Regulation of the Glycophorin C-Protein 4.1 Membrane-to-Skeleton Bridge and Evaluation of Its Contribution to Erythrocyte Membrane Stability*

The band 3-ankyrin-spectrin bridge and the glycophorin C-protein 4.1-spectrin/actin bridge constitute the two major tethers between the erythrocyte membrane and its spectrin skeleton. Although a structural re-quirement for the band 3-ankyrin bridge is well established, the contribution of the glycophorin C-protein 4.1 bridge to red cell function remains to be defined. In order to explore this latter bridge further, we have identified and/or characterized five stimuli that sever the linkage in intact erythrocytes and have examined the impact of this rupture on membrane mechanical properties. We report here that elevation of cytosolic 2,3-bisphosphoglycerate, an increase in intracellular Ca 2 1 , removal of cell O 2 , a decrease in intracellular pH, and activation of erythrocyte protein kinase C all promote dissociation of protein 4.1 from glycophorin C, leading to reduced retention of glycophorin C in deter-gent-extracted spectrin/actin skeletons. Significantly, where mechanical studies could be performed, we also observe that rupture of the membrane-to-skeleton bridge has little or no impact on the mechanical properties of the cell, as assayed by ektacytometry and nickel mesh filtration. We, therefore, suggest that, although regulation of the glycophorin C-protein 4.1-spectrin/actin bridge likely occurs physiologically, the role of the tether and the associated regulatory changes remain to be established. The shape and deformability

The shape and deformability of the human red blood cell (RBC) 1 membrane are thought to be maintained by a combination of protein-protein and protein-lipid interactions. Prominent among the former are associations within the spectrin-actin skeleton (often termed horizontal interactions), and bridges between the membrane skeleton and various membrane-spanning proteins (frequently called vertical interactions) (1,2). The horizontal interactions are considered important, since they maintain the architecture and stability of the protein network underlying the lipid bilayer (3)(4)(5). The vertical interactions are thought to be essential, since they permit stress sharing between the lipid bilayer and protein skeleton (6), and because they prevent aggregation of membrane-spanning proteins and the consequent vesiculation of protein free lipid domains (7,8).
Although several minor integral membrane proteins are believed to attach to the membrane skeleton (9 -11), the band 3-ankyrin-spectrin bridge and the glycophorin C-protein 4.1-spectrin/actin bridge are thought to constitute the major tethers between the membrane skeleton and the lipid bilayer (12). Evidence that the band 3-ankyrin-spectrin bridge is critical to membrane mechanical properties derives primarily from analyses of cells with genetic defects in one or more members of this bridge. Thus, mutations leading to reduced expression or abnormal folding of band 3, ankyrin, or spectrin generally lead to membrane vesiculation and hereditary spherocytosis (3,4,(13)(14)(15)(16). Other studies aimed at manually disrupting the band 3-ankyrin-spectrin bridge also document a causal relationship between rupture of the bridge and loss of normal cell morphology and stability (6).
Evidence for the importance of the glycophorin C-protein 4.1-spectrin/actin bridge to membrane shape and flexibility also stems from investigations of natural defects in the bridging components. Thus, deficiencies in either protein 4.1 or glycophorin C commonly result in elliptocytic cells with compromised mechanical properties. However, although such observations might initially be construed to suggest that the protein 4.1-mediated bridge to the membrane skeleton is critical to membrane stability, observations by Narla and colleagues (17)(18)(19)(20) raise questions regarding the validity of such an hypothesis. Thus, quantitative investigations of the above elliptocytic cells have revealed that a natural deficiency in glycophorin C is invariably accompanied by a decrease in protein 4.1 (17,18) and that reconstitution of only the spectrin-actin binding domain of protein 4.1 into protein 4.1-deficient cells fully restores the membrane's mechanical properties (19,20). Because the above reconstitution procedure does not re-establish protein 4.1's attachment to glycophorin C or the lipid bilayer, these data suggest that the protein 4.1-mediated bridge has little impact on membrane stability.
In order to examine the contribution of the glycophorin C-protein 4.1-spectrin/actin tether to red cell membrane structure in greater detail, we have undertaken to manually disrupt the bridge without compromising protein 4.1's established role in stabilizing the spectrin-actin junctional complex. For this purpose, we have explored methods to break the protein 4.1-mediated tether in intact cells and have examined the impact of this rupture on membrane mechanical properties. We report here four novel and one previously published protocol for weakening/severing the protein 4.1 linkage to glycophorin C. Although not all of the protocols allow for an unambiguous evaluation of the role of the glycophorin C linkage in establishing membrane mechanical properties, the collective results support the contention that the glycophorin C-protein 4.1-spectrin/actin bridge does not contribute prominently to membrane stability.
Protocols for Rupturing the Glycophorin C-Protein 4.1 Bridge in Intact RBCs-RBCs were sedimented at 800 ϫ g for 5 min, washed twice in phosphate-buffered saline (PBS, pH 7.4) and added to an ␣-cellulose-Sigma Cell column to remove white cells and platelets. Collected RBCs were then washed with HEPES buffer (125 mM NaCl, 3 mM KCl, 16 mM HEPES, 10 mM glucose, 2 mM CaCl 2 , 1 mM MgCl 2 , pH 7.4) and resuspended in the same buffer prior to subjection to one of the following protocols for severing glycophorin C-protein 4.1 interactions. In the first methodology, washed RBCs were incubated at 40% hematocrit for 1 h at 37°C with increasing concentrations of phosphoenolpyruvate. Phosphoenolpyruvate is rapidly transported into erythrocytes and converted to 2,3-bisphosphoglycerate (21), which is reported to inhibit protein 4.1 binding to the red cell membrane in vitro (22). In a second protocol, RBCs at 5% hematocrit were loaded with Ca 2ϩ by treatment with 1 M A23187 and various calcium-EGTA buffers (23) for 30 min at 37°C. The influx of Ca 2ϩ would be expected to down-regulate protein 4.1-membrane interactions based on its ability to prevent 4.1 binding to band 3 and glycophorin C in vitro (24,25). Third, RBCs at 5% hematocrit were incubated for 30 min at 37°C with 0.1 M phorbol ester (PMA dissolved in Me 2 SO) in order to activate protein kinase C (26). As controls, the inactive analog of PMA, 4␣-PDD, or an equal volume of Me 2 SO was added to analogous cell suspensions. Whenever protein kinase or phosphatase inhibitors were also employed, they were added to the red cell suspensions 30 min prior to PMA treatment. Fourth, RBCs suspended at 5% hematocrit in deoxygenated HEPES buffer were bubbled with N 2 gas for ϳ5-30 min (depending on the volume of the suspension) at room temperature to deoxygenate the cells. To evaluate the reversibility of this treatment, reoxygenation was carried out by mixing the deoxygenated RBCs in oxygenated HEPES buffer. Finally, RBCs at 5% hematocrit were equilibrated for 3 h in MES buffers (27) of pHs ranging from pH 5.5 to 7.4.
Preparation of Membrane Skeletons-After the desired incubations, packed RBCs were collected and mixed with an equal volume of PBS. Membrane skeletons were then isolated by incubating the suspensions in two volumes of 1% C 12 E 8 (Nikko Chemical Co.) in PBS supplemented with 0.5 mM dithiothreitol, 20 g/ml leupeptin, 20 g/ml pepstatin A, 1 mM EDTA, and 80 g/ml phenylmethanesulfonyl fluoride. The mixture was loaded onto a 35% sucrose cushion and centrifuged at 85,500 ϫ g for 90 min to separate the solubilized membrane components from the pelleted membrane skeletons. The pellet was solubilized in SDS buffer, and its protein concentration was measured using the bicinchoninic acid assay.
Electrophoretic and Immunoblot Analysis of Membranes and Membrane Skeleton Components-Samples dissolved in SDS sample buffer were separated on 10% Laemmli gels, as described elsewhere (28). After transfer to nitrocellulose membranes, nonspecific protein binding was blocked by incubating the nitrocellulose (Bio-Rad) for 1 h in 5% milk-TBST (50 mM Tris, 200 mM NaCl, 0.05% Tween 20, pH 7.5). The nitrocellulose membranes were then incubated with a mouse monoclonal anti-glycophorin C antibody (BRIC 10, a kind gift from D. J. Anstee, International Blood Group Reference Library, Bristol, United Kingdom), anti-actin, anti-adducin (a kind gift from V. Bennett, Duke University, Durham, NC), anti-band 3, anti-protein 4.1, or anti-glycophorin A in 1% milk-TBST for 1 h. The resulting nitrocellulose membranes were washed three times with 0.1% milk-TBST, incubated with horseradish peroxidase-conjugated secondary antibody in 1% milk-TBST for 30 min, and washed three times with 0.1% milk-TBST. Finally, the membranes were incubated with ECL chemiluminescent reagent (Amersham Pharmacia Biotech) and exposed to x-ray film. When reprobing the same nitrocellulose sheet was necessary, the nitro-cellulose membrane was incubated with stripping buffer (100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris-HCl, pH 6.7) for 45 min at 70°C, washed three times with TBST, and then relabeled with a different antibody and developed, as described above. Quantitation of the glycophorin C content in the membrane skeletons by densitometry was invariably performed by normalizing its staining intensity to that of actin in the same gel lane. Using this method, the glycophorin C content of control membranes at pH 7.4 was arbitrarily set at 100%.
Analysis of the Phosphorylation of Membrane Proteins-RBCs suspended at 50% hematocrit in HEPES buffer were incubated with 0.5 mCi of 32 PO 4 /ml of packed cells for 2 h at 37°C. The cells were washed twice with HEPES buffer before PMA treatment and SDS-polyacrylamide gel electrophoresis, as described above. Identification of phosphorylated membrane polypeptides was made by analyzing the resulting polyacrylamide gel on a phosphorimager (Cyclone, Packard Instrument Co.).

Measurement of 2,3-BPG Content in RBC-
The concentration of 2,3-BPG in deproteinized extracts of phosphoenolpyruvate-treated erythrocytes was measured using a Sigma diagnostic kit. Briefly, red cells were suspended at 50% hematocrit in PBS, mixed with three volumes of trichloroacetic acid, and pelleted at 8000 ϫ g. The supernatant was neutralized with NaOH and mixed with a triethanolamine assay mixture containing NADH, ATP, phosphoglycerate mutase, 3-phosphoglycerate phosphokinase, and glyceraldehyde-3-phosphate dehydrogenase. The conversion of NADH to NAD was initiated by addition of phosphoglycolate, and the amount of 2,3-BPG in the extracts was calculated from the amount of NADH consumed, using the absorbance at 340 nm to quantitate NADH consumption.
Deformability and Stability Measurements-After the desired treatment, packed RBCs were suspended in 4% polyvinylpyrrolidone (M r 360,000) in PBS and examined by the ektacytometry, a laser diffraction method described elsewhere (29). In brief, suspended cells were subjected either to increasing osmolality (from 50 to 500 mosmol/kg) at constant shear stress or to constant osmolality at increasing shear stress (0 -250 dynes/cm 2 ), and the axial ratio of the deformed cells was quantitated by laser diffraction and designated as their deformability index (DI). A measurement of DI as a function of osmolality and shear stress can provide information on both the hydration of the cell and the flexibility of the membrane.
For stability measurements, packed RBCs were lysed in 40 volumes of 10 mM sodium phosphate (pH 7.4) and pelleted at 20,000 ϫ g. After centrifugation, the lysed cells were resealed by adding 20 volumes of PBS and incubating for 45 min at 37°C. The resealed ghosts were washed one time in PBS, resuspended in a 35% dextran-phosphate buffer, and analyzed by ektacytometry at constant high shear stress (785 dynes/cm 2 ), as described above. The decline in deformability with time derives from the mechanical fragmentation of the membranes into nondeformable vesicles. The half-time for this shear-induced fragmentation to reach completion can be used as a measure of the membrane's mechanical stability (30).
Filterability Measurements-Analysis of the rate of filtration of an RBC suspension through a highly uniform nickel mesh filter was conducted by a gravity-based, vertical tube method (31) that measures the rate of passage of erythrocytes through the filter as a function of hydrostatic pressure (Tsukusa Sokken Co., Tokyo, Japan). The pore diameter of the nickel mesh used in this study was 4.6 m. From a height (pressure)-time curve obtained during the filtration analysis, a pressure-flow rate relationship can be determined. The filtration was started at a pressure of 150 mm H 2 O, and the percentage of the flow rate (ml/min) of the RBC suspension (0.1% hematocrit) relative to that of the suspending medium at 100 mm H 2 O was taken as the index of RBC filterability.

RESULTS
A variety of observations in the literature, stemming primarily from studies on purified membrane components, have suggested that the glycophorin C-protein 4.1 bridge to the membrane skeleton might be physiologically regulated (22,24,(32)(33)(34). However, except for the impact of protein kinase C phosphorylation on this bridging function (35), none of the potential regulatory pathways has been validated in intact erythrocytes. As a tool for investigating the significance of the glycophorin C-protein 4.1-spectrin/actin bridge to membrane mechanical properties, we have undertaken to characterize methods that might sever this bridge in intact erythrocytes. Although not all of the hypothetical mecha-nisms could be validated in situ, those that were confirmed (or newly discovered) will be briefly characterized below, since they reveal much about the possible regulation of the glycophorin C-protein 4.1-spectrin/actin bridge in vivo. Where such treatments also shed light on the function of the glycophorin C-protein 4.1 bridge, they will be further evaluated for their impacts on cell stability and deformability.
Effect of Elevated 2,3-Bisphosphoglycerate-In a previous study (22), we demonstrated that 2,3-BPG inhibited protein 4.1 binding to stripped inside-out erythrocyte membrane vesicles. More detailed studies with antibodies to band 3 and glycophorin C further established that protein 4.1 binding to both membrane-spanning proteins was specifically prevented by BPG. In an effort to test the impact of elevated 2,3-bisphosphoglycerate on protein 4.1's bridging function in situ, we incubated RBCs with PEP, a membrane-permeable glycolytic substrate that is rapidly metabolized to 2,3-bisphosphoglycerate and lactate in situ (21). Calibration of this method in whole erythrocytes reveals that cytoplasmic BPG levels can be elevated in direct proportion to the concentration of phosphoenolpyruvate added (Fig. 1A). By 25 mM external PEP, intracellular BPG rises to 9.3 mM, essentially twice its normal value.
To evaluate the influence of BPG on the protein 4.1-glycophorin C linkage in situ, PEP loaded RBCs were injected into a 1% solution of the nondenaturating detergent, C 12 E 8 , and their membrane skeletons were isolated on a sucrose density gradient. As shown in Fig. 1B, at low PEP concentrations (0 -10 mM), where the total intracellular BPG content was Յ6 mM, little change in glycophorin C retention in the membrane skeletons was observed. However, as PEP concentration was elevated, cytosolic BPG increased and glycophorin C retention in the membrane skeletons was strongly hindered. By 9 mM BPG (25 mM PEP), only 48 Ϯ 8% (mean Ϯ S.D., n ϭ 3) of the original glycophorin C remained in the membrane skeletons. Because the protein 4.1 content of these skeletal preparations was essentially unaltered (data not shown), we conclude that BPG can sever the glycophorin C-protein 4.1 bridge without displacing protein 4.1 from the spectrin skeleton.
Unfortunately, PEP loading was also observed to induce hemoglobin oxidation (probably due to depletion of NADH via the lactate dehydrogenase reaction), leading to hemichrome deposition on the membrane (data not shown). Because hemichrome precipitation can independently rigidify the membrane (36), evaluation of the treated cell's mechanical properties was not pursued. Nevertheless, the biochemical data do suggest that endogenous changes in BPG, such as those that occur during oxygenation/deoxygenation (37), might have the potential to regulate the protein 4.1-glycophorin C bridge in vivo.
Effect of Ca 2ϩ Loading-Previous studies from our laboratory (24) and others (38) have shown that Ca 2ϩ plus calmodulin can block protein 4.1 binding to glycophorin C. In fact, a recent crystallographic structure of the N-terminal 30-kDa domain of protein 4.1 reveals that calmodulin associates at a central site on protein 4.1 and that calmodulin's occupancy of this site should obstruct protein 4.1's interaction with glycophorin C, p55, and band 3 (39). To determine whether Ca 2ϩ influx might regulate the glycophorin C-protein 4.1 bridge in vivo, fresh erythrocytes were treated with a Ca 2ϩ ionophore (1 M A23187) and equilibrated with various concentrations of Ca 2ϩ using Ca 2ϩ -EGTA buffers. As shown in Fig. 2A, increasing intracellular Ca 2ϩ concentration indeed releases glycophorin C from its attachment to the membrane skeleton. Although no change in skeletal composition could be detected up to 0.1 mM extracellular Ca 2ϩ , by 0.3 mM Ca 2ϩ a 14% (n ϭ 4) decline in glycophorin C content was observed, and by 1 mM Ca 2ϩ , glycophorin C retention in the membrane skeletons was reduced by 75% (n ϭ 2). Curiously, adducin binding was affected similarly to glycophorin C, whereas the band 3 and protein 4.1 contents in the membrane skeletons remained essentially unchanged. We conclude from these data that Ca 2ϩ can indeed modulate the glycophorin C-protein 4.1 bridge in situ, as predicted by studies in simplified systems (24,38), but that measurable changes only occur when intracellular Ca 2ϩ concentrations exceed those experienced by healthy cells.
Analysis of the mechanical properties of the Ca 2ϩ -loaded erythrocytes was complicated by unwanted side effects of Ca 2ϩ on several unrelated red cell systems. Even when internal and external [K ϩ ] were balanced to prevent Ca 2ϩ -activated K ϩ efflux and cell shrinkage (40,41), Ca 2ϩ stimulation of RBC phospholipase C (42,43), the phospholipid scramblase (44), the transglutaminase (45) (note the protein 4.1 near the top of the gel in Fig. 2A that was presumably cross-linked through ␥-glutamyl-⑀-lysine bridges by the RBC transglutaminase), and adducin release ( Fig. 2A) either preceded or coincided with the Ca 2ϩ -induced release of glycophorin C. Thus, although the maximum deformability index (DI max ) and filterability of the K ϩ balanced erythrocyte suspensions declined by 15% (Fig.  2B) and 44% (Fig. 2C) at 500 M Ca 2ϩ , respectively, in comparison to the DI max and filterability of normal erythrocytes, it was not possible to assign the decline to a rupture of the protein 4.1 bridge. In fact, since no changes in deformability or filterability were observed at 0.3 mM Ca 2ϩ , where protein 4.1 release from glycophorin C had already begun ( Fig. 2A), the absence of any causal relationship between the two processes would seem more plausible.
Effect of Oxygen Content-During our search for physiological modulators of the glycophorin C-protein 4.1 bridge, we reasoned that RBCs might be capable of responding to entrapment in the microvasculature by becoming more flexible. Based on this speculation, we examined the impact of RBC deoxygenation on the integrity of the glycophorin C-protein 4.1 linkage. Deoxygenated RBCs were generated by bubbling N 2 through the cell suspension, and membrane skeletons were isolated by plunging the deoxygenated cells directly into C 12 E 8 detergent solution before any reoxygenation could occur. As seen in Fig.  3A, less glycophorin C was retained in membrane skeletons from deoxygenated cells than oxygenated cells. Densitometric analysis, in fact, indicates that glycophorin C's content was reduced 30 Ϯ 16% (n ϭ 4) following O 2 removal. Restoration of O 2 by washing the cells in O 2 -saturated buffer also restores the glycophorin C content to its normal level (Fig. 3A), demonstrating that O 2 -mediated rupture of the 4.1 bridge is fully reversible. To further confirm that RBC oxygenation can regulate protein 4.1's bridging function, RBCs were treated with 10 mM dithionite to chemically consume the erythrocyte's O 2 , and the extracted membrane skeletons were again examined for glycophorin C retention. As expected (Fig. 3A), dithionite also severs the glycophorin C-protein 4.1 linkage.
Analysis of the mechanical properties of the dithionitetreated cells was unfortunately hindered by the known generation of hemichromes during dithionite treatment (46). Simple deoxygenation of the cells, however, was not compromised by any known limitation or side effect; therefore, normally deoxygenated cells were examined for changes in filterability and deformability. As displayed in Fig. 3B, deoxygenation similar to that achieved in Fig. 3A causes no change in the rate of RBC filtration through a 4.6-m nickel mesh filter. This result suggests that, although the availability of O 2 may modulate the Osmotic deformability was then measured by subjecting the RBCs to a moderate shear stress while increasing osmotic pressure. C, evaluation of the filterability of Ca 2ϩ -plus A23187-treated cells. RBCs were diluted to 0.1% hematocrit in the above high K ϩ HEPES buffer and allowed to flow through a 4.6-m pore-sized nickel mesh filter. The height of the column of suspended erythrocytes was recorded on-line as a function of time, and the filtration rate was derived assuming first order kinetics.

FIG. 3.
A, immunoblot analysis of erythrocyte membrane skeletons derived from deoxygenated RBCs. Erythrocytes were bubbled with nitrogen gas or treated with 10 mM sodium dithionite for 5 to 30 min at 37°C to promote deoxygenation. For reoxygenation studies, previously deoxygenated erythrocytes were washed several times in oxygenated buffer. Membrane skeletons were then isolated and analyzed by SDS-PAGE, followed by immunoblotting with anti-glycophorin C and antiactin antibodies. The anti-actin antibody staining was conducted to ensure even loading of protein in each gel lane. Sodium sulfate was used as a negative control for sodium dithionite, since it does not alter O 2 content. Nitrogen purging for periods as short as 5 min induced measurable changes in glycophorin C retention, whereas bubbling for Ͼ30 min caused no additional changes. B, effect of deoxygenation on the filterability of RBCs. RBCs at 0.1% hematocrit were deoxygenated for 1 h and allowed to flow through a 4.6-m pore-sized nickel mesh filter as described in Fig. 2. Filtration rates were recorded on-line by monitoring the height of the RBC suspension as a function of filtration time.
integrity of the 4.1 bridge, penetration of the deoxygenated erythrocytes through a narrow pore is not altered. In a previous study, examination of deoxygenated cells in an ektacytometer revealed no reproducible changes in deformability (47). We, therefore, doubt that deoxygenated cells have improved rheological properties over oxygenated cells.
Effect of pH-Because of the abundance of anion transporters (band 3) in the erythrocyte membrane, hydroxide can rapidly equilibrate across the membrane, leading to facile changes in intracellular pH. During RBC transit through exercising muscles and the spleen, serum pH can decline to Ͻ6.8, especially when normal blood flow is retarded, raising the question whether pH might be exploited to regulate membrane mechanical properties in vivo. To test whether the protein 4.1 bridge might be subject to pH regulation, RBCs were equilibrated at various external pHs, and examined for changes in membrane skeletal interactions, as described above. Although no change in glycophorin C content could be detected at pH values above 7.4, as pH was decreased below neutrality, the glycophorin C bridge gradually dissolved. By an external pH of 5.5, 45 Ϯ 11% (n ϭ 3) of attached glycophorin C molecules could be readily extracted from the membrane skeletons (Fig. 4A). Significantly, the largest change in glycophorin C extractability was observed between pH 6.8 and 6.0 (Fig. 4B), where pH changes might occur under conditions of stress. These data suggest that the glycophorin C-4.1 bridge to the membrane skeleton might be modulated by pH.
To determine whether the observed rupture of the glycophorin C-protein 4.1 bridge to the membrane skeleton affects the deformability or filterability of the cells, RBCs were equilibrated at various pH levels and examined in both the ektacytometry and nickel mesh filtration apparatus. As shown in Fig.  5A, a decrease in external pH promotes a minor decline in cell hydration (as indicated by the right shift in the osmotic ektacytometry scans; see also Ref. 27), but no significant change in DI at optimum hydration. This result suggests that the intrinsic deformability of the membrane is not altered by partial rupture of the protein 4.1 bridge. Significantly, a similar conclusion is also supported by data on red cell filterability (Fig.  5B). Thus, except for a very modest decrease in filterability due to the aforementioned hydration, no significant decline in filtration rate was observed. It is, therefore, plausible that the pH-induced dissociation of glycophorin C and protein 4.1 occurs with few or no mechanical consequences.
Effect of Stimulation of Protein Kinase C-Activation of protein kinase C in human erythrocytes can be induced by insulin and perhaps other hormones (48,49). Stimulation of erythrocyte protein kinase C leads to phosphorylation of three prominent membrane skeletal proteins: adducin, dematin, and protein 4.1 (26,50,51). Phosphorylation of protein 4.1 has been specifically shown to inhibit its association with stripped inside-out erythrocyte membrane vesicles (34), suggesting that phosphorylation might regulate the glycophorin C-protein 4.1-spectrin/actin bridge in situ. To explore this possibility, RBCs were incubated in 0.1 M PMA, an activator of protein kinase C, and the polypeptide content of the derived membrane skeletons was examined as described above. As noted by Gratzer and colleagues (35), PMA promotes release of glycophorin C from membrane skeletons (Fig. 6A). In our hands, the glycophorin C content was reduced 58 Ϯ 13% (n ϭ 4), and adducin levels were also somewhat diminished. Although lower concentrations of PMA promoted less dissociation, higher concentrations exerted no additional effect (data not shown). Furthermore, the availability of Ca 2ϩ seemed to have no impact on membrane skeleton protein composition, indicating that the PMA-stimulated dissociation of glycophorin C and ␤-adducin was caused by a Ca 2ϩ -independent protein kinase C isoform (Fig. 6A).
To confirm that dissociation of glycophorin C and ␤-adducin was a consequence of their phosphorylation, the influence of FIG. 4. A, immunoblot analysis of erythrocyte membrane skeletons derived from RBCs incubated for 30 min at 37°C at the indicated pHs. After incubation, membrane skeletons were isolated and analyzed by SDS-PAGE, followed by immunoblotting with anti-glycophorin C and anti-actin antibodies. The anti-actin antibody staining was conducted to ensure even loading of protein in each gel lane. B, densitometric analysis of glycophorin C retention in membrane skeletons derived from RBCs incubated at various pHs. The band intensities of glycophorin C and actin from three independent Western blots were measured by densitometry. The ratio of glycophorin C/actin in control skeletons at pH 7.4 was set at 100%, and all other ratios are reported as a percentage of this value.

FIG. 5. Evaluation of the deformability and filterability of RBCs equilibrated at various pHs.
Erythrocytes were incubated at the indicated pHs for 3 h at 37°C, and osmotic deformability was measured by subjecting the RBCs to a moderate shear stress while increasing osmotic pressure (A). Alternatively, equilibrated cells were further diluted to 0.1% hematocrit and allowed to flow through a 4.6-m pore-sized nickel mesh filter. The height of the erythrocytes suspension was recorded as a function of filtration time as the cells were forced by hydrostatic pressure though the nickel mesh filter (B). several modulators on both processes was also compared. First, RBCs were metabolically labeled with 32 PO 4 and the impact of PMA, 4␣-PDD (an inactive analog of PMA), staurosporine (a protein kinase inhibitor), and calyculin A (a phosphatase inhibitor) on the stimulated incorporation of 32 PO 4 into adducin, 4.1, and 4.9 in the presence or absence of calcium was examined. Importantly, 0.1 M PMA induced the phosphorylation of the above proteins, whereas 4␣-PDD did not, regardless of the availability of extracellular Ca 2ϩ (Fig. 7A). And as expected, staurosporine inhibited the PMA-induced phosphorylation, whereas calyculin A enhanced it (Fig. 7B). To verify that these phosphorylation changes were indeed responsible for the observed changes in protein interactions, RBCs incubated with the same kinase/phosphatase inhibitors were again stimulated with PMA and examined for changes in skeletal protein composition (Fig. 6B). In agreement with the phosphorylation studies, staurosporine completely blocked the effect of PMA on the dissociation of glycophorin C, whereas calyculin A enhanced it. As above, 4␣-PDD had no effect on retention of glycophorin C in the membrane skeletons.
By restaining the above blots with antibodies to other components of the membrane skeletons, it was also possible to demonstrate that band 3, protein 4.1, and glycophorin A remained unaffected by PMA treatment (Fig. 6A). Thus, as with the other modulators of the glycophorin C bridge, only glycophorin C and possibly adducin respond with a change in skeleton association. These data would suggest that regulation by protein kinase C is not random, but instead is focused on the bridging interaction that is modulated by many other erythrocyte variables.
To determine whether the aforementioned PMA-induced changes in protein interactions exert an impact on red cell deformability, RBCs were treated with PMA and subjected to increasing osmolality during shear stress. As revealed by the ektacytometry, no significant changes in the DI either as a function of osmolality (Fig. 8A) or shear stress (Fig. 8B) were observed in response to PMA, even for incubation periods up to 2 h. Although higher concentrations of PMA (above 3 M) did lead to decreases in DI (data not shown), we attribute these changes to nonspecific effects, since maximal phosphorylation is attained at 0.1 M PMA (26) and higher concentrations of PMA do not further enhance protein phosphorylation or dissociation of glycophorin C or ␤-adducin from the skeletons (data not shown).
In order to obtain a more sensitive measure of the effect of PMA on membrane mechanical properties, resealed ghosts were prepared from PMA-treated or untreated cells and examined for their mechanical stability during exposure to increasing shear stress. As seen in Fig. 8C, the half-time of shear-induced fragmentation of ghosts prepared from cells incubated with PMA was identical to that from cells incubated with 4␣-PDD (inactive analog) and control cells incubated in buffer. These data confirm that RBCs treated with PMA retain their normal deformability and stability, despite the release of part of their glycophorin C and adducin from the membrane skeletons.
In a final attempt to identify a mechanical consequence of the PMA-induced release of glycophorin C and adducin from the membrane skeletons, we examined the effect of PMA on the filterability of RBCs through a 4.6-m nickel mesh. Following exposure to 0.1 M PMA for 30 min at 37°C, where the dissociation of glycophorin C and ␤-adducin was previously shown to be maximal, the filtration rate of the treated cells was found to decrease by less than 5% (Fig. 9). After a much longer incubation time (2 h), an average decrease in filterability of ϳ13% (compared with RBCs treated with the same concentration of 4␣-PDD) was observed (Fig. 9). However, since the longer incubation time exerted no further effect on retention of glycophorin C or ␤-adducin in the membrane skeletons, such a decrease in filterability may not be related to the integrity of membrane-to-skeleton bridges. More likely, the decrease in filterability could derive from other unrelated effects of protein kinase C, such as loss of phospholipid asymmetry (52) or increase in ion channel activity (53). DISCUSSION The human erythrocyte is commonly considered biologically inactive, essentially unable to respond to signals from within and outside the cell with meaningful changes in cellular properties. We have, however, demonstrated that fresh intact erythrocytes do indeed sense changes in 2,3-bisphosphoglycer-ate (a modulator of hemoglobin-O 2 affinity), calcium, O 2 levels, intracellular pH, and protein kinase C activity, and respond to these changes by altering the number of bridges connecting glycophorin C to the membrane skeleton. Although each of these membrane rearrangements could admittedly be entirely fortuitous, collective consideration of the data may suggest otherwise. Thus, none of the above modulators affects the band 3-ankyrin-spectrin bridge and none of the stimulants alters the content of protein 4.1 in the membrane skeletons. In fact, except for minor changes in adducin retention, all of the changes are specific for the glycophorin C-protein 4.1 bridge. Furthermore, the most prominent modulatory effects appear to be situated near the physiological ranges of the above stimulants and several of the stimulants might be most prominent in erythrocytes retarded in deoxygenated tissues. Although we cannot yet assign a function to the modulations, we nevertheless hypothesize that the ability to regulate the integrity of the glycophorin C bridge to the junctional complex might somehow improve the fitness of the circulating erythrocyte.
The most unanticipated observation from this study was that dissociation of glycophorin C from the membrane skeletons does not impact membrane deformability, stability, or whole cell filterability. Several explanations of this finding are worth considering. First, the glycophorin C bridge may not be highly important to membrane mechanical properties. Although protein 4.1-deficient erythrocytes are abnormally shaped and mechanically unstable, reconstitution of the spectrin-actin binding fragment of protein 4.1 into these cells restores normal behavior (19). Since the glycophorin C binding domain is not contained within this fragment, one can easily argue that the glycophorin C bridge to the junctional complex is unimportant to membrane stability. In this scenario, the instability of glycophorin C-deficient membranes might simply be attributed to the concomitant partial deficiency in protein 4.1 (54).
Second, the lack of impact of glycophorin C dissociation could derive from the short duration of our treatments. Thus, release of glycophorin C from the spectrin skeleton could enable only slow weakening of other interactions more critical to membrane stability. Although we extended our investigations of membrane mechanical properties for at least 2 h following treatment, red blood cells must circulate for ϳ120 days in vivo; hence, the true significance of the glycophorin C interaction may only emerge when a cell is stressed for longer periods of time.
A third possible explanation for the absence of obvious mechanical consequences is that complete release of glycophorin C  were then further diluted to 0.1% hematocrit and allowed to flow through a 4.6-m pore-sized nickel mesh filter. Filtration rates were recorded on-line as described in Fig. 2. was never achieved. Whether part of the protein 4.1 population is inaccessible to each type of regulation or whether other components render the retained population of glycophorin C less extractable cannot be ascertained from the data, but clearly not all copies of glycophorin C are subject to similar modulation. Perhaps the extractable fraction of glycophorin C was already contributing little to membrane stability.
Fourth, it should not be ignored that ektacytometry and nickel mesh filtration could easily be insensitive to the mechanical changes induced by rupture of the protein 4.1 bridge. Thus, PMA treatment has been shown to delay the onset of the discocyte to echinocyte transition upon ATP depletion or Ca 2ϩ entry (55) and to directly cause erythrocyte morphological changes (56), albeit at concentrations much higher than those required to induce maximal protein phosphorylation. Based on these observations, it would be surprising if PMA did not modulate some mechanical properties of the membrane. Whether such mechanical changes only occur at excessive PMA concentrations or whether our current techniques are too insensitive to detect them must obviously await further investigation.
Finally, the glycophorin C-protein 4.1 bridge could serve a function unrelated to membrane mechanical properties. Instead, the tether to the membrane could simply position the spectrin/actin skeleton closer to the bilayer where it could better interact with other membrane components such as metabolic enzymes, ion channels/transporters, adhesion receptors, or phospholipid flippases/scramblases, etc. The impact of changes in BPG, Ca 2ϩ , O 2 , pH, and protein kinase C activity would then presumably be manifested in the regulation of these other membrane functions. It will be important to evaluate whether bilayer-to-skeleton tethers can indeed participate in the regulation of nonstructural membrane functions.