Proc. Nati. Acad. Sci. USA Vol. 77, No. 2, pp. 1177-1180, February 1980 Medical Sciences Calcium dependence of phalloidin-induced liver cell death (toxin/plasma membrane/cytochalasin B/microfilaments/scanning electron microscopy) AGNES B. KANE, ELLORA E. YOUNG, FRANCIS A. X. SCHANNE, AND JOHN L. FARBER* Department of Pathology and the Fels Research Institute, Temple University School of Medicine, Philadelphia, Pennsylvania 19140 Communicated by Hans Popper, December 10, 1979

ABSTRACT The role of Ca2+ in toxic liver cell death was Phalloidin is a toxic bicyclic heptapeptide isolated from the studied with primary cultures of adult rat hepatocytes. Within mushroom Amanita phalloides (16, 17). Treated animals die 1 hr of exposure to phalloidin, a bicyclic heptapeptide isolated within a few hours with a hemorrhagic necrosis of the liver from the mushroom Amanita phalloides, at 50 ,g/ml, 60-70% of the cells were dead (trypan blue stainable). There was no loss characterized by numerous nonfatty vacuoles (18). Phalloidin of viability of the same cells exposed to phalloidin in culture is active on rat hepatocytes in uitro, where it produces easily medium devoid of Ca2+. A marked structural alteration of the observed deformations of the cell surface that accompany the surface of the phalloidin-treated hepatocytes characterized by death of the cells (14, 15). The protrusions or evaginations of innumerable evaginations seen by scanning electron microscopy the plasma membrane seen by scanning electron microscopy occurred in the presence or absence of Ca2+. Pretreatment of are felt to to the of the the cells with cytochalasin B at Ig/ml10 prevented the surface correspond invaginations plasma alteration and the death of the cells in Ca2+ medium. Exposure membrane and subsequent vacuolization of the cells in the in- of the cells to phalloidin in the absence of Ca2+ followed by tact liver as seen by transmission electron microscopy (14, 15, exposure to cytochalasin B and then to Ca2+ also prevented the 18). The molecular basis of this injury most likely lies in the cell death. These results suggest a two-step mechanism by which interaction of phalloidin with actin filaments intimately asso- phalloidin causes liver cell death. Initially phalloidin interacts ciated with the plasma membrane (19-21). Phalloidin accel- in a Ca2+-independent process with cell membrane-associated erates the polymerization of G-actin to filamentous structures actin. The second step is a Ca2+-dependent process that most likely represents an increased influx of Ca2+ across a compro- (Ph-actin) that are more resistant than F-actin to the destabi- mised cell membrane permeability barrier and down the steep lizing effects of 0.6 M KI (22-25). Liver plasma membranes concentration gradient that exists between the outside and in- isolated from phalloidin-treated rats contain increased fila- side of the cell. These results strengthen the hypothesis that mentous structures (26). [3H]Desemethylphalloin, a phallotoxin disturbances in Ca2+ homeostasis induced in vivo by a variety chemically similar to phalloidin, binds to isolated rat liver of hepatotoxins are causally related to liver cell death. plasma membranes with a dissociation constant identical to that of the binding of the same toxin to muscle actin (25, 26). These A possible role for calcium in the production of liver necrosis data suggest that phalloidin-binding sites in the rat liver plasma was suggested by investigations of Judah and colleagues membrane are predominantly actin isolated together with the showing that some anti-inflammatory drugs give considerable membrane fragments. protection to the liver against the action of hepatotoxins (1-6). In the present report we have used this ability of phalloidin Although calcium had been known for many years to accu- to kill rat hepatocytes in culture to assess the role of Ca2+ in toxic mulate in necrotic tissues, these studies showed an early increase liver cell death. It is shown that phalloidin interacts with isolated in the calcium content of the cells that was prevented by pro- hepatocytes to produce alterations of the cell surface inde- methazine and that could be related to inhibition of various pendently of extracellular Ca2+. However, the death of the cells enzyme activities. Antihistamines such as promethazine were is known to block ion movements, and it was suggested that the absolutely dependent upon the presence of Ca2+ in the cul- is ture medium. These results suggest that the death of liver cells primary lesion a loss of the semipermeable properties of the in the intact animal is very likely to be similarly dependent upon cells with consequent entry of lethal concentrations of Ca2+. calcium ions. Subsequent studies documenting alteration in calcium ho- meostasis with carbon tetrachloride supported this hypothesis MATERIALS AND METHODS (7-11). The antihistamine compounds also protected against CCI4-induced liver necrosis (6). With galactosamine a corre- Isolated hepatocytes were prepared from the livers of nonfasted lation was shown among early plasma membrane injury, an female Wistar rats (150-200 g) by the method of collagenase increased Ca2+ content, and the ability of uridine to prevent perfusion as described by Laishes and Williams (27). Colla- and to reverse these changes while preventing the liver cell genase (type 1, Sigma) at 100 units/ml in 250 ml of Hanks' death (12, 13). balanced salt solution at pH 7.4 (28) (Flow Laboratories, Such accumulations of Ca2+ could always be explained, McLean, VA) was recirculated through the liver for 15 min at however, as simply a passive equilibration of Ca2+ concentra- 37°C at a flow rate of 32 ml/min. A yield of 2-4 X 108 cells with tions in cells lethally injured by different and as yet unexplained 85-90% viability (trypan blue exclusion) was obtained. mechanisms. The ability to maintain isolated hepatocytes in The hepatocytes were plated in plastic multiwell chambers culture for several days has provided a means for more directly (Costar, Cambridge, MA) with wells 1.6 cm in diameter at a exploring the role of Ca2+ in toxic liver cell death. In particular, density of 5 X 104/cm2 in Williams' medium E (29) (Flow the effects of phalloidin on such cells (14, 15) provide an ideal Laboratories) containing 10% inactivated (560C for 10 min) experimental model. fetal calf serum (Flow Laboratories), garamycin at 50 Mg/ml, and insulin at 0.02 units/ml. Williams' medium E plus serum, The publication costs of this article were defrayed in part by page garamycin, and insulin is referred to below as complete Wil- charge payment. This article must therefore be hereby marked "ad- liams' medium. After incubation at 370C in a humidified at- vertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. * To whom correspondence should be addressed. 1177 Downloaded by guest on September 24, 2021 1178 Medical Sciences: Kane et al. Proc. Natl. Acad. Sci. USA 77 (1980) mosphere of 5% C02/95% air for 90 min to allow the cells to In the presence of 3.6 mM CaCl2, phalloidin at 50 ug/ml attach, the cultures were rinsed three times with pre-warmed killed 60-70% of the cells within 1 hr as evidenced by their Hanks' balanced salt solution and incubated for 15-20 hr in inability to exclude trypan blue (Table 1). This represents complete Williams' medium. Prior to treatment of the cultures 60-90% of the hepatocytes, because at least 20% of the cells are with phalloidin, cytochalasin B, or the Ca2+ ionophore A23187, probably Kupffer cells as evidenced by the presence of Fc re- the cells were rinsed four times with prewarmed Ca2+-free ceptors (unpublished data). In the absence of extracellular Ca2 , Hanks' balanced salt solution and placed in Williams' medium the same concentration of phalloidin had no effect on cell via- E without CaC12 (made under special order by Flow Labora- bility (Table 1). Without any calcium in the medium the cells tories) and serum. Where indicated CaCI2 was added to give remained viable in the presence of phalloidin for at least 8 hr. a final concentration of 3.6 mM. If the cells were exposed to phalloidin for 1 hr in the absence Phalloidin (Boehringer Mannheim) was dissolved in Ca2 - of Ca2+, washed six times, and then exposed to medium con- free Hanks' solution and added to the cultures at a concentration taining Ca2+, 54% of the cells took up trypan blue within 2 hr of 50 Mg/ml. A23187 was a gift from R. L. Hammill (Eli Lilly (Table 1). We have shown previously (32) that this use of trypan Research Laboratories, Indianapolis, IN). It was dissolved in blue to assess the viability of cultured hepatocytes correlates absolute ethanol (1 mg/ml) and added to the cultures at a final exactly with two independent assays, the hydrolysis of fluo- concentration of 10 yg/ml. Ethanol alone did not affect cell rescein diacetate and the plating efficiency. viability. Cytochalasin B (Sigma) was dissolved in Ca2 -free In order to interpret these results it is necessary to know Hanks' solution and added to the cultures at a concentration of whether the interaction of phalloidin with the liver cells is de- 10 Yg/ml. pendent upon extracellular Ca2+. This interaction produces Cell viability was assayed by trypan blue exclusion. Trypan morphological alterations in the surface of the hepatocytes blue (0.4% in 0.15% NaCl, GIBCO) was added directly to the easily appreciated with scanning electron microscopy (14, 15). cultures at a final concentration of 0.01%. Within 10 min the Fig. 1A illustrates the appearance by scanning electron mi- attached cells that excluded the dye were counted by using a croscopy of a representative hepatocyte 1 hr after exposure to 10-mm2 eyepiece grid in an inverted microscope at X200 phalloidin at 50 Mg/ml in the presence of extracellular Ca2+. magnification. In all cases, the number of attached cells was The normal surface architecture has been altered by the pres- 90-95% of the number of cells present at the start of the ex- ence of numerous blebs or protrusions presumably representing periment. Trypan blue exclusion is expressed as the percentage evaginations of the plasma membrane. These evaginations are of the number of unstained cells in untreated cultures. All most impressive where the cells are attached to the substrate, measurements were made by counting three fields per culture where they spread out along the surface of the plastic coverslip. with 200 cells per field and up to five fields per culture with Such changes were seen uniformly in 70-80% of the cells. This fewer than 200 viable cells per field. Three separate cultures alteration in the surface morphology of the cultured hepatocytes were used for each value in the tables. Such data, therefore, was not dependent upon extracellular Ca2 . Fig. 1B illustrates represent the mean + SD of three separate cultures and a total a similar effect'of phalloidin on the surface morphology in the of 9-15 fields. absence of Ca2+ in the culture medium. The number, size, and In preparation for scanning electron microscopy the isolated distribution of the surface evaginations are indistinguishable hepatocytes were plated onto no. 11/2 round plastic tissue culture from those seen with Ca2+. Again, virtually all the cells were coverslips 2.5 cm in diameter (Lux Scientific) at a density of 5 similarly affected. The absence of phalloidin-induced liver cell X 104 cells per cm2. The coverslips were placed in plastic death in culture medium devoid of Ca2+ cannot, therefore, be multiwell chambers and treated with phalloidin as described explained by an inability of phalloidin to alter the plasma above. After treatment the plastic coverslips layered with cells membrane morphology of the hepatocytes. were fixed in 1% glutaraldehyde in a 0.1 M sodium cacodylate This plasma membrane lesion is felt to be a consequence of buffer, pH 7.2 (300 milliosmolar) at room temperature for 2 the interaction of phalloidin with'actin intimately associated hr. The coverslips were rinsed twice in the cacodylate buffer with the cytoplasmic surface of these membranes. It is con- and postfixed in OS04 in the same buffer; The rinsed cells were ceivable, however, that this plasma membrane alteration is not then dehydrated in a graded series of ethyl alcohol solutions. the lethal injury in phalloidin toxicity, there being some as yet The coverslips in 100% alcohol were quickly transferred to a undefined disturbance that is dependent upon extracellular previously cooled critical point apparatus (Sorvall). Critical Ca2+. The role of the interaction of phalloidin with actin in the point drying was carried out according to the method of An- generation of lethal cell injury was explored with the use of derson (30). The dried samples were mounted on specimen cytochalasin B. stubs and coated with a thin layer of gold on a rotating stage in a Denton Vacuum 502 apparatus. The specimens were exam- ined in an ETEC Autoscan microscope and the images we-re Table 1. Calcium dependence of phalloidin-induced recorded on Polaroid P/N 55 film. liver cell death % of control viability RESULTS Medium Medium Immediately after preparation, isolated rat hepatocytes lose Treatment plus Ca2+ minus Ca2+ viability if cultured without serum. However, after 12-15 hr None 100 : 3.7 95.4 ± 5.1 in complete medium plus 10% fetal calf serum, the hepatocytes Phalloidin (50 38.3 + 3.0 103 + 4.7 can be maintained for up to 24 hr in medium devoid of serum Atg/ml) Phalloidin (50 in Ca2+-free and concentration was qg/ml Ca2+. The of total Ca2+ in this medium medium, then washed and Ca2+ less than 20,uM as determined by atomic spectros- absorption added) 46.0 + 1.8 copy. Alternatively, the Ca2+ concentration outside the cells was made 3.6 mM by the addition of CaCl2. The Ca2+ con- Rat hepatocytes were exposed to phalloidin for 1 hr at 37°C. Via- bility was assayed by trypan blue exclusion. In the third centration in the cytosol of the liver cells must be on the order experiment the hepatocytes were exposed to phalloidin at 50 ,g/ml in Ca2+-free of 1 MM, because slight displacements from this value would medium for 1 hr; the cells were washed six times, then exposed to most likely result in net mitochondrial Ca2+ uptake at the ex- medium containing 3.6 mM CaCl2 and assayed for viability after 10 pense of ADP phosphorylation (31). min. Downloaded by guest on September 24, 2021 Medical Sciences: Kane et al. Proc. Natl. Acad. Sci. USA 77 (1980) 1179

Cytochalasin B by virtue of its ability to interact with actin Table 2. Prevention by cytochalasin B of phalloidin-induced microfilaments (33, 34) has been reported to both prevent and liver cell death reverse the effect of phalloidin on the polymerization of actin % of control viability (35). Table 2 indicates that pretreatment of the hepatocytes with Medium Medium cytochalasin B at 10 yg/ml in the presence of Ca2+ completely Treatment plus Ca2+ minus Ca2+ prevented the toxicity of phalloidin. In this case there was no None 100 ± 3.2 101 ±3.2 alteration in the appearance of the cell surface. When this al- Phalloidin (50 Ag/ml) 40.6 + 2.5 104 i 2.4 teration was allowed to develop by exposing the cells to phal- Cytochalasin B (10 Ag/ml) for 30 loidin at 50 ,ug/ml in the absence of extracellular Ca2+ (so that min, then phalloidin (50 jig/ml) the cells would be injured but not killed) and these cells were forlhr 93.2±4.1 96.8±4.0 then treated with cytochalasin B again in the absence of Ca2 Phalloidin (50 gg/ml) for 1 hr in Ca2+-free medium, then cytochalasin B (10 tg/ml) for 30 min in same medium followed by addition of Ca2+ 101 + 4.0 Rat hepatocytes were exposed to phalloidin for 1 hr at 370C. In the third treatment the cells were pretreated for 30 min at 370C with cytochalasin B at 10 gg/ml before addition of phalloidin. In the fourth treatment the cells were exposed to phalloidin at 50 Mg/ml in Ca2+-free medium for 1 hr, then treated with cytochalasin B for 30 min, and finally exposed to 3.6 mM CaCl2. Viability was assayed by trypan blue exclusion. there was no loss of viability of the cells upon addition of Ca2+ (Table 2). Cytochalasin B (10 ,g/ml) for 1 hr had no effect by itself on the viability of the hepatocytes in the presence or ab- sence of Ca2+. The experiments with cytochalasin B and phalloidin imply that the lethal cell injury is, indeed, dependent upon the interaction of phalloidin with actin and that this injury is not necessarily irreversible, even in the presence of such distortions of the cell surface as are illustrated in Fig. 1. In order to document directly the lethal consequences of an influx of Ca2+, calcium homeostasis was perturbed in the cul- tured hepatocytes without the necessity of disrupting the in- tegrity of the plasma membrane. A23187 is a Ca2+ ionophore whose specific biological activity is to create Ca2+ channels and, thereby, to overcome the permeability barrier represented by cellular membranes (36). Table 3 indicates that within 1 hr of the exposure of hepatocytes to A23187 at 10,g/ml, 95% of the cells were dead. In the absence of a large Ca2+ gradient across the plasma membrane-i.e., in the absence of Ca2+ in the culture medium-the same concentration of A23187 had no effect on the viability of the liver cells. Addition of Ca2+ to the culture medium at any time after addition of A23187 resulted in the deterioration of the cells within 5-10 min, with again virtually all of the cells taking up trypan blue. Cytochalasin B did not prevent the killing by A23187 either after pretreatment or after an initial exposure to the ionophore in the absence of Ca2+. DISCUSSION The data presented above allow the following reconstruction of the events underlying phalloidin-induced liver cell death. The earliest clearly definable alteration is the interaction of phalloidin with plasma membrane-associated microfilaments, leading to the formation of Ph-actin (19-25). This event occurs

FIG. 1. Scanning electron micrographs of cultured rat hepato- Table 3. Calcium dependence of the toxicity of A23187 cytes treated with phalloidin at 50 gg/ml for 1 hr in the presence of % of control viability 3.6 mM CaCl2 in the culture medium (A) or in the absence of extra- cellular Ca2+ (B). (X1400.) In A, numerous evaginations of the plasma Medium Medium membrane are most noticeable adjacent to the attached surface of the Treatment plus Ca2+ minus Ca2+ hepatocyte where the cells can anchor to the plastic substrate. Many None 100 ±3.7 96.2+5.1 smaller evaginations are also visible on the free surface of the cell. In 5.8 + 0.8 98.3 + 4.4 B, both cells exhibit the same alteration of the surface as seen in A, A23187 with large evaginations of the plasma membrane spreading along the Rat hepatocytes were exposed to A23187 at 10 ,g/ml for 1 hr at plastic substrate and smaller blebs on the free surface. 37°C. Viability was assayed by trypan blue exclusion. Downloaded by guest on September 24, 2021 1180 Medical Sciences: Kane et al. Proc. Natl. Acad. Sci. USA 77 (1980) on the cytoplasmic side of the plasma membrane. Becap- there 3. McLean, A. E. M., Ahmed, K. & Judah, J. D. (1964) Ann. N.Y. is no evidence that there is any specificity for liver cell actin, Acad. Sci. 116,986-989. the specificity for liver cell toxicity with phalloidin presumably 4. McLean, A. E. M., McLean, E. K. & Judah, J. D. (1965) Int. Rev. relates to the as yet undefined mechanism whereby phalloidin Exp Pathol. 4, 127-157. recognizes a liver cell and is transported across the plasma 5. Rees, K. R. (1962) in Ciba Symposium on Enzymes and Drug of with Action, eds. Mongar, J. L. & de Reuck, A. V. S. (Churchill, Lon- membrane. The interaction phalloidin liver cell actin don), pp. 344-358. is not dependent upon extracellular Ca2+, and it leads to the 6. Rees, K. R., Sinha, P. & Spector, W. G. (1961) J. Pathol. Bacteriol. alterations in the surface morphology of the cells illustrated in 81, 107-118. Fig. 1. Pretreatment of the liver cells with cytochalasin B pre- 7. Reynolds, E. S., Thiers, R. E. & Vallee, B. L. (1962) ]. Biol. Chem. vents or reduces the interaction of phalloidin and actin as evi- 237,3546-3551. denced by the prevention of the altered surface morphology 8. Reynolds, E. S. (1963) J. Cell Biol. 19, 139-157. and subsequent cell death (Table 2). Although the interaction 9. Reynolds, E. S. (1964) Lab. Invest. 13, 1457-1470. with actin is necessary for phalloidin-induced cell death, this 10. Smuckler, E. A. (1966) Lab. Invest. 15, 157-166. interaction does not necessarily produce irreversible injury. If 11. Moore, L., Davenport, G. R. & Landon, E. J. (1976) J. Biol. Chem. the hepatocytes were treated in the absence of Ca2+ initially 251, 1197-1201. with phalloidin for 1 hr and then with cytochalasin B for l/2 12. Farber, J. L., Gill, G. & Konishi, Y. (1973) Am. J. Pathol. 72, hr, 53-60. there was no cell death upon addition of Ca2+ (Table 2). It 13. El Mofty, S. K., Scrutton, M. C., Serroni, A., Nicolini, C. & Farber, would seem that lethal cell injury is dependent more on the J. L. (1965) Am. J. Pathol. 79,579-596. continued interaction of the Ph-actin with the plasma mem- 14. Frimmer, M. (1975) in Pathogenesis and Mechanisms ofLiver brane than it is on the mere presence of the membrane lesions Cell Necrosis, ed. Keppler, D. (MTD Press, Lancaster, England) illustrated in Fig. 1. pp. 163-174. Whereas the development of phalloidin-induced membrane 15. Frimmer, M. (1977) Naunyn Schmiedebergs Arch. Pharmakol. injury is not dependent upon extracellular Ca2+, the mecha- 297, S15-S19. nisms connecting such injury with the death of the cells are 16. Frimmer, M. (1971) Naunyn Schmiedebergs Arch. Pharmakol. clearly dependent upon extracellular Ca2+. Taking into account 269, 152-163. the Ca2+ dependency of both A23187- and phalloidin-induced 17. Wieland, Th. & Wieland, 0. (1972) Microbiol. Toxins 8, 249- cell death, the simplest interpretation of these data is that 280. 18. Agostini, B., Govindan, V. M. & Hofman, W. (1975) in Patho- phalloidin injury alters the permeability properties of the genesis and Mechanisms of Liver Cell Necrosis, ed. Keppler, plasma membrane to Ca2+, allowing influx of Ca2+ down the D. (MTD Press, Lancaster, England) pp. 175-192. steep electrochemical gradient that must exist between the 19. Wieland, Th. & Faulstich, H. (1978) Crit. Rev. Biochem. 5, outside and the inside of the cells in the presence of 3.6 mM 185-260. extracellular Ca2+. Although we clearly have no indication of 20. Lengsfeld, A. M., Low, I., Wieland, Th., Dancker, P. & Hassel- how much Ca2+ must actually enter the cell in order to kill it, bach, W. (1974) Proc. Natl. Acad. Sci. USA 71, 2803-2807. our data would suggest that such an entry most likely occurs and 21. Low, I. & Wieland, Th. (1974) FEBS Lett. 44,340-343. is, at this point, the most obvious candidate for the event that 22. Dancker, P., Low, I., Hasselbach, W. & Wieland, Th. (1975) couples the plasma membrane injury with the requirement for Biochim. Biophys. Acta 400, 407-414. extracellular Ca2+. The dependency on Ca2+ in phalloidin- 23. L6w, I., Dancker, P. & Wieland, Th. (1975) FEBS Lett. 54, induced cytotoxicity that a 263-265. implies specific membrane alter- 24. Lutz, G., Glossman, H. & Frimmer, M. (1972) Naunyn ation rather than a generalized membrane-perturbing effect Schmiedebergs Arch. Pharmakol. 273, 341-351. mediates the response of the liver cells. 25. Faulstich, H., Schafer, A. J. & Weckauf, M. (1977) Hoppe Seylers Our primary focus of concern still remains the role of Ca2+ Z. Physiol. Chem. 358,181-186. in liver cell necrosis in the intact animal or human. The results 26. Frimmer, M., Kroker, R. & Porstend6rfer, J. (1974) Naunyn of the present study strengthen the hypothesis that disturbances Schmiedebergs Chem. 358,181-186. in calcium homeostasis induced by a variety of hepatotoxins are 27. Laishes, B. A. & Williams, G. M. (1976) In Vitro 12,521-532. causally related to liver cell death. The results also imply that 28. Hanks, J. H. & Wallace, R. E. (1949) Proc. Soc. Exp. Biol. Med. the further use of the in vitro system to assess the ability of a 71, 195-200. variety of drugs to prevent Ca2+ fluxes induced by phalloidin 29. Williams, G. M., Weisburger, E. K. & Weisburger, J. H. (1971) may suggest specific therapeutic agents to be used in vivo. Exp. Cell Res. 69, 106-112. 30. Anderson, T. F. (1951) Trans. N.Y. Acad. Sci. 13, 130-131. The authors express their appreciation to Joseph T. Martin for his 31. Nicholls, D. G. (1978) Biochem. J. 176,463-474. expert assistance in the preparation of the scanning electron micro- 32. Schanne, F. A. X., Kane, A. B., Young, E. E. & Farber, J. L. (1979) graphs. This work was supported by grants from the National Institutes Science 206,700-702. of Health-CA 12093, CA 12923, and AM 19154. 33. Spudich, J. A. & Lin, S. (1972) Proc. Natl. Acad. Sci. USA 69, 441-446. 1. Gallagher, G. H., Gupta, D. N., Judah, J. D. & Rees, K. R. (1956) 34. L6w, I. & Dancker, P. (1976) Biochim. Biophys. Acta 430, J. Pathol. Bacteriol. 72, 193-201. 366-374. 2. Judah, J. D., Ahmed, K. & McLean, A. E. M. (1964) in Ciba 35. Low, I., Dancker, P. & Wieland, Th. (1975) FEBS Lett. 54, Symposium on Cellular Injury, eds. de Reuck, A. V. S. & Knight, 263-265. J. (Churchill, London), pp. 187-205. 36. Pressman, B. C. (1976) Annu. Rev. Biochem. 45,501-530. Downloaded by guest on September 24, 2021