Assembled clathrin in erythrocytes.

Clathrin cages were isolated from rat erythrocytes. These structures exist in the intact cell as demonstrated by immunofluorescence and were not formed during the isolation procedure. The cages were largely devoid of membrane but contained the assembly protein complex and both the 50-kDa kinase (pp50) and casein kinase II activities found previously in clathrin-coated vesicles.

* This research was supported by Grant GM-31579 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
. D. Bar-Zvi, B. Mosley, and D. Branton, manuscript in preparation. with either antibodies to the clathrin light chains or antibodies to the assembly protein.

EXPERIMENTAL PROCEDURES
Materials-Poly-DL-lysine and ATP were obtained from Sigma.
Carrier-free "Pi was obtained from Du Pont-New England Nuclear.
[ T -~~P ] A T P was prepared as described previously (Bar-Zvi and Branton, 1986). Bicinchoninic acid protein reagent was purchased from Pierce Chemical Co. Calf brain coated vesicles were prepared as described previously (Bar-Zvi and Branton, 1986).
Antibodies-Anti-clathrin light chain antibodies were obtained from a rabbit injected with native clathrin triskelions purified from bovine brain coated vesicles. The antibodies were affinity-purified on a column containing native clathrin light chains coupled to Sepharose. Rabbit antibodies were raised against polypeptides eluted from the 100-110-kDa region of an SDS'-polyacrylamide gel on which highly purified bovine brain coated vesicles had been resolved. These antibodies were likewise purified on an affinity column containing 100-110-kDa proteins coupled to Sepharose. Both anti-clathrin light chain and anti-100-110-kDa antibodies specifically immunoprecipitated the corresponding antigens and identified them on immunoblots of coated vesicles.' Erythrocyte Preparation-The following steps were performed at 4 "C. Blood was obtained from male white rats (250-400 g) by heart puncture. Approximately 10 ml of fresh blood was mixed with 3 ml of ice-cold anticoagulant solution containing 75 mM sodium citrate and 38 mM citric acid. Cells were washed by brief centrifugation in Hepes-buffered saline (20 mM Hepes-NaOH, pH 7.4,130 mM NaCl), and leukocytes and platelets were removed by adsorption with a mixture of a-cellulose and Sigmacell (1:l). Erythrocytes were separated from reticulocytes and remaining white blood cells by centrifugation through a 45/70% Percoll step gradient for 10 min at 2500 rpm in a Sorvall SS34 rotor (without braking). Erythrocytes were recovered in the pellet, while reticulocytes banded at the 45/70% Percoll interface, and a mixture of lysed cells and remaining leukocytes banded at the 0/45% Percoll interface. Percoll was removed by washing the cells twice in Hepes-buffered saline. Human blood from healthy donors was obtained from the Red Cross and treated as above, with omission of the Percoll gradient step.
Clathrin Cage Preparation-Washed erythrocytes were resuspended in KC1 isolation buffer containing 1 mg/ml phenylmethylsulfonyl fluoride (Bar-Zvi and Branton, 1986) and lysed by freezing in liquid nitrogen and thawing or, alternatively, by addition of 0.5% Triton X-100. Lysed cells were centrifuged at 12,000 X g for 20 min, and the supernatant was recentrifuged at 100,000 X g for 30 min. The pellet from this step was homogenized in the isolation buffer and recentrifuged at both low and high speeds as above. The resulting pellet was centrifuged at 12,000 X g for 5 min to remove aggregated material. The supernatant was applied to a 7.5-30% glycerol gradient in KC1 isolation buffer and centrifuged at 24,000 rpm for 2 h (brake off) in a Beckman SW 27.1 rotor. Fractions of 1 ml each were obtained from gradient tubes by piercing the bottom. Fractions containing clathrin cages were identified by SDS-PAGE. Cages were sedimented from diluted fractions by centrifugation at 100,000 X g for 1 h. This procedure routinely yielded approximately 5 pg of clathrin cages/ml of packed cells.
Immunofluorescence-Fresh rat blood cells were washed in phosphate-buffered saline and applied to glass coverslips previously coated with 1 mg/ml poly-L-lysine. After a 30-min incubation, attached cells were fixed overnight a t 4 "C in 3% HCHO in Small's cytoskeleton buffer (Small, 1981). Cells were made permeable by incubation for 3 min a t room temperature with 0.2% Triton X-100 in Tris-buffered saline containing 10 mM Tris-HCI, pH 7.5, 150 mM NaCl and then incubated in the same buffer containing 10% neonatal calf serum for 1 h. Individual coverslips were then incubated with diluted, affinitypurified rabbit antibodies for 3 h a t room temperature. Coverslips were rinsed with three changes of Tris-buffered saline containing 0.1% Tween 20 and then incubated with affinity-purified fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Boehringer Mannheim) for 1 h. After rinsing in three changes of Tris-buffered saline containing 0.1% Tween 20, coverslips were mounted in polyvinyl alcohol containing 1 mg/ml p-phenylenediamine and observed with an epifluorescence-equipped Zeiss microscope.
Other Procedures-In vitro phosphorylations were carried out as described previously (Bar-Zvi and Rranton, 1986). Immunoprecipitations and in vivo phosphorylations were performed as described elsewhere,' with the exception that the washing solution was made with KC1 isolation buffer in place of NaF isolation buffer. Immunoprecipitated proteins were quantified in Coomassie Blue-stained gels with the aid of a Hoefer GS 300 scanning densitometer. SDS-PAGE was carried out in 10% polyacrylamide gels as described by Laemmli (1970). Immunoblots were performed as described by Olmsted (1981). Protein was determined according to Smith et al. (1985) using bovine serum albumin as standard. Relative phospholipid concentrations were determined by diphenylhexatriene fluorescence (Prasad et al., 1985).

RESULTS AND DISCUSSION
A polypeptide of the same electrophoretic mobility as the 180-kDa clathrin heavy chain was sedimented when an erythrocyte lysate was centrifuged at 100,000 X g for 30 min. Further fractionation of this high speed pellet in a glycerol gradient produced a peak which cosedimented with purified coated vesicles isolated from either rat reticulocytes or bovine brain (Fig. 1). The protein components of the peak fractions (lanes 8-10] resembled those of rat reticulocyte coated vesicles' and included major polypeptides of 180 kDa (clathrin heavy chain), 100-110 and 50 kDa (assembly protein), and 33 kDa (clathrin a-light chain) and a minor 30-kDa polypeptide (clathrin P-light chain) (Fig. 1). The identities of the 180-, 100-110-, and 33-kDa polypeptides were confirmed by their cross-reaction on immunoblots with antibodies raised against bovine brain clathrin heavy chain, assembly protein 100-110-kDa polypeptides, and clathrin light chains, respectively (Fig.

2).
The principal light chain of clathrin shown to be present in erythrocytes was the a-light chain, although traces of the Mr l X 1 6 9 180-' 110- Erythrocytes were purified from 20 ml of rat blood and lysed by freeze-thaw in KC1 isolation buffer. The high speed pellet was applied to a 7.5-30% glycerol gradient and centrifuged as described under "Experimental Procedures." One-ml fractions were collected from the gradient, and protein was precipitated from 0.3 ml of each fraction by the addition of trichloroacetic acid and Triton X-100 to 10% (w/ v) and 0.1% (w/v), respectively. Protein precipitates were collected by centrifugation a t 12,000 X g for 5 min, washed with cold (-20 "C) acetone, resuspended in SDS-PAGE sample buffer, heated, and analyzed by SDS-PAGE. T, top of gradient. clathrin fraction from the glycerol gradient was resolved by SDS-PAGE, and proteins were transferred to nitrocellulose. The blot was incubated with the following affinity-purified antibodies to bovine brain coat proteins. Lane 1, anti-clathrin heavy chain; lone 2, anti-100-110-kDa polypeptides of the assembly protein; lone 3, anticlathrin light chains; lone 4, nonimmune rabbit I&. @-light chain were also detected. Analysis by densitometry of Coomassie Blue-stained reticulocyte coated vesicle proteins indicated a molar ratio of at least 10 clathrin a-light chains per &light chain,'

33-
in agreement with Davis and Bennett (1985), who reported one type of clathrin light chain in erythrocyte triskelions. The molar ratio of clathrin monomer to assembly protein, calculated from densitometric scans of the 180-and 100-110-kDa gel band regions, is 5.5 2 1.0 (n = 5); by comparison, we obtained a ratio of 1.9 f 0.4 (n = 5) for these proteins in brain coated vesicles, similar to the stoichiometries reported by Zaremba and Keen (1983) and Pearse and Robinson (1984). Davis and Bennett (1985) suggested that all erythrocyte triskelions should be soluble, given the presence of an excess of uncoating protein. This is not in agreement with our findings (Fig. 1). Davis and Bennett (1985) purified clathrin from human erythrocytes lysed in 7.5 mM Napi, 0.5 mM EDTA. However, we were able to isolate cages successfully from both human and rat erythrocytes lysed in Davis and Bennett's buffer (not shown), suggesting that differences in the cell source or lysis buffer do not account for the difference in results. We suspect that clathrin cages existed in I ' nrythrocyte lysate of Davis and Bennett but probably embled during the subsequent chromatography steps \r111ch our isolation scheme avoids.
Negative staining of peak fractions from the glycerol gradient revealed a multitude of assembled clathrin structures (Fig. 3). Most of these structures appeared to be cages, while about 20% of them (n = 128) appeared to contain material within the cage (Fig. 3, arrowheads). Because both coated vesicles and cages composed of clathrin and assembly protein may show an electron-dense core region (Vigers et al., 1986 and footnote 3), it was not possible to distinguish between these two types of structures based on their appearance. The lack of membrane in the erythrocyte clathrin cages was confirmed by the fractionation of brain coated vesicles and erythrocyte cages, preincubated with the fluorescent lipid probe diphenylhexatriene, on a glycerol gradient. Although the diphenylhexatriene fluorescence cosedimented with the brain W. Bazari and D. Branton, manuscript in preparation.
~~ coated vesicles, no fluorescence peak was detected in association with erythrocyte clathrin cages (Fig. 4), suggesting that the majority of these structures lack membrane.
To rule out the possibility that the clathrin cages which we isolated originated in contaminating non-erythrocyte blood cells, we compared the quantity of clathrin and assembly protein immunoprecipitated from the purified erythrocyte fraction with that immunoprecipitated from the crude fraction which contained lymphocytes and reticulocytes, in addition to erythrocytes (Table I). Whereas greater than 99% of the white blood cells and reticulocytes were removed from whole blood by the erythrocyte purification procedure (Beutler et al., 1976;Rennie et al., 1979), such purification reduced the amount of immunoprecipitated coat protein only slightly, suggesting that the clathrin and assembly protein indeed originated in erythrocytes and not in contaminating cells. Furthermore, unlike the high speed pellet prepared from reticulocytes, the erythrocyte pellet did not contain ribosomes,

FIG. 3. Electron microscopy of erythrocyte cages.
Erythrocyte cages were prepared as described in Fig. 1. A peak clathrincontaining fraction was negatively stained with 3% uranyl acetate. Bars, 100 nm.

Immunoprecipitation of coated vesicle proteins from unfractionuted blood cells and purified erythrocytes
Rat blood cells were washed twice in Hepes-buffered saline by centrifugation for 10 min at 12,000 X g. The washed cells are referred to above as unfractionated cells. Erythrocytes were further purified by removal of white blood cells by absorption to cellulose, followed by centrifugation through a 45/70% Percoll step gradient as described under "Experimental Procedures." Equivalent amounts of cells were lysed in KC1 buffer supplemented with 0.5% Triton X-100 and centrifuged at 12,000 X g for 20 min. Proteins from samples of either 0.35 or 0.7 ml of the supernatants were immunoprecipitated with antibody to either clathrin light chains or to the 100-110-kDa polypeptides of the assembly protein. Immunoprecipitates were resolved by SDS-PAGE and stained with Coomassie Blue. Clathrin and assembly protein were quantified by scanning the 180-and 100-110-kDa bands, respectively, with a Hoefer GS 300 scanning densitometer. confirming that the pellet was not derived from a cell containing intracellular organelles (not shown). About one-half of the total clathrin and one-third of the assembly protein were found to be in the assembled form by immunoprecipitation of these protein complexes from erythrocytes following cell fractionation (not shown). To test for possible post-homogenization assembly of the erythrocyte coat proteins, 12sI-triskelions from rat liver were mixed with the erythrocyte lysate. Immediate contact with erythrocyte contents was ensured in this experiment by addition of labeled triskelions to the lysis buffer prior to the addition of erythrocytes. Following lysis, the incorporation of the label into assembled structures was analyzed (Table 11). Essentially no incorporation of labeled triskelions was observed, indicating that no assembly occurred during lysis and subsequent steps. This conclusion is in agreement with Goud et al. (1985), who showed that under similar conditions there is no assembly of clathrin during the lysis and fractionation of rat brain cortex. However, the added clathrin assembled as expected, following dialysis against 20 mM Mes-NaOH, 2 mM CaCI2 buffer (Table  11). The possibility that assembled clathrin is disassembled during the steps subsequent to erythrocyte lysis would not be consistent with the stability of coated vesicles in the lysis buffer; coated vesicles isolated in this buffer remained in an assembled state for at least 2 months at 4 "C.4 The existence of clathrin cages in the intact erythrocyte was further supported by immunofluorescent staining using affinity-purified antibodies to the light chains of clathrin or to the assembly protein. Staining with either antibody produced a punctate distribution of fluorescence (Fig. 5), in agreement with the typical punctate patterns shown for the localization of clathrin heavy chain (Anderson et al., 1978), clathrin light chains (Lisanti et al., 1982), and the 100-kDa polypeptides of the assembly protein (Robinson and Pearse, 1986) in other cells. After staining with either anti-clathrin light chain or anti-100-110-kDa antibodies, the number of fluorescent dots per cell varied from cell to cell. Furthermore, more than one-half of the cells in a typical microscope field were stained with the anti-clathrin light chain antibody (panel A), while only about one-eighth were stained by the anti-100-110 kDa antibody (panel B ) . Although the erythrocyte ghosts were difficult to see by phase contrast, staining of ghosts on ' D. Bar-Zvi, unpublished observation. identically prepared coverslips with antibodies to spectrin showed that ghosts were distributed at a similar density on all coverslips and were made equally permeable to antibodies by detergent extraction (not shown). Thus, the disparity in the proportion of cells stained by our two antibodies appears to be real and accords with the measured excess of clathrin over assembly protein in erythrocytes (see above).
Erythrocyte clathrin cages possess both of the protein kinase activities (50-kDa kinase and casein kinase 11) shown to be associated with clathrin-coated vesicles. Incubation of erythrocyte cages with M$'-[y-32P]ATP resulted in the la-' beling of the 50-kDa polypeptide (Fig. 6, lane 1 ), which is consistent with the previously described autophosphorylation of the assembly protein (Pauloin et al., 1982;Campbell et al., 1984;Keen et al., 1987). When phosphorylation was assayed in the presence of polylysine, a stimulator of casein kinase I1 (Bar-Zvi and Branton, 1986), both of the clathrin light chains and to a lesser extent, the 100-110-kDa assembly polypeptides were labeled ( l a n e 2). Casein kinase was shown to preferentially phosphorylate the clathrin @-light chain (Usami et al., 1985;Schook and Puszkin, 1985;Bar-Zvi and Branton, 1986). Although both erythrocyte clathrin light chains were labeled in the presence of polylysine, when the higher abundance of the clathrin a-light chain ( Fig. 1) was taken into account, the specific activity of the phosphorylated @-light chain was at least 10 times higher than that of the a-light chain. Because the clathrin cages lack membrane (Figs. 3 and 4) but possess casein kinase I1 activity (Fig. 6), we conclude that casein kinase I1 is associated with the coat proteins and not with the membranes of coated vesicles.

TABLE I1
Clathrin i s not assembled during the handling of erythrocyte cell lysate Clathrin was purified from rat liver (Bar-Zvi and Branton, 1986) and iodinated with Bolton-Hunter reagent (Hanspal et al., 1984). Two pg of labeled triskelions were added to the erythrocyte lysis buffer (KC1 isolation buffer with 0.5% Triton X-100 and 1 mg/ml phenylmethylsulfonyl fluoride). One portion of the buffer was added to an erythrocyte pellet (erythrocyte lysates), while an equal portion was not mixed with erythrocytes but was otherwise treated identically (lysis buffer). Samples were incubated on ice for 2 h and then centrifuged at 100,000 X g for 30 min (without dialysis) or dialyzed overnight against 20 mM Mes-NaOH, 2 mM CaC12, pH 6.2, before centrifugation (with dialysis). Radioactivity was determined in supernatant and pellet fractions using a Beckman y-counter. [ The presence in erythrocytes of cages composed of clathrin and assembly protein is unexpected, and the role of clathrin in mature erythrocytes is not clear. Two possibilities may be suggested that clathrin is nonfunctional while gradually disappearing from the cell or that it maintains a residual function. In agreement with the first possibility, erythrocytes do not possess intracellular organelles and are not believed to carry out membrane transport processes or receptor-mediated endocytosis. Reticulocytes, by contrast, are very active in the endocytosis of transferrin (for hemoglobin synthesis) and other ligands. Maturation of reticulocytes to erythrocytes involves degradation of many proteins and cell components (Rappoport et al., 1974;Van Bockxmeer and Morgan, 1979). Further changes occur during the aging of mature erythrocytes: cells become more dense (Rennie et al., 1979), and the plasma membrane becomes more rigid (Shiga et aZ., 1979). ATP-dependent proteolysis activity declines rapidly upon reticulocyte maturation and decreases further with the age of the erythrocytes (Speiser and Etlinger, 1982). The activities of several metabolic enzymes, including pyruvate kinase, acetylcholinesterase and phosphoglycerate kinase decrease as a function of cell age (Kadluboski and Agutter, 1977;Cohen et al., 1976;Rennie et al., 1979). Consistent with these changes, 1 2 FIG. 6. Erythrocyte cages possess two protein kinase activities. Erythrocyte cages were incubated for 10 min at 20 "C with 1 mM MgClz and 50 p~ [y-3ZP]ATP without ( l a n e I ) or with (lane 2 ) 50 pg/ml polylysine. Reactions were quenched by the addition of electrophoresis sample buffer, samples were resolved by SDS-PAGE, and labeled protein was visualized by autoradiography. Arrowheads indicate the phosphorylated 100-110-, 50-, 33-, and 30-kDa coat polypeptides.
FIG. 5. Immunofluorescent localization of erythrocyte coat proteins. Fixed erythrocyte ghosts were prepared as under "Experimental Procedures" and stained with affinity-purified antibodies to clathrin light chains ( A ) , or 100-110-kDa polypeptides of the assembly protein ( B ) , or with nonimmune IgG (C).
All antibodies were used at an IgG concentration of 5 pg/ml. Bars, 5 pm. the quantity of assembled clathrin in rat erythrocytes is less than 10% of that found in rat reticulocytes.' It is thus possible that the disparity in the proportion of cells stained by our two antibodies (Fig. 5) results from a difference in the rate of degradation of the two proteins during reticulocyte maturation and erythrocyte aging.
On the other hand, a possible role for clathrin is suggested by observations of recycling and down-regulation of insulin receptors in erythrocytes (Peterson et al., 1983). The number of insulin receptors per cell has been shown to decrease exponentially with erythrocyte age (Dons et aL, 1981;Wilson and Peterson, 1986). In contrast, the maximal extent of insulin receptor down-regulation decreases linearly with age of the cell (Wilson and Peterson, 1986), which suggests that some factor other than the copy number of the receptor itself participates in its regulation. Since receptor recycling and down-regulation have been shown to involve clathrin-coated membranes in other cell types , it is likely that some of the coat proteins are involved in these processes in erythrocytes. cellent technical assistance.