Proc. Nati. Acad. Sci. USAVol. 87, pp. 9310-9314, December 1990Cell Biology

Transformed and nontransformed cells differ in stability andcell cycle regulation of a binding activity to the murinethymidine kinase promoter

(cycloheximide/DNA binding protein/labile protein/late GI phase/restriction point)

DAVID W. BRADLEY*, QING-PING Dout, JUDITH L. FRIDOVICH-KEILt, AND ARTHUR B. PARDEEtt

*Program on Cell and Developmental Biology, tDepartment of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, and Division ofCell Growth and Regulation, Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115

Contributed by Arthur B. Pardee, September 14, 1990

ABSTRACT A DNA binding activity to an upstream re-gion of the murine thymidine kinase gene is regulated differ-ently in a transformed and nontransformed cell line pair.Differences in regulation were observed (i) after serum levelswere reduced, (ii) when serum levels were returned to initialhigh levels, and (iii) while protein synthesis was inhibited. Afterreduction of serum levels, the binding activity was unstable innontransformed BALB/c 3T3 clone A31 cells but was signif-icantly more stable in benzo[a]pyrene-transformed BALB/c3T3 cells. After serum concentration was returned to highlevels, the kinetic pattern of the binding activity differedbetween nontransformed and transformed cells. While proteinsynthesis was inhibited, the binding activity was unstable innontransformed cells and stable in transformed cells. Partialinhibition of protein synthesis-a more stringent condition totest instability-prevented the induction of the binding activityin nontransformed cells. Previously, the labile protein hypoth-esis set forth the criterion that a protein regulating the onset ofDNA synthesis should be unstable in nontransformed cells andstable in transformed cells. TheDNA binding activity describedhere satisfies this criterion.

The transition from Go phase, a quiescent, nongrowing state,to G1 phase of the cell cycle is the first of several stepspreceding the initiation ofDNA synthesis. In nontransformedcells, the binding of extracellular growth factors to theirplasma membrane receptors is required for this transition (1).Subsequent steps in G1 include the activation of intracellularsecond messengers, kinases, and early response genes (2). Atthe G1/S boundary, the final molecular events regulating theonset of DNA synthesis occur (3) and during S phase acomplete copy of the genome is made (4). When compared tonontransformed cells, transformed cells are often less depen-dent on growth factors (5, 6) or they can have alteredregulation of second messengers, kinases (7), and earlyresponse genes (8).

Since a cell coordinately regulates initiation at many sitesof DNA replication (9) at the start of S phase (10, 11), themolecular events at the G1/S boundary are of particularinterest. Kinetic experiments showed that reduction of serumlevels or inhibition of protein synthesis before a restrictionpoint in late G1 prevented or lengthened the time of entry toS in nontransformed cells (10). However, after this restrictionpoint, the transit from G1 to S was insensitive to serumdeprivation and protein synthesis inhibition (10). Based onthese results and others (11, 12), we proposed that the levelof a regulatory protein passes above a critical threshold at therestriction point and commits a cell to DNA synthesis (13).Protein synthesis inhibition and serum deprivation lengthen

the time of G1 because the protein is unstable under theseconditions and cannot reach the critical threshold level.A major difference was observed in transformed cells: the

duration of G1 was not lengthened by protein synthesisinhibition or serum deprivation (11, 14). Hypothetically, theprotein is stabilized in transformed cells and is able to reachits critical threshold level despite serum deprivation or pro-tein synthesis inhibition.

Recently, we have used a "reverse approach" to study theregulatory events in late G1 (15). Instead of following themolecular events sequentially after stimulation with growthfactors, one works backwards by studying the regulation ofthe early molecular events of S phase (15). Thymidine kinasewas chosen as a model system for the reverse approachbecause its activity and its mRNA increase coordinately withDNA synthesis (16), and these increases are blocked bypartial inhibition of protein synthesis, as is DNA synthesis(17).Thymidine kinase gene expression is regulated by both

transcriptional and posttranscriptional mechanisms (16, 18,19). Deletion analysis performed on the mouse thymidinekinase promoter indicated that a region between 24 and 174bases upstream from the translation start site was essentialfor expression of reporter genes (J.L.F.-K., unpublisheddata). When DNase I footprint and mobility-shift DNAbinding assays with nuclear extracts prepared from nontrans-formed A31 cells were used, a binding activity to a consensussequence located in three binding sites (MT1, MT2, and MT3)of this region was found (21). Designated Yi, the DNAbinding activity was absent in quiescent cells but presentduring late G1 and S phases (21).The data presented here demonstrate that regulation of Yi

binding activity differs between nontransformed and trans-formed cells under three experimental conditions. First,during serum starvation, Yi binding activity was unstable innontransformed cells and stable in transformed cells. Sec-ond, after serum stimulation, the kinetics of induction for Yibinding was different in nontransformed and transformedcells. Third, during protein synthesis inhibition, Yi bindingwas unstable in nontransformed cells and stable in trans-formed cells. Passing a further test for instability in nontrans-formed cells, the induction of Yi binding was blocked bypartial inhibition of protein synthesis. Therefore, Yi bindingactivity satisfies major criteria set forth for the hypotheticalprotein of the restriction point-it is unstable in nontrans-formed cells and stable in transformed cells.

MATERIALS AND METHODSCell Culture and Cell Synchronization. BALB/c 3T3 clone

A31 cells (A31 cells) (22) and benzo[a]pyrene-transformed

tTo whom reprint requests should be addressed.

9310

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 9311

BALB/c 3T3 cells (BPA31 cells) (5) were grown at 370C with10% C02/90% air in Dulbecco's modified Eagle's mediumsupplemented with 4 mM glutamine, penicillin (10 units/ml),streptomycin (10 gg/ml), 10 mM Hepes buffer (pH 7.3), and10% bovine calf serum. Cultures did not contain myco-plasma, as determined by the MycoTect kit (GIBCO). Forsynchronization experiments, cells were plated at 5 x 105cells per 150-cm2 tissue culture dish and were grown for 72 hr.Then cells were rinsed once with Dulbecco's modified Ea-gle's medium and Dulbecco's modified Eagle's medium wasadded with 0.2% bovine calf serum for 96 hr (BPA31 cells) or0.4% bovine calf serum for 60 hr (A31 cells). Under theseconditions, both A31 and BPA31 cells became growth ar-rested as determined by [methyl-3H]thymidine incorporationrates (see below). After growth arrest, serum levels wereincreased to 10% and cells proceeded through the cell cycle.DNA and Protein Synthesis Rate Determination. Cultures

parallel to those in 150-cm2 dishes were grown in 24-wellplates to measure DNA and protein synthesis rates. Aftergrowing cells initially plated at 5 x 103 cells per well, thenumber of cells per cm2 of plating area was indistinguishablefrom the 150-cm2 dishes.DNA synthesis rates were monitored by first adding

[methyl-3H]thymidine (4 ttCi/ml; 1 Ci = 37 GBq) to individ-ual wells and stopping incorporation after 1 hr with 150 mM(final concentration) ascorbic acid. After at least 10 min inascorbic acid, wells were rinsed twice with cold phosphate-buffered saline (40C), rinsed once with MeOH, fixed inMeOH at room temperature for 15 min, and allowed to airdry. After incubation with 0.2 M NaOH at 37°C overnight,wells were processed for scintillation counting.

Protein synthesis rates were measured by adding 10-20,uCi of L-[35S]methionine per ml to individual wells for 1 hr.Samples were precipitated in 10% trichloroacetic acid, sol-ubilized in 0.2 M NaOH, and assayed by scintillation count-ing. Background values were obtained by adding L-[35S]me-thionine and immediately processing the sample and byadding L-[P5S]methionine to wells without cells.

Preparation of Nuclear Extracts and Mobility-Shift DNABinding Assays. Nuclear extracts were prepared as described

(27) by Dounce homogenizing cells while in buffer I (10 mMHepes, pH 7.9/1.5 mM MgCl2/10 mM KCI/0.3 M sucrose/0.1 mM EGTA/0.5 mM dithiothreitol/0.5 mM phenylmeth-ylsulfonyl fluoride), pelleting nuclei after spinning 30 min inan Eppendorf centrifuge (model 5415, top speed), extractingpellets in buffer II [20 mM Hepes, pH 7.9/25% (vol/vol)glycerol/0.42 M NaCl/1.5 mM MgCl2/0.2 mM EDTA/0.1mM EGTA/0.5 mM dithiothreitol/0.5 mM phenylmethylsul-fonyl fluoride], spinning again in an Eppendorf centrifuge,and finally dialyzing the supernatant against buffer III (20mM Hepes, pH 7.9/20% glycerol/100 mM KCl/0.2 mMEDTA/0.5 mM dithiothreitol/0.5 mM phenylmethylsulfonylfluoride).The synthetic MT1 oligonucleotide 5'-TCGAGTCT-

TCGTCCCGCCCCCTTTT-3', which contains the MT1 se-quence of the murine thymidine kinase promoter (21), washybridized to its complement and end-labeled with deoxy-cytidine 5'-[a-32P]triphosphate by using the Klenow fragmentof DNA polymerase to a specific activity of 1-10 x 105cpm/ng. Levels of Yi binding activity in nuclear extractswere measured with the mobility-shift DNA binding assay(20) by adding saturating amounts of end-labeled MT1 oligo-nucleotide (=100,000 cpm; =0.5 ng) to 20 gg of nuclearextract in the presence of poly(dFdC) (0.05 mg/ml) as anonspecific competitor. Samples were electrophoresedthrough 4% polyacrylamide at 170 V. Polyacrylamide gelswere fixed and dried, and XAR-5 Kodak film was exposed at-70°C for 1-5 days.To test the mobility-shift DNA binding assay for linearity,

the MT1 oligonucleotide was added to increasing amounts ofnuclear extracts (up to 25 gg) and Yi binding activity wasmeasured. A linear relationship existed between the intensi-ties of the Yi bands and the amounts of extracts used.

RESULTSYi Binding Activity Is Stabilized in Transformed Cells

During Serum Starvation. Yi binding activity was monitoredin nontransformed A31 cells and transformed BPA31 cellsduring 96 hr of serum starvation. Nuclear extracts were

F . A3IrIIA-BPA3 1

C~~~~~~~~~~~~~~a~~~~~~~~~~~- o ' la la_nC cW )U W) C) W.)U CU CU CUW)

Q O- - c (N "en rs O _ _ N

S~l~liwWWWSW"fmILA....

* t %d %_0 %w * ,

FIG. 1. Stability of Yi binding to theMT1 oligonucleotide during serum star-vation of A31 and BPA31 cells. The mo-bility-shift DNA binding assay was car-ried out on nuclear extracts from expo-nential A31 and BPA31 cells or from A31and BPA31 cells cultured in low concen-

free - trations of serum (0.4% for A31 cells and0.2% for BPA31 cells) for the indicatedDNa )A _ _number of days. No nuclear extract wasadded for the last lane.

Cell Biology: Bradley et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990)

prepared from exponentially growing cells and at half-dayintervals after serum levels were reduced. Yi binding activity,as determined by the mobility-shift DNA binding assay,decreased to undetectable levels 1 day after serum with-drawal for A31 cells (Fig. 1). In contrast, high levels of Yibinding activity were detected throughout a 4-day period ofserum starvation for BPA31 cells (Fig. 1). This disappearanceof Yi binding activity in serum-starved A31 cells and itspersistence in serum-starved BPA31 cells was observed inmultiple experiments.The top two bands in the mobility-shift DNA binding assay

are labeled as SP1 (Fig. 1) because experiments with an SP1consensus site and point mutation suggest that these twobands are due to a murine SPi-like activity (21). These bandsremained relatively constant through all these experiments,serving as an internal control to demonstrate that equalamounts of nuclear extracts were used for each lane. As acontrol for growth arrest, [methyl-3H]thymidine incorpo-ration rates were measured in parallel cultures of cells athalf-day intervals over the same 4-day period. After 2.5 daysof serum starvation, the rates of [methyl-3H]thymidine in-corporation decreased 50-fold for BPA31 cells (to 0.0075 cpmper cell per hr) and decreased 10-fold for A31 cells (to 0.02cpm per cell per hr).

Yi Binding Activity in Transformed Cells After SerumStimulation. A31 and BPA31 cells were synchronized asdescribed in Materials and Methods. Nuclear extracts fromcells were prepared at the indicated times (Fig. 2) after serumlevels were increased, and Yi binding activity was measuredby the mobility-shift DNA binding assay. The detailed ap-pearance of Yi binding activity in A31 cells at late G1 and Sphases has been reported elsewhere (21) and is shown here asa control (Fig. 2). In contrast to A31 cells, the levels of Yibinding activity for BPA31 cells was always detected duringserum starvation and was at higher levels within 4 hr afterserum stimulation (Fig. 2).

A31

L. $U. =.G _0~ QC 0.

BPA3 1 mO-., " &

.

=- ~z .-- \0.0

01- "t 00 ~-

4

r_

SPl

Yi

freeDNA

FIG. 2. Yi binding activity after serum stimulation of growth-arrested BPA31 cells. The mobility-shift DNA binding assay wascarried out on nuclear extracts prepared from synchronized A31 andBPA31 cells before (O hr) and at various times after serum stimula-tion. No nuclear extract was used for the last lane.

The synchronization of the A31 and BPA31 cells wasverified by scintillation counting of incorporated [methyl-3H]thymidine. Results were similar to those reported previ-ously (23); [methyl-3H]thymidine incorporation rates in-creased after 9.5 hr for BPA31 cells and after 11 hr for A31cells.

Yi Binding Activity in Transformed Cells Is Stable DuringProtein Synthesis Inhibition. Cycloheximide (1.0 ,g/ml) wasadded to A31 and BPA31 cells 14 hr after serum stimulation.In confirmation of previous results (10), protein synthesis, asmeasured by L-[35S]methionine incorporation, was inhibitedby -95% (data not shown). It has been shown that cyclo-heximide inhibits protein synthesis in A31 cells and BPA31cells equally and that the inhibition occurs within 10 min afterit is added (11). Yi binding activity in A31 cells was dramat-ically reduced by 4 hr after protein synthesis inhibition (Fig.3). A similar drop-offin Yi binding activity was also observedwhen cycloheximide was applied to A31 cells 12 or 17 hr afterserum stimulation (data not shown). The SP1-like bindingactivity remained constant, demonstrating that the effect ofcycloheximide on Yi binding is not a general phenomenonapplicable to all DNA binding activities.

In BPA31 cells, Yi binding activity remained constant forat least 7 hr after application of cycloheximide (Fig. 3). SinceBPA31 cells progress from Go to S phase 1.5-2 hr faster thanA31 cells (23), cycloheximide was also added to BPA31 cells12 hr after serum stimulation. Identical results were obtained(data not shown).

Partial Inhibition of Protein Synthesis in NontransformedCells Blocks Appearance ofYi Binding Activity. In controls, Yibinding activity increased between 8 and 14 hr after serumstimulation (Fig. 4). Cycloheximide was applied to A31 cells8 hr after serum stimulation, and Yi binding activity wasassayed at 14 hr. The increase in Yi binding activity wasblocked by cycloheximide (0.1 ,g/ml) (Fig. 4). This low levelof cycloheximide inhibited protein synthesis by 60-80% andblocked DNA synthesis by >90%o (data not shown) as re-ported (10). The strong effects of low concentrations ofcycloheximide on the induction of Yi binding were seen inmultiple experiments.

DISCUSSIONPrevious kinetic experiments demonstrated that reducedserum levels and inhibition of protein synthesis preventedDNA synthesis only if applied before a restriction point inlate G1 (10). The labile protein model proposed that highserum levels and high protein synthesis rates enabled acritical quantity of an unstable, regulatory protein to besynthesized at the restriction point (10). Hypothetically, ifconditions (i.e., serum deprivation and protein synthesisinhibition) prevent this regulatory protein from reaching acritical threshold level, DNA synthesis is blocked.Here we have presented three lines of evidence demon-

strating that Yi binding disappears, suggesting that a proteinis unstable in nontransformed cells: (i) lowering serum con-centration levels results in a reduction of Yi binding activityto undetectable levels, (ii) inhibition of protein synthesiscauses a sharp decrease in Yi binding activity, and (iii) partialinhibition of protein synthesis blocks induction of Yi bindingactivity at late G1/S.

Unlike nontransformed cells, growth was not arrested intransformed cells after partial inhibition of protein synthesisor reduction of serum levels (11, 14). Accordingly, weexpected the hypothetical regulatory protein to be stable intransformed cells. Stability was monitored under the sameconditions used for nontransformed cells-serum depriva-tion and protein synthesis inhibition. During serum starva-tion, Yi binding activity was more stable in transformedBPA31 cells (Fig. 1). This could be caused by a longer half-life

9312 Cell Biology: Bradley et al.

k A

0. v .: aidw..:4i i f .'A Orf.4.- 4, kz-4 k.-

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 9313

time I A31 --Iafter U ..4

cycloheximide o - N en t 'nm Nfaddition _ _ O* -- 4

I -BPA31 -

.2e nrA n V

-m s = = =

m £

4. *

freeDNA

FIG. 3. Stability of Yi binding afterinhibition of protein synthesis. The mo-bility-shift DNA binding assay was car-ried out on nuclear extracts from syn-chronized A31 and BPA31 cells 14 hrafter serum stimulation (marked as 0 hr),and for 1-7 hr after the addition of cy-cloheximide at 14 hr. No nuclear extractwas added for the last lane. Because alarge number of samples had to be proc-essed in a short period of time, cell ex-tracts were spun 5 min instead of 30 min.No difference in Yi binding activity wasseen for the different centrifugationtimes.

for the protein complex or by continued synthesis. Also,during strong inhibition of protein synthesis Yi binding ac-tivity was more stable in transformed BPA31 cells (Fig. 3).Two other proteins that are stabilized in transformed cells arep68 (24) and p53 (25). Protein stabilization in transformed

[cycloheximide]at 14 hrs

0 00-0-

itww°SPYI UP

I..iA

s

Yib>

freeDNA-""

FIG. 4. Prevention of Yi binding in A31 cells by partial inhibitionof protein synthesis. The mobility-shift DNA binding assay wasperformed with A31 nuclear extracts prepared before serum stimu-lation (O hr), 8 and 14 hr after serum stimulation, and 14 hr after serumstimulation for cells that had 0.1 ,ug or 1.0 ,g of cycloheximide perml added at 8 hr.

cells could be part of a general mechanism that causes thetransformed phenotype.

Interestingly, the abrupt decrease of Yi binding activity inS-phase A31 cells treated with cycloheximide (Fig. 3) was notobserved for exponentially growing cycling A31 cells (datanot shown). This may reflect a difference in the regulation ofS phases of constantly cycling cells as opposed to cellsimmediately after serum stimulation. The sharpness of thedrop-off is most likely due to the type of molecular interac-tions. For example, reduction in levels of one overabundantcomponent of a cooperative complex could result in thesekinetics.

Yi binding activity is present in serum-starved BPA31cells, in which both thymidine kinase activity and DNAsynthesis rates are at low levels. This observation does notprove that the function of Yi binding activity is unrelated tothymidine kinase activity and/orDNA synthesis. As outlinedin the Introduction, there are several levels of regulation forDNA synthesis, as there are for thymidine kinase. Sincetransformed BPA31 cells have an increase of thymidinekinase activity and DNA synthesis rates at S phase, some ofthese levels of control must persist after transformation. Theregulation ofgene expression for thymidine kinase in A31 andBPA31 cells can be used as a model system to determinewhich levels of control remain intact and which levels arederegulated in transformed cells.

In addition to the studies on stability of Yi binding activity,we also have shown that the kinetics of Yi appearance afterserum stimulation vary between transformed cells and non-transformed cells. Similar differences between transformedand nontransformed cells in binding activities to the histonepromoter have been observed (26). The proteins causing Yibinding activity may regulate a set ofgenes that participate inS phase. Since the consensus sequence to which Yi binds ispresent in other S-phase promoters (D.W.B., unpublisheddata), the regulation of S-phase genes might be coordinatedby this group of proteins.The Yi binding activity has a major property expected for

the hypothetical regulatory protein(s) of late G1-stabiliza-

,-a

x

0

L"'.

Cell Biology: Bradley et al.

.f

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990)

tion in transformed cells. Activation of the Yi complex couldbe among the final of a series of events that regulate theinitiation of DNA synthesis in nontransformed cells. Thealtered regulation of Yi binding activity in BPA31 cells mightbe part of the mechanism that makes these transformed cellsless dependent on growth factors.

We thank Dr. Anne Crozat for critically reading this manuscript.This work was supported by National Institutes of Health GrantGM24571 (to A.B.P.) and the Richard B. Carter Fellowship (toD.W.B.).

1. Stiles, C. D. (1983) Cell 33, 653-655.2. Pardee, A. B. (1989) Science 246, 603-608.3. Pardee, A. B., Coppock, D. L. & Yang, H. C. (1986) J. Cell

Sci. Suppl. 4, 171-180.4. Laskey, R. A., Fairman, M. P. & Blow, J. J. (1989) Science

246, 609-614.5. Holley, R. W., Baldwin, J. H., Kiernan, J. A. & Messmer,

T. 0. (1976) Proc. Natl. Acad. Sci. USA 73, 3229-3232.6. Cherington, P. V., Smith, B. L. & Pardee, A. B. (1979) Proc.

Natl. Acad. Sci. USA 76, 3937-3941.7. Nishizuka, Y. (1984) Nature (London) 308, 693-698.8. Campisi, J., Gray, H. E., Pardee, A. B., Dean, M. & Sonen-

shein, G. E. (1984) Cell 36, 241-247.9. Nakayasu, H. & Berezney, R. (1989) J. Cell Biol. 106, 1-11.

10. Rossow, P. W., Riddle, V. G. H. & Pardee, A. B. (1979) Proc.Natl. Acad. Sci. USA 76, 4446-4450.

11. Campisi, J., Medrano, E. E., Morreo, G. & Pardee, A. B.(1982> Proc. Natl. Acad. Sci. USA 79, 436-440.

12. Campisi, J., Morreo, G. & Pardee, A. B. (1984) Exp. Cell Res.152, 459-466.

13. Pardee, A. B., Medrano, E. E. & Rossow, P. W. (1981) in The

Biology of Normal Human Growth, eds. Ritzdn, M., Aperia,A., Hall, K., Larsson, A., Zetterberg, A. & Zetterstrom, R.(Raven, New York), pp. 59-70.

14. Medrano, E. E. & Pardee, A. B. (1980) Proc. Natl. Acad. Sci.USA 77, 4123-4126.

15. Knight, G. B., Gudas, J. M. & Pardee, A. B. (1988) in GeneExpression and Regulation: The Legacy ofLuigi Gorini, eds.Bissell, M., Deho, G., Sironi, G. & Torriani, A. (Elsevier,Amsterdam), pp. 325-332.

16. Knight, G. B., Gudas, J. M. & Pardee, A. B. (1989) Jpn. J.Cancer Res. 80, 493-498.

17. Coppock, D. L. & Pardee, A. B. (1985) J. Cell. Phys. 124,269-274.

18. Coppock, D. L. & Pardee, A. B. (1987) Mol. Cell. Biol. 7,2925-2932.

19. Lieberman, H. B., Lin, P. F., Yeh, D. B. & Ruddle, F. H.(1988) Mol. Cell. Biol. 8, 5280-5291.

20. Fried, M. & Crothers, D. M. (1981) Nucleic Acids Res. 9,6505-6525.

21. Dou, Q.-P., Fridovich-Keil, J. L. & Pardee, A. B. (1991) Proc.Natl. Acad. Sci. USA, in press.

22. Todaro, G. J. & Green, H. (1963) J. Cell Biol. 17, 299-313.23. Riddle, V. G. H. & Lehtomaki, D. M. (1981) Cancer Res. 41,

1778-1783.24. Croy, R. G. & Pardee, A. B. (1983) Proc. NatI. Acad. Sci. USA

80, 4699-4703.25. Halevy, O., Hall, A. & Oren, M. (1989) Mol. Cell. Biol. 9,

3385-3392.26. Holthuis, J., Owen, T. A., Van Wijnen, A. J., Wright, K. L.,

Ramsey-Ewing, A., Kennedy, M. B., Carter, R., Cosenza,S. C., Soprano, K. J., Lian, J. B., Stein, J. L. & Stein, G. S.(1990) Science 247, 1454-1457.

27. Dignam, J. D., Lebowitz, R. M. & Roeder, R. G. (1983) Nu-cleic Acids Res. 11, 1475-1489.

9314 Cell Biology: Bradley et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0