Calcium-dependent calmodulin-binding proteins associated with mammalian DNA polymerases α

Biochimica et Biophysica Acta, 951 (1988) 315-321 Elsevier

315

BBA 91892

Calcium-dependent calmodulin-binding proteins associated with m a m m a l i a n D N A p o l y m e r a s e s a R u s s e l l A. H a m m o n d a, K i m b e r l y A. F o s t e r a, M a r t i n W. B e r c h t h o l d b, M a x G a s s m a n n b, A n d r e w M. H o l m e s c, U l r i c h H i i b s c h e r b a n d N e a l C. Brown a Department of Pharmacology, University of Massachusetts Medical School, Worcester, MA ( U.S.A.), l, Department of Pharmacology and Biochemistry, University of Ziirich, Ziirich (Switzerland) and c Department of Biochemistry, Uniformed Services of Health Sciences, Bethesda, MD (U.S.A.) (Received 21 September 1988)

Key words: DNA polymerase a; Calmodulin binding protein; Calcium ion; (Mammalian mitotic cycle)

Complex, muitiprotein forms of bovine (calf thymus), hamster (Chinese hamster ovary cell), and human (HeLa) cell DNA polymerase a (Pola) were analyzed for their content of calmodulin-binding proteins. The approach used an established autoradiographic technique employing 12SI-labeled calmodulin to probe proteins in denaturing SDS-polyacrylamide gel electropherograms. All three Pola enzymes were associated with discrete, Ca2+-dependent calmodulin-binding proteins. Conventionally purified calf thymus Pola holoenzyme contained three prominent, trifluoperazine-sensitive species with apparent molecular masses of approx. 120, 80 and 48 kDa. The 120 and 48 kDa species remained associated with the polymerase • primase core of the calf enzyme during immunopurification with monoclonal antibodies directed specifically against the polymerase subunit. The patterns of the calmodulin-binding proteins displayed by conventionally purified preparations of hamster and human Pola enzymes were similar to each other and distinctly different from the pattern of comparable preparations of calf thymus Pola. Immunopurified preparations of the human and hamster Polas retained significant calmodulin-binding activity of apparent molecular masses of approx. 55, 80 and 150-200 kDa.

Introduction Abbreviations: Cam, calmodulin; CamBP, calmodulin-binding protein; Pola, DNA polymerase a; MAB, monoclonal antibody; CHO, Chinese hamster ovary; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; TFP, trifluoperazine; EGTA, ethyleneglycol-O,O'-bis(2-aminoethyl) N, N, N ', N '-tetraacetic acid. Correspondence: N.C. Brown, Department of Pharmacology, University of Massachusetts Medical School, Worcester, MA 01655, U.S.A.

C a 2+ is an essential stimulatory modulator of the prereplicative, G 1 phase of the mammalian mitotic cycle (for a review, see Ref. 1). The stimulatory effect of Ca 2÷ on the G1 --, S transition is derived in part through its reaction with Ca 2÷specific intracellular receptors (for a review, see Ref. 2). One of the major receptors involved in the

0167-4781/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)

316

transduction of the prereplicative Ca 2+ signal is the calcium-binding protein, calmodulin (Cam) [2-4]. Cam is a small ( M r 16700), strictly conserved eukaryotic protein which serves as a sensor of change in the intracellular concentration of Ca 2+ [2]. The utility of Cam as a sensor and transducer of Ca 2 +-specific signals is derived from its high affinity for Ca 2 + and the dramatic change in conformation it undergoes upon reaction with Ca 2+. The Cam-Ca 2+ conformer can strongly bind and perturb the function of a wide variety of target calmodulin-binding proteins (CamBPs) for which the Ca2+-deficient conformer displays little or no affinity. Although the dependence of the G~ --* S transition on Cam is well documented [4-9], the Camspecific target(s) and mechanisms basic to this dependence have not been identified. One of the systems which we have considered as a possible Cam target is the multiprotein DNA Pola 'holoenzyme' complex which constitutes the enzymatic core of the machinery responsible for S-phasespecific replicative DNA synthesis (for a review, see Ref. 10). Specifically, we hypothesize that Cam, through its reaction with specific CamBP receptors during G 1, positively and directly modulates the formation a n d / o r functional state of a replication-'competent,' S-phase-specific form of DNA Pola. Our first step in the pursuit of this hypothesis has been simply to determine whether mammalian Pola is associated significantly with a CamBP which might serve as a Pola-specific, Cam-Ca 2+ target. The results of this initial investigation, which we describe below, indicate that mammalian Polas are strongly associated with discrete, Ca2+-dependent CamBPs.

Materials and Methods Materials. Sepharose 4B and Protein A-Sepharose C1-4B were from Pharmacia. Murine MAB17 [11] was provided by Dr. Lucy Chang. Murine MAB-287-38 [12] was isolated as described in Ref. 13 from the medium supernatant of hybridoma CRL-1644 (American Type Culture Collection). Conventionally purified HeLa cell DNA Polet (DNA-cellulose fraction, Ref. 14) was provided by Dr. Earl Baril. Conventionally purified CHO Polet

was the DNA-cellulose fraction IV [15] produced in this laboratory. Murine DNA primase [16] was provided by Dr. Ben Tseng, Hela-S, and CHO-S cells used for the immunopurification of the respective Polc~s were grown in suspension culture and processed as described in, respectively, Refs. 14 and 15. Trifluoperazine (TFP) was provided by Smith, Kline and French. Methods. Pola activity was assayed with activated calf thymus DNA as described in Ref. 17, and DNA primase activity was assayed as described in Ref. 23. MAB-17 was cross-linked with dimethylsuberimidate to protein A-Sepharose CL-4B as described in Ref. 11. MAB-287-38 was covalently linked to CNBr-activated Sepharose 4B as described in Ref. 13. Protein was determined by the method of Bradford [18] with a bovine serum albumin standard. Calf thymus Polc~ was extracted and immunopurified through the G-50 step as described in Ref. 11. Calf thymus Pola was conventionally purified through step V (RC-c0, as described in Ref. 17. Immunopurification of Pol~s on MAB-287-38-Sepharose followed the procedure described in Ref. 13. Detection o f Ca m BPs. Bovine testis Cam (Pharmacia) was labeled with 125I using the Bolton Hunter [19] procedure described in Ref. 25; the 125I-labeled Cam had a specific activity of (1-2)108cpm/~g and a radiopurity of more than 99.5%. Protein preparations were denatured and subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) as described by Laemmli [20] on 8.8 or 10% mini gels (Mighty Small, Bio-Rad; thickness, 1.5 mm). The gels were processed, subjected to 1251-Cam overlay in the presence of either Ca 2+ (1 mM) or the chelating agent, EGTA (2 mM), and autoradiographed as described in Ref. 21. For 125I-Cam overlay of gel blots, the Laemmli gels were prepared as described above and blotted to nitrocellulose (Hybond C) sheets; the blots were subjected to overlay in the presence of either Ca 2+ (1 mM) or E G T A (2 raM) by the method of Flanagan and Yost [22]. The blot overlay method was modified by the inclusion of 5% non-fat dry milk in the quenching step and deletion of Tween 20 from the post-overlay, wash step. Protein bands were detected visually by staining gels and gel blots with, respectively, Coomassie brilliant blue [20] and Amido black [22].

317

R ~

CamBPs in conventionally purified bovine Pola The primary target of our investigation was calf thymus DNA Pola. Our first experiments analyzed the proteins in a preparation of calf Pola purified approx. 15 000-fold in the presence of ATP, EDTA and proteinase irthibitors [17]. Purification consisted of: (i) ammonium sulfate fractionation; (ii) adsorption and batch elution from phosphocellulose; (iii) isolation in the void fraction of Sephacryl S-300; (iv) adsorption and gradient elution on Bio-Rex 70; and (v) readsorption and gradient elution on phosphocellulose. RC-a, the product of step v, is a multiprotein 'holoenzyme' complex capable of priming and executing DNA synthesis on circular single-stranded DNA templates [17]. The denaturing gel electropherogram of Fig. 1A displays the polypeptides of RC-a - a collection of at least 10 distinct species ranging in size from 200 to 45 kDa. Fig. 1B displays the ]25I-Cam overlays of gel blots of RC-a (right lanes) and the relatively crude product of the phosphocellulose batch step ii (left lanes). Comparison of the autoradiograms of gels probed in the presence of Ca 2+ or in its absence, in EGTA, indicates the presence of strong Ca2+-dependent CamBP signals in both

A

RCALPHA c.B

P.CELL +

_

B

RCALPHA +

_

116-92-- I 66_

200--

45-Fig. 1. Polypeptide composition and CamBPs of conventionally purified calf thymus Pola. (A) 10 #g of RC-a (fraction v) was analysed by S D S - P A G E on an 8.8% gel and visualized with Coomassie brilliant blue (CBB). (B) 10/~g each of P-cell fraction ii and RC-a fraction v were electrophoresed on 10% SDS polyacrylamide gels and subjected to t25I-CAM-blot overlay analysis in the presence ( + ) and abseace ( - ) of Ca 2+, as described in Materials and Methods. Autoradiography was for 72 h.

fractions. The early phosphocellulose fraction displayed strong Ca2+-dependent Cam-binding signals at the respective positions of -- 120, = 75-80, --48 and < 30 kDa. The signal pattern of the RC-a preparation suggests that ' significant amounts of the major 120 and 80 kDa signals of fraction ii survived the subsequent size selection and successive chromatographic steps through which it passed to generate RC-a. During the aging and handling of RC-a, the level of the 80 kDa species appeared to increase relative to that of the 120 kDa species, suggesting that the 80 kDa species is not a unique protein but a signal-positive polypeptide derived proteolytically from the 120 kDa protein (Hammond, R.A., Foster, K.A. and Brown, N.C., unpublished results). The 48 kDa CamBP of fraction ii is not remarkably strong in the relatively 'old' RC-a preparation used in the experiment of Fig. 1. However, in fresher preparations of RC-a, we have found the 48 kDa signal to be considerably more intense relative to that of the 120 kDa species, suggesting its deterioration upon prolonged storage at - 80 o C and upon freezing and thawing (Hammond, R.A., Foster, K.A+ and Brown, N.C., unpublished results).

CamBPs in calf thymus Pola polymerase -primase produced by immunopurification Are the CamBPs of conventionally purified RC-a firmly bound in a CamBP-Pola complex? We attempted to address this question by analyzing the CamBP content of bovine Pola immunopurified with agaroses coupled to monoclonal antibodies (MABs) directed specifically at the Pola catalytic subunit. We employed two different systems. The first was the system devised by Chang et al. [11]; it employed: (a) a murine MAB directed against bovine Pola; (b) adsorption in low salt; (c) extensive washing in 1 and 2 M NaC1; and (d) elution with 3.2 M MgC12. The second system was devised by Nasheuer and Grosse [13]; it employed: (a) MAB-287-38, a Pola-neutralizing MAB raised against human Pola [12]; (b) adsorption in low salt; (c) extensive washing in 150 mM KC1; and (d) elution with 1 M KC1 at pH 12.5. As expected, both methods yielded active polymerase, primase, and as shown in the denaturing gel electropherograms in Fig. 2A, the complex produced by each method was similar. Each pre-

318

A

of the primase heterodimer [11,23]: (c) a diffuse b a n d in the 60 70 kDa position: a n d (d) various m i n o r b a n d s of relatively low intensity at positions u n d e r 45 kDa. The results of overlay analysis of the two i m m u n o p u r i f i e d enzymes are summarized in Fig. 2B, Both p r e p a r a t i o n s clearly displayed two m a j o r Ca 2 +-dependent C a m B P signals - one at the position of approx. 120 k D a a n d a n o t h e r in the 48 k D a region. The 48 k D a signal was c o i n c i d e n t with a broad, stain-reactive b a n d in the region occupied by the smaller primase subunit. In contrast, little stain-reactive material coincided with the approx. 120 k D a signal. To control for the possibility that the 120 a n d 48 k D a C a m B P signals might have b o u n d n o n specifically to m u r i n e M A B (IgG)-agarose, we performed a sham i m m u n o p u r i f i c a t i o n of calf Polc~ o n a c o m p a r a b l e m u r i n e anti-fl-galactosidase MAB-agarose; the eluate c o n t a i n e d no detectable P o l a activity or CamBPs.

B

MAB 17

MAB 2 8 7

CaB

CBa

M A B 17

÷

MAB 2 8 7

--

+

--

200_ 116 _ 97--

9;'

68-

68

43--

43

i! Fig. 2. Polypeptide composition and CamBPs of immunopurifled calf thymus Pola. The enzymes was purified on MAB-17Sepharose and MAB-287-38-Sepharoseas described in Materials and Methods. (A) 6 p.g of the product of each column was analysed by SDS-PAGE on 10% gels and visualized with Coomassie brilhant blue (CBB). (B) Gels identical to those of (A) were subjected to blot overlay analysis in the presence (+) or absence ( - ) of Ca 2+ as described in Materials and Methods.

CamBPs in hamster and human Pola p a r a t i o n displayed: (a) a m a j o r MAB-reactive p o l y p e p t i d e of approx. 1 3 0 - 1 4 0 kDa, which represents a proteolysed form of the 1 6 0 - 1 8 0 k D a catalytic (polymerase) s u b u n i t ; (b) peptides in the 4 5 - 6 0 k D a range expected for the 48 • 58 s u b u n i t s

A DNA

Fig. 3 summarizes the results of analysis of D N A P o l a p r e p a r a t i o n s derived from cultured ( C H O ) cells. (The results of analysis of e q u i v a l e n t p r e p a r a t i o n s of Polc~ of h u m a n (HeLa) cells were nearly identical to those o b t a i n e d with C H O cell

CELL. 287 ELUATE CBB

4iii!iiii!iii!ii~i

C

B 287

287

INPUT

ELUATE

4-

-

4-

-

ii~iiiiiii~ 200

_~;ii!i!i~i

200

_200 _

97

--

97--

-- 6 8 - --

42

~iiii~i!ii!iI

43-

ii:~:~:::

43

ii Fig. 3. CamBPs in CHO Pola. (A) Enzyme was purified conventionally through the DNA cellulose step as described in Ref. 15; 25 /~g of this material was electrophoresed on 8.8% SDS polyacrylamide gels and subjected to overlay analysis in the presence ( + ) and absence ( - ) of C a 2 + a s described in Materials and Methods. Autoradiography was for 72 h. (B) Enzyme was immunopurified with MAB-287-38-Sepharose as described in Materials and methods, and approx. 10/.tg of this material was analysed by SDS-PAGE on 10% gels and visualized with Coomassie brilliant blue (CBB). (C) 10/tg of the immunopurified enzyme (eluate, right lanes) and 20 ,ttg of the extract used as the starting material for irnmunopurification (input, left lanes) were subjected to blot overlay analysis in the presence ( + ) and absence ( - ) of C a 2 ÷ a s described in Materials and Methods. Autoradiography was for 72 h.

319 Pola, and therefore, are not shown). Fig. 3A displays the results of gel overlay analysis of a partially purified DNA cellulose fraction prepared by a conventional technique [15]. The DNA cellulose fraction displayed several distinct, Ca2+-depen dent signals - a major signal at approx. 55 kDa and several weaker signals at positions representing higher and lower molecular weights; unlike comparably purified calf Pola, this preparation did not display a pattern combining a dominant pair of Ca2÷-dependent signals at 120 and 48 kDa. Fig. 3B displays a stained SDS-polyacrylamide gel electropherogram of the immunopurified Pola-primase complex of CHO cells. The gel displayed a major 200 kDa protein, minor 150-170 kDa bands and a pattern of smaller stain-reactive polypeptides roughly comparable to that displayed by immunopurified calf Pola. Fig. 3C displays the results of Cam blot analysis of the immunopurified CHO Pola and the crude extract used as its source. The crude extract (input, left lanes), like the DNA-cellulose fraction in A, displayed a major Ca2+-dependent signal at approx. 55 kDa and slightly less intense signals at approx. 80 kDa and in the 150-170 kDa region (cf. arrowheads). The immunopurified Polaprimase (right lanes) retained discernible Ca 2+dependent signals at each of the positions (arrowheads) noted in the crude input - including those of the 150-170 kDa region, in which the putative polymerase-specific polypeptides are distributed. Comparison of the signal patterns of the crude and immunopurified fractions suggests that immunopurification intensified the signal of the 150-170 kDa doublet relative to that of the 55 kDa species.

Sensitivity of the calf thymus CamBPs to TFP The antipsychotic drug, TFP, binds Cam in a Ca2+-dependent manner, occluding the site at which it binds many of its target proteins [24]. To examine the TFP sensitivity of the calf proteins, we exposed blot overlays of fraction ii to the inhibitor at a concentration of 200 /~M. The resuits of this experiment, which are not shown, indicated that 200 btM TFP completely blocked the signals generated by all three major CamBPs.

Is the thymus 48 kDa CamBP the smaller subunit of the DNA primase 45-48 heterodimer? The answer to this question is not resolved and is not likely to be resolved unequivocally until we succeed in isolating the 48 kDa CamBP in significant amounts in pure form. Evidence supporting an affirmative answer includes the coincidence of the primase-specific 48 kDa subunit [23,26] and the presence of considerable 48 kDa CamBP in the primase-rich, high-salt wash [23] of MAB-17Sepharose/Pola-primase complex (results not shown). Evidence supporting a negative answer includes our failure to detect: (i) a CamBP signal coincident with the 48 kDa subunit of immunopurified CHO (and HeLa) Pola-primase (cf. Fig. 3) or highly purified, Pola-free mouse DNA primase [16] (results not shown); and (ii) interference of the Cam 48 kDa interaction by an antibody directed against the 48 kDa calf primase subunit [23]. Discussion

Do specific, Cae+-dependent CamBPs strongly associate with mammalian DNA Pola? The results of our experiments strongly support an affirmative answer to this question. In the case of the calf thymus enzyme, RC-a, at least two distinct species of CamBP (120 and 48 kDa signals) consistently accompany the polymerase, following it through salt fractionation, three steps of ion-exchange chromatography, and size selection by gel filtration (cf. Fig. 1). Per se, the 'conventional' copurification of Pola and CamBPs, although extensive, does not exclude the possibility that the CamBP signals are not physically associated with the polymerase. However, when the results of conventional copurification are considered in the context of the finding that significant amounts of the CamBP signals also accompany the enzyme core through purification with two different Polaspecific immunoadsorbents (cf. Fig. 2), the latter possibility becomes remote. If the CamBPs were not strongly associated with a Pola multiprotein complex, they would be expected to behave as they did in our sham experiment with anti-flgalactoside MAB-Sepharose and simply not adsorb to the Pola-specific IgG-agarose matrix.

320 In the case of C H O (and HeLa) Polas, three classes of CamBP signals were observed in crude and partially purified enzyme preparations (Fig. 3A). All three classes of CamBPs apparently survived Pola-specific immunopurification (Fig. 3B, C) - two distinct species at approx. 55 and approx. 80 kDa and a doublet in the 150-170 kDa range where the polymerase subunit and its various proteolysed forms are normally found. At present, we do not know what, relationship, if any, the 150-170 kDa CamBPs have to the polymerase core subunit. Nor do we know what relationship, if any, the 5 5 / 1 5 0 - 1 7 0 kDa species might have to the 4 8 - 1 2 0 kDa calf CamBPs, although we speculate on a possible relationship below.

kDa species, which we find in calf Polc~ and the > 120 kDa species which we detect in immunopurified C H O and HeLa Polas are equivalent to the 'variable,' 120 kDa replitase CamBP. We are attempting to resolve the latter hypotheses by assessing the affiliation of the bovine and hamster CamBPs with the respective Polas and replitases as a function of the mitotic cycle. We also are attempting to isolate the major Pola-associated bovine and hamster CamBPs in amounts sufficient for the characterization of their structure and for the production of antibodies with which to probe their identity and assess their possible role(s) in Pola 'metabolism' and function.

Acknowledgements How might the Pola-associated CamBPs be related to a GFspecific Cam-dependent modification of Pola? An attractive experimental basis to address this question emerges from consideration of the unpublished results of recent experiments of Reddy (Prem Veer Reddy, G., personal communication) with a complex form of Pola derived from nuclei of synchronized embryonic Chinese hamster lung fibroblast cells. The target of Reddy's study has been the multienzyme D N A biosynthetic 'replitase' complex formed specifically in the cell nucleus during the G 1 phase [27]. Prompted by the detection of Cam in the replitase and the observation that the Cam inhibitor, TFP, inhibits the G 1 --~ S transition [9], Reddy has examined the CamBP content of the nuclear replitase isolated during the G~ ~ S period. He has observed that the replitase is associated with two distinct Ca 2+dependent CamBPs, a = 55 and = 120 kDa species. The signal of the 55 kDa species was 'constituitive' and readily detectable throughout G~, whereas the signal of the 120 kDa species appeared to vary from barely detectable early in G~ to strong in late G~. Reddy also observed that the development of the 120 kDa CamBP signal in the replitase was sensitive to the Cam inhibitor, TFP. Considering Reddy's results, we hypothesise: (i) that the 55 kDa CamBP which we observe in the C H O (and HeLa) Polas and the slightly smaller, 48 kDa species which we detect in calf Pola are equivalent to the 'constituitive', 55 kDa replitase-specific CamBP; and (ii) that the 120

We are grateful to Dr. Lucy Chang for the gift of calf-specific anti-Pola MAB, to Dr. Earl Baril for the gift of the HeLa D N A polymerase, to Dr. Ben Tseng for the gift of mouse D N A primase, to Dr. Prem Reddy for the communication of unpublished results, and to Dr. Naseema Khan for valuable assistance in the immunopurification of C H O and HeLa Pola. This work was supported by USPHS research grant GM28775 to NCB, and Swiss National Science Foundation grants 3.6040.87 to U.H. and 3.634-0.87 to M.W.B.

References 1 Metcalfe, J., Moore, J., Smith, "G. and Hesketh, T. (1986) Br. Med. Bull. 42. 405-412. 2 England, P. (1986) Ibid. 375-383. 3 Cheung, W., Lynch, T. and Wallace, R. (1978) Adv. Cyclic Nucleotide Res. 9, 233-251. 4 Rasmussen, C. and Means, A. (1987) EMBO J. 6, 3961-3968. 5 Boynton, A., Whitfield, J. and MacManus, J. (1980) Biochem. Biophys. Res. Commun. 95, 745-749. 6 Chafouleas, J., Bolton, W., Boyd, A. and Means, A. (1982) J. Cell Biol. 95, 26a. 7 Means, A. and Chafouleas, J. (1983) Cell Biol. Internat. Rep. 7, 481-482. 8 Thyberg,J. and Palmberg, L. (1987) Biol. Cell 60, 125-132. 9 Reddy, G.P.V. and Gray, M. (1987) Fed. Proc. 46, 2207. 10 Kaguni, L. and Lehman, I. (1988) Biochim. Biophys. Acta 950, 87-101. 11 Chang, L., Rafter, E., Augl, C. and Bollum, F. (1984) J. Biol. Chem. 259, 14679-14687. 12 Tanaka, S., Hu, S-Z., Wang, T. and Korn, D. (1982) J. Biol. Chem. 257, 8386-8390.

321 13 Nasheuer, H-P. and Grosse, F. (1987) Biochemistry 26, 8458-8466. 14 Vishwanatha, J., Coughlin, S., Weslowski-Owen, M. and Baril, E. (1986) J. Biol. Chem. 261, 6619-6628. 15 Khan, N. and Brown, N. (1985) Mol. Cell. Biochem. 68, 169-179. 16 Tsen 8, B. and Ahlem, C. (1983) J. Biol. Chem. 258, 9845-9849. 17 Ottiger, H-P., Frei, P., H~issig, M. and Hiibscher, U. (1987) Nucleic Acids Res. 15, 4789-4807. 18 Bradford, M. (1976) Anal. Biochem. 72, 248-254. 19 Bolton, A. and Hunter, W. (1973) Biochem. J. 133, 5290-539. 20 Laemmli, U. (1970) Nature 227, 680-685.

21 Glenney, J. and Weber, K. (1980) J. Biol. Chem. 255, 10551-10562. 22 Flanagan, S. and Yost, B. (1984) Anal. Biochem. 140, 510-519. 23 Holmes, A., Cheriathundam, E., Bollum, F. and Chang, L. (1986) J. Biol. Chem. 261, 11924-11930. 24 Levin, R. and Weiss, B. (1977) Mol. Pharmacol. 13,690-697. 25 Chafouleas, J., Dedman, J., Munjaal, R. and Means, A. (1979) J. Biol. Chem. 254, 10262-10267. 26 Nasheuer, H-P. and Grosse, F. (1988) J. Biol. Chem. 263, 8981-8988. 27 Noguchi, H., Reddy, G.P. and Pardee, A. (1983) Cell 32, 443-451.