Topoisomerase IV Can Support oriC DNA Replication in Vitro*

THE JOURNAL OF BIOLOGICAL. CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269, No. 23, Issue of June 10, pp. 16371-16375, 1994 Printed in U.S.A.

Topoisomerase IV Can Support oriC DNA Replication in Vitro* (Received forpublication, March 1, 1994)

Hiroshi Hiasa and Kenneth J. Marians From the Molecular Biology Program, Memorial Sloan-Kettering Cancer Center, New York, New York 10021

Escherichia coli has two typeI1 topoisomerases, DNA for topologically separating the daughter chromosomes at the gyrase and topoisomerase IV (Topo IV). Top0 IV is re- terminal stages of DNA replication in E. coli. quired for the decatenation of the linked daughter chroThus, the primary responsibilities of gyrase seem to be the DNA replication, maintenance of a critical superhelical density requiredfor inimosomesattheterminalstagesof whereas gyrase, because its of ability to convert to negatiation and theremoval of the positive supercoils generated by tive supercoils the positive supercoils generated by rep- replication fork progression. BecauseTop0 IV can relaxpositive lication fork progression in a circular chromosome, is supercoils (17), it should be capable of substituting for gyrase required to support nascent chain elongation. Using an during nascent chain elongation. We have investigated this oriC DNA replication systemin vitro, we show that Top0 question using oriC plasmidDNA replication reconstituted in IV,which can relax positive supercoils, can also support vitro with purified proteins and show here that thisis, in fact, replication fork progression. This activity is only obthe case. IV to temserved at substoichiometric ratios of Top0 Top0 IV-supported oriC DNA replication is nearlyas efficient plate, at higher ratios, the template becomes relaxed Top0 IV as that supportedby gyrase, although substoichiometric levels and initiation of DNA replication cannot occur. of Top0 IV are required to avoid relaxation of the template was capable of supporting bidirectional DNAreplication from oriC, although, unlike the case with gyrase, some before initiation can occur. In the presence ofTop0IV, most templates replicate bidirectionally, although some appear to templates apparently replicated unidirectionally. This suggests that either gyrase itself or a certainminimum replicate unidirectionally, suggesting that either gyrase itself superhelical densityis required for proper initiation ofor a particular superhelical density is requiredfor correct initiation at 0%. DNA replication from oriC. MATERIALS AND METHODS Replication Proteins and DNAs-E. coli replication proteins were as The action of topoisomerases is likely to be required for all described previously (18-20). Thepreparation of Top0 IV was described stages of the replication of the circular chromosome of Esch- by Peng and Marians(21j. Two types of minichromosomes, pBROTB535 erichia coli. Initiation of replication at oriC requiressuperhelic- types I and I1 (6 kb) (22)’ were prepared according to Marians et al. (23). ity (1).Progression of replication forks around the circular These are plasmid templates carryingthe AatII-EcoRV oriC fragment chromosome generates positivesupercoils that must be re- (nucleotides567-1402) of pCM959 (24) inserted in either orientation moved (2,3).And, the replicating daughterchromosomes must into the PuuII siteof pBR322 DNA. In addition,the plasmids carry two be topologically disengaged at the terminal stagesof DNA rep- TerB sequences (25)that are oriented to exclude the passage of replication forks between them and that are separated by 1 kb of DNA lication (2, 3). inserted into BamHI- and HindIII-digestedpBR322. The type I and I1 E. coli has four topoisomerases. Of these, DNA gyrase and plasmids differ by the orientation of the oriC fragment. topoisomerase IV (Topo IV)’ are clearly involved in supporting oriC DNA Replication-StandardoriC DNA replicationreactions DNA replication. Mutations in the genes encoding either DNA were as described previously (22, 26). Gyrase andTop0 IV were added gyrase (gyrA and gyrB) or Top0 IV (parC andparE) are con- to the reaction mixturesas indicated in the figure legends. Replication ditionally lethal (4-7). parC and parE mutations result in a productswereanalyzed by agarosegelelectrophoresisaccordingto Hiasa and Marians (22). Any changes in conditionsare indicated inthe terminal partitioning defect where the daughternucleoids re- figure legends. main unseparated at the midpoint of the cell, leading t o the Pulse-chase Analysisof Replication Products-Pulse-chase analysis production of anucleate cells and cells with multiploidy. Gyrase for oriC DNA replication was performed as described previously (26). pl) were mutations generally display an immediate-stop DNA replica- Briefly,reaction mixtureslackinganytopoisomerase(37.5 was added followedby an incubated at 30 “C. After2 min, [Q-~~PI~ATP tion phenotype (8, 91, although some mutant alleles display additional 1min of incubation. This radioactive pulse wasthen chased either an initiation(10, 11)or par (12, 13) defect. by the addition of 4 mM unlabeled dATP. Top0 IV and gyrase were added Our biochemical studies demonstrated that only Top0 IV,and at the same time as the unlabeled dATP. Aliquots (5 pl) were removed not gyrase, could decatenate replicating oriC plasmid daughter at the indicated times after the chase, and the DNA products were DNA molecules in vitro (14), even though gyrase could unlink analyzed by 0.7%alkaline agarose gel electrophoresis (22). these multiply linked DNA molecules in isolated reactions(15). RESULTS This, taken together with the phenotype of parC and parE mutations and the observation ofAdams et al. (16) that Top0 IV, Top0 N Can Support oriC DNA Replication-Plasmid DNAs but not gyrase, unlinked plasmid catenanes arising from DNA carrying oriC are efficient templates for DnaA- and oriC-dereplication inuiuo, argues thatTop0 IV is primarilyresponsible pendent DNA replication(1).Reaction mixtures contain, in addition to the template and DnaA, DnaB, DnaC, DnaG, HU, *These studies weresupported by National Institutes of Health the single-strandedDNA-binding protein, theDNA polymerase Grant GM 34558. The costsof publication of this article were defrayed I11 holoenzyme, and DNAgyrase(1). Completed daughter DNA be molecules are produced if the reaction mixture also contains in part by the payment of page charges. This article must therefore hereby marked “aduertisement” in accordance with 18 U.S.C. Section DNA ligase, DNApolymerase I, and RNaseH (1).The template 1734 solely toindicate this fact. The abbreviations used are: Topo, topoisomerase; 1I:II di, form11: form I1 DNA dimers; nt, nucleotide; kb, kilobase(s). H. Hiasa and K. J. Marians, submittedfor publication.

’

16371

Topoisomerase N-dependent oriC DNA Replication

16372

form 11-form I1 DNA dimers (1I:II di) (Fig. 2B). This is in contrast t o the products formed in the presence of only DNA gyrase. Because gyrase cannot decatenate the linked daughter of replication (14, DNAmolecules formed at the terminal stages 261, the products accumulate as highly linked 1I:II di and late replicative intermediate (LRI) (26). Denaturing gel electrophoresis demonstrated thatboth Top0 IV and gyrase supported leading and lagging strand DNA synthesis. This was revealed by the presence of two distinctly sized populations of DNA, one that centered about 0.4 kb in length and represented theOkazaki fragments, andone significantly larger population that consisted of the leading strands (Fig. 2C). The leading strand populations produced in the two replication systems, however, were distinctive. Leading strands made during gyrase-stimulated replication as observed previously centered about 3 kb, or half unit-length, 0 for bidirectionally replicating oriC templates(22). A significant 0 100 200 fraction of the leading strands made during Top0 IV-stimulated Top0 (fmol) replication tended to be closer to unit length, suggesting that some percentage of these templates were replicating unidirecFIG.1. Both DNA gyrase and Top0IV support oriC DNA replitionally. This is also consistent with the appearanceof multication. Standard oriC DNA replication reactions contained the indigenome length DNA products in the Top0 IV system. We have cated amountsof either gyrase or Top0 N.Incorporation of [(U-~~PI~AMP intoacid-insolubleproductwasmeasured.Total DNA synthesis is shown previously that thisDNA product, which is produced via given. Reaction mixtures contained35 fmol of template DNA(420pmol a rolling circle mechanism, occurs only on oriC templates that as nt). 0, DNA gyrase; 0, Top0 n! replicateunidirectionally (22). Longer incubations with the Top0 IV system showed proportionally greater amountsof rollmust be supercoiled for efficient initiation (1,27). DNA gyrase ing circle product (data not shown). acts to allow replication fork progression and, by maintaining Because Top0 IV has no supercoiling activity, the expected negative superhelicity in the plasmid population, to ensure completed monomer product in theTop0 IV-stimulated replicathat initiation occurs on the largest fraction of template mol- tion system was form 1', not form I.This was assessed by ecules possible. examining the mobility on neutral agarose gels run in either The supercoiling activity of gyrase perse is not required for the presence or absence of ethidium bromide of product DNA replicationforkprogression in a circulartemplate.In eu- from replication reactions thatincluded both DNA polymerase karyotes, this is supported by topoisomerases that can relax I, RNase H, and DNA ligase to allow sealing of the nascent positive supercoils but that cannot introduce supercoils (28). DNA (Fig. 3). In the absence of ligase, the monomer product presence of The discovery of Top0 IV (7, 17) revealed the existence of an was form I1 because its mobility did not shift in the enzyme in E. coli other than gyrase that could operate on posi- the ethidium bromide. On the other hand, when ligase was tive supercoils (171, although, like its eukaryotic counterparts, present, thepresence of ethidium bromide in thegel caused the Top0 IVcannot supercoil. We, therefore, asked whetherTop0 IV monomer product t o have significantly greater mobility, indiform I' DNA. could support replication fork progression in circular tem- cating that it was in fact, Both DNA Gyrase and Top0 N Support the Elongation of plates. The dependence on either Top0 IV or DNA gyrase of nucle- Nascent Chains from Preformed Early Replicative datapresentedinthe previoussection otide incorporation in oriC DNA replication reactions was as- Intermediate-The sessed (Fig.1).Gyrase stimulated DNA synthesis maximally by clearly showed that Top0 IV could support replication fork pro6-fold at a ratio to template of 3:l. Top0 IV was also able to gression in a circular template. The initial rates of DNA synstimulate DNA synthesis, although the extent of stimulation thesis of the Top0 IV- and gyrase-stimulated reactions were was slightly lessthan for gyrase. Unlike gyrase, stimulationby similar. However, because the initial rate reflects both initiaa good indicator elongation, this was not Top0 IV could be observed in only a very narrow range of en- tion and nascent chain zyme t o template ratios. Maximal stimulation occurred at a of the relative efficiencies of the two enzymes in supporting ratio of 1:l.When this ratio was exceeded, a dramatic inhibi- only replication fork progression. In order to address this question ofDNA synthesis occurred. Presumably, inhibition oc- tion, we compared the ability of the two enzymes to support the curred because excess Top0 IV led torelaxation of the template elongation of nascent chains in a preformed early replicative intermediate (Fig. 4). before initiation could occur. oriC DNA replication reactions were assembled in the abThe initial time course ofTop0 IV- and gyrase-stimulated DNA replication was examined (Fig. 2). Both reactions showed sence of any topoisomerase and incubated for 2 min at 30 "C. A T Pthen added, and DNA products were pulsea slight initial lag followed by a linear phase (Fig. 2 A ) . The [ ( U - ~ ~ P I ~was initial rateof Top0IV-stimulated DNAreplicationwas 67% that labeled for 1 min. Either DNA gyrase or Top0 IV was then added along with a chase of unlabeled dATP, and the reaction of gyrase-stimulated DNA replication. Examination of the DNA products by native gel electrophore- followed kinetically. The DNA formed during the pulse wasbetween 300-600 nt sis revealed the expected patterns (Fig. 2B). In the absence of absence of any addedtopoisomerase, this DNA any topoisomerase, only early replicative intermediates (ERI) in length. In the accumulated. These are molecules on which initiation has oc- was notelongated during thechase. The addition of either Top0 curred, but no extensive nascent chain elongation has ensued. IV or gyrase during the chase clearly allowed nascent chain Alkaline agarose gel electrophoresis showed the majority of elongation, as evinced by the increase in length of the early these nascent strands tobe 300-600 nt in length(Fig. 2C). In replication products (Fig. 4). Based on the size of the leading the presence of only Top0IV, the majority of the DNA was strand in each case after 1 min of chase, gyrase-supported present as resolved monomer product, form I1 DNA, and some nascent chainelongation appeared tobe about 20% faster than 150

Topoisomerase N-dependent oriC DNA Replication

16373

’ ” “ 1

A-

-c

c

0

E,

v

.-

v) v)

Q,

5 c

z1 v)

a z

0

I

-I

n ”.

time (min)

FIG. 2. Initial time course of Top0 IV- and gyrase-supported replication reactions. Standard oriC replication reactions were increased in size by 3-fold and containedeither no topoisomerase (A), Top0 IV (35 fmo1/35 fmol template) (0) or.gyrase(115 fmo1/35 fmol template) (0).Aliquots were withdrawnat the indicated times and used to either measure DNA synthesis (panelA ) or to analyze the nascent DNA products by either neutral agarose gel electrophoresis (panel B ) or by denaturing alkaline gel electrophoresis (panel C). Gel electrophoreticanalyses were as described by Hiasa and Marians (22). Size markers were HindIII-digested bacteriophage A DNA. RC, rolling circle product; LRI, late replicative intermediate; ERI, early replicative intermediate; 1121 di,form 11-form I1 DNA dimers.

12

8

4

0

B. none

Top0

time (min)

0

2

4

6

8

1

I

0 0 1

Gyrase

Top0 IV 2

4

6

8

1

0 0

..””

2

__

4 .

.

6

8

1

0

~”

RC -

LRI -

1

-

ERI -

form II

11:lI di

C. none

Top0

time (min)

0

2

4

6

Top0 IV 6

1

0

0 2 4 6 8 1 0

Gyrase 0

2

4

6

8

1

0

kb

-9.6 -6.4 -4.4 - 2.0

-0.6

Top0 IV-supported nascent chain elongation. In addition, gyrase supported chain elongation more efficientlythan ‘Ibpo IV. The percentage of label that could be chased into larger products was 42 and 33% for the gyrase- and Top0 IV-supported reactions, respectively. Interestingly, in both cases, the leading

strands grew to nearly unit length. This suggests that proper bidirectional replication fork progression wasdisrupted by this staged incubation procedure. Topoisomerase N-supported DNA Replication from oric IS Bidirectional-Some of the data described thus far suggested

oriC DNA Replication

Topoisomerase IV-dependent

16374

that a significant fraction of templates in the Top0 IV-supported system were replicating unidirectionally. To examine this question directly, we used a template designed to reveal, by simple inspection of the leading strandproducts, the directionality of DNA replication. This template (22) consists of pBR322 DNA carrying oriC and, roughly 180 “C away from oriC, two TerB sequences (25) separated by 1 kb and oriented to exclude the passage of replication forks between them. In the presence of Tus, the replication fork arrest protein (29, 301, bidirectional replication generates two distinctly sized leading strands,one 2 kb in length that is synthesized by the leftward-moving fork, and one 3 kb in length that issynthesized by the rightwardmoving fork. Replication products generated in thepresence of Tus in the Top0 IV- and gyrase-stimulated replication reactions,as well as in a reaction where both topoisomerases were present, were fractionated by alkaline-agaroseelectrophoresis (Fig. 5). In each case, two distinct leading strandsof 2 and 3 kb in length could be observed, indicating that, under all conditions examined, replicationwas bidirectional from oriC. However, the balance of rightward and leftward initiation appeared to be affected by Top0 IV. In thepresence of gyrase alone, the molar ratio (determined by analysis with a Fuji BAS 1000 PhosphorImager) of the 2EtBr

ligase

-

and 3-kb fragments was 1:1, indicating equivalentinitiation. In the presence of Top0 IV alone, this ratiowas 1:1.6, indicating a significant bias for rightward initiation under these conditions. In the presence of both topoisomerases, the ratio was N . 4 . Thus, theobserved unit lengthleading strands produced in the presence of Top0 IV and in the absence of Tus are likely the consequence of this bias for rightward initiation. The righthand replication fork presumably executes acomplete circuit of the template before the left-hand fork is either formed or released from the origin. DISCUSSION

Replication fork progression in a circular template thatreplicates as a theta structure requiresa topoisomerase to remove the positive supercoils generated. In E. coli, two of the four topoisomerases could possibly accomplish this task, DNA gyrase, which could convert the positive supercoils directly to negative ones, and Top0 IV, which could relax them (17). GeTop0 IV

- + +

Gyrase

+ - +

&

-4.4

+

- +

-

2.3

1

- 2.0

- + -

.

c

d

- 0.6

1 2 3 4 FIG.3. The final product of Top0 N-supported oriC DNA replication is form 1’. Standard oriC replication reactions containing Top0 IV (35 frnoV35 fmol template), either in the presence (lanes 2 and 4 ) or absence (lanes I and 3 ) of DNA polymerase I, RNase H, and DNA

1 2 3 FIG.5. Top0 N-supported oriC DNA replication is bidirectional. DNA products generated in standard oriC replication reactions in the presence of Tus (TusDNA = 20:l) and either in the presenceof ligase were analyzed by neutral agarosegel electrophoresis either in thegyrase (115 fmo1/35 fmol template) alone (lane 1), Top0 IV (35 fmo1/35 presence (lanes 3 and 4 ) or absence (lanes 1 and 2 ) of 0.5 pg/ml ethidium fmol template) alone (lane Z ) , or both topoisomerases (lane 3 ) were bromide (Et&) in both the gel and theelectrophoresis buffer. Total DNA analyzed by denaturing alkaline-agarosegel electrophoresis. Total DNA synthesis in the reactions shown was 93 pmol (lanes I and 3 ) and 106 synthesis for the reactions shown in lanes I 3 was 102, 46, and 101 pmol (lanes 2 and 4 ) . pmol, respectively. Size markers were as in the legend Fig. to 2.

Top0

time (min)

none 0 1 2.5 5 10 15

Top0 IV 0 1 2.5 5 10 15

Gyrase 0

1 2.5

5 10 15

Q ! FIG.4. Both Top0 N and DNA gyrase support elongation of nascent chains from early replication intermediates. This pulse-chase analysis was as described under “Materials and Methods.” The times indicated are post-chase. Top0 IV (35 fmo1/35 fmol template) orgyrase (115 fmo1/35fmol template)were added at the start of the chase. DNA productswereanalyzed by electrophoresis through 0.7% denaturing alkaline agarose gels. Size markers wereas in the legend to Fig. 2.

- 9.6

- 6.4 - 4.4 - 2.0 - 0.6

Topoisomerase N-dependent oriC DNA Replication netic data (8, 9) indicate that it is DNA gyrase that accomplishes this in thecell. We have shown here that, in vitro,Top0 IV is nearly as capable as gyrase at this task. In our initial studies on Top0 IV (141, we noted that Top0 IV could support oriC replication in vitro only poorly.At the time, we failed to appreciate the narrow ratio of Top0 IV to template at which DNA synthesis could be supported. Top0 IV levels in a excess of a stoichiometric ratio to template result in profound inhibition ofDNA replication. It is likely that this inhibition results from a relaxation of the template, thereby preventing the initiationof DNA replication, which is dependent on supercoiling (1, 27). Top0 IV seemed nearly as efficient as gyrase in stimulating DNA synthesis, supporting replication fork movement at 80% the rate ofDNA gyrase. Because the optimal ratio of topoisomerase to DNA was %fold higher for gyrase than Top0 IV, this suggests that the turnover number of 'Ibpo IV may be significantly higher than for gyrase. Distinct differences could be observed between the Top0 IVand gyrase-stimulated reactions. Because Top0 IV, but not gyrase, could decatenate the replicating daughter DNA molecules, the DNA products were different, with Top0 IV producing completely resolved monomer forms and gyrase multiply linked dimer forms. Several observations suggested a role forDNA gyrase during the early stages of replication that was distinct from that of Top0 IV. Gyrase was significantly more efficient at allowing templates initiated in the absence of any topoisomerase to complete replication and the extent of rightward and leftward initiation, which is equivalent in the presence of gyrase, was skewed significantly toward the right-hand side in thepresence of Top0 IV. Initiation at oriC appears to occur near or to the left of the A+T-rich 13-mer at the left-hand border of oriC.2 Thus, the right-hand fork must eitherdisplace the DnaA-oriC complexor wait for it to dissociate before it can leave the oriC region. The left-hand fork does not face a similar problem. However, our observations suggest that in thepresence of Top0 IV alone, the left-hand fork takes so long to release from oriC that therighthand fork makes it all the way around the template first. This does not occur in thepresence of DNA gyrase alone. Therefore, either gyrase itself is required for the timely release of the two replication forks from the origin or a certain critical level of supercoiling is required. The data presented here demonstrate that Top0IV could substitute for DNA gyrase in vivoin supporting replication fork movement. Yet, apparently it does not do so. Most mutant gyr

16375

alleles display an immediate-stop DNA synthesis phenotype (8, 9). We have determined3 that there areabout 400 molecules of Top0 IV in thecell, roughly equivalent to the amount of gyrase (31). If this Top0 IV were freely available to substitute for gyrase, one might expect a delayed stop phenotypefor the gyrase mutants, reflecting the fact that Top0 IV could take over in supporting replication fork progression, but could not generate thesuperhelicity required for the next round of initiation. This may implythat Top0 IV is sequestered in some manner in the cell. Acknowledgment-We thank David Valentine for the artwork. REFERENCES 1. Funnell, B. E., Baker, T. A., andKornberg, A. (1986) J. Biol. Chem. 261, 5616-5624 2. Gellert, M. (1981)Annu. Reu. Biochem. 50,879-910 3. Wang, J. C. (1985) Annu. Rev. Biochem. 5 4 , 6 6 5 4 9 8 4. Gellert, M., O'Dea, M.H., Itoh, T., and 'Ibmizawa,J.-I.(1976)Proc. Natl. Acad. Sci. U. S. A. 73, 44744478 5. Gellert, M., Mizuuchi, K., O'Dea, M. H., Itoh, T., and Tomizawa, J . 4 . (1977) Proc. Natl. Acad. Sci. U.S. A. 74,4772-4776 6. Sugino, A,, Peebles, C. L., Kreuzer, K. N., and Cozzarelli, N. R. (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 4767-4771 7. Kato, J., Nishimura, R., Niki, H., Hiraga, S., and Suzuki, H. (1990) Cell 64, 393-404 8. Filutowicz, M., and Jonczk, P. (1982) Mol. Gen. Genet. 191, 282-287 9. Keuzer, K. N., and Cozzarelli, N. R. (1979) J. Bacteriol. 140, 424435 10. Fairweather,N. E , Orr, E., and Holland, I. B. (1980)J. Bacteriol. 142,153-161 11. Filutowicz, M. (1980) Mol. Gen. Genet. 177, 301-309 12. Steck, T. R. and Drilica, K. (1984) Cell 36, 1081-1088 13. Steck, T. R.,Pruss, G. J., Manes, S. H., Burg, L., and Drilica, K. (1984) J. Bacteriol. 158, 3 9 7 4 0 3 14. Peng, H., and Marians, K. J. (1993) Proc. Natl. Acad. Sci. U.S. A. 90, 85718575 15. Marians, K. J. (1987) J. B i d . Chem. 262,10362-10368 16. Adams, D. E., Shekhtman, E. M., Zechiedrich, E. L., Schmid., M. B., and Cozzarelli, N. R., (1992) Cell 71, 277-288 17. Kato, J., Suzuki, H., and Ikeda, H. (1992) J. Biol. Chem. 267,25676-25684 18. Minden, J. S., and Marians, K. J. (1985) J. Biol. Chem. 260,9316-9325 19. Parada, C. A,, and Marians, K. J. (1991) J . Biol. Chem. 266, 18895-18906 20. Wu,C. A,, Zechner, E. L., and Marians, K. J. (1992) J. Biol. Chem. 267, 40304044 21. Peng, H. and Marians, K. J. (1993) J. Biol. Chem. 268, 24481-24490 22. Hiasa, H. and Marians, K. J. (1994) J. Biol. Chem. 269,6058-6063 23. Marians, K. J., Soeller, W., and Zipursky, S. L. (1982) J . Biol. Chem. 257, 5656-5662 24. Bohk, J.-J.,and Messer, W. (1983) Gene (Amst.)24, 265-279 25. Hill, T. M., Pelletier, A. J.,Tecklenburg, M., and Kuempel, P. L. (1988) Cell 55, 459466 26. Hiasa, H., DiGate, R. J.,and Marians, K. J. (1994) J . Biol. Chem. 269, 20932099 27. Baker, T. A. and Kornberg, A. (1988) Cell 55, 113-123 28. Wang, J. C. (1991) J. Biol. Chem. 266,6659-6662 29. Pelletier, A. J., Hill, T. M., and Kuempel, P. L. (1989) J . Bacteriol. 171, 17391741 30. Hidaka, M., Akiyama, M., and Horiuchi, T. (1988) Cell 55,467-475 31. Staudenbauer, W. L., and Orr, E. (1981) Nucleic Acids Res. 9, 358S3603

H. Peng, and K. J. Marians, unpublished data.