EXPRESSION OF THE CATALYTIC ACTIVITY OF PLASMINOGEN ACTIVATOR UNDER PHYSIOLOGIC CONDITIONS
JUDITH AGGELER, JANET RISCH and ZENA WERB
Laboratory o f Radiobiology LRI02 and Department o f Anatomy, University of California, San Francisco, CA 94143 (U.S.A.)
(Received September 18th, 1980) (Revised manuscript received January 8th, 1981)
Key words. Plasminogen activator," Salt inhibition; Urokinase; (Fibroblast)
We have investigated the factors governing the plasminogen-dependent fibrinolysis catalyzed by the serine pro- teinase, plasminogen activator (EC 3.4.21 .-), under physiologic conditions. We found that live rabbit fibroblasts digested much less fibrin than predicted by cell-free assay of the secreted plasminogen activator. The reduced catalytic activity of plasminogen activator expressed by cells growing on fibrin was regulated by the salt concen- tration of culture medium. The plasminogen activators of cells from several mammalian species were inhibited by physiologic salt concentrations (0.15 M NaCI) in cell-free assays. CaCI2 and KC1, but not D-glucose, were also effective inhibitors. The catalytic activity of purified human urokinase and of plasmin was unaffected by increased ionic strength. Plasminogen activators secreted both spontaneously and in response to stimulation by the tumor promoter, 12-O-tetradecanoyl-phorbol-I 3-acetate, were inhibited by O. 15 M NaCI. Physiologic salt con- centration appeared to function by interacting with plasminogen activator, or plasminogen, and a third compo- nent, possibly a reversible inhibitor. One consequence of this regulation of plasminogen activator under physiol- ogic conditions is the limitation of plasminogen-dependent fibrin degradation by living cells.
Introduction
Plasminogen activator (EC 3.4.21.-), a proteinase that converts plasminogen to plasmin, is found in many animal tissues and cell lines in several molecular forms [1-4] . It has been postulated that the amount of tissue proteinase activity expressed by cells in vivo depends on enzyme synthesis and secretion, activa- tion of latent forms and modulation by inhibitors [5]. Plasma zymogens involved in coagulation and fibrinolysis are also regulated by a variety of me- chanisms [6]. Increases in plasminogen activator pro- duction and activity have been documented at dif- ferent stages of embryologic development [7-9] , during the course of viral and chemical transforma- tion [10-14], and after treatment with the tumor promoter, 12-O-tetradecanoyl-phorbol-13-acetate
Abbreviation: SDS, sodium dodecyl suffate.
[2,14--16]. Plasminogen activator activity can also be modulated by cellular inhibitors [17-19]. Plasmi- nogen activator may play an important role in con- nective tissue degradation by activating plasmin and other proteinases [20] or may have specific proteo- lytic activity of its own [21]. Two immunologically distinct forms of plasminogen activator have been described in human and hamster cell lines [2-4] . It is not known whether these represent separate gene products or different stages in enzyme processing (e.g., inactive precursor and active enzyme), although certain transformed cells can apparently make either or both immunologic forms.
To assess the potential expression of plasminogen activator in a physiologic context, we compared enzyme activities under various assay conditions with plasminogen-dependent fibrinolysis by live cells. We report that plasminogen activator activity can be altered by changing the salt concentration of the
assay mixture and that activity in the presence of physiologic salt more closely resembles the activity of living cells.
Experimen tal Cells were grown in 75 c m 2 plastic flasks (Costar)
in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and subcultured twice weekly by trypsinization (0.05% trypsin). To collect large pools of enzyme, confluent flasks were washed five-times with Hanks' balanced salt solution and then incubated for 24-48 h at 37°C with 5 ml of serum- free Dulbecco's medium containing 0.2% lactalbumin hydrolysate. These conditioned media were stored at -20°C until assayed for enzyme activity. In some experiments 20-100 ng/ml 12-O-tetradecanoyl-phor- bol-I 3-acetate was included during incubation.
Rabbit fibroblasts (R535) derived in our labora- tory from the knee synovium of 4 - 6 week-old rab- bits were used between passages 2 and 5 [5]. A mouse macrophage line (P388D1) and a hamster kid- ney fibroblast line (BHK-21) were obtained from the American Type Culture Collection and a human skin fibroblast line (GM 1391) was obtained from the Human Genetic Mutant Cell Repository. Chinese hamster ovary fibroblasts were originally provided by Dr. R. Tobey (Los Alamos Scientific Laboratories) and V79-S171 hamster lung fibroblasts by Dr. A.V. Carrano (Lawrence Livermore Laboratory).
Plasminogen activator was assayed by the method of Unkeless et al. [10] using an l:SI-labeled fibri- nogen substrate repurified and radioactively labeled by the chloramine-T method [22]. l~5I-labeled fibrinogen (20 /.tg/well; (2 -5)" 104 cpm/well) was plated into 24 well Costar culture dishes and dried. Plates were incubated with calf serum (0.5 ml/well) for 2 h at 37°C to convert fibrinogen to fibrin, and then washed three times with 0.9% NaC1 before use.
All plasminogen activator assays were done in 0.05 M Tris,HC1 buffer (pH 7.8) in a final volume of 0.5 ml/well. The plasminogen substrate, isolated from bovine plasma according to the method of Chibber et al. [23] was used at a final concentration of 6 nM. Samples assayed in the absence of plasminogen always showed less than 2% fibrinolysis. One unit of activity is defined as the amount degrading 5% (approx. 1 #g) of the available fibrin substrate in 1 h
in the presence of plasminogen. Total available fibrin was determined with trypsin. In some experiments direct conversion of 12SI-labeled plasminogen to plas- rain was assessed by SDS polyacrylamide gel electro- phoresis and autoradiography [5].
For studies of urokinase inhibitors, mouse livers were excised and weighed, then 9 ml H20 was added per g liver and the mixture was homogenized with a glass homogenizer. The liver lysate was acid treated [17] by adjusting the pH to 3.0 with 1 M HC1. After 1 h at 22°C, the homogenate was readjusted to pH 7.0 with I M NaOH, and particulate material removed by centrifugation (10000Xg, 20 rain). The super- natant fluid contained inhibitory activity.
12-O-Tetradecanoyl-phorbol-13-acetate (Consoli- dated Midlands Co., Brewster, N.Y.) was dissolved at 1 mg/ml in ethanol and stored at -20°C. Urokinase (Lot No. 902497 Calbiochem Co., La Jolla, CA) was stored at -20°C as a stock solution of 1000 Plougll U/ml. TPCK-trypsin was obtained from Worthington Biochemical Co. (Freehold, N.J.). Cell culture media were obtained from Grand Island Biological Co., Grand Island, N.Y. Other materials were obtained from standard sources.
Results and Discussion
Plasminogen activator expression by live cells During investigation of factors regulating pro-
teinase secretion by cultured cells, we found that varying the assay conditions used to measure enzyme could produce activities differing up to 100-fold. In particular, the plasminogen-dependent fibrinolysis ob- served for living cells plated directly on a fibrin sub- strate differed significantly from the activity of the same cells measured by assaying a small amount of medium or cell extract in a cell-free system. When rabbit fibroblasts plated directly onto 12Sl-labeled fibrin were incubated with plasminogen in serum-free medium for 6 h, they lysed 30% of the fibrin sub- strate (plasminogen activator activity = 5 U/well). However, medium assayed separately on a cell-free fibrin substrate showed plasminogen activator activity of 53 U/well or 10-times as much as the cells them- selves. (50 #1 of medium lysed 19.9% of the fibrin substrate within 1.5 h at 37°C.) Cells or medium incu- bated without plasminogen showed no release ot
2 s I-labeled fibrinopeptides.
64
To understand this activation of plasminogen acti- vator in the cell-free assay, we compared the activity of plasminogen activator under various assay condi- tions. Fibrinolysis (cpm of 12sI released) increased linearly with added sample when less than 10% con- ditioned medium was present in the assay (Fig. 1, inset), but when more than 20% medium was present, plasminogen activator activity actually decreased (Fig. 1). Wells containing more than 40% conditioned medium showed greater than 95% reduction in plas- minogen activator activity (U/ml), a result similar to growing cells directly on the fibrin (100% medium).
Effect of salt on plasminogen activator assays One consequence of increasing the amount o f con-
ditioned medium in the assay was to raise the ionic strength; therefore we tested the effect of increasing ionic strength on the plasminogen activator activity of small aliquots (50 /zl) of conditioned medium. Raising the ionic strength of the assay with NaC1
o01 U
200 3
5 ~ \ u o 20 .o 60 80 "E ~ Medium (% Total)
100
E o
0 20 40 60 80
Medium in Assay (% Total)
Fig. 1. Inhibition of plasminogen activator by culture medium. Rabbit fibrobtasts were incubated in serum-free medium containing 20 ng/m112-O-tetradecanoyl-phorbol-13- acetate at 37°C for 48 h to produce conditioned medium containing plasminogen activator. Plasminogen activator activity (U/ml) was determined with increasing proportions of the assay volume (up to 80%) made up of medium. Points are averages of duplicate wells counted after 1 h. (Inset) The same data expressed as cpm of 12s I released without cor- rection for the volume of medium assayed. Total fibrinolysis was 22 000 cpm/well.
alone decreased plasminogen activator activity as effectively as adding conditioned medium. Physiol. ogic saline (0.15 M NaC1) decreased plasminogen acti- vator activity by 98% relative to a salt-free control (Fig. 2). Direct inhibition of rabbit plasminogen acti- vator by 0,15 M NaC1 was also seen when conversion of 12SI-labeled plasminogen to heavy and light chains of plasmin was monitored by SDS polyacrylamide gel electrophoresis and autoradiography (data not shown). In contrast to the plasminogen activator in the conditioned culture medium, purified urokinase activity was not decreased by NaC1 alone (Fig. 2).
To test whether the inhibition of fibrinolysis by physiologic saline in this assay was due to an irrevers- ible effect on plasminogen activator, plasminogen or plasmin, we preincubated conditioned medium from rabbit fibroblasts in the presence or absence of plas- minogen and then assayed this medium under condi- tions of either low or physiologic ionic strength. Con- ditioned medium preincubated without plasminogen showed full activity when assayed without added salt (132 U/ml) and complete inhibition when 0.15 M
>,
>
E .>_ ~
0 c-
O a_
200'
1S0
100 ~
60
4s O
I---
30 .~-
0 0.1 0.2 0.3 0.4 NaCI Conc, (M)
Fig. 2. Inhibition of plasminogen activator by NaCI. Rabbit fibroblasts were incubated in serum-free medium containing 20 ng/ml 12-O-tetradecanoyl-phorbol-13-acetate at 37°1] for 48 h to produce conditioned medium containing plasminogen activator. Plasminogen activator activity (U/ml) was deter- mined when 50 /~1 aliquots of medium were assayed in 0.05 M Tris (pH 7.8) in the presence of increasing NaC1 con- centrations (e), compared with the activity of urokinase (2 Plough units/ml) under the same assay conditions (o); results with urokinase are expressed as percentage of total counts released in 1 h.
NaC1 was added to the assay (0.4 U/ml), indicating that the plasminogen activator in the medium was not irreversibly inhibited by preincubation at 37°C. In addition, when conditioned medium was preincu- bated in the presence of plasminogen and then assayed in the absence of salt, fibrinolytic activity was increased (217 U/ml), indicating that neither plasminogen nor plasmin was irreversibly inactivated.
Although physiologic concentrations of NaC1 alone (0.I 5 M) duplicated the inhibition produced by complete medium, we also tested other medium com- ponents for their inhibitory potential. KC1 gave results that were very similar to those with NaC1, but CaC12 was a more effective inhibitor of the rabbit fibroblast plasminogen activator on a molar basis (Fig. 3). D-Glucose was not inhibitory up to 0.5 M (90 mg/ml) (Fig. 3), but 1 M D-glucose produced more than 50% inhibition, and Dextran 250 was an effective inhibitor at 90 mg/ml (data not shown). Bovine serum albumin (7 mg/ml) did not inhibit
o
0
C
0
E 8
lO0
50
0 0.1 0.2 0.3 0.~, 0.5
Effector Concentration (M)
Fig. 3. Effect of chloride salts and D-glucose on plasminogen activator. BHK-21 cells were incubated for 24 h at 37°(2 in serum-free medium. Aliquots (50 ~ul) of conditioned medium were assayed in duplicate in 0.05 M Tris (pH 7.8) in increas- ing concentrations of NaCI (o), KC1 (o), CaCI~ (~), or D-glu- cose (~).
65
activity. Although lysine binds to plasminogen and inhibits the activation by plasminogen activator [24], the lysine concentration in culture medium (1.5 raM) was considerably lower than that required to inhibit fibrinolysis. Thus, it is probable that the marked plas- minogen activator inhibition seen with culture medium or physiologic saline is dependent on ionic strength and that other medium components are rela- tively unimportant. In view of the observation of anion-dependent inhibition of activation of partially purified plasminogen in an esterase assay [25], the chloride anion may be the important species in this effect. Gadjewski and Markus [26] noted inhibition of streptokinase induced plasminogen activation by bicarbonate buffers, but little effect of NaC1, borate or Tris. In assays measuring conversion of plasmi- nogen to plasmin, Dang and Reich [24] reported approximately 50% inhibition by 30 mM NaC1, a result consistent with our data. Unkeless et al. [10] also noted inhibition of fibfinolysis by NaC1 (15% by 0.1 M; 98% by 0.5 M). The very pronounced plas- minogen activator inhibition by heavy metal salts including ZnC12, YCla, YbCla, FeCI3, and HgC12 [24] may be mediated by a different mechanism.
Effect of salt on urokinase and other molecular forms of plasminogen activator
To determine whether the reduction of activity by physiologic saline occurred with plasminogen activa- tor from other cell types, we tested enzymes from a variety of sources. Purified human urokinase was not inhibited by physiologic NaC1 concentrations (Fig. 2); the slight increase in activity in the presence of NaC1 may indicate greater stability of plasmin under these assay conditions. In contrast to urokinase, some inhibitory activity by physiologic saline was found for all the other plasminogen activators tested (Table I). Plasminogen activator secreted by human and rabbit fibroblasts was inhibited 80-99% by 0.15 M NaC1. The plasminogen activator secreted by mouse peritoneal macrophages and by a mouse macrophage cell line (P388D1) showed less inhibition (20-50%), in keeping with the greater expression of plasminogen activator by living macrophages [27, 28]. Cell-associated plasminogen activator was also inhibited when a rabbit fibroblast lysate was assayed in the presence of 0.15 M NaC1 (81.6 (5/106 cells were reduced to 11.2 U).
66
TABLE I
EFFECT OF NaC1 ON PLASMINOGEN ACTIVATOR FROM VARIOUS CELL TYPES
Cultured cells were incubated for 24 h at 37°C in serum-free medium with or without 100 ng/ml 12-O-tetradecanoyl-phorbol-13- acetate to obtain conditioned media. Mouse peritoneal macrophages (thioglycollate-elicited) were harvested and cultured [28,29] and macrophage-conditioned medium was prepared as above. Units were calculated from fibrinolysis after 2 h assay incubation at 37°C. For assay, 50-,ul aliquots of medium were assayed in duplicate in the presence or absence of 0.15 M NaCI. Inhibition of plasminogen activator by NaCI is expressed as [1-(activity with 0.15 M NaC1)/(activity without NaC1)] X 100%.
Species Cell type Plasminogen activator activity (control)
Because purified urokinase was resistant to salt inhibition, whereas the plasminogen activator secreted by cultured cell lines was sensitive, we examined three hamster fibroblast lines (BHK-21, kidney; Chinese hamster ovary, and V79, lung) that are known to have plasminogen activators with dif- fering molecular weights and immunologic speci- ficities. Christman et al. [2] have shown that a ham- ster kidney line makes plasminogen activator (uro- kinase) of Mr 75 000 that is immunologically similar to the plasminogen activator of Chinese hamster ovary cells and distinct from the plasminogen activa- tor produced by lung cells (M r 50,000). The plas- minogen activator secreted by the three hamster cell lines tested here all showed decreased activity when assayed in the presence of 0.15 M NaC1 (Table I), indicating that salt inhibition was common to enzymes that differ in molecular weight and anti- genicity and that the lack of inhibition of purified urokinase was probably not due to differences in molecular structure.
Effect of salt on plasminogen activator induced by 12-O-tetradecanoyl-phorbol-13-acetate
The tumor-promoting phorbol ester, 12-O-tetra- decanoylphorbol-13-acetate, increases plasminogen
activator production in a variety of cultured cells [2, 14-16] . The plasminogen activator secreted after treatment of rabbit fibroblasts and mouse macro- phages (P388D1) with this tumor promoter was strongly inhibited by 0.15 M NaC1 in the cell-free fibrinolytic assay (Table I). Christman et al. [2] found that this promoter increased plasminogen activator activity in six of 15 hamster lines, always inducing the M r 50 000 lung form of plasminogen activator regardless of the original form of enzyme produced. None of the three hamster lines treated with promoter in the present study showed increased plasminogen activator secretion when assayed in Tris buffer alone, but the BHK-21 (kidney) line was stimulated 3-fold when assayed in the presence of 0.I 5 M NaC1 (Table I), suggesting loss of salt inhibi- tion.
Relationship o f inhibitors to salt inhibition of plas- minogen activation
Because purified urokinase was fully active in the presence of physiologic salt while the urokinase-like plasminogen activator secreted by BHK-21 or Chinese hamster ovary cells was inhibited, we attempted to determine whether the decreased activity under high salt conditions reflected the presence of a salt-sen-
sitive inhibitor or accessory molecule in the condi- t ioned medium. Several tissue and cellular inhibitors of urokinase have been described [I 7 -19 ,31 ]. Inhibi- tory factors may be present in plasma, because chloride ions have been reported to inhibit activation of partially enriched plasminogen preparations, but not of purified plasminogen [25], and a plasma inhi- bitor of plasminogen activator has been identified [29,30]. When purified urokinase was incubated with 50 /A aliquots of condit ioned medium from rabbit fibroblasts, fibrinolysis was not decreased, but when 0.15 M NaC1 was included in this assay, urokinase activity was reduced by 70 -90% (Table II). Fibrinol- ysis by plasmin was slightly enhanced in the presence
of 0.15 M NaC1, indicating that the inhibitor was not affecting the fibrinolytic activity of plasmin (Table II). We also found that mouse liver homogenates, a rich source of plasminogen activator inhibitor [19], produced NaCl-dependent inhibition of urokinase (Table III). The ability of cell homogenates and medium to inhibit purified urokinase at physiologic salt concentration makes it l ikely that the salt-depen.
TABLE II
INHIBITION OF UROKINASE AND PLASMIN BY RABBIT FIBROBLAST-CONDITIONED MEDIUM
Rabbit synovial fibroblasts (5 • 106) were incubated in 75 cm 2 flasks for 48 h at 37°C in serum-free medium. Aliquots (50 td) of conditioned medium were then assayed in dupli- cate in the presence or absence of 0.15 M NaC1. For assays with urokinase, 5 Plough units were used in each well. Plas- min was prepared by incubating 5/~g of plasminogen with 10 Plough units of urokinase for 1 h at 18°C for each well. Results are expressed as percentage of fibrin degraded after 1 h at 37°C. Inhibition by NaC1 is expressed as [1-(fibrinol- ysis with 0.15 M NaC1)/(fibrinolysis without NaC1)] X 100%.
Exogenous Medium Fibrinolysis Inhibi- enzyme in the tion
EFFECT OF NaC1 ON INHIBITION OF UROKINASE BY MOUSE LIVER LYSATES
Mouse liver homogenates were prepared and acid-treated as described in the Experimental Section. For assays, aliquots ol homogenate and urokinase with 50 mM Tris-HC1 buffer, pH 7.8, were added to 12Si_labele d fibrin plates in the presence or absence of 0.4 M NaC1. To start the assays bovine plas- minogen was added to a final concentration of 6 nM. Fibri- nolysis was defined as the percentage of 12SI-labeled fibrin solubilized at 1 h.
dent inhibition of plasminogen activator in these sam- ples is mediated by either an enzyme/inhibi tor or a plasminogen/inhibitor mechanism rather than by conformational changes in the plasminogen activator molecular itself. In view of reported salt-dependent inhibition of plasminogen activation by chloroform [25] and streptokinase [26], we favor the formation of a plasminogen/inhibitory factor-complex as the most likely possibility. Such a mechanism suggests that only plasminogen activation would be regulated by these factors and not the other possible proteo- lytic activities of plasminogen activator [21]. This hypothesis remains to be tested. Control of the activ- i ty of a variety of other neutral proteinases by coor- dinately synthesized inhibitors has also been sug- gested [1,20].
Conclusions
Although a variety of secreted neutral proteinases can degrade interstitial connective tissue components in cell-free systems, it has been difficult to demon- strate secretion of a proteinase capable of acting directly on the extracellular matrix surrounding cells. It is probable that degradation of interstitial proteins
68
accompanies such pathologic conditions as malig- nant metastasis and such normal functions as tissue remodeling during growth and development. Degrada- tive enzymes can be synthesized and secreted by vari- ous cell types, including fibroblasts and macrophages, that may interact with each other to stimulate or inhibit proteinase secretion. The potential for cascade effects in such systems requires stringent regulation to effect ordered proteolysis. Under physiologic con- ditions, such plasma and tissue proteinases as the coagulation factors, plasminogen ac'tivator, and col- lagenase are usually found in inactive forms that may be regulated by protein inhibitors or by substances such as anionic lipids [6,32]. It is not yet clear whether secreted plasminogen activator ever becomes active in vivo, other than in its role in lysis of blood clots. The plasminogen activator activity that mediates fibrinolysis by living cells may be associated with the cell surface [21] and may be active only in a narrow pericellular zone [32]. Because so little is known about the molecular mechanisms by which connective tissue cells interact with and regulate degradation of their pericellular proteins, further study of these proteinases under physiological condi- tions may shed light on this important regulatory problem.
Acknowledgment
This work was supported by the U.S. Department of Energy, a predoctoral fellowship from the National Science Foundation to J. Aggeler, and an NIH traineeship (AM-07175) to J. Risch.
References
1 Christman, J.K., Silverstein, S.C. and Acs, G. (1977) in Proteinases in Mammalian Cells and Tissues (Barrett, A.J., ed.), pp. 90-149, North-Holland Publishing Co., Amster- dam
2 Christman, J.K., Copp, R.P., Pedrinan, L. and Whalen, C.E. (1978) Cancer Res. 38, 3854-3860
3 Vetterlein, D., Young, P.L., Bell, T.E. and Roblin, R. (1979) J. Biol. Chem. 254,575-578
4 Vetterlein, D., Bell, T.E., Young, P.L. and Roblin, R. (1980) J. Biol. Chem. 255, 3665-3672
5 Werb, Z. and Aggeler, J. (1978) Proc. Natl. Acad. Set. U.S.A. 75, 1839-1843
6 Fugikawa, K., Heimark, R.L., Kurachi, K. and Davie, E.W. (1980) Biochemistry 19, 1322-I330
7 Beers, W.H., Strickland, S. and Reich, E. (1975) Cell 6, 387-394
8 Strickland, S., Reich, E. and Sherman, M.I. (1976) Cell 9,231-240
9 Linney, E. and Levinson, B.B. (1977) Cell 10, 297-304 10 Unkeless, J.C., Tobia, A., Ossowski, L., Quigley, J.P.,
Rifkin, D.B. and Reich, E. (1973) J. Exp. Med. 137, 85-- II1
11 Unkeless, J., DanG, K., KeUerman, G,M. and Reich, E. (1974) J. Biol. Chem. 249, 4295-4305
12 Ossowski, L., Quigley, J.P., Keilerman, G.M. and Reich, E. (1973) J. Exp. Med. 138, 1056-1064
13 Rifkin, D.B. and Pollack, R. (1977) J. Cell Biol. 73, 47- 55
14 Goldfarb, R.H. and Quigley, J.P. (1978) Cancer Res. 38, 4601-4609
15 Wigler, M. and Weinstein, I.B. (1976) Nature 259, 232- 233
16 Vassalli, J.-D., Hamilton, J. and Reich, E. (1977)Cell 11, 695 -705
17 Loskutoff, D.J. and Edginton, T.S. (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 3903-3907
18 Seffert, S.C. and Gelehrter, T.D. (1978) Proc. Natl. Acad. Sci. U.S.A. 75,6130-6133
19 Izaki, S., Goldstein, S.M., Fukuyama, K. and Epstein, W.L. (1979) J. Invest. Dermatol. 73,561-565
20 Werb, Z., Mainaxdi, C.L., Vater, C.A. and Harris, E.D., Jr. (1977) N. Engl. J. Med. 296, 1017-1023
21 Quigley, J. (1979)Cell 17,131-141 22 Gordon, S., Werb, Z. and Cohn, Z.A. (1976) in In Vitro
Methods in Cell-Mediated and Tumor Immunity (Bloom, B.R. and David, J.R., eds.), pp. 341-352, Academic Press, New York
