Comparison of the phosphorylation events in membranes from proliferating vs. quiescent endothelial cells

JOURNAL OF CELLULAR PHYSIOLOGY 130:228-244 (1987)

Comparison of the Phosphorylation Events in Membranes From Proliferating vs.

Quiescent Endothelial Cells ANDRIUS KAZLAUSKAS AND PAUL E. DiCORLETO*

Atherosclerosis Section, Department of Cardiovascular Research, The Cleveland Clinic Foundation, Cleveland, Ohio 44106 (A. K., P. E.D.), and Department of Chemistry,

Cleveland State University, Cleveland, Ohio 44115 (A. K.)

In an attempt to elucidate the intracellular events regulating the proliferation of endothelial cells (EC), we have compared the phosphorylation events in membranes prepared from proliferating (sparse) and quiescent (confluent) EC. Triton-solubilized membranes from sparse and confluent EC were incubated at pH 6.5 in the presence of divalent cations and [32P]ATP. Membrane proteins were then separated by SDS-PAGE and the radiolabeled phosphoproteins visualized by autoradiography. The overall kinase activity per milligram protein was 1.7 2 0.2-fold greater in membranes prepared from proliferating than from quiescent cells. The extent of phosphorylation was dramatically elevated in sparse over confluent samples for four phosphoproteins having the following approximate molecular masses: 180,100,97, and 55 kDa. The 180 and 100 kDa phosphoproteins exhibited 3.6- and 7.4-fold higher labeling, respectively, in sparse than in confluent membranes and both were phosphorylated on serine residues exclusively. The 97 kDa phosphoprotein was 11.6-fold higher in sparse membranes and contained both phosphoserine (p- ser) and phosphotheronine (p-thr), the latter comprising 61% of the radioactivity. The 55 kDA phosphoprotein contained 62% p-ser, 16% p-thr, and 22% phosphotyrosine (p-tyr) and was 2.3-fold higher in sparse membranes. Of these four phosphoproteins, only the 55 kDa protein was phosphorylated in confluent samples to an appreciable degree. Whereas the p-ser and p-thr content of the 55 kDa band increased moderately in sparse vs. confluent sample (1.8-fold increase), the tyrosine residues of this protein in sparse membranes were radiolabeled to a much greater extent relative to confluent membranes (5.4-fold increase). Analysis of the cofactor requirements of the FC membrane kinase(s) revealed that Mn2+ is the optimum cofactor and that Mg2+ can replace MnL+ only for the kinase acting on the 100 kDa band. This suggests the presence of multiple EC membrane kinases. In the presence of both cofactors, the phosphorylation pattern is similar to the pattern obtained with Mn2+ alone. The kinase activity acting on all four phosphoproteins was independent of Ca2+, CAMP, cGMP, and phorbol 12-myristate 13-acetate. The mechanism responsible for the difference in kinase activity of proliferating vs. quiescent cells was not due to an inhibitor or enhanced phosphatase activity in confluent cells; the phosphorylation patterns obtained with sparse solubilized membranes and a mixture of sparse and confluent solubilized membranes were similar. The observed differences in phosphorylation events between sparse and confluent membranes occurred in multiple strains of two types of EC-pig aortic and bovine aortic-but were not apparent in membranes prepared from proliferating and quiescent human foreskin fibroblasts or 3T3 cells. Sparse endothelial cells made quiescent by serum deprivation exhibited reduced kinase activity with a phosphoprotein pattern similar to that of confluent cells; therefore, the enhanced kinase activity in sparse membranes may be growth-dependent.

The growth control of cultured endothelial cells (EC) is highly density-dependent such that at high cell density EC are extremely quiescent, forming a characteris- t i c cobblestone monolayer (Jaffe et al., 1973; Lewis et al., 1973; Booyse et al., 1975; Gimbrone, 1976; Schwa&,,

1978). This density-dependent inhibi t ion of growth is characteristic of many types of normal mammalian cells;

Received 7, 1986; accepted Au@st4, 1986. *TO whom reprint requests/correspondence should be addressed.

c 1987 ALAN R LISS. INC

ENDOTHELIAL CELL PHOSPHORYLATION 229

however, unlike fibroblasts, smooth muscle cells, and other connective tissue cells, confluent monolayers of EC cannot be stimulated to reenter the cell cycle by the addition of fresh serum or appropriate mitogens (Hau- denschild et al., 1976; Westermark, 1976; Holley et al., 1977; Vogel et al., 1980). A second unique characteristic of cultured EC is that they are able to proliferate equally well in either serum- or plasma-supplemented medium (Haudenschild et al., 1976; Davies and Ross, 1978; Kaz- lauskas and DiCorleto, 1985), demonstrating their growth independence from platelet-released mitogens or other exogenous growth factors. Though several EC growth factors have been described including an acidic (Lemmon et al., 1982; Gordon et al., 1983; Maciag et al., 1984a) and a basic (Gospodarowicz et al., 1976, 1983) form of fibroblast growth factor, EC growth factor (Ma- ciag et al., 1982, 1984b), and several other acidic mitogens (Barritault et al., 1982; D’Amore and Klagsbrun, 19841, most laboratories successfully culture arterial EC in plasma- or serum-containing media without further supplementation. Several growth inhibitors specific for EC have been reported (Heimark and Schwartz, 1985; Schumacher et al., 1985), including one that is a membrane component of confluent but not subconfluent EC (Heimark and Schwartz, 1985). The mechanism of action of these growth inhibitors has not yet been determined.

Numerous advances have been made recently in the elucidation of the intracellular events regulating the growth of cultured cells. This research has focused primarily on cells for which growth requirements have been well defined, such as 3T3 cells and fibroblasts. Binding of polypeptide growth factors to specific receptors on such cells stimulates numerous kinases, including intrinsic receptor kinases (Cohen et al., 1980; Ek et al., 1982; Pike et al., 1983; Frackelton et al., 1984), a Ca2+-phospholipid-dependent kinase (Habenicht et al., 1981; Berridge and Irvine, 1984; Nishizuka, 1984), S6 protein kinase (Thomas et al., 19821, and protein tyrosine kinases (Cooper et al., 1982). Protein phosphoryla- tiorddephosphorylation is a primary mechanism by which intracellular events are controlled by the environ- ment of a cell (Cohen, 1982), and it is likely that the initiation and propagation of a mitogenic signal are at least in part mediated by the action of phosphotransfer- ases. Research examining EC growth has largely been directed to defining exogenous growth requirements and describing characteristics in culture rather than deline- ating the intracellular events involved in regulating EC growth. We have chosen to compare the phosphorylation events in proliferating vs. quiescent EC in an attempt to elucidate the mechanisms regulating EC growth. We report that numerous proteins are phosphorylated to a markedly greater extent in proliferating EC membranes. Four of these phosphoproteins and the kinases responsible for their phosphorylation have been examined.

MATERIALS AND METHODS Materials

[Y-~~PIATP (10-40 Ci/mmole) and [3H]thymidine (6.7 Ci/mmole) were purchased from New England Nuclear (Boston, MA). Carrier-free [32P]H3P04 (HC1-free in H2O) was purchased from ICN Radiochemicals (Irvine, CAI. Unless otherwise indicated, all other chemicals were

purchased from Sigma (St. Louis, MO). Media and its supplements were purchased from Grand Island Biolog- ical Co. (Grand Island, NY). Fetal bovine serum was purchased from Hyclone (Logan, UT).

Cell culture Bovine aortic EC (Schwartz, 1978), pig aortic EC

(Schwartz, 1978), and human foreskin fibroblasts (Ross, 1971) were isolated according to previously described methods. Swiss 3T3 mouse embryo cells were obtained from the American Type Culture Collection (Rockville, MD) CCL 92. Cells were characterized as EC by their nonoverlapping cobblestone culture morpholcgy, binding of factor VIII antibody, manifestation of scavenger (acetylated low-density lipoprotein) receptor (Voyta et al., 19841, production of a platelet-derived growth factor (PDGF)-like protein, and absence of PDGF receptors (DiCorleto and de la Motte, 1985; Kazlauskas and Di- Corleto, 1985). Cells were maintained in a humidified, 95% air:5% C02 incubator at 37°C and were used between passages 2 and 20. Culture medium consisted of a 1: l mixture of Dulbecco-Vogt modified Eagle’s medium and Ham’s F12 with the following supplements: sodium bicarbonate (0.24%, v/v), penicillin (100 U/ml), strepto- mycin (100 pg/ml), nonessential amino acids (0.1 mM), sodium pyruvate (1 mM), and L-glutamine (4 mM). Cul- ture medium was supplemented with 5% (v/v) fetal bovine serum. Cells were subcultured with trypsin-EDTA [0.05% trypsin (Worthington, Freehold, NY), 0.02% EDTA] and maintained in fresh 5% fetal bovine serum- containing medium, which was replaced every 2-3 days.

Membrane isolation Pig aortic EC and bovine aortic EC were plated at an

initial cell density of 5,000 cells/cm2 into 150-mm tissue culture dishes (Nunc, Grand Island, NY, or Lux, Naper- ville, L) in culture medium containing 5% fetal bovine serum. 3T3 cells and human fibroblasts were plated at initial cell densities of 2,000 cells/cm2 and 3,000 cells/ cm2, respectively. Two to three days after plating, cells were sparse (= 1.1 x lo4 cells/cm2) and proliferating (verified by autoradiographic determination of labeled nuclei using [3H]thymidine), and membranes isolated from these cells were termed sparse membranes. Repli- cate dishes were allowed to continue growing for 4-5 days until they became confluent ( = 3.9 x lo4 cells/cm2). After waiting 1 additional day to ensure that the cells were quiescent (verified by determining the percent labeled nuclei), membranes were isolated and termed confluent. The membrane isolation procedure was previously described by Pike et al. (1983). Cells were rinsed three times with ice-cold phosphate-buffered saline, scraped off the dish with a cell scraper (Costar, Cam- bridge, MA) in the presence of 6 ml of ice-cold scraping buffer [5 mM 4-(2-hydroxyethyl)-l-piperazine-ethane sulfonic acid (Hepes), 1.2 mM MgC12,5 mM 0-mercaptoeth- anol, pH 7.41, and homogenized at 0°C with 20 strokes of the tightly fitting pestle in a Dounce homogenizer. Homogenized cells were centrifuged for 60 min at 42,OOOg (18,000 rpm) at 4°C in a Beckman 52-21 centri- fuge, JA-20.1 rotor. The supernatant was discarded and the membrane pellet resuspended in ice-cold 20 mM Hepes, pH 7.4, and stored at -70°C. Although this isolation procedure yielded a particulate fraction that, in addition to membranes, contained nuclei, organelles,

230 KAZLAUSKAS AND DICORLETO

and traces of cytoplasmic components, we refer to it as a membrane fraction, as do other groups (Pike et al., 1983). Protein concentration was determined by the method of Bradford (1976) using BSA as a standard.

Kinase assay The standard assay mixture contained -50 pg of

membrane protein isolated from sparse or confluent cells, 60 mM 2-(N-morpholino)ethane sulfonic acid (pH 6.51, 100 mM NaC1, 0.2% Triton X-100, 10 mM MnC12, and 5 mM p-nitrophenyl phosphate in a final volume of 60 pl. After the samples were preincubated for 10 min at O"C, 5-7.5 pM [32P]ATP(5-15 $3) was added and the reaction tubes were placed in a 37°C water bath and incubated for 3 min. Reactions were terminated by the addition of 35 p1 of sodium dodecyl sulfate (SDS) electrophoresis buffer [5% (w/v) SDS, 200 mM dithiothreitol, 250 mM Tris HC1,5.0 mM ethylenediaminetetraacetate (EDTA), 8 M urea, 0.002% bromophenyl blue, 30% su- crose] followed by boiling for 3 min at 100°C. Proteins were separated by SDS-PAGE according to Laemmli (1970) utilizing 5-10% linear gradient separating gels with a 3% stacking gel unless otherwise specified. Gels were run at 10 mA overnight (12-15 hr), stained with Coomassie blue, destained (10% glacial acetic acid, 50% methanol), rehydrated (10% glacial acetic acid, 10% methanol, 1% glycerol), and dried onto thick filter paper. Autoradiography was performed with Kodak XK-1 x- ray films with regular intensifying screens at -70°C. The extent to which specific proteins were radiolabeled was quantitated by excising the appropriate gel band and counting it in the absence of scintillant in a Packard (Downers Grove, IL) scintillation spectrophotometer. To correct for background radioactivity, an equivalent-sized gel piece containing no radiolabeled proteins, and therefore representative of the background level of radioactivity, was excised and quantitated. This value was subtracted from the radiolabeled proteins of that lane.

Other procedures The total kinase activity of sparse and confluent EC

membranes was compared by quantitating the radioactivity incorporated into all the proteins in an SDS polyacrylamide gel lane, including the unresolved proteins running with the dye front of a 5-10% gel. Unhydro- lyzed [32P]ATP runs in front of the dye front in this type of gel.

To measure phosphatase activity, the EC membranes were first phosphorylated in the standard kinase assay lacking p-nitrophenyl phosphate, and then further incorporation of radiolabel was inhibited by the addition of a 1,000-fold excess of unlabeled ATP, and the incubation at 37°C was continued. The phosphatase assay was terminated by the addition of SDS electrophoresis buffer, and samples were analyzed by SDS-PAGE and autoradiography as described above.

To analyze the phosphoamino acid content of EC membrane phosphoproteins, the gel band of interest was excised, digested, and hydrolyzed, and the resultant phosphoamino acids were resolved by thin-layer electrophoresis in the first dimension followed by ascending chromatography in the second dimension, as previously described (Cooper and Hunter, 1981; Ek et al., 1982).

Pig aortic EC were made quiescent at a sparse cell density by the following procedure. Cells were plated at

a density of 10,000 cells/cm2 in culture medium containing 0.1% fetal bovine serum, and after 8 hr the medium was replaced with serum-free medium. Membranes were isolated from these cells after 48 hr in serum-free medium, at which time the rate of [3H]thymidine incorporation into DNA was 5.3% of the rate of cells at a similar cell density (1.2 x lo4 cells/cm2) but in 5% fetal bovine serum.

The percentage of a population of cells that was proliferating was determined autoradiographically by measuring the percentage of labeled nuclei. Cells for membrane isolation and percent labeled nuclei determi- nations were simultaneously plated at the same cell density and cultured in an identical manner, except that, in place of the 150 mm dishes, six-well plates were used in the percentage of labeled nuclei assay. Cells were incubated for 24 hr with 2 pCi/well (1 pCi/ml) 13H]thymidine during time intervals corresponding to the harvesting of cells for membrane isolation at sparse and confluent cell densities. Following the 24 hr labeling period, cells were processed as described by Glenn and Ross (1981), except the trichloroacetic acid wash of fixed-cell monolayers was omitted. The percentage of labeled nuclei was determined by counting at least 300 cells in each well.

RESULTS The growth of EC is highly density-dependent: At low

cell density the cells grow rapidly until they attain con- fluence or saturation density, at which point they cease proliferating, even in the presence of fresh serum-containing medium (Fig. 1). In an attempt to elucidate the means by which EC regulate their growth, we have compared the phosphorylation events occurring in membranes prepared from proliferating (sparse) and quiescent (confluent) pig aortic EC. Membranes from sparse cells were isolated from EC that had been cultured for 2-3 days following plating and were therefore exponen- tially proliferating (Fig. l), and EC that had been in culture for 5-6 days after plating and were therefore quiescent (Fig. 1) were harvested to prepare membranes from confluent cells. To verify that the population of sparse cells was actively proliferating whereas the confluent cells were quiescent, the percentage of nuclei labeled after incubation with [3H]thymidine was determined by autoradiography and found to be 90.0 * 3.0% (n=3) for sparse vs. 1.9 & 1.0% (n=4) for confluent pig aortic EC. Membranes from these two populations of cells were prepared and the kinase activity was compared using equal quantities of membrane protein from sparse and confluent cells. Membranes were solubilized and incubated with [32P]ATP and Mn2+ and the protein components separated on the basis of molecular mass by SDS-PAGE. The level of total phosphotransferase activity was lower in confluent membranes than in sparse membranes: In a typical experiment, a 1.7 & 0.2 (n=4)- fold greater incorporation of [32P]ATP into protein was observed in the sparse than in confluent membranes. Although few differences were noted between the Coo- massie-stained sparse and confluent membrane samples, the autoradiograms of the SDS polyacrylamide gels revealed several differences between growing and quiescent EC in the extent of phosphorylation of specific protein bands (Fig. 2). Included among the proteins that were radiolabeled to a markedly greater extent in prolif-


80 T

40 Ji z 0 j 30 y1

y1 U

20

10

0 0 2 4 6 8 10

DAYS AFTER CULTURING CELLS

Fig. 1. Growth of pig aortic EC in 5% FBS. Pig aortic EC were plated into six-well plates at a density of 5,000 cells/cm2 in 2 ml culture medium containing 5% fetal bovine serum (FBS). Culture medium was aspirated and replaced with fresh 5% FBS-containing culture medium every 2-3 days for the duration of the experiment. At the indicated time intervals, cells were detached with trypsin and counted in a hemacytometer. Data points represent the mean of three wells; error bars represent the SD of the mean.

erating cells were two in the 250 kDa range and 180, 130, 100, 97, and 55 kDa proteins as well as several broad bands. Although the extent of phosphorylation of all of these phosphoproteins may reflect important differences between proliferating and quiescent EC, only the 180, 100, 97, and 55 kDa phosphoproteins will be discussed in this report. Some proteins, for example, the doublet of approximately 44 kDa, were labeled to similar extents in both samples (cpm sparsekpm confluent = 0.91). No phosphoproteins were labeled to a dramatically greater extent in confluent than in sparse membranes. All samples were separated via single-dimension gel electrophoresis, so it is possible that some of the radiolabeled bands are composed of several unresolved proteins.

Gel bands containing each of the four phosphoproteins were excised, and the amount of radiolabel incorporated was quantitated by measuring the Cerenkov radiation.

S c s C

M W X I O - ~

- 200

4

4 < - 97.4

- 68

4

- 43

I - 25.7 Fig. 2. Phosphorylation of sparse and confluent EC membranes. Equal quantities bf membrane protein isolated from sparse and confluent endothelial cells were incubated in the presence of [32P]ATP and Mn2+ at pH 6.5 as described in Materials and Methods. Membrane proteins were separated on a 5-10% SDS polyacrylamide gel. A representative portion of the Coomassie-stained gel is shown on the left. The arrowheads point to the protein bands corresponding to the four indicated radiolabeled proteins of the autoradiogram. The right portion is an autoradiogram of the dried gel. Arrowheads on the right point to the four phosphoproteins of the following molecular masses: 180, 100, 97, and 55 kDa. S, sparse sample; C, confluent sample; MW, molecular weight. The autoradiogram was exposed for 90 min.


TABLE 1. Phosphorylation of proteins from membranes of sparse vs. confluent endothelial cells

Fold stimulation (sparse/confluent)l Phosphoprotein Experiment No. (kDa) 1 2 3 4 5 6

180 3.5 ND2 ND ND ND 3.7 100 ND 7.4 4.2 6.4 15.3 3.5 97 ND 21.0 10.9 9.0 ND 5.5 55 2.0 2.1 ND ND 2.7 2.4

'Radiolabeled proteins were visualized by autoradiography, excised, and quantitated. An equivalent-sized gel slice that did not contain specifically radiolabeled proteins, and was therefore representative of the background level of radioactivity, was excised from each lane and quantitated and the value subtracted from the radiolabeled phosphoproteins of that lane to correct for background radioactivity. 'ND, not determined.

The resulting data from six experiments are presented in Table 1. Although variability was observed between experiments in the quantitative difference of phosphorylation between sparse and confluent membranes, the differences in proliferating vs. quiescent samples were greatest for the 97 kDa protein.

To examine the nature of the four phosphoproteins as well as the kinases that act upon them, the identity of the amino acids that were radiolabeled on each of the four phosphoproteins in sparse samples was determined by phosphoamino acid analysis. The results are shown in Figure 3 and Table 2. The 180 and 100 kDa phosphoproteins contain only p-ser, the 97 kDa contains both p- ser and p-thr, and the 55 kDa band contains all three phosphoamino acids. Analysis of the 55 kDa band from confluent samples revealed that, similar to sparse samples, all three hydroxyamino acids were radiolabeled. The quantitative results of the phosphoamino acid analysis are listed in Table 2. In addition to comparing the relative distribution of radiolabel among the three amino acids, we also quantitated the fold increase (sparselconfluent) of the 55 kDa band and found that, although the amount of radiolabel increased in duplicate determina- tions of both p-ser and p-thr, the increase in p-tyr was the most marked 1.8, 1.8, 5.4, p-ser, p-thr, p-tyr, respectively.

In an attempt to characterize further the nature of the EC membrane kinases, their cofactor and cyclic nucleotide requirements were examined. In the absence of added divalent cations or with only Ca2+ (10 mM), phosphotransferase activity was undetectable (data not shown). As is illustrated in Figure 4, Mn2+ was the best cofactor, which could be partially replaced by M$+; however, in the presence of M$+ only, the phosphorylation of the 180, 97, and 55 kDa bands was markedly reduced (Fig. 4, cf. lanes 3 and 5). Note that in lane 5 the 97 kDa protein is the lightly labeled band just below the prominently phosphorylated 100 kDa band. The phosphorylation pattern in the presence of both man- ganese and magnesium was similar to the phosphorylation pattern in the presence of Mn2+ alone (Fig. 4, cf. lanes 1 and 3 or 2 and 4). The addition of Ca2+ (10 mM) to the reaction mixture containing Mn2+ and M$' did not cause any noticeable changes (cf. lanes 7,8 with lanes 1,2). The samples in Figure 4 were separated on a 7.5-20% gradient gel instead of the 5-10% gradient gel used routinely. Although this reduced the resolution of the 97 and 100 kDa bands, it allowed us to observe that there were no lower-molecular-mass proteins phosphor-

ylated in the presence of either Mn2+ or M$+ plus Mn2+. However, in the presence of M$+ alone, three new phosphoproteins were observed having the following approximate molecular masses: 30, 20, and 17 kDa. As is illustrated in Figure 5, the addition of cyclic AMP or cyclic GMP (10 or 100 pM) to the standard reaction mixture containing Mn2+ did not affect the phosphorylation of the 100,97, or 55 kDa phosphoproteins in either sparse or confluent samples, whereas the 180 kDa in sparse samples was phosphorylated to a lesser extent in the presence of CAMP. The addition of the cyclic nucleotides resulted in the slightly enhanced phosphorylation of two other bands (- 250 and 40 kDa). Although the phosphorylation of each of these bands in both sparse and confluent samples was enhanced, the 250 kDa band in sparse samples was labeled to a greater extent: 15.7- and 7.7-fold increase sparselconfluent in the presence of 100 pM of CAMP and cGMP, respectively. The negligible kinase activity in the presence of' Ca2+ was not enhanced by the active tumor promotor phorbol 12-myristate-13-acetate (0.1 pM not shown), nor was the phosphotransferase activity of the standard kinase assay of either sparse or confluent samples affected by the addition of 0.1 pM phorbo1-12-myristate-13-acetate (Fig. 5).

As a part of characterizing the EC membrane kinases, we determined the time course of phosphorylation of the endogenous EC proteins. The time course of the phosphorylation of the 100, 97 and 55 kDa proteins was measured. Figure 6A is an autoradiogram of the SDS polyacrylamide gel, and the quantitation of the radiola- be1 incorporated into each of the four phosphoproteins is graphically summarized in Figure 6B. Maximal phosphorylation of sparse EC membranes occurred between 3 and 5 min a t 37"C, and a small decrease was observed between 5 and 15 min. The decline is suggestive of the presence of active phosphatases. Following addition of [32P]ATP, the extent of phosphorylation of membrane proteins prepared from confluent EC was not enhanced when the standard kinase assay was shortened (1 min) or lengthened to 5 min, and between 5 and 15 min the extent of phosphorylation decreased slightly, as was noted for sparse membranes (Fig. 6B). The presence of active phosphatases in both sparse and confluent EC membranes was further suggested by the observation that, as the incubation time at 37°C was increased, the reaction mixture became progressively more yellow, which was most probably due to the hydrolysis of the competitive phosphatase inhibitor p-nitrophenyl phosphate (Pike et al., 1983).

The effect of membrane concentration on the phosphotransferase activity was examined, and the results are shown in Figure 7. Note that, with the exception of the 55 kDa band, none of the EC phosphoproteins was ap- preciably phosphorylated in confluent samples even at high membrane concentrations. The kinase activity of membranes from proliferating cells increased linearly to - 50 pg of membrane protein, whereupon the extent of phosphorylation at higher membrane concentrations remained constant for the 180 kDa band, decreased for the 100 and 55 kDa bands, and continued to increase for the 97 kDa phosphoprotein (Fig. 7B). This curious find- ing that increasing the concentration of membrane protein, which contained both enzyme and substrate, sus- tained or diminished the extent of phosphorylation may

Fig. 3. Phosphoamino acid analysis of the endogenous proteins phosphorylated in EC membranes. EC membrane proteins were subjected to the standard phosphorylation assay. Radiolabeled proteins were visualized by autoradiography, excised from the dried polyacrylamide gel, digested, and hydrolyzed, and the phosphoamino acid content was analyzed as described in Materials and Methods. Only a portion of the resulting autoradiogram is shown; the origin of each panel is to the bottom right; the first dimension was to the left, and the second dimension was run in the upward direction. P-ser, phosphoserine; P- thr, phosphothreonine; P-tyr, phosphotyrosine. A, 180 kDa band, auto-

radiogram was exposed for 45 hr; B, 100 kDa band, autoradiogram was exposed for 94 hr; C, 97 kDa band, autoradiogram was exposed for 94 hr; D, 55 kDa band from sparse sample, autoradiogram was exposed for 71 hr, cpm, 613, p-ser; 134, p-thr; 304, p-tyr; E, 55 kDa band from confluent sample, autoradiogram was exposed for 80 hr, cpm, 301, p- ser; 62, p-thr; 43, p-tyr. The relative positions of the phosphoamino acids indicated in A are the same in all panels. The radiolabeled spots to the right of the separated phosphoamino acids are most probably incompletely hydrolyzed peptides as described by Cooper et al. (1983).

Fig. 4. Effect of divalent cations on the phosphorylation of endogenous EC membrane proteins. The standard kinase assay was performed in the presence of 10 mM of each of the followin divalent cations: lanes 1 and 2, Mn2+ and Mg2+. lanes 3 and 4, Mn'+; lanes 5 and 6, Mg2+; lanes 7 and 8, Mn2+, Mg'+, and Ca2+. The left lane of each pair contains sparse membranes; the right contains confluent membranes. The broad radioactive band at the dye front is most probably [32P]ATP. Proteins were separated on a 7.5-20% gradient

SDS polyacrylamide gel. Arrowheads point to the four phosphoproteins routinely observed the small arrows point to the novel phosphorylations observed in the presence of Mg2+. The molecular mass standards used and their molecular masses are as follows: myosin, 200,000; phosphorylase B, 97,400; bovine serum albumin, 68,000; ovalbumin, 43,000; a-chymotrypsinogen, 25,700; P-lactoglobulin, 18,400; lysozyme, 14,300. The autoradiogram was exposed for 1 hr.


Fig. 5. Effect of cyclic nucleotides and active tumor promoter on the EC membrane kinase activity. The kinase assay was performed in the standard manner with no additions, control; with phorbol-12-myristate-13-acetate (0.1 pM) + TPA; with cGMP, 10 pM, 100 pM; with CAMP, 10 pM, 100 pM. Sparse samples are to the left of each pair; confluent samples are on the right. The bars to the right indicate the

position of the following molecular mass markers (top to bottom): 200, 97.4, 63, and 43 kDa. Arrowheads point to the four routinely observed phosphoproteins; small arrows denote the cyclic nucleotide-inducible bands. The gel with the control, + TPA, and + cGMP samples was exposed for 1 hr; the + CAMP samples are from a different gel, which was exposed for 90 min.

236 KAZLAUSKAS AND DrCORLETO

TABLE 2. Phosphoamino acid analysis of the radiolabeled EC membrane moteins'

Phosphoprotein (kDa) P-ser (%) P-thr (%) P-tyr (%)

180 100 0 0 100 100 0 0 97 39 2 3 61 k 3 0 55 (sparse) 62 + 5 12 + 1 26 5 5* 55 (confluent) 74 + 1 16 + 1 11 + 0

'Radiolaheled proteins were visualized hy autoradiography, appropriate gel hands excised, the phosphoproteins digested, and hydrolyzed, and the phosphoamino acids analyzed as described in Materials and Methods. Phosphoamino acid analysis was not performed on the 180, 100, and 97 kDa hands from confluent membranes because the extent of phosphorylation was minimal. The data presented are calculated as follows: (cpm of single phosphoamino aciditotal cpm in all three phosphoamino acids) x 100; mean i SD (n = 2). P-ser, phosphoserine; P-thr, phosphothreonine; and P-tyr, phosphotyrosine. *The difference between the phosphotyrosine content of the sparse and confluent 55 kDa phosphoprotein is statistically different (P < 0.05).

indicate that a n endogenous phosphatase was becoming more active as the ratio of phosphatase to phosphatase inhibitor (a constant) increased.

The results from the preceding experiments suggested the presence of an active phosphatase in EC membranes, and this possibility was examined by monitoring dephosphorylation of EC membrane phosphoproteins. Membranes were phosphorylated in the standard reaction mixture lacking phosphatase inhibitor for 3 min at 37 "C; then the further incorporation of radiolabeled phosphate was inhibited by the addition of a 1,000-fold excess of unlabeled ATP. Incubation at 37°C was prolonged for the indicated time interval. The first two lanes in Figure 8A contain duplicate sparse samples, which were phosphorylated in the presence of phosphatase inhibitor. From a quantitative comparison of the extent of phosphorylation in the presence and absence of phosphatase inhibitors (Fig. 8B), it is apparent that less labeled phosphate is incorporated into protein in the absence of p-nitrophenyl phosphate. The time course of dephosphorylation indicates that active phosphatases existed in the EC membrane preparations and that the four phosphoproteins under study were not equivalent substrates (Fig. 8A, B). The 97 kDa band, containing the greatest percentage of p-thr, was the poorest substrate, whereas the p-ser-containing proteins were very good substrates. Alternatively, inherent differences between the phosphoproteins other than the phosphorylated amino acids might determine to what extent a substrate is dephosphorylated by a phosphatase.

We examined whether the difference in phosphorylation events between proliferating and quiescent cells was EC-specific. The kinase activities of a second species of EC (bovine aortic) as well as two non-EC cell lines (human foreskin fibroblasts and 3T3 cells) were examined. Membranes from proliferating, sparse (percent labeled nuclei = 71.9 f 4.5, n = 4; 73.1 & 1.3, n = 6; 73.0 & 6.4, n = 6 for bovine aortic EC, human fibroblasts, and 3T3 cells, respectively) and quiescent, confluent (percent labeled nuclei = 5.6 f 2.3, n = 6; 5.9 & 1.6, n = 6; 20.4 5 11.0, n = 6 for bovine aortic EC, human fibroblasts, and 3T3 cells, respectively) cells were prepared for each cell type following the identical procedures used to prepare porcine EC membranes, and phosphotransferase activity was assayed. The results are depicted in Figure 9. In human fibroblast membranes, a 55 kDa protein was phosphorylated, but the extent of phosphorylation was similar in sparse and in confluent membranes. With the exception of an -220

0 10 30 5 0 70 90 110 I30 150

TIME (minutes)

Fig. 6. Time course of EC membrane phosphorylation. The standard kinase assay was carried out for the indicated time intervals. A, the autoradiogram illustrates the radiolabeled proteins. The time indicated at the top of each pair of lanes is in minutes. Lane on the left of each pair contains sparse membranes; lane on the right contains confluent membranes. Arrowheads denote the four routinely observed phosphoproteins. The autoradiogram was exposed for 2 hr. MW, molecular weight. Bagel bands containing the appropriate phosphoprotein were excised and quantitated by Cerenkov counting. Closed symbols, sparse samples; open symbols, confluent samples.

B

I

R 10 I

0

4 180 kDa

F ' ~ ' " ' " " ' ' " ' ' ~ D I 0 10 20 30 40 50 60 70 80 90 100

MEMBRANE PROTEIN (pg)

Fig. 7. Phosphorylation of EC membrane proteins as a function of the position of the following molecular mass standards (top to bottom): membrane protein concentration. The indicated amounts of mem- 200, 97.4, 63, and 43 kDa. The left lane of each pair contains sparse branes were phosphorylated in the standard kinase assay. A, autora- membranes, the right lane confluent membranes. Autoradiogram was diogram illustrating the radiolabeled proteins resolved on a 7.5% SDS exposed for 90 min. B, gel bands containing the appropriate phospho- polyacrylamide gel. Numbers above each pair of lanes indicate the proteins were excised and quantitated by Cerenkov counting. Closed amount of protein per sample in micrograms. Arrowheads point to the symbols, sparse samples; open symbols, confluent samples. position of the four routinely observed phosphoproteins. Bars indicate


i OPEN SYYBOLSI*PNPP UOSEDSYYBOLS~-PNPP

TIME (minutes)

Fig. 8. Phosphatase activity in sparse EC membranes. Membrane proteins were phosphorylated in the standard kinase assay, further incorporation of radiolabel was abated by the addition of a 1,000-fold excess of unlabeled ATP, and the incubation at 37 "C was prolonged for the indicated time interval. A, autoradiogram illustrating the radiolabeled proteins. The first two lanes on the left differed from the others in that these samples contained the phosphatase inhibitor p-nitrophenyl phosphate whereas the others did not. The time indicated above

each pair of samples is in minutes. Each pair of samples is a duplicate sparse membrane sample. Arrowheads denote the four routinely labeled phosphoproteins. The autoradiogram was exposed for 1 hr. MW, molecular weight. B, gel bands containing the appropriate phosphoproteins were excised and quantitated by Cerenkov counting. Open sym- bol, with p-nitrophenyl phosphate (PNPP); closed symbols, without PNPP.


Fig. 9. Comparison of the phosphorylation of EC and non-EC membranes. Membranes prepared from sparse and confluent pig aortic EC (PAEC), 3T3 cells, human fibroblasts (HF), and bovine aortic EC (BAEC) were phosphorylated in the standard manner, and a collection of autoradiograms illustrate the radiolabeled proteins. S, sparse sample; C, confluent sample. The bars correspond to the position of the following molecular mass standards (top to bottom): 200, 97.4, 68, and 43 kDa; the arrowheads denote the position of the four routinely observed EC

phosphoproteins. The autoradiogram containing lanes 1-6 was exposed for 90 min; lanes 7 and 8 were exposed for 1 hr; and lanes 9-12 were exposed for 2 hr. Note that the extent of phosphorylation of the 180 kDa band in PAEC samples is similar in sparse and confluent samples. This is due in part to the migration of this phosphoprotein as several bands (see also Figs. 5 and 61, which tends to diminish the difference between sparse and confluent.


Fig. 10. Phosphorylation of a mixture of sparse and confluent EC membranes. Membranes were phosphorylated in the standard manner described in Materials and Methods. S/2 indicates duplicate lanes containing sparse sample alone (25 pg); S/2 + C/2 denotes duplicate lanes containing a mixture of 25 pg sparse and 25 pg confluent membrane protein (50 pg in each lane); S and C mark single lanes each containing 50 pg sparse or confluent sample, respectively. Arrowheads point to the position of the four prominent phosphoproteins routinely observed. MW, molecular weight. The autoradiogram was exposed for 90 min.


Fig. 11. EC phosphorylation is growth-dependent. Membranes were phosphorylated in the routine manner, as described in Materials and Methods. The two lanes marked SFM contain membranes (50 pg) prepared from sparse cells made quiescent by serum deprivation. Con- trol lanes contain membranes (50 pg) prepared from sparse, rapidly

proliferating cells (S) or confluent, quiescent cells (C). Arrowheads denote the position of the four prominent phosphoproteins. Samples were separated on a 7.5-20% gradient SDS polyacrylamide gel resulting in the diminished resolution of the 100 and 97 kDa bands. MW, molecular weight. The autoradiogram was exposed for 1 hr.


kDa band, which is greater in confluent than in sparse samples, the phosphorylation pattern of sparse and confluent human fibroblast membranes was found to be very similar in three different experiments. Numerous proteins were phosphorylated in 3T3 cell membranes, several of which had molecular masses similar to the EC phosphoproteins: 180, 130, and 100 kDa; however, based only on the similarity in molecular mass, determined by one-dimensional SDS-PAGE, it is impossible to determine whether the 3T3 cell phosphoproteins are identical to the EC phosphoproteins. In addition, the extent to which these 3T3 cell membrane proteins were phosphorylated was quite similar in sparse and confluent samples. The major difference in the phosphorylation pattern of sparse and confluent 3T3 cell membranes consisted of two bands (- 65 and 45 kDa) that were present to a greater extent in confluent samples. These results were in sharp contrast to the phosphorylation pattern of bovine aortic EC: Not only were the phosphorylated proteins in bovine aortic EC membranes of molecular masses similar to those radiolabeled in pig aortic EC but the proteins in proliferating membranes were radiolabeled to a dramatically greater extent. The only difference between the phosphorylation patterns of the two types of EC was that the 97 kDa phosphoprotein in bovine aortic EC sparse samples was phosphorylated to a lesser extent than the 100 kDa band, whereas in pig aortic EC samples the two bands were labeled to a similar extent. Differences between phosphorylation of proteins in sparse membranes was greater than that in confluent membranes by 5.3-, 8.4, and 2.7-fold for the 100, 97, 55 kDa bands, respectively. These -fold differences were similar to the differences observed between sparse and confluent samples of pig aortic EC summarized in Table 1. The 130 kDa phosphoprotein present in sparse pig aortic EC samples was also present in bovine aortic EC sparse samples.

As a first step toward elucidating the mechanism responsible for the difference in the kinase activity of membranes from proliferating vs. quiescent EC, we have examined the possibility that confluent membranes contain either a kinase inhibitor or a potent phosphatase(s). Solubilized membranes prepared from proliferating and quiescent EC were mixed, and the extent of phosphorylation was compared to an equivalent quantity of sparse membranes alone. As is shown in Figure 10, the phosphorylation pattern was very similar in the mixed and sparse alone samples for the 100, 97, and 55 kDa phosphoproteins, and there was a slight decrease in the 180 kDa phosphoprotein in mixed samples. Quantitation of the extent of phosphorylation of the 100 and 97 kDa bands verified this observation: 1,150 f 131, 599 f 63 (cpm incorporated, sparse membrane) vs. 912 f 44,659 f 62 (cpm incorporated, mixture of sparse and confluent) for 100 and 97 kDa, respectively. Clearly, the lack of phosphorylation of the 100,97, and 55 kDa bands in confluent samples was not due to the presence of a kinase inhibitor or a phosphatase. A portion of the 180 kDa complex does decrease in the mixed samples, suggesting a phosphatase and/or kinase inhibitor acting on this phosphoprotein, which may partially explain its lack of phosphorylation in confluent samples.

The enhanced kinase activity of membranes prepared from sparse EC may reflect the growth rate or the low cell density of these cells. To begin to distinguish be-

tween these two possibilities, EC at low cell density were made quiescent by serum deprivation prior to membrane isolation. As is illustrated in Figure 11, a striking difference was observed between the kinase activities of membranes prepared from proliferating EC at low cell density and membranes from replicate cul- tures rendered quiescent by serum removal. The phosphorylation pattern of sparse quiescent EC membranes was similar to that observed in confluent quiescent membranes. This result is consistent with the possibility that the observed enhanced phosphotransferase activity reflects the growth fraction of the EC preparation rather than a density-dependent phenomenon.

DISCUSSION A fruitful approach to the examination of intracellular

events that may be regulating proliferation is to study changes in intracellular enzyme activities following stimulation of a target cell with a specific mitogenic agent. Many such experiments have focussed on changes in a specific growth factor receptor following binding of the appropriate mitogen. Examination of what regu- lates the PDGF-inducible PDGF receptor tyrosine kinase activity as well as the identification of specific substrates of this kinase is one such example (Cooper et al., 1982; Ek and Heldin, 1984; Kazlauskas and Di- Corleto, 1986). A similar approach to the elucidation of the intracellular events controlling EC growth has been limited by the fact that the plasma membrane receptors with which putative EC mitogens presumably interact remain for the most part uncharacterized. Of the var- ious known EC mitogens, a specific plasma membrane receptor has been described only for EC growth factor (ECGF; Schreiber et al., 1985). The mechanism by which ECGF relays its mitogenic stimulus through the ECGF receptor to the interior of the cell is a t present unknown; no tyrosine kinase activity has been reported to be associated with this receptor.

The goal of the present study was to begin to elucidate intracellular events that mediate EC proliferation. We chose to focus on phosphorylation events and have capi- talized on the strict density-dependent growth inhibition of EC at confluent cell density and compared the phosphorylation events of confluent, quiescent and sparse, proliferating EC. Whereas the overall kinase activity of membranes prepared from proliferating cells was slightly greater than that of quiescent cell membranes, there were a number of proteins phosphorylated to a dramatically greater extent in sparse than in confluent membranes. By examining the cofactor specificity of the kinases, as well as the phosphoamino acid content of the endogenous substrates, we have determined that there are at least three distinct kinases responsible for the phosphorylation differences in sparse vs. confluent cells: a serine- and a threonine-specific kinase (or two kinases each of which phosphor lates only one of the two amino acids) that prefeds) MS', a serine-specific kinase that can utilize either Mn2+ or M2 ' as a cofactor, and a tyrosine-specific kinase.

Several preliminary reports have recently appeared that discuss phosphorylation events in EC. Brotherton (1985) examined changes in protein phosphorylation that may be associated with the stimulation of prostacyclin biosynthesis. Following stimulation of [32P]-metaboli- cally labeled human umbilical vein EC with either


thrombin or the calcium ionophore A23187, increased phosphorylation of a number of proteins including a 57 kDa species was observed. Similarly, Bormann and Jaffe (1985) using several different phosphorylation assays reported that stimulation of human umbilical vein EC with thrombin resulted in increased phosphorylation of numerous proteins, including 91, 59, and 54 kDa species. In addition, they showed that the 59 kDa substrate was the cytoskeletal protein vimentin, and that it was a substrate for both CAMP-dependent protein kinase and protein kinase C. We have determined that the phosphorylation of the four principal EC proteins that we have studied in sparse cell membranes was independent of CAMP, although it is possible that the 55 kDa protein is a substrate for endogenous membrane-associated kinases as well as for the CAMP-dependent kinase. In an attempt to examine the mechanism constituting the ability of the endothelium to regulate local circulation, Mackie et al. (1985) examined the endogenous kinase activity and substrates of EC derived from microvascu- lature as well as aorta. They reported that a CAMP- dependent kinase phosphorylates a 57 and a 54 kDa protein, that a 55 kDa protein is the substrate of C d phospholipid-dependent protein kinase; and that a 100 and a 59 kDa protein are phosphorylated by a Cdcal- modulin-dependent kinase. (Note that Mackie et al. reported several other substrates for each type of kinase, but we have mentioned only the phosphoproteins having molecular masses similar to those of the EC phosphoproteins discussed in this report.) Although the above-cited studies have examined the role of phosphorylation in certain specialized functions of the endothelium, they differ from our work in that we focussed on phosphorylation events that may play a role in EC proliferation.

The comparison of phosphorylation events in sparse and confluent cultured cells has previously been under- taken to search for phosphorylation events associated with both cell density and growth state. Wehner et al. (1977) compared the density-dependent phosphorylation of whole-cell homogenates prepared from a variety of cell lines. They reported a 110 kDa phosphoprotein whose phosphorylation state increased as cell density increased. The phosphorylation of this protein was independent of growth, since both transformed cells, which continue to proliferate a t high cell density, and normal cells, which are quiescent at high cell density, displayed a highly phosphorylated form of the 110 kDa protein relative to the corresponding cell homogenate at low cell density. The 110 kDa substrate and its kinase were not membrane-bound, which may explain why this phosphoprotein was not detected in confluent EC membranes in our studies. In an attempt to elucidate phosphorylation events that may be growth-dependent, Branton (1980) compared the phosphorylation events occurring in membranes prepared from proliferating and quiescent chick embryo fibroblasts. Although he does report several differences between the two populations of cells, none cor- responded to the differences observed between sparse and confluent EC.

Phosphorylation experiments utilizing crude enzyme preparations are inherently complex; the extent of phosphorylation of any given substrate is determined by the action of numerous factors, including kinases, phosphatases, and ATPases as well as substrate availability. The experiments presented in this report were designed to

focus specifically on the role of kinases in determining the phosphorylation state of endogenous substrates. We have demonstrated that phosphatases are not primarily responsible for the differences observed in the extent of radiolabeled phosphate present in endogenous substrates from proliferating vs. quiescent cells. The degree to which ATPases, as well as the interaction between kinases, phosphatases, and ATPases, influence the phosphorylation state of these substrates remains to be as- sessed. In preliminary experiments, equal quantities of histone were phosphorylated to a 2.0-fold greater extent in proliferating than in quiescent samples, suggesting that the substrate availability alone cannot account for the differences observed (unpublished observations).

The identities of the endogenous EC membrane kinases and their substrates remain unknown. Although we have demonstrated that these kinases were unaf- fected by cyclic nucleotide, Ca2+, and active tumor promotor, it still remains to be determined whether these EC kinases belong to one of the other known classes of kinases or constitute their own class. In addition, cytoplasmic phosphotransferase activity as well as the determination of the subcellular location and distribution of the described kinases await elucidation.

One question raised by our experiments with EC con- cerns the mechanism by which the phosphorylation events observed in proliferating EC are suppressed at quiescence. Our results indicate that it is not due to a kinase inhibitor or a phosphatase in the confluent EC membranes. Perhaps the simplest explanation is that confluent EC membranes lack either the active kinases andlor the appropriate substrates. The phosphorylation difference could also be maintained if, in confluent membranes, the endogenous substrates are fully phosphorylated such that radiolabeled phosphate cannot be incorporated. However, when confluent membranes were preincubated for 5 min at 37°C (in the absence of phosphatase inhibitors) immediately prior to measuring the kinase activity, giving endogenous phosphatases the op- portunity to remove phosphate, the phosphorylation pattern of confluent membranes was not affected (not shown). Alternatively, the phosphorylations observed in the sparse membranes may be autophosphorylations of the responsible kinases, which are present at much reduced levels in confluent cells.

In summary, we have utilized an in vitro assay to demonstrate that EC possess a number of kinases that exhibit markedly enhanced activity in sparse vs. confluent cells. The phosphorylation pattern in sparse vs. confluent cells is EC-specific and appears to reflect the growth state rather than the density of the cells. Our results are consistent with the possibility that certain protein kinases play a role in the regulation of EC proliferation.

ACKNOWLEDGMENTS We would like to thank Ms. Muriel Daly and Mr.

Endry Ritly for providing excellent secretarial and pho- tographic assistance. Human foreskin fibroblasts were prepared from tissue provided by the perinatal Research Center (NIH USPHS MOlRR00210), Cleveland Metro- politan General Hospital. This work was supported by grants from the National Heart, Lung, and Blood Insti- tute, National Institutes of Health (HL-29582 and HL-


34727). P.E.D. is the recipient of a Research Career Development Award (HL-01561) from the N.I.H.

LITERATURE CITED Barritault, D., Plouet, J., Courty, J., and Courtois, Y. (1982) Purifica-

tion, characterization, and biological properties of the eye-derived growth factor from retina: Analogies with brain-derived growth factor. J. Neurosci. Res., 8477-490.

Berridge, M.J., and Irvine, R.F. (1984) Inositol trisphosphate, a novel second messenger in cellular signal transduction. Nature, 312:315- 321.

Booyse, F.M., Sedlak, B.J., and Rafelson, M.E. (1975) Culture of arterial endothelial cells. Characterization and growth of bovins aortic cells. Thromb. Diath. Haemorrh., 34:825-839.

Bormann, B.J., and Jaffe, E.A. (1985) Endogenous substrates of protein kinases in thrombin stimulated endothelial cells (abstract). Fed. Proc., 44:8362.

Bradford, M.M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem., 72:248-254.

Branton, P.E. (1980) Protein phosphorylation and the proliferation of chick fibroblasts: analysis by in vitro phosphorylation using isolated plasma membranes and whole cell homogenates. Can. J. Biochem., 58:30-39.

Brotherton, A.F.A. (1985) Thrombin and A23187 induce specific phosphorylation in cultured endothelium (abstract). Fed. Proc., 44:8363.

Cohen, P. (1982) The role of protein phosphorylation in neural and hormonal control of cellular activity. Nature, 296613-620.

Cohen, S., Carpenter, G., and King, L., Jr. (1980) Epidermal growth factor-receptor-protein kinase interactions. Co-purification of receptor and epidermal growth-factor-enhanced phosphorylation activity. J . Biol. Chem., 255:48344842.

Cooper, J.A., and Hunter, T. (1981) Changes in protein phosphorylation in Rous sarcoma virus-transformed chick embryo cells. Mol. Cell. Biol., 1:165-178.

Cooper J.A., BowcnPopc, D.F., Raines, E., Ross, R., and IIuiiter, T. (1982) Similar effects of platelet-derived growth factor and epidermal growth factor on the phosphorylation of tyrosine in cellular proteins. Cell, 31:263-273.

Cooper, J.A., Sefton, B.M., and Hunter, T. (1983) Detection and quan- tification of phosphotyrosine in proteins. In: Methods in Enzymology. S.P.Colowick and N.O. Kaplan, eds. Academic Press, New York, Vol. 99, pp. 387-405.

D’Amore, P.A., and Klagsbrun, M. (1984) Endothelial cell mitogens from retina and hypothalamus: Biochemical and biological simiiari- ties. J. Cell Biol., 99:1545-1549.

Davies, P.F., and Ross, R. (1978) Mediation of pinocytosis in cultured arterial smooth muscle cells and endothelial cells by platelet-derived growth factor. J . Cell Biol., 79563-671.

DiCorleto, P.E., and de la Motte, C.A. (1985) Characterization of the adhesion of the human monocytic cell line U937 to cultured endothelial cells. J . Clin. Invest., 75:1153-1161.

Ek, B., and Heldin, C.-H. (1984) Use of an antiserum against phosphotyrosine for the identification of phosphorylated components in human fibroblasts stimulated by platelet-derived growth factor. J . Biol. Chem., 259:11145-11152.

Ek, B., Westermark, B., Wasteson, A,, and Heldin, C.-H. (1982) Stimu- lation of tyrosine-specific phosphorylation by platelet-derived growth factor. Nature, 295:419-420.

Frackelton, A.R., Jr., Tremble, P.M., and Williams, L.T. (1984) Evi- dence for the platelet-derived growth factor-stimulated tyrosine phosphorylation of the platelet-derived growth factor receptor in vivo. Immunopurification using a monoclonal antibody to phosphotyrosine. J . Biol. Chem., 259:7909-7915.

Gimbrone, M.A. (1976) Culture of vascular endothelium. In: Progress in Hemostasis and Thrombosis. T.H. Spaet, ed. Grune and Stratton, New York, Vol. 3, pp. 1-28.

Glenn, K.C., and Ross, R. (1981) Human monocyte-derived growth factoxfs) for mesenchymal cells: Activation of secretion by endotoxin and concanavalin A. Cell, 25503-615.

Gordon, P.B., Sussman, I.I., and Hatcher, V.B. (1983) Long term culture of human endothelial cells. In Vitro, 19:661-671.

Gospodarowicz, D., Brown, K.D., Birdwell, C.R., and Zetter B.R. (1978) Control of proliferation of human vascular endothelial cells. Charac- terization of the response of human umbilical vein endothelial cells to fibroblast growth factor, epidermal growth factor, and thrombin. J . Cell. Biol., 77774-778.

Gospodarowicz, D., Cheng, J., and Lirette, M. (1983) Bovine brain and pituitary fibroblast growth factors: Comparison of their abilities to

support the proliferation of human and bovine vascular endothelial cells. J. Cell. Biol., 971677-1685.

Habenicht, A.J.R., Glomset, J.A., King, W.C., Nist, C., Mitchell, C.D., and Ross, R. (1981) Early changes in phosphatidylinositol and ara- chidonic acid metabolism in quiescent Swiss 3T3 cells stimulated to divide by platelet-derived growth factor. J. Biol. Chem., 25612329- 12335.

Haudenschild, C.C., Zahniser, D. Folkman, J., and Klagsbrun, M. (1976) Human vascular endothelial cells in culture. Lack of response to serum growth factors. Exp. Cell Res., 98175-183.

Heimark, R.L., and Schwartz, S.M. (1985) The role of membrane-membrane interactions in the regulation of endothelial cell growth. J. Cell Biol., 1001934-1940.

Holley, R.W., Armour, R., Baldwin, J.H., Brown, K.H., and Yeh, Y.C. (1977) Density-dependent regulation of growth of BSC-1 cells in cell culture: Control of growth by serum factors. Proc. Natl. Acad. Sci. U.S.A., 745046-5050.

Jaffe, E.A., Hoyer, L.W., and Nachman, R.L. (1973) Synthesis of anti- hemophilic factor antigen by cultured human endothelial cells. J. Clin. Invest., 52:2757-2764.

Kazlauskas, A., and DiCorleto, P.E. (1985) Cultured endothelial cells do not respond to a platelet-derived growth factor in an autocrine manner. Biochim. Biophys. Acta, 846405-412.

Kazlauskas, A., and DiCorleto, P.E. (1986) A comparison of platelet- derived growth factor-dependent tyrosine kinase activity in sparse and confluent fibroblasts. J . Cell Physiol., 126225-236.

Laemmli, U.K. (19701 Cleavage of structural proteins during the as- sembly of the head of bacteriophage T4. Nature, 227680-685.

Lemmon, S.K., Riley, M.C., Thomas, K.A., Hoover, G.A., Maciag, T., and Bradshaw R.A. (1982) Bovine fibroblast growth factor: Compari- son of brain and pituitary preparation. J. Cell Biol., 95:162-169.

Lewis, L.J., Hoak, J.C., Maca, R.D., and Fry, G.L. (1973) Replication of human endothelial cells in culture. Science., 181 :453-454.

Maciag, T., Hoover, G.A., and Weinstein, R. (1982) High and low molecular weight forms of endothelial cell growth factor. J. Biol. Chem., 2575333-5336.

Maciag, T., Hoover, G.A., Stemerman, M.B., and Weinstein, R. (1984a) Factors which stimulate the growth of human umbilical vein endothelial cells in vitro. In: The Endothelial Cell. E. Jaffe, ed. Martinus Nijhoff, Boston, pp. 87-96.

Maciag, T., Mehlman, T., Friesel, R., and Schreiber, A.B. (1984b) Hep- arin binds endothelial cell growth factor, the urinciDal endothelial cell mitogen in bovine brainy Science, 225:932-935.

Mackie, K., Lai, Y., Nairin, A.C., Greengard, P., Lazo, J., and Pitt, B.R. (1985) Protein phosphorylation in cultured endothelial cells (ab-

*

stract). Fed. Proc., 44:2904. Nishizuka, Y. (1984) The role of protein kinase C in cell surface signal

transduction and tumor promokion. Nature, 308:693-698. Pike, L.J., Bowen-Pope, D.F., Ross, R., and Krebs, E.G. (1983) Charac-

terization of platelet-derived growth factor-stimulated phosphorylation in cell membranes. J . Biol. Chem., 2589383-9390.

Ross, R. (1971) The smooth muscle cell. 11. Growth of smooth muscle cells in culture and formation of elastic fibers. J. Cell Biol., 50:172- 186.

Schreiber, A.B., Kenney, J., Kowalski, W.J., Friesel, R., Mehlman, T., and Maciag, T. (1985) Interaction of endothelial cell growth factor with heparin: Characterization by receptor and antibody recognition. Proc. Natl. Acad. Sci. U.S.A., 82:6138-6142.

Schumacher, B.L., Grant, D. and Eisenstein, R. (1985) Smooth muscle cells produce an inhibitor of endothelial cell growth. Arteriosclerosis, 5:llO-115.

Schwartz, S.M. (1978) Selection and characterization of bovine aortic endothelial cells. In Vitro, 14:966-980.

Thomas, G., Martin-Perez, J., Siegman, M., and Otto, A.M. (1982) The effect of serum, EGF, PGF and insulin of S6 phosphorylation and the initiation of protein and DNA synthesis. Cell, 30:235-242.

Vogel, A,, Ross, R., and Raines, E. (1980) Role of serum components in density-dependent inhibition of growth of cells in culture. Platelet- derived growth factor is the major serum determinant of saturation density. J. Cell Biol., 85:377-385.

Voyta, J.C., Via, D.P., Butterfield, C.E., and Zetter, B.R. (1984) Identi- fication and isolation of endothelial cells based on their increased uptake of acetylated low density lipoprotein. J . Cell Biol., 99:2034- 2040.

Wehner, J.M., Sheppard, J.R., and Malkinson, A.M. (1977) Density- dependent phosphorylation of a specific protein in cultured endothelial cells. Nature, 266:842-844.

Westermark, B. (1976) Density dependent proliferation of human glial cells stimulated by epidermal growth factor. Biochem. Biophys. Res. Commun., 69:304-310.

I

Documents

Comparison of the phosphorylation events in membranes from proliferating vs. quiescent endothelial cells