schmidt-Ulrich. R., Fischer. H. and Wallach, D.F.H. (1974) Biochim. Biophys. Acta 363,39-53 21 Carraway, K.L., Triple& R.B. and Anderson, D.R. (1975) Biochim. Biophys. Acta 379, 571-581 22 Steck. T.L. (1972) J. Mol. Biol. 66. 295-305 23 Wang. K. and Richards, F.M. (1974) J. Biol. Chem. 249. 8005-8018 24 Girotti. A.W. (1975) Biochem. 14.3377-3383 25 Paste, G., Papabadjopoulos, D., Jacobson, K. and Vail, W.J. (1975) Biochim. Biophys. Acta 394, 520-539 26 Steck, T.L. (1974) J. Cell Biol. 62, l-19 27 Ji, T.H. and Nicolson. G.L. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 2212-2216
Clinica Chimica Acta, 71 (1976) 485494 0 Elsevier Scientific Publishing Company,
Amsterdam
-Printed
in The Netherlands
CCA 8043
PHOTODYNAMIC MODIFICATION OF PROTEINS IN HUMAN RED BLOOD CELL MEMBRANES, INDUCED BY PROTOPORPHYRIN
A.F.P.M.
DE GOEIJ,
Sylvius Laboratories, (Received
R.J.C.
VAN STRAALEN
Wasseuarseweg
and J. VAN STEVENINCK
72, Leiden
*
(The Ne~herlo~ds~
May 6th, 1976)
Illumination of erythrocytes or erythrocyte membranes with visible light in the presence of protoporphyrin causes photodynamic damage of the cell membrane. This process is reflected a.o. by a mutilated ultrastructure and changes of the physical properties of the membrane proteins. I~umination in the presence of pro~porphyr~ causes association of membrane proteins, leading to blurring of the protein bands in electropheretograms, disappearance of bands and the appearance of protein aggregates on top of the gels. The formation of large protein aggregates is also indicated by Sephadex gel filtration of the solubilized membrane proteins. In kinetic studies it appeared that spectrin and the bands 2.1, 2.2,2.3 and 6 are most susceptible and that band 3 is least susceptible to this cross-linking reaction. Experimental results indicate that this cross-linking is caused by direct photooxidation of membrane proteins. Peroxidation of unsaturated fatty acids is not involved in the process. The significance of this process for studies on membrane structure and on photodynamic membrane damage is discussed.
Introduction Oxidation of cellular constituents during illumination with visible light in the presence of protoporphyrin as photosensitizer produces presumably the skin lesions in erythropoietic protoporphyria (EPP) and certainly photohemolysis of EPP red blood cells [lf. It was shown previously that protoporphyrininduced photohemolys~ can be attributed to photooxidation of membrane con~ituents [l-4]. Experiment evidence indicates, that medication of * To whom correspondence should be addressed.
486
membrane proteins is an essential link in the disturbance of membrane integrity, leading to increased kation permeability and colloid osmotic hemolysis [1,41. Moreover it seemed likely that the membrane proteins are also the primary target of protoporphyrin-induced photooxidation. Experimental results contradict both cholesterol peroxidation [5] and peroxidation of unsaturated fatty acid side chains [l] as primary events during photooxidation, although a possible role of unsaturated fatty acids could not yet be fully excluded. The protein modification is reflected by profound changes of the architecture of the erythrocyte membranes and of the physical properties of the membrane proteins [ 41. A more detailed study of membrane protein cross-linking seemed to be of crucial importance to elucidate further the molecular background of the photodynamic effects of protoporphyrin. Methods In all experiments heparinized blood was utilized. The red blood cells were spun down shortly after collection of the blood and washed three times in buffered isotonic NaCl solution [ 11. Hemoglobin-free ghosts were prepared according to Weed et al. [6]. Photooxidation was studied with 40% red blood cell suspensions or with 80% ghost suspensions in the presence of 20 pg/ml respectively 100 E.cg/ml protoporphyrin. Illumination was carried out as described previously [ 41. K’ leakage from intact cells was measured with a flame-photometer. The rate of photooxidation was measured manometrically, with differential respirometer units with air as the gas phase. The effects of malonaldehyde and glutaraldehyde on membrane proteins was studied by incubation of ghosts with these agents in a buffered solution, pH 7.2, containing Tris, 4.8 mM; EDTA, 0.84 mM; and NaCl, 10.2 mM (buffer A), prior to SDS gel electrophoresis. Malonaldehyde was prepared by acid hydrolysis of malonaldehyde bis(dimethy1 acetal). The effect of HzOz was studied on 30% red blood cell suspensions in isotonic NaCl, containing 1.2 mM sodium azide and varying concentrations of HzOz. Sodium azide was added to inhibit catalase [ 7 $1. Malonaldehyde was measured with thiobarbituric acid, according to Ottolenghi 191. Lipid peroxides were measured with the KCNS method described by Robey and Wiese [lo]. The influence of linoleic acid oxidation products on membrane proteins was studied by incubating ghosts with pretreated linoleic acid at varying concentrations. Pretreatment of linoleic acid was accomplished by illumination of a 0.2 M solution of the fatty acid in ethanol, in the presence of 150 pg protoporphyrin/ml, for one hour. Lipid extraction from ghosts was performed with acetone/water (9 : 1, v/v) or with n-butanol, according to Zwaal and Van Deenen [ 111. Lipids were measured as described by Hooghwinkel and Van Niekerk [12]. Sephadex chromatography was performed on Sephadex G-200, with membranes solved in 1% sodium dodecyl sulphate in buffer A. Sodium dodecyl sulphate gel electrophoresis of membrane proteins was performed with the meth-
487
od described by Fairbanks et al. [X3], after ~lubilization of the stroma in 10 mM Tris/HCl, containing 1% sodium dodecyl sulphate, 40 mM dithiothreitol and 1 mM EDTA. Gel densitometric scans were recorded on a Zeiss PMQ II spectrophotometer with scanning device. Results The effect of illumination of ghosts in the presence of protoporphyrin on membrane proteins is shown in Figs. 1 and 2. The protein bands obtained after SDS electrophoresis became progressively blurred and increasing amounts of protein failed to enter the gels. With 3% polya~l~ide gels (instead of the normally used 5.6%) the results were similar. In kinetic studies with g~du~ly increasing illum~ation times it appeared that the spectrin bands (1 and 2), band 2.1, 2.2, 2.3 and 6 were more susceptible to this cross-linking mechanism than the other proteins, whereas band 3 was relatively resistant, disappearing only after longer illumination periods. The glycoproteins disappeared with about the same velocity as band 3 (Fig. 3), with a concomitant accumulation of Schiff-positive material on top of the gels.
Fig. 1. Photodynamic effect of pmtoporphyrin on electropheretograms of membrane proteins, Ghost eoncentration: 80%. Protoporphyrin concentration: 100 &g/ml. Iiiumination time in the presence of protoporphyrin: a, 0 min: b. 6 min: c. 15 min: d. 30 min: e, 45 min: f: control, illuminated in the absence of protoporphyrin during 45 min.
b
II
0
G--
Relaiive moblllty
1.0
1.0
0.5 Relatwe
mobMy
Fig. 2. Photodynamic effect of protoporphyrin on densitometric Experimental conditions as described in the legend of Fig. 1.
w?ans of Coomassie Blue stained gels.
This phenomenon of protoporphyr~-induced cross-linking of membrane proteins was not restricted to ghost preparations. When normal cells, with protoporphyrin added to the medium, or EPP red blood cells were illuminated till about 60% of the intracellular K’ had leaked out of the cells, with subsequent ghost preparation and SDS electrophoresis, a similar cross-linking pattern was observed. To investigate the nature of the formed protein complex the illuminated ghosts were treated with several agents, prior to electrophoresis. Treatment during 30 min at 37°C with 40 mM dithiothreitol, prior to addition of SDS had no effect on cross-linking. Treatment with 8 M urea and incubation during 3 min at 100°C in the presence of SDS also failed to abolish cross-linking. Further, pre~~ubation of the ghosts with 40 mM dithiothreitol, ~-ethylm~e~ide or iodoacetate before photoo~dation had no effect on the ultimate results. only illumination in the presence of 1% SDS protected strongly against cross-linking.
489
Fig. 3. Photodynamic effect of pmtoparphyrin eonditicms as described in ttre legend of Fig. 1.
on densitametrie swns of PAS-stained gels. Experimental
In control experiments, however, an SO% inhibition of photooxidation of ghosts was found under these circumstances. The results of Sephadex G-200 filtration of membranes, solved in buffer A containing 1% SDS, are shown in Fig. 4. With normal ghosts two protein peaks were obtained: one in the void volume and a retarded peak, with an apparent protein concentration ratio of about 0.5. After illumination in the presence of protoporphyrin the protein peak in the void volume increased at the expense
Fjg. 4. Saphadex 0 200 fiitratbn of erythrowte membranes w&d in buffer A with 1% SDS. a, control. not illuminated; b, ilEumhated derring 30 tin in the presence of protoporpbyrin.
490
of the retarded peak, with a shift of the apparent concentration ratio to 1.5. Moreover, there appeared a third peak at an intermediate position. Further experiments were designed to discriminate between primary photooxidation of membrane proteins or primary oxidation of other membrane constituents as cause of the observed protein cross-linking. In model experiments it appeared that glucose, galactose and neuraminic acid were not susceptible to photooxidation by protoporphyrin. This virtually excludes the possibility that carbohydrate residues could be involved in protein cross-linking. During illumination of ghosts in the presence of protoporphyrin a very slow lipid peroxidation could be detected [l], leading to the formation of trace amounts of malonaldehyde, up to concentrations of 0.001 mM. The possible role of lipid peroxidation in this context was studied in further experiments. As a first approach the influence of exogenous m~onaldehyde and glutaraldehyde on membrane proteins was studied. At a malonaldehyde concentration of 10 mM a slow cross-linking of membrane proteins was observed, with small amounts of protein unable to enter the gels and a preferential disappearance of spectrin and band 4 (Fig. 5). Even after an incubation period of 2 h, however, cross-linking was limited as compared to the cross-linking caused by protoporphyrin-induced photooxidation. At much lower malonaldehyde concentrations (1 mM) cross-linking was not noticeable. Glutaraldehyde on the other hand caused a very pronounced cross-linking
1.0
io
1.0 d
b 1.0;
__-A
L._-_.A
0.5
0
Relative
I.0
mobility
0
Relatwe
mobiltty
Fig. 5. Effect of m~o~dehyde and ~u~r~dehyde on membrane proteins Densitometric scans of Coomask Blue stained gels a, control. untreated; b, ghosts incubated during 30 min with 10 mM glutat!aldehyde; c, ghosts incubated during 2 h with 10 mM malonaldehyde: d, ghosts incubated during 2 h with 10 mM giutaraldehyde.
491
of the membrane proteins at a concentration of 10 mM (Fig. 5). Again spectrin and band 4 were more susceptible than any of the other proteins. The effect of malonaldehyde generated inside the membrane of intact cells was studied by evoking lipid peroxidation by H,O,, as described by Stocks and Dormandy [7,8]. With this procedure malonaldehyde concentrations up to 0.05 mM were reached in the cell suspensions. No cross-linking of membrane proteins was observed in these experiments, however. In this context it is relevant that in suspensions of intact cells, illuminated in the presence of protoporphyrin, the malonaldehyde concentration never exceeded 0.0001 mM (one order of magnitude less than in illuminated ghost suspensions), despite the high degree of protein cuss-linking [1,5]. Peroxidation of un~t~at~ fatty acids yields primarily fatty acid hydroperoxides and some secondary products, a.o. m~onaldehyde. As pointed out by Matsushita and coworkers, not only malonaldehyde but also the hydroperoxide itself and .other secondary products could be involved in protein modification [ 14-161. To study this possibility linoleic acid peroxidation products were prepared as described in the method section, with subsequent incubation of ghosts with these oxidation products during 30 min at 37°C. The malonaldehyde concentration in the final incubation mixture varied from 0.01 to 0.1 mM, whereas the lipid hydroperoxide concentration was at least two orders of magnitude higher than the lipid peroxide concentration found in ghosts, illuminated in the presence of protoporph~in [ 5 ]. Again, no protein cross-linking could be detected. Finally the effect of lipid extraction prior to photooxidation was investigated. In several experiments it appeared that four extractions of ghosts with acetone/water (9 : 1, v/v) or with n-butanol according to Zwaal and Van Deenen removed at least 90% of the membrane lipids. Illumination of these pretreated ghost preparations in the presence of protoporphyrin yielded cross-linking patterns, undistinguishable from the patterns obtained with fresh ghosts. Discussion The failure of increasing amounts of protein to enter into 5.6% and 3% polyacrylamide gels during SDS electrophoresis indicates the formation of large protein aggregates. This is supposed by the results with Sephadex G-200, Separation of membrane proteins in only two major fractions with this technique is in agreement with the results of gel filtration described in literature ]17-191. With normal, untreated ghosts spectrin is excluded from the gel, appearing in the void volume [18]. Most of the other proteins appear as a rather broad, retarded fraction, Fig. 4 agrees with this pattern for untreated ghosts. After photooxidation the amount of protein excluded from the gel is greatly enhanced, indicating the formation of large protein aggregates. As described by Bhakdi et al. and Carraway et al. some non-covalent bonds in proteins are not disrupted by SDS (refs. 20 and 21). The failure of 8 M urea and of SDS at 100°C to disrupt the bonds makes it quite conceivable, however, that the protein ~~egation observed after photooxidation may be caused by covalent cross-linking of polypeptide chains. Reduction of the illuminated ghost preparation with dithiothreitol prior to
492
addition of SDS and electrophoresis, and treatment of the ghosts with sulfhydry1 reagents before photooxidation had no effect on the final protein pattern on the electrophorograms. Thus formation of disulfide bridges apparently does not play an essential role in the cross-linking reaction. In previous papers it was shown that peroxidation of membrane lipids is not involved in the effectuation of photohemolysis. Thus, if lipid peroxidation would be the cause of protein cross-linking, this would imply that also this profound disturbance of the normal protein pattern is not an essential link in the chain of events leading from photooxidation of membrane constituents via increased kation permeability to photohemolysis. Although this seemed very unlikely, several experiments, designed to elucidate a possible role of lipid peroxidation in the process of protein cross-linking, were conducted. Toxic effects of lipid peroxidation are usually ascribed to malonaldehyde, a major secondary product of lipid peroxidation. Exogenous malonaldehyde, in a concentration of four orders of magnitude higher than the concentration found during illumination in the presence of protoporphyrin, caused only a limited cross-linking. Much lower concentrations (1 mM) had no significant effect. Malonaldehyde, generated inside the membrane structure by Hz02 treatment, caused no cross-linking at a concentration of 0.05 mM. These figures should be compared to the maximal malonaldehyde concentration of 0.001 mM in photooxidized ghosts and of about 0.0001 mM in photooxidized intact cells, with extensive cross-linking in both cases. These observations virtually exclude the possibility that malonaldehyde, formed as a secondary product of lipid peroxidation, could be involved in in the aldehyde-induced crossmembrane protein cross-linking. Moreover, linking spectrin and band 4 are stronger affected than the other proteins, whereas during, photooxidation spectrin and band 6 are more susceptible to cross-linking. Studies of Matsushita et al. have shown that protein modification can also be caused by fatty acid hydroperoxides and possibly secondary products other than malonaldehyde [ 14-161. Therefore in some experiments ghosts were incubated with oxidation products of linoleic acid, as described by these authors. Again no cross-linking was found. Finally the observation that extraction of at least 90% of the membrane lipids prior to photooxidation did not inhibit cross-linking, strongly supports the conclusion that peroxidation of fatty acid side chains is not involved in this process. It seems likely that the cross-linking of membrane proteins and the concomitant change of membrane architecture as described in a previous paper [4] are of crucial importance in protoporphyrin-induced photodynamic membrane damage. Cross-linking experiments can be considered as a possible approach to nearest neighbor analysis of membrane proteins. Considering studies in recent literature, the interpretation of the experimental results is rather difficult however. As discussed above, with malonaldehyde and glutaraldehyde as cross-linking reagents, spectrin and band 4 were more sensitive than the other membrane proteins. This is in accordance with previous studies of Steck [22]. With dimethyl dithiobispropionimidate on the other hand, spectrin and band 3 were most easily cross-linked [ 231. Finally, in protoporphyrininduced photodynamic cross-
493
linking spectrin and band 6 were most susceptible, whereas band 3 was less susceptible than any of the other proteins. Therefore the cross-linking pattern depends on the cross-linking reagent. The only common feature in all these studies is extensive involvement of spectrin in the reaction. This would support the conclusion that spectrin is in close association with many of the other membrane proteins. In a previous paper we have shown that the first photodynamic effect on the architecture of the membrane is particle aggregation, as observed in freeze-etch electron microscopy [4]. Poste et al. [ 251 presented experimental evidence, indicating that a similar redistribution of lectin receptors can be evoked in membranes of mammalian cells by disruption of microtubules and microfilaments at the cytoplasmic face of the membrane. An “anchor” function with respect to intramembrane proteins has been attributed to spectrin [26,27]. Therefore it is tempting to speculate that oxidation and cross-linking of spectrin is a primary effect during photooxidation, leading to impairment of this anchor function and thus to particle aggregation. In a recent paper Girotti described similar experimental results with human erythrocyte membranes, illuminated in the presence of bilirubin as sensitizer [24]. In this system Girotti found the same kind of protein cross-linking as described in this paper. This indicates that this protein modification, not reported before in this kind of studies, may be of crucial importance to understand photodynamic membrane damage in general. Acknowledgement The authors assistance.
wish to thank
Miss Karmi Christianse
for her expert
technical
References 1 Schothorst, A.A.. Van Steveninck, J.. Went, L.N. and Suurmond, D. (1972) Clin. Chim. Acta 3Y, 161-170 2 Schothorst, A.A., Van Steveninck, J., Went, L.N. and Suurmond, D. (1970) Clin. Chirn. Acta 28, 4149 3 Schothorst. A.A.. Van S .veninck, J.. Went, L.N. and Suurmond, D. (1971) Clin. Chim. Acta 33, 207-213 4 De Goeij, AF.P.M., Ververgaert, P.H.J.Th. and Van Steveninck. J. (1975) Clin. Chim. Acta 62. 287292 5 De Goeij, A.F.P.M. and Van Steveninck, J. (1976) Clin Chim. Acta 68, 115-122 6 Weed, R.L. Reed, C.F. and Berg, G. (1963) J. Clin. Invest. 42, 581-588 7 Stocks, J. and Dormandy. T.L. (1971) Br. J. Haematol. 20,95-111 8 Dormandy, T.L. (1971) Br. J. Haematol. 20. 457-461 9 Ottolenghi, A. (1959) Arch. Biochem. Biophys. 79, 355-363 10 Robey, R.F. and Wiese.H.K. (1945) Ind. Eng. Chem. Anal. Ed. 17, 425-426 11 Zwaal. R.F.A. and Van Deenen, L.L.M. (1968) Biochim. Biophys. Acta 163,44-49 12 Hooghwinkel, G.J.M. and Van Niekerk, H.P.G.A. (1960) Proc. Acad. Sci Amsterdam, Ser. B 63, 475-483 13 Fairbanks, G., Steck. T.L. and Wallach, D.F.H. (1971) Biochemistry 10, 26062617 14 Matsushita, S. and Kobayaahi. M. (1970) Agr. Biol. Chem. 34. 825-829 15 Gamage, P.T., Mori. T. and Matsushita, S. (1971) Agr. Biol. Chem. 35, 33-39 16 Garnage, P.T. and Matsushita. S. (1973) Agr. Biol. Chem. 37, l-8 17 Lenard, J. (1970) Biochemistry 9. 1129-1132 18 Marchesi, S.L.. Steers, E., Marchesi. V.T. and Tillack. T.W. (1970) Biochemistry 9, 50-57 19 Carraway, K.L and Shin, B.C. (1972) J. Biol. Chem. 247. 2102-2108
494 20 Bhakdi, S.. Kniifermann. H.. Schmidt-Ulrich. R., Fischer. H. and Wallach, D.F.H. (1974) Biochim. Biophys. Acta 363,39-53 21 Carraway, K.L., Triple& R.B. and Anderson, D.R. (1975) Biochim. Biophys. Acta 379, 571-581 22 Steck. T.L. (1972) J. Mol. Biol. 66. 295-305 23 Wang. K. and Richards, F.M. (1974) J. Biol. Chem. 249. 8005-8018 24 Girotti. A.W. (1975) Biochem. 14.3377-3383 25 Paste, G., Papabadjopoulos, D., Jacobson, K. and Vail, W.J. (1975) Biochim. Biophys. Acta 394, 520-539 26 Steck, T.L. (1974) J. Cell Biol. 62, l-19 27 Ji, T.H. and Nicolson. G.L. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 2212-2216
