THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 8, Issue of March 15, pp. 44564459,1959 Printed in U.S.A.

Platelet-derived Growth Factor Production by Human Umbilical Vein Endothelial Cells Is Regulated by Basic Fibroblast Growth Factor*

(Received for publication, July 11,1988)

Stella KourembanasSg and Douglas V. FallerTII From the TDivision of HematologylOncology, Dann-Farber Cancer Institute and the Divisions of 7lMedicine and $.Neonntology, The Children's Hospital, Harvard Medical School, Boston, Massachusetts 021 15

Human umbilical vein endothelial cells express the B chain gene (c-sis) of platelet-derived growth factor (PDGF). PDGF acts as a mitogen for mesenchymal cells and has recently been reported to be a potent vasocon- strictor (Berk, B. C., Alexander, R. W., Brock, T. A., Gimbrone, M. A., Jr., and Webb, R. C. (1986) Science 232, 87-90). Regulation of c-sis therefore, may play an important role in maintaining and controlling the integrity of the vasculature. We report here that basic fibroblast growth factor (bFGF), a growth factor for endothelial cells in culture, significantly decreases the amount of PDGF-like protein secreted by these cells. Levels of c-sis mRNA increase more than 10-fold in the absence of bFGF. This effect is specific for the B chain of PDGF as transcript levels of other growth factors synthesized by endothelial cells, such as trans- forming growth factor /3 are not affected by bFGF. c- sis mRNA levels increase 12 h after bFGF removal as the cells begin to accumulate in Go stage of the cell cycle and remain elevated for at least 96 h. Transcript levels fall again upon re-exposure of the cells to bFGF. bFGF may regulate c-sis mRNA levels and PDGF pro- tein production by endothelial cells either directly, or indirectly via cell cycle arrest at Go. This effect is not due to nonspecific cessation of cell growth because arresting the cells at S phase or in metaphase does not result in increased c-sis transcript levels.

PDGF' is an ubiquitous growth factor that is involved in cell proliferation and directed cell movement. It has been implicated in several physiologic processes including embry- ogenesis and development (2, 3), as well as in abnormal proliferation of cells that occurs in atherosclerosis (4), neo- plasia (5), and pulmonary fibrosis (6). PDGF was discovered in 1974, when it was observed that material from platelets was the principle source of mitogens in serum responsible for the growth of many cell types in culture (7, 8). In the vascu- lature, the endothelium forms the permeability barrier be- tween the circulating blood cells and the underlying mesen- chyme which is composed of fibroblasts and smooth muscle cells. As such, the endothelial cell layer may modulate peri-

* This work was supported by research grants from the National Cancer Institute (CA37169 and S07RR05526) and from the American Cancer Society. 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.

Recipient of a Parker B. Francis Foundation Fellowship. 11 Recipient of an American Cancer Society Senior Faculty Award. The abbreviations used are: PDGF, platelet-derived growth factor;

bFGF, basic fibroblast growth factor; TGFB, transforming growth factor @; PPP, platelet-poor plasma.

vascular events via a paracrine mechanism, involving the secretion of mediators such as PDGF that regulate the prolif- eration and migration of adjacent cells. In addition to PDGF, endothelial cells secrete other factors such as TGFB (9) and bFGF (9, 10) that may also be involved in the control of mesenchymal proliferation.

PDGF is predominantly a heterodimer composed of A and B chains that are 40% homologous to each other. In homodi- mers, either chain is mitogenic to fibroblasts (11, 12). The B chain is homologous to the v-sis oncogene of the Simian Sarcoma Virus (13-15). The gene which is the cellular hom- ologue of v-sis, c-sis, can be expressed by various normal cell types, including human umbilical vein endothelial cells (16, 17). Previous studies have shown c-sis expression by human endothelial cells to be regulated by factors including TGFB, phorbol esters (18), and thrombin (19). In the experiments presented here, we show that basic FGF, a growth factor for endothelial cells in culture, also regulates c-sis transcript levels, resulting in decreased amounts of PDGF production and secretion by these cells.

MATERIALS AND METHODS

Cell Culture-Primary cultures of human umbilical vein endothe- lial cells were obtained from fresh human umbilical cords using established methods (20). They were grown on 1% gelatin-coated plates in the presence of heparin and endothelial cell growth supple- ment (Sigma) in M199 medium (Gibco) with 20% fetal calf serum (Hyclone, Logan, UT) at 37 "C, 5% CO, in a NAPCO incubator. Cells were passaged by trypsinization with 1% trypsin/EDTA (Gibco) every 4-5 days.

RNA Anulysis-Total cellular RNA was prepared by the method of Chirgwin et al. (21), from early passage endothelial cells grown in the presence or absence of recombinant bFGF (generous gift of California Biotechnologies, Mountain View, CA). Twenty micrograms of RNA per lane were separated by electrophoresis on 1.2% agarose gels containing formaldehyde and transferred to nitrocellulose mem- brane by blotting. The filters were hybridized with 32P-labeled probes specific for the sis gene (1300-base pair PstI fragment of Simian Sarcoma Virus Clone C60) (22) or the PDGF A chain gene (1300- base pair cDNA clone) (23). The TGFB probe used was the 1050-base pair EcoRI fragment corresponding to XP-Cl (24) coding for human TGF,. The actin probe was the 800-base pair PstI fragment of the mouse @-actin gene. Hybridizations were carried out in 10% dextran, 5 X saline sodium citrate (5 X SSC) (0.75 M NaC1, 0.075 M sodium citrate) and 50% formamide at 42 "C for 18-24 h. The final washing of the blots was in 0.2 X SSC at 55 "C. Autoradiography was per- formed at -80 "C with an intensifying screen for 24-48 h. A densi- tometer scanner (Helena Laboratories, Beaumont, TX) was used to quantitate relative amounts of transcript levels. Transcript levels of c-sis or TGFB were normalized to the levels of actin transcripts, which did not change under the various cell growth conditions.

Cell Proliferation Assays-For the time-course studies of PDGF production, M199 media, supplemented with 5% platelet-poor plasma (PPP) and without additional growth factors, were conditioned by monolayers of human umbilical vein endothelial cells. For 5 X 10' cells, 5 ml of media were used. At 1 h, the media were removed and replaced with fresh 5% PPP/M199. Every 8 h, all the media were

4456

Regulation of PDGF by bFGF 4457

removed and concentrated &fold using Amicon Microconcentrators (Amicon). These samples were then assayed for PDGF activity.

To evaluate intracellular accumulation of PDGF, extracts from endothelial cells were obtained. Endothelial cells a t near confluence were rinsed with M199 and the media were replaced with new media containing recombinant bFGF at 10 ng/ml (in 20% fetal calf serum/ MI99 plus heparin) or without bFGF (5% PPP/M199 or 20% fetal calf serum/M199 plus heparin). A t 24 h, the cells were washed twice with 1 X phosphate-buffered saline, trypsinized, and centrifuged (300 X g, 5 min, 21 "C). The pellet was washed with phosphate-buffered saline, and lo7 cells were resuspended in 10 mM sodium phosphate buffer (pH 7.4) containing 0.15 M NaCI. Cells were lysed by rapidly freezing and thawing several times. Nuclei and cell debris were pelleted by centrifugation (50,000 X g, 10 min, 4 "C), and the clear supernatant (referred to as crude cell extract) was obtained. These extracts were added to media in final concentrations of up to 10% to assay for growth-promoting properties.

PDGF activity in conditioned media or in cell extracts was evalu- ated by proliferation assays using BALB/c-3T3 murine fibroblasts. Fibroblasts were grown to confluence in 96-well plates (Costar) and serum-starved over 7 days in Dulbecco's Modified Eagle's Media, 10% calf serum, without media change. The endothelial cell-condi- tioned media from the different collection time points were used directly on these quiescent cells while the endothelial cell extracts were diluted with 5% PPP/M199 down to a 10% final concentration. [3H]Thymidine (Du Pont-New England Nuclear) at a concentration of 5 pCi/ml of medium was added to each well. After 30-36 h of incubation, cells were washed and harvested, and acid-precipitable counts were collected on glass fiber filters. All analyses were per- formed in triplicate. M199 media containing 5% PPP was used as a negative control, while human PDGF purified from platelets (Collab- orative Research, Bedford, MA) or recombinant (Amgen, Thousand Oaks, CA) was added to some cultures to a final concentration of 300 units/ml of medium. Anti-PDGF antibody that recognizes both A and B chains (Collaborative Research, Bedford, MA) was used to neutralize mitogenic activity of purified PDGF to establish standard curves for quantitation of PDGF present in conditioned media or in the cell extracts.

Flow Cytornetry-To evaluate the distribution of endothelial cells in the cell cycle we used flow cytometry (Epics IV, Coulter Electronics, Hialeah, FL). Cells were grown to 75% confluence and then exposed to fresh media containing bFGF or without exogenous bFGF. DNA content per cell was evaluated at desired intervals over a 24-h period by staining the DNA with Chromomycin & (Sigma); 1 ml of 2% Chromomycin in phosphate-buffered saline was used per 5 X 10' cells. Cells were kept a t 4 "C, and fluorescence was assessed within 1 h of staining.

Cells were arrested in the S phase of DNA replication with a sublethal dose of yirradiation. Metaphase arrest was achieved using colcemid (Sigma) at 0.05 pg/ml in normal growth media for 24 h. RNA was subsequently purified as above, and Northern analyses were performed.

RESULTS

Endothelial cells at near confluence (75%) were exposed to different growth factors, and the effect on PDGF transcript levels was examined by Northern analysis.

Levels of c-sis mRNA in human umbilical vein endothelial cells varied when the cells were grown in the presence or absence of bFGF. Low levels of c-sis mRNA were detected in cells grown in the presence of basic FGF (Fig. 1, Panel A ) . In contrast, transcript levels increased more than 10-fold in the absence of basic FGF. Quantitation of the difference in c-sis transcript levels was provided by normalizing to the level of actin mRNA on the same filter. Actin transcript levels re- mained the same in the presence or absence of bFGF, as shown on Panel B of Fig. 1. The specificity of the c-sis response to bFGF is shown in Panel C of the same figure: mRNA levels in endothelial cells of another growth factor, TGFB, were not affected by the presence or absence of bFGF. As a control for the amount of RNA loaded in each lane, the same filter was hybridized with an actin-specific probe (Panel Dl .

c-sis mRNA levels increased within 12 h after bFGF re-

A kEGE - +

ms - 1 8 s -

C hEGE - +

211s - 18s -

D kEQE - + v9m

FIG. 1. Basic FGF down-regulates c-sis transcript levels. Human umbilical vein endothelial cells were grown to confluence, as outlined under "Material and Methods," and placed in fresh media containing 20% fetal calf serum with 10 ng/ml recombinant bFGF (+ bFGF) or without bFGF (- bFGF). At 24 h, cells were harvested and RNA was purified via a cesium chloride gradient using the method of Chirgwin et al. (21). 20 pg of total cellular RNA from cells exposed to each condition was separated by electrophoresis on a formaldehyde agarose gel, transferred to nitrocellulose membrane, and subsequently hybridized with a radiolabeled 3zP c-sis cDNA probe (Panel A ) . After autoradiography was performed, the filter was stripped and reprobed with a radiolabeled actin cDNA probe (Panel B ) . Panel C demon- strates the specificity of the effect of bFGF on c-sis. A different filter containing RNA samples from the same preparations were hybridized to a cDNA probe for TGFB. The blot was subsequently reprobed with an actin probe to normalize for the amounts of RNA loaded in each lane, as shown in Panel D. The positions of migration of the 28 S and 18 S ribosomal RNA are indicated in each panel.

moval and remained elevated at 24 h (Fig. 2). Although not shown here, c-sis transcript levels remained high for at least 96 h. In contrast, very low levels of c-sis mRNA were found in cells grown in the presence of bFGF over the same time intervals (Fig. 2, lune C). Readdition of bFGF to the endothe- lial cell cultures at any point over this time course, down- regulated c-sis mRNA within 24 h (data not shown).

In order to correlate changes in the amount of PDGF

4458 Regulation of PDGF by bFGF

A

28s - 18s-

* . .

0 2 6 1 2 2 4 C

HOURS

FIG. 2. Time course of PDGF mRNA regulation by basic FGF. At time 0, endothelial cells grown to confluence were exposed to their normal growth media (20% fetal calf serum plus heparin) but without endothelial cell growth supplement or basic FGF. Total cellular RNA was extracted at the indicated times (0, 2,6, 12, and 24 h), and RNA blot analysis was performed to evaluate the time required for induction of c-sis transcript levels. Lane C contains 20 pg of total cellular RNA from endothelial cells that were incubated for 24 h with recombinant bFGF at 10 ng/ml. The positions of migration of the 28 S and 18 S ribosomal RNA species are indicated.

protein produced in response to bFGF with the observed suppression of c-sis mRNA, proliferation assays were per- formed on BALB/c-3T3 fibroblasts with conditioned media obtained from endothelial cells. This biological assay was more sensitive in detecting small amounts of PDGF than immunoprecipitation or Western blot analysis. The endothe- lial cells were maintained in 5% platelet-poor plasma without bFGF and the conditioned media were used to stimulate fibroblasts, as described under “Materials and Methods.” The increase in c-sis mRNA levels occurred 12 h after bFGF removal. In order to determine if that increase in mRNA levels correlated temporally with production of PDGF-like biological activity, conditioned media were collected every 8 h. The mitogenic activity detected in the 20-28-h period was found to be %fold greater than that obtained at 1-9 h (Fig. 3A). Other factors released by endothelial cells can also stim- ulate fibroblast proliferation (25,26). However, the mitogenic activity found in the conditioned media under the above conditions was predominantly PDGF-like since it was almost fully neutralized with an anti-PDGF antibody. Crude cell extracts obtained from endothelial cells grown in the presence or absence of bFGF showed no difference in mitogenic activity when tested on a similar proliferation assay (Fig. 3B) indi- cating that the PDGF-like substance was not stored intracel- lularly but was secreted into the media (27).

bFGF is a mitogen for endothelial cells and the effect of its removal on the growth of endothelial cells was examined by flow cytometry. Under normal growth conditions in the pres- ence of bFGF, 60% of cells are in Go/Gl stage of the cell cycle at any one time and 40% in S/GZ + M. Upon removal of bFGF for 12 h, a drastic reduction of the early S phase is observed (Fig. 4). At this time, 75% of cells have accumulated in Go/GI and 25% in late S and GP + M. The percent of cells present at different stages of the cycle were calculated by measuring the area under the curve. Although the increase in PDGF production was coincident with the entry of more cells into Go/Gl, this effect on PDGF production was not simply due to arrest of cell growth. Growth arrest of the cells in S phase of DNA replication (with y-irradiation) or in metaphase (with colchicine) did not result in increased transcript levels (data not shown).

DISCUSSION

The data presented here demonstrate that PDGF produc- tion by human endothelial cells is regulated at a pretransla-

1-9 HRS 20-28 HRS 20-28 HRS

B 9 AM-PDGE

0.5 2.5 5 10 10 PDGF Crude C e l l Extract (%of total media) e + + +

FIG. 3. Proliferation assay for PDGF activity in endothelial cell-conditioned media and endothelial cell extracts. Panel A, confluent endothelial cells were rinsed with M199 media twice and then incubated in 5% platelet-poor plasma/h4199. Media were col- lected every 8 h, concentrated B-fold, and then assayed for mitogenic activity on BALB/c-3T3 fibroblasts by quantitating the amount of [3H]thymidine incorporated as outlined under “Material and Meth- ods.” The proliferation of fibroblasts a t 1-9 h uersus that at 20-28 h is plotted. The medium collected at 20-28 h was also treated with anti-PDGF antibody (+ Anti-PDGF) prior to use in the proliferation assay. Panel B, crude cell extracts (CCE) were obtained from endo- thelial cells grown in the presence of 10 ng/ml bFGF (CCE/bFGF) or in its absence (CCEIPPP) for 24 h as outlined under “Materials and Methods.” Increasing amounts of crude cell extracts (0.5,2.5,5.0, and 10% of total media (5% PPP/M199)) were used on quiescent BALB/ c-3T3 fibroblasts to obtain a maximal proliferative response. The maximal proliferation achieved using 10% of crude cell extracts from either condition equaled that found using 300 units purified PDGF/ ml (column labeled PDGF) and was fully neutralized with anti-PDGF antibody. The antibody also fully neutralized the proliferation in- duced by purified PDGF.

tional level by bFGF. Transcript levels of c-sis are reduced at least 10-fold when endothelial cells are grown in the presence of bFGF. In addition, the PDGF protein secreted by these cells is coordinately reduced in response to bFGF. The intra- cellular levels of PDGF remain unchanged after exposure to bFGF. Other investigators have also shown that PDGF pro- duced by endothelial cells is not stored intracellularly, as the amount of PDGF-like activity detected in conditioned media does not increase upon cell lysis (27).

The mechanism by which bFGF down-regulates PDGF production is currently under study. I t may be a direct effect by bFGF at the level of c-sis gene transcription or mRNA stabilization. Alternatively, the up-regulation may be indirect and mediated via cell cycle arrest at Go. This effect on c-sis mRNA levels and protein secretion is not the nonspecific result of arrest of cell growth, as arresting the cells at other phases of the cell cycle does not result in increased c-sis transcript levels. In addition, the increased secretion of PDGF is not a preterminal release event, as described by others (27), because readdition of bFGF results in drastic reduction of c-

Regulation of PDGF by bFGF 4459

+ bFGF - bFGF

1 c 2 c I C 2 c DNA CONTENT

FIG. 4. DNA content analysis of endothelial cells by flow cytometry. Endothelial cells a t near confluence were exposed to

with 10 ng/ml bFGF (+ bFGF) or without bFGF (- bFGF). DNA fresh media containing M199 with 20% fetal calf serum plus heparin

content per cell was evaluated at 0, 6, and 12 h of exposure to the two different conditions by staining the DNA of each cell with Chromomycin A,. Cell number uers'sus DNA content is plotted on the figure with IC and 2C indicating one and two chromosomal comple- ments, respectively, per cell.

sis mRNA within 24 h. The effect of bFGF on endothelial cells appears to be specific for c-sis mRNA and is not due to an overall decrease in mRNA accumulation, as transcript levels of constitutively expressed genes, such as actin, or other growth factors, such as TGFB, are not affected. It is also pertinent to point out that studies in progress indicate that PDGF-A mRNA levels are not modulated by bFGF in parallel with c-sis mRNA levels (data not shown).

It is apparent that perivascular events are closely linked to the precise interaction of such mesenchymal growth factors as PDGF, TGFB, and bFGF, all of which are produced by endothelial cells (9, 10, 25). Endothelial cells do not have receptors for PDGF and hence do not respond to it. However, they do have receptors for TGFB and bFGF and respond to these factors (9, 10, 27). It thus seems that the PDGF they secrete in response to various stimuli might control the pro- liferative state of other neighboring cells rather than being involved in an autocrine self-regulation. Factors such as thrombin and phorbol esters have been shown to increase modestly the amount of PDGF produced by endothelial cells (18, 19). In addition, TGFB has been shown to increase c-sis mRNA 10-fold in human renal microvascular endothelial cells and to increase the amount of secreted PDGF-like protein about 3-5-fold by these cells (18). Other investigators have shown that TGFB inhibits the growth of various types of endothelial cells in culture including bovine aortic and capil- lary endothelial cells (1, 28). In the present study we have shown that bFGF, known to be a potent mitogen for endothe- lial cells, decreases transcript levels of c-sis resulting in re- duced amounts of PDGF produced by these cells. In one case, TGFB inhibits vascular endothelial cell proliferation and stim- ulates PDGF production, whereas in the other case, bFGF stimulates endothelial cell growth but down-regulates PDGF production by these cells. Hence, the control exerted by these

factors on PDGF may be mediated via the movement of endothelial cells in the cell cycle. One can speculate that PDGF production by endothelial cells is closely linked to their proliferative state, and this in turn may be controlled by other growth factors in a paracrine or autocrine manner. Evidence against a direct relationship between growth arrest in Go and production of PDGF is the finding that a large fraction of cells are always present in Go/G~ even in the presence of bFGF, yet PDGF production is very low. Clearly, the regula- tion of PDGF by bFGF is complex and may be mediated indirectly via the proliferative state of the cells or alterna- tively, more directly by the growth factor.

Acknowledgments-We thank Dr. Robert L. Hannan for providing some of the cell lines used in this study and for helpful advice; also Dr. Michael Epstein for critical reading of the manuscript

REFERENCES 1. Takehara, K., LeRoy, E. C., and Grotendorst, G. R. (1987) Cell

2. Goustin, A. S., Betsholtz, C., Pfeifer-Ohlsson, S., Persson, H., Rydnert, J., Bywater, M., Holmgren, G., Heldin, C.-H., Wes- termark, B., and Ohlsson, R. (1985) Cell 41,301-312

3. Seifert, R. A., Schwartz, S. M., and Bowen-Pope, D. F. (1984) Nature 311,669-671

4. Ross, R. (1986) N. Engl. J. Med. 314,488-500 5. Johnsson, A., Betsholtz, C., Heldin, C.-H., and Westermark, B.

6. Martinet, Y., Rom, W. N., Grotendorst, G. R., Martin, G. R., and

7. Ross, R., Glomset, J. A., Kariya, B., and Harker, L. (1974) Proc.

8. Kohler, N., and Lipton, A. (1974) Exp. Cell. Res. 8 7 , 297-301 9. Hannan, R. L., Kourembanas, S., Flanders, K. C., Rogelj, S. J.,

Roberts, A. B., Faller, D. V., and Klagsbrun, M. (1989) Growth Factors, in press

10. Schweigerer, L., Neufeld, G., Friedman, J., Abraham, J. A., Fiddes, J. C., and Gospodarowicz, D. (1987) Nature 326 , 257- 259

11. Antoniades, H. N. (1981) Proc. Natl. Acad. Sci. U. S. A . 78,

12. Deuel, T. F., Huang, J. S., Profftt, R. T., Boenziger, J. U., Chang, D., and Kennedy, B. B. (1981) J. Biol. Chem. 266,8896-8899

13. Heldin, C.-H., and Westermark, B. (1984) Cell 37,9-20 14. Deuel, T. F., and Huang, J. S. (1984) Blood 64,951-958 15. Stiles, C. D. (1983) Cell 33 , 653-655 16. Barrett, T. B., Gajdusek, C. M., Schwartz, S. M., McDougall. J.

49,415-422

(1985b) Nature 317,438-440

Crystal, R. G. (1987) N. Engl. J. Med. 317 , 202-209

Natl. Acad. Sci. U. S. A. 71,1207-1210

7314-7317

K., and Benditt, E. Pi (1984) Proc. Natl. Acad. Sci. Ur S.' A. 81,6772-6774

17. Collins, T., Ginsburg, D., Boss, J. M., Orkin, S. H., and Pober, J.

18. Daniel, T. O., Gibbs, V. C., Milfay, D. F., and Williams, L. T.

19. Daniel, T. O., Gibbs, V. C., Milfay, D. F., Garovoy, M. R., and

20. Gordon, P. B., Sussman, I. I., and Hatcher, V. B. (1982) In Vitro

21. Chirgwin, J. M., Przybyla, A. E., McDonald, R. J., and Rutter, W. J. (1979) Biochemistry 18, 5294-5299

22. Gelmann, E. P., Petri, E., Cetta, A., and Wong-Staal, F. (1982) J. Virol. 41, 593-604

23. Betsholtz, C., Johnsson, A., Heldin, C-H., Westermark, B., Ling, P., Urdea, M. S., Eddy, R., Shows, T. B., Philpott, K., Mellor, A. L., Knott, T. J., and Scott, J. (1986) Nature 320,695-699

24. Derynck, R., Jarrett, J. A., Chen, E. Y., Eaton, D. H., Bell, J. R., Assoian, R. K., Roberts, A. B., Sporn, M. B., and Goeddel, D. V. (1985) Nature 316 , 701-705

25. Vlodavsky, I., Fridman, R., Sullivan, R., Sasse, J., and Klagsbrun, M. (1987) J. Cell. Physiol. 131,402-408

26. DiCorleto, P. E. (1984) Exp. Cell Res. 163 , 167-172 27. Fox, P. L., and DiCorletto, P. E. (1984) J. Cell. Physiol. 1 2 1 ,

28. Muller, G., Behrens, J., Nussbaumer, U., Bohlen, P., and Birch- meier, W. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 5600-5604

S. (1985) Nature 3 1 6 , 748-750

(1987) J. Biol. Chem. 2 6 2 , 11893-11896

Williams, L. T. (1986) J. Biol. Chem. 261,9579-9582

19,661-671

298-308