The cytoskeletal system of nucleated erythrocytes. I. Composition and function of major elements

We have studied the dogfish erythrocyte cytoskeletal system, which consists of a marginal band of microtubules (MB) and trans-marginal band material (TBM). The TBM appeared in whole mounts as a rough irregular network and in thin sections as a surface-delimiting layer completely enclosing nucleus and MB. In cells incubated at 0 degrees C for 30 min or more, the MB disappeared but the TBM remained. MB reassembly occurred with rewarming, and was inhibited by colchicine. Flattened elliptical erythrocyte morphology was retained even when MBs were absent. Total solubilization of MB and TBM at low pH, or dissolution of whole anucleate cytoskeletons, yielded components comigrating with actin, spectrin, and tubulin standards during gel electrophoresis. Mass-isolated MBs, exhibiting ribbonlike construction apparently maintained by cross-bridges, contained four polypeptides in the tubulin region of the gel. Only these four bands were noticeably increased in the soluble phase obtained from cells with 0 degrees C- disassembled MBs. The best isolated MB preparations contained tubulin but no components comigrating with high molecular weight microtubule- associated proteins, spectrin, or actin. Actin and spectrin therefore appear to be major TBM constituents, with tubulin localized in the MB. The results are interpreted in terms of an actin- and spectrin- containing subsurface cytoskeletal layer (TBM), related to that of mammalian erythrocytes, which maintains cell shape in the absence of MBs. Observations on abnormal pointed erythrocytes containing similarly pointed MBs indicate further that the MB can deform the TBM from within so as to alter cell shape. MBs may function in this manner during normal cellular morphogenesis and during blood flow in vivo.     

Spectrin band 1 and 2 are antigenically distinct and are not composed of subunits.

Human erythrocyte membranes and freshly isolated spectrin were separated into their constituent peptides by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The peptides were electrophoresed from slices of such gels into agarose gels containing anti-spectrin antibodies and Triton X-100. In fresh preparations, precipitin arcs were observed only against peptides migrating as bands 1 and 2. It was found that bands 1 and 2 did not cross-react. There were two major arcs from band 1 and one principal arc from band 2, plus minor splitting of these arcs. None of the band 1 arcs fused with band 2 arcs. In fresh erythrocyte ghosts only bands 1 and 2 reacted with anti-spectrin; bands 2.1, 3, and 5, in particular, showed no precipitin arcs. However, in aged ghosts, arcs appeared in the band 3 region; in aged isolated spectrin, arcs appeared in the band 2.1 region; and in trypsin-degraded spectrin, reactive species occurred in all molecular weight classes. It is concluded that spectrin has no subunits smaller than 220,000 molecular weight and that bands 1 and 2 are immunochemically distinct.     

Photodynamic action of bilirubin on human erythrocyte membranes. Modification of polypeptide constituents.

The photodynamic action of bilirubin on isolated human erythrocyte membranes (ghosts) has been studied. When incorporated into ghosts (pH 8.0,10 degrees) the bile pigment photosensitizes in blue light the peroxidation of unsaturated lipids, as evidenced by a positive color reaction with 2-thiobarbituric acid. Accompanying lipid peroxidation was the disappearance of most of the major membrane proteins (Coomassie Blue staining in sodium dodecyl sulfate-polyacrylamide gel electrophoresis) and appearance of polypeptide photoproducts of greater size (mol wt greater than 250,000). The association of membrane proteins (presumably by cross-linking) was insignificant when bilirubin-ghost suspensions were kept in the dark, or when ghosts were irradiated in the absence of bilirubin. Electrophoretic bands 1 and 2 (Fairbanks, G., Steck, T.L., and Wallach, D. F.H (1971), Biochemistry 10, 2606) diminished rapidly during the photoreaction, whereas band 3 and the three sialoglycoproteins disappeared at a much slower rate. Dispersal of membrane consituents by treatment with sodium dodecyl sulfate prior to irradiation resulted in relatively little peroxidation and no noticeable formation of high molecular weight polypeptide complexes. The possibility that malonaldehyde, a product of lipid peroxidation, is involved in cross-linking during irradiation was studied by incubating ghosts with exogenous malonaldehyde. Although the reagent did cross-link membrane proteins (electrophoretic bands 1, 2, 2.1 2.2, and 4.1 diminished most rapidly and high molecular weight bands appeared), the reaction could only be demonstrated with malonaldehyde concentrations several orders of magnitude greater than those detected in irradiation experiments. If malonaldehyde cross-linking occurs, it does not appeare to be the predominant mechanism of polypeptide association during irradiation of bilirubin-containing ghosts.     

The effect of endogenous proteases on the spectrin binding proteins of human erythrocytes

We have demonstrated that in human erythrocyte ghosts endogenous proteolytic activity is responsible for the digestion of the spectrin binding proteins (bands 2.1 to 2.6). The pH optimum, cofactor requirements and inhibitor sensitivity have been established. Our results indicate that proteolysis of bands 2.1 to 2.6 and the formation of 3′, a fragment containing an active spectrin binding site, can occur through two enzymatic pathways: a cascade of consecutive proteolytic cleavages of the spectrin binding proteins inhibited by phenylmethylsulfonyl fluoride or a Ca²⁺-stimulated, phenylmethylsulfonyl fluoride-insensitive, EDTA-inhibited cleavage of band 2.1 to band 2.3, followed by digestion to band 3′ by phenylmethylsulfonyl fluoride-inhibitable enzymes. These findings may provide the techniques necessary to prevent proteolysis of the spectrin binding proteins during purification and reconstitution experiments and provide insight into how they are formed in vivo.     

Purification of the (Ca2+-Mg2+)-ATPase from human erythrocyte membranes using a calmodulin affinity column.

The (Ca2+-Mg2+)-ATPase from human erythrocyte membranes has been solubilized in Triton X-100 and purified on a calmodulin affinity chromatography column in the presence of phosphatidylserine, to limit the inactivation of the enzyme. The enzyme was purified at least 150 times when compared with the original ghosts and showed a specific activity of 3.8 mumol.mg-1.min-1. In sodium dodecyl sulfate-polyacrylamide gels, a single major band was visible at a position corresponding to a molecular weight of about 125,000; a minor band (11% of the total protein) was present at a position corresponding to Mr = 205,000. Upon incubation of the purified preparation with [32P]ATP, both bands were phosphorylated in proportion to their mass, suggesting that both were active forms of purified ATPase.     

Vesicle release from rat red cell ghosts and increased association of cell membrane proteins with cytoskeletons induced by cadmium

When rat red cell ghosts were incubated with 0.1-0.5 mM CdCl2 in 10 mM Tris-HCl (pH 7.4) at 37 degrees C for 30 min, they became irregular in shape and released small vesicles. The release of vesicles was dependent on the incubation temperature and Cd2+ concentration. The maximum release occurred at 37 degrees C in the presence of 0.2 mM Cd2+. The protein composition of Cd2+-induced vesicles was similar to that of the vesicles released from ATP-depleted red cells. Upon incubation with 0.1-0.2 mM Cd2+, more than 90% of the Cd2+ added to the incubation buffer was recovered in ghosts and 15-20% of the ghost Cd2+ was located on the cytoskeletons prepared by washing ghosts with 0.5% Triton X-100 solution containing 0.1 M KCl and 10 mM Tris-HCl (pH 7.4). Moreover, the cytoskeletons prepared from Cd2+-treated ghosts markedly contained cell membrane proteins, bands 2.1, 3, 4.2 and 4.5, and glycophorins. The association of bands 3 and 4.2 with cytoskeletons increased with increasing concentrations of Cd2+ added to the incubation buffer and saturated at 0.2 mM Cd2+. The solubilization of cytoskeletal proteins, bands 1, 2 and 5, from ghosts at low ionic strength was almost completely suppressed by preincubation of ghosts with 0.1 mM Cd2+. HgCl2, PbCl2 and ZnCl2 at 0.2 mM each also produced an increased association of cell membrane proteins with cytoskeletons, whereas CaCl2 and MgCl2 did not.     

Integral membrane protein interaction with Triton cytoskeletons of erythrocytes

The organization of erythrocyte membrane lipids and proteins has been studied following the release of cytoplasmic components with the non-ionic detergent Triton X-100. After detergent extraction, a detergent-resistant complex called the erythrocyte cytoskeleton is separated from detergent, solubilized lipid and protein by sucrose buoyant density sedimentation. In cytoskeletons prepared under isotonic conditions all of the major erythrocyte membrane proteins are retained except for the integral protein, glycophorin, which is quantitatively solubilized and another integral glycoprotein, band 3, which is only 60% removed. When cytoskeletons are prepared in hypertonic KCl solutions, band 3 is fully solubilized along with bands 2.1 and 4.2 and several minor components. The resulting cytoskeletons have the same morphology as those prepared in isotonic buffer but they are composed of only three major peripheral proteins, spectrin, actin and band 4.1. We have designated this peripheral protein complex the 'shell' of the erythrocyte membrane, and have shown that the attachment of band 3 to the shell satisfies the criteria for a specific interaction. Although Triton did affect erythrocyte shape, cytoskeleton lipid content and the activity of membrane proteases, there was no indication that Triton altered the attachment of band 3 to the shell. We suggest that band 3 attaches to the shell as part of a ternary complex of bands 2.1, 3 and 4.2.     

Calmodulin. An activator of human erythrocyte (Ca2+ + Mg2+)ATPase phosphorylation

The effect of purified calmodulin on the calcium-dependent phosphorylation of human erythrocyte membranes was studied. Under the conditions employed, only one major peak of phosphorylation was observed when solubilized membrane proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The molecular weight of this phosphorylated protein band was estimated to be 130000 and in the presence of purified red blood cell calmodulin, the rate of phosphorylation of this band was increased. These data suggest that calmodulin activation of (Ca2+ + Mg2+)-ATPase could be a partial reflection of an increased rate of phosphorylation of the (Ca2+ + Mg2+)-ATPase of human erythrocyte membranes.     

Identification of Surface Proteins on Viable Plasmodium knowlesi Merozoites

Viable merozoites of Plasmodium knowlesi were isolated and the proteins that were labeled on intact merozoites by lactoperoxidase-catalyzed radioiodination were identified. Sodium dodecyl sulfate polyacrylamide gel electrophoresis and autoradiography of Triton soluble extracts of labeled merozoites demonstrated eight major bands ranging in apparent molecular weight from 150,000 D to 22,000 D. Exposure of intact merozoites to trypsin (10 μg/ml) for 10 min resulted in the loss of the two highest molecular weight proteins (150,000 D and 105,000 D) and the appearance of two new bands at 70,000 D and 62,000 D. Trypsin treatment under these conditions also removed the receptor(s) for merozoite attachment to erythrocytes. Therefore, these high molecular weight proteins are candidates for the merozoite component that attaches to erythrocytes. There was no evidence that the labeled membrane components were serum or erythrocyte membrane components, two potential contaminants in the preparation. Anti-rhesus erythrocyte antibody did not precipitate labeled merozoite proteins. Furthermore, the immunoprecipitation of labeled merozoite proteins by rhesus anti-merozoite serum was not inhibited by erythrocyte ghosts.     

Modification of human erythrocyte ghosts with transglutaminase

Guinea pig liver transglutaminase was shown to catalyze the incorporation of dansylcadaverine and putrescine into two major protein fractions of human erythrocyte ghosts. As judged by sodium dodecylsulfate gel electrophoresis under reducing conditions, one of these is a high molecular weight polymer which may contain spectrin. The other corresponds to band 3, an 88, 000 dalton polypeptide. Amine substrates of transglutaminase were synthesized with specific properties to further explore this useful enzymatic technique of covalently labelling proteins in erythrocyte ghosts and in other biological membranes.     

Binding of ras p21 to bands 4.2 and 6 of human erythrocyte membranes

The direct binding protein(s) of ras p21 was (were) investigated in inside-out vesicles of human erythrocyte ghosts using the pure v-Kirsten (Ki)-ras p21 synthesized in E. coli. The bound ras p21 was detected immunochemically using an anti-v-Ki-ras p21 monoclonal antibody, ras p21 bound to vesicles. Prior digestion of the vesicles with trypsin reduced this binding significantly. When ras p21 was laid over vesicle proteins immobilized on a nitrocellulose sheet by transfer from the gel of SDS-polyacrylamide gel electrophoresis, ras p21 bound to bands 4.2 and 6. ras p21 binding to these proteins was reduced by prior incubation of ras p21 with the purified band 4.2 or 6 protein. These results indicate that v-Ki-ras p21 can bind directly to bands 4.2 and 6 of human erythrocyte membranes as far as tested in an in vitro cell-free system.     

Effects of calmodulin on the (Ca2+ + Mg2+)ATPase partially purified from erythrocyte membranes.

The stimulation of the (Ca²⁺ + Mg²⁺)ATPase of erythrocyte ghosts by calmodulin was observed not only in intact ghosts, but also in the solubilized (Triton X-100) and partially purified, reconstituted (phosphatidylserine liposomes) forms. Since the solubilized form of the enzyme migrated on Sepharose 6B at a position corresponding to a molecular weight of about 150,000, these results show that calmodulin stimulates by direct interaction with the ATPase complex. Additionally, the effects of calmodulin on erythrocyte ghosts prepared by the Dodge-EDTA method (hypotonic ghosts) and by the method of Ronner et al. (involving lysis followed by an isotonic wash repeated several times) were compared (P. Ronner, P. Gazzotti, and E. Carafoli, 1977, Arch. Biochem. Biophys. 179, 578–583). The (Ca²⁺ + Mg²⁺)ATPase of the hypotonic ghosts was low and was stimulated by added calmodulin while that of the isotonic ghosts was high and changed only slightly upon calmodulin addition; this difference in response to calmodulin persisted in the solubilized and reconstituted forms. Hypotonic ghosts bound ¹²⁵I-labeled calmodulin, while isotonic ghosts did not. This comparison of two types of ghosts showed that isotonic ghosts possess an intact calmodulin-(Ca²⁺ + Mg²⁺)ATPase complex, and that the calmodulin remained with the ATPase during solubilization and reconstitution. The isotonic preparation is a particularly useful method of preparing ghosts with an intact calmodulin-ATPase complex, since it requires no special equipment and produces an enzyme activity which is stable to freezing.     

Characterization of immunoglobulin binding to isolated human erythrocyte membranes: evidence for selective, temperature-induced binding of naturally occurring autoantibodies to the cytoskeleton

Human plasma contains naturally occurring autoantibodies to the predominant components of the erythrocyte membrane: band 3 and spectrin bands 1 and 2 of the cytoskeleton. The titer of cytoskeletal plasma autoantibodies increases in various hemolytic conditions, suggesting that opsonization of the cytoskeleton may play an important role in the clearance of hemolyzed (not senescent) erythrocytes from the circulation. In this study, we use Alexa Fluor 488 goat anti-human IgG conjugate (Molecular Probes, Eugene, OR, USA), to characterize plasma immunoglobulin binding to erythrocyte membranes from osmotically hemolyzed cells ('ghosts'). The results show that exposure of ghosts to plasma results in 4-fold more immunoglobulin binding to the cytoskeleton than is bound to the proteins contained within the lipid bilayer. Preincubation of the ghosts at 37 degrees C causes 8-fold more immunoglobulin binding to the cytoskeleton compared to bilayer proteins. This temperature-induced change resulted from selective immunoglobulin binding to the cytoskeleton, with no change in immunoglobulin binding to bilayer proteins. However, the rate of increase in cytoskeletal antigenicity at 37 degrees C did correlate with the rate of a conformational change in band 3, a transmembrane protein which serves as a major membrane attachment site for the cytoskeleton. The results of this study suggest that the cytoskeleton is the primary target in the opsonization of hemolyzed erythrocyte membranes by naturally occurring plasma autoantibodies. The conformational changes which occur in ghosts at 37 degrees C are associated with selective exposure of new immunoglobulin binding sites on the cytoskeleton, and with a change in the structure of band 3. We propose a model suggesting that opsonization of the cytoskeleton occurs prior to the decomposition of hemolyzed erythrocytes at 37 degrees C.     

Solubilization of the binding protein from Ehrlich ascites cells and erythrocytes to Pseudomonas aeruginosa cytotoxin.

The binding protein for pore-forming Pseudomonas aeruginosa cytotoxin was solubilized from Ehrlich ascites cell plasma membranes and rabbit and bovine erythrocyte ghosts using nonionic and zwittergent detergents. Analysis of solubilized plasma membranes from Ehrlich cells by a ligand-blot technique after separation by SDS-PAGE/electrophoretic transfer to nitrocellulose or affinity chromatography showed a protein of 70 kDa molecular mass, which binds to cytotoxin. The binding protein solubilized from rabbit erythrocyte ghosts showed a molecular mass of 50 kDa and that from bovine ghosts 55 kDa according to the former test. The binding proteins could be characterized as acidic. They contain a glycan moiety which is, however, not involved in the interaction of cytotoxin with the binding site.     

Characterization of a (Ca2+ + Mg2+)-ATPase activator bound to human erythrocyte membranes

Incubation of human erythrocyte ghosts with an equal volume of 0.2 mM EDTA in isotonic KCl decreased both the activity and Ca2+ sensitivity of the (Ca2+ + Mg2+)-ATPase remaining associated with the membrane. Readdition of the EDTA-extract activated the (Ca2+ + Mg2+)-ATPase activity. The activator activity was trypsin sensitive, heat stable and retained by a phenothiazine affinity column, consistent with properties expected of calmodulin. However, unlike calmodulin, the activity was not retained by DEAE Sephadex A-50 and it eluted from Sephacryl S-200 as heterogeneous peaks of activator activity of apparent molecular weight between 107,000 and 178,000. Nevertheless, the activator in the EDTA extract both before and after gel filtration contained calmodulin, as determined by radioimmunoassay and by its activation of calmodulin - deficient phosphodiesterase. SDS-gel electrophoresis of the activator isolated by gel filtration showed a protein of Mr 56,000 in addition to a low molecular weight protein corresponding to calmodulin. It is suggested that the red cell membrane contains a calmodulin binding protein which tightly binds calmodulin as a polymeric complex in a Ca2+-independent manner.     

Association of a receptor and G-protein-regulated phospholipase C with the cytoskeleton.

Approximately 98% of turkey erythrocyte phospholipase C (PLC) is cytosolic and is released by hypotonic lysis of the cells and extensive washing of the resultant erythrocyte ghosts. Well washed turkey erythrocyte ghosts retain a fraction of tightly associated PLC, which is activated by the P2y-purinergic receptor and G-protein present in ghost membranes. The particulate PLC is sufficient to couple to all the available purinergic receptor-regulated G-protein. In contrast to ghosts, turkey erythrocyte plasma membrane preparations contain no detectable PLC. To investigate the subcellular location of the ghost-associated PLC, cytoskeletons were prepared by Triton X-100 extraction of turkey erythrocyte ghosts. The ghost-associated PLC was quantitatively recovered in cytoskeleton preparations. Cytoskeleton-associated PLC was solubilized by sodium cholate extraction, partially purified, and shown to reconstitute with PLC-free plasma membrane preparations in an agonist and guanine nucleotide-dependent fashion, indicating that the cytoskeleton-associated PLC is G-protein-regulated. Dissociation of erythrocyte ghost cytoskeletons with the actin-binding protein DNase 1 resulted in a dose-dependent inhibition of agonist and guanine nucleotide-stimulated PLC responses in ghosts and caused release of PLC from ghost or cytoskeleton preparations. These data demonstrate the specific association of a receptor and G-protein-regulated PLC with a component of the detergent-insoluble cytoskeleton and indicate that the integrity of the actin cytoskeleton is important for localization and effective coupling of PLC to the relevant G-protein.     

Studies on toxin of Aspergillus fumigatus. XXII. Fashion of binding of Asp-hemolysin to human erythrocytes and Asp-hemolysin-binding proteins of erythrocyte membranes

The fashion of binding of Asp-hemolysin to human erythrocytes and the isolation of Asp-hemolysin-binding proteins from erythrocyte membranes were investigated by the immunocytochemical technique and affinity chromatography. Asp-hemolysin bound best at a pH range from 5 to 7. The erythrocytes treated with Asp-hemolysin showed diffuse, ring-like or cap-like staining by the peroxidase-labeled antibody method under the light microscope. The distribution of Asp-hemolysin on the erythrocyte surface was clearly observed as patches or caps in the scanning electron microscope. The erythrocyte ghosts were extracted with 1% sodium deoxycholate-0.1 M Tris-HC1 buffer (pH 7.5) containing 0.2 M NaCl and 1 mM EDTA, and the extract was chromatographed on an affinity column consisting of Asp-hemolysin attached to activated thiol-Sepharose 4B. Four proteins present in the membrane extract were retained by activated thiol-Sepharose 4B and eluted with 50 mM cysteine as toxin-membrane components. Sodium dodecyl sulfate polyacrylamide gel electrophoresis indicated that the polypeptides correspond to band 2.1, one protein of the 2 region, band 3 and band 7 in the Steck nomenclature system.     

Similarities between the Mr 245,000 calmodulin-binding protein of the dogfish erythrocyte cytoskeleton and alpha-fodrin

The Mr 245,000 calmodulin-binding protein of the dogfish erythrocyte cytoskeleton (D245) has been compared with human erythrocyte spectrin and mammalian brain fodrin [J. Levine and M. Willard (1981) J. Cell Biol. 90, 631-643]. Mammalian erythrocyte alpha-spectrin, brain alpha-fodrin, and D245 are all localized in the cell surface-associated cytoskeleton, and have similar molecular weights. Like mammalian erythrocyte spectrin, D245 was extracted from erythrocyte ghosts under low-ionic-strength conditions. However, D245 failed to bind an antibody which reacted strongly with both subunits of human erythrocyte spectrin. Unlike mammalian erythrocyte alpha- and beta-spectrin, D245 bound calmodulin in the absence of urea both in a "gel-binding" assay and in situ using azidocalmodulin [D.C. Bartelt, R.K. Carlin, G.A. Scheele, and W.D. Cohen (1982) J. Cell Biol. 95, 278-284]. Striking similarities were noted between D245 and alpha-fodrin in that both exhibited (a) comparable calcium-dependent calmodulin binding properties, (b) strong reactivity with two different anti-fodrin antibody preparations, (c) similar reactivity with antibody to brain CBP-I, now believed to be fodrin, (d) proteolytic degradation yielding an Mr 150,000 calmodulin-binding fragment, and (e) lack of reactivity with an anti-spectrin antibody. A protein with calmodulin-binding and anti-fodrin-binding properties similar to D245 was detected in cytoskeletal preparations of chicken erythrocytes. Moderate and consistent cross-reactivity of anti-fodrin with human erythrocyte alpha-spectrin was also observed. The data indicate that D245 is functionally and immunologically more closely related to alpha-fodrin than to alpha-spectrin of the mammalian erythrocyte.     

Spin-labeled erythrocyte membranes: direct identification of nitroxide-conjugated proteins

The covalent incorporation of a spin-labeled analog of N-ethyl maleimide into erythrocyte membrane proteins has been monitored by electron paramagnetic resonance spectroscopy and the individually labeled proteins detected by immunoblotting techniques, using an anti-nitroxide antibody, following electrophoretic separation of the membrane components. Spin-label was primarily found in the high molecular weight bands (I and II) with no incorporation in proteins with molecular weights less than 35,000. Increasing the reaction time between the spin-label and ghosts altered both the observed labeling pattern and the epr spectra with an increase in the proportion of strongly-immobilized species.     

Glyceraldehyde-3-phosphate dehydrogenase in the isolated human erthrocyte membrane: selective displacement by bilirubin.

When unsealed erythrocyte ghosts in 6 mm phosphate buffer (pH 8.0, 4 °C) were incubated with bilirubin in excess of 0.1 mm and washed with buffer, a single polypeptide component (band 6 in sodium dodecyl sulfate-polyacrylamide-gel electrophoresis) diminished and was recovered in the supernatant fraction. Release of this component was virtually complete at 1 mm initial bile pigment. Since band 6 was believed to be the protomer of membrane-bound glyceraldehyde-3-phosphate dehydrogenase (G3PD), assays for this enzyme in bilirubin-treated ghosts were carried out. These revealed that enzymatic activity decreased concurrently with the disappearance of band 6. The molecular weight of the eluted polypeptide was found to be 36,000, in agreement with the known value for the G3PD protomer. When Mg²⁺-resealed ghosts were washed after exposure to 1 mm bilirubin, less than 20% of the G3PD was eluted, which is consistent with the fact that the enzyme is attached to the cytoplasmic face of the membrane. NAD⁺ in concentrations up to 2 mm displaced no more than 15% of the G3PD from unsealed ghosts. However, after preincubation with NAD⁺ (1 mm) followed by bilirubin (1 Mm) and washing, loss of G3PD was similar to that observed in the absence of cofactor. Since NAD⁺ did not attenuate release of the enzyme, it appears unlikely that such release is attributable to binding of bilirubin at the active site. Protoporphyrin acted similarly to bilirubin on unsealed ghosts, whereas rose bengal had a more pronounced effect, removing all enzymatic activity when the dye concentration reached 0.2 mm. Electrophoretic analysis of ghosts after rose bengal treatment, however, revealed a diminution not only of band 6 but bands 1, 2, and 5 as well.