References and Notes

1. W. 0. Roberts and R. H. Olson, Rev.Geophys. Space Phys. 11, 731 (1973).

2. W. N. Hess, The Radiation Belt and Magneto-sphere (Blaisdell, Waltham, Mass., 1968); J. G.Roederer, Ed., Physics and Chemistry inSpace [Springer-Verlag, New York, 1970 (vols.1 and 2), 1972 (vol. 4), and 1973 (vols. 6 and7)]; B. M. McCormac, Ed., Earth's Magneto-spheric Processes (Reidel, Dordrecht, Nether-lands, 1972); E. R. Dyer and J. G. Roederer,Ed., The Magnetosphere (Reidel, Dordrecht,Netherlands, 1972).

3. International Magnetospheric Study-Guide-line for United States Participation (NationalAcademy of Sciences, Washington, D.C.,1973).

4. Scientific Uses of the Space Shuttle (NationalAcademy of Sciences, Washington, D.C., inpress).

5. A. J. Hundhausen, Coronal Expansion andSolar Wind (Springer-Verlag, New York,1972); C. P. Sonett, P. J. Coleman, Jr., J. M.Wilcox, Eds., Solar Wind (NASA SP-308,National Aeronautics and Space Administra-tion, Washington, D.C., 1972)

6. Afte? W. J. Heikkila, in Critical Problemsof Magnetospheric Research, E. R. Dyer, Ed.[IUCSTP (Interunion Commission of Solar-Terrestrial Physics) Secretariat, NationalAcademy of Sciences, Washington, D.C.,1972], p. 67.

7. For up-to-date reviews of magnetosphericconfiguration and current research topics, seeE. R. Dyer, Ed., Critical Problems of Mag-netospheric Research (IUCSTP Secretariat,National Academy of Sciences, Washington,D.C., 1972).

8. See review by J. G. Roederer [Rev. Geophys.Space Phys. 10, 599 (1972)].

9. See review by M. Sugiura, in Critical Prob-lems of Magnetospheric Research, E. R. Dyer,Ed. (IUCSTP Secretariat, National Academyof Sciences, Washington, D.C., 1972), p. 195.

10. See review by G. Morfill and M. Scholer(Space Sci. Rev., in press).

11. See review by L. A. Frank, in Critical Prob-lems of Magnetospheric Research, E. R. Dyer,Ed. (IUCSTP Secretariat, National Academyof Sciences, Washington, D.C., 1972), p. 53.

12. R. H. Eather, Rev. Geophys. Space Phys. 11,155 (1973).

13. E. W. Hones, Jr., J. R. Asbridge, S. J. Bame,M. D. Montgomery, S. Singer, S.-I. Akasofu,J. Geophys. Res. 77, 5503 (1972).

14. See reviews by J. P. Heppner [in CriticalProblems of Magnetospheric Research, E. R.Dyer, Ed. (IUCSTP Secretariat, NationalAcademy of Sciences, Washington, D.C., 1972),p. 107] and by D. A. Gurnett (in ibid., p. 123).

15. See, for example: S. E. DeForest and C. E.Mcllwain, J. Geophys. Res. 76, 3587 (1971);J. G. Roederer and E. W. Hones, Jr., ibid.75, 3923 (1970).

16. See, for example, D. L. Carpenter, K. Stone,J. C. Siren, T. L. Crystal, ibid. 77, 2819 (1972).

17. See popular review by G. Haerendel and R.Lust [Sci. Amer. 219, 80 (Nov. 1968)].

18. F. S. Mozer, Pure Appi. Geophys. 84, 32(1971).

19. V. S. Bassolo, S. M. Mansurov, V. P. Sha-bansky, Issled. Geomagn. Aeron. Fiz. Solntsa23, 125 (1972); E. Friis-Christensen, K. Lassen,J. Wilhjelm, J. M. Wilcox, W. Gonzalez,D. S. Colburn, J. Geophys. Res. 77, 3371(1972).

20. L. Svalgaard, Bull. Amer. Astron. Soc. 4, 393(1972).

21. F. S. Mozer and W. D. Gonzalez, J. Geophys.Res. 78, 6784 (1973).

22. D. P. Stern, ibid., p. 7292.23. V. M. Vasyliunas, in Earth's Magnetospheric

Processes, B. M. McCormac, Ed. (Reidel,Dordrecht, Netherlands, 1972), p. 60.

24. For instance, P. M. Banks and T. E. Holzer,J. Geophys. Res. 74, 6317 (1969).

25. See review by C. R. Chappell [Rev. Geophys.Space Phys. 10, 951 (1972)].

26. See review by D. L. Carpenter and C. 0.Park [ibid. 11, 133 (1973)].

27. D. L. Carpenter, J. Geophys. Res. 71, 693(1966).

28. E. W. Hones, Jr., Rev. Geophys. Space Phys.,in press.

29. H. Alfven and C.-G. Falthammar, CosmicalElectrodynamics (Oxford Univ. Press, FairLawn, N.J., ed. 2, 1963).

30. J. A. Jacobs, Geomagnetic Micropulsations(Springer-Verlag, New York, 1970).

31. See review by C. T. RusseUl and R. L.McPherron (Space Sc. Rev., in press).

32. V. M. Vasyliunas and R. A. Wolf, Rev.Geophys. Space Phys. 11, 181 (1973).

33. S.-I. Akasofu,- Polar and Magnetospheric Sub-storms (Reidel, Dordrecht, Netherlands, 1968);G. Rostoker, Rev. Geophys. Space Phys. 10,157 (1972).

34. See, for instance, T. Yeh and W. I. Axford,1. Plasma Phys. 4, 207 (1970).

35. J. G. Roederer, Dynamics of GeomagneticallyTrapped Radiation (Springer-Verlag, NewYork, 1970); J. B. Cladis, G. T. Davidson,L. L. Newkirk, Eds., The Trapped RadiationHandbook (U.S. Defense Nuclear Agency pub-lication DNA-2524H, Washington, D.C., 1971).

36. J. I. Vette, A. B. Lucero, J. A. Wright,Inner and Outer Zone Electrons (NASA SP-3024, National Aeronautics and Space Admin-istration, Washington, D.C., 1966).

37. T. G. Northrop, The Adiabatic Motion ofCharged Particles (Wiley-Interscience, NewYork, 1963).

38. H. H. Heckmann and P. J. Lindstrom, J.Geophys. Res. 77, 740 (1972); M. Schulz andG. A. Paulikas, ibid., p. 744.

39. M. Schulz and L. J. Lanzerotti, Particle Diffu-sion in the Radiation Belts (Springer-Verlag,New York, 1973).

40. R. M. Thorne, in Critical Problems of Mag-netospheric Research, E. R. Dyer, Ed.(IUCSTP Secretariat, National Academy ofSciences, Washington, D.C., 1972), p. 211.

41. R. S. White, Rev. Geophys. Space Phys. 11,595 (1973).

42. Supported under grants from the NationalScience Foundation and the National Aero-nautics and Space Administration.

Genetic Control of the CellDivision Cycle in Yeast

A model to account for the order of cell cycle events

is deduced from the phenotypes of yeast mutants.

Leland H. Hartwell, Joseph Culotti,John R. Pringle, Brian J. Reid

Mitotic cell division in eukaryotes isaccomplished through a highly repro-ducible temporal sequence of eventsthat is common to almost all higher or-

ganisms. An interval of time, GJ, sep-arates the previous cell division fromthe initiation of DNA synthesis. Chro-mosome replication is accomplished dur-ing the DNA synthetic period, S, which

46

typically occupies about a third of thecell cycle. Another interval of time, G2,separates the completion of DNA syn-thesis from prophase, the beginning ofmitosis, M. A dramatic sequence ofchanges in chromosome structure andof chromosome movement character-izes the brief mitotic period that resultsin the precise separation of sister

chromatids to daughter nuclei. Mitosisis followed by cytokinesis, the parti-tioning of the cytoplasm into two daugh-ter cells with separate plasma mem-branes. In some organisms the cycle iscompleted by cell wall separation.

Each of these events occurs duringthe cell division cycle of the yeast,Saccharomyces cerevisiae (1) (Fig. 1).However, two features which distin-guish the cell cycle of S. cerevisiaefrom most other eukaryotes are par-ticularly useful for an analysis of thegene functions that control the celldivision cycle. First, the fact that bothhaploid and diploid cells undergo mi-tosis permits the isolation of recessivemutations in haploids and their analy-sis by complementation in diploids.Second, the daughter cell is recog-nizable at an early stage of the cellcycle as a bud on the surface of theparent cell. Since the ratio of bud sizeto parent cell size increases progres-sively during the cycle, this ratio pro-

Dr. Hartwell is a professor, Mr. Culotti andMr. Reid are graduate students, and Dr. Pringleis a postdoctoral fellow in the department ofgenetics, at the University of Washington, Seattle98195.

SCIENCE, VOL. 183

on

Sep

tem

ber

17, 2

009

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

vides a visual marker of the positionof the cell in the cycle.We have taken advantage of these

features of the S. cerevisiae cell cyclein the isolation and characterization of150 temperature-sensitive mutants ofthe cell division cycle (cdc mutants).These mutants are temperature-sensi-tive in the sense that they are unableto reproduce at 36°C (the restrictivetemperature) but do grow normally at23 °C (the permissive temperature); theparent strain from which they werederived reproduces at both tempera-tures. These mutations define 32 genes,each of whose products plays an essen-tial role in the successful completionof one event in the mitotic cycle (2).Although our genetic dissection of thecell cycle is in its early stages, thephenotypes of the mutants already ex-amined provide information on the in-terdependence of events in the cycle.We shall discuss the conclusions thatcan be derived from the mutant pheno-types in the context of the followingquestion: How are the events budemergence, initiation of DNA syn-thesis, DNA synthesis, nuclear migra-tion, nuclear division, cytokinesis, andcell separation coordinated in the yeastcell cycle so that their sequence isfixed? WVhile it is not necessarily thecase that all events in the cell cycleare ordered relative to one another ina fixed sequence, it is reasonable to as-sume that these events are, since theirproper order is essential for the pro-duction of two viable daughter cells.

It has been pointed out by Mitchi-son that two possible mechanisms existfor ordering a fixed sequence of cellcycle events relative to one another(3). First, there may be a direct causalconnection between one event and thenext. In this case, it would be neces-sary for the earlier event in the cycleto be completed before the later eventcould occur. For example, the "product"of the earlier event might provide the"substrate" for the later event, as ina biochemical pathway, or the com-pletion of the earlier event might acti-vate the occurrence of the later event.We shall refer to this model as the"dependent pathway model" (Fig. 2).A second possibility is that there is

not a direct causal connection betweentwo events, but that they are orderedby signals from some master timingmechanism. In this model it would notbe necessary for the earlier event tobe completed before the later eventcould occur, although the two events11 JANUARY 1974

NM

Ct to~~~'0Fig. 1. The sequence of events in the celldivision cycle of yeast: iDS, initiation ofDNA synthesis; BE, bud emergence; DS,DNA synthesis; NM, nuclear migration;mND, medical nuclear division; IND, latenuclear division; CK, cytokinesis; CS, cellseparation. Other abbreviations: G1, timeinterval between previous cytokinesis andinitiation of DNA synthesis; S, period ofDNA synthesis; G2, time between DNAsynthesis and onset of mitosis; and M, theperiod of mitosis.

would normally occur in the properorder because of the activity of thetimer. This model has appeared fre-quently in the literature in one guise oranother, and two specific ideas havebeen presented concerning a possibletiming mechanism. One invokes theaccumulation of a specific division pro-tein (4) and another a temporal se-quence of genetic transcriptions (5).We shall refer to this model as the "in-dependent pathways model" (Fig. 2).

It is important to note that these twopossible models relate, strictly speak-ing, to the dependence or independenceof events in the cell cycle taken two ata time. It is quite possible that the cellcycle is controlled by a combination ofthe two models, with some events re-lated to one another in a dependent

dependent pathway model

A - B --C - D --E --F

independent pathways model

A

/ B

C

K~~~~~~ r~~~~~E-~~~~~~~~F

Fig. 2. Two models to account for theordering of cell cycle events.

pathway and others in independentpathways.

It should be possible to distinguishbetween these fundamentally differentmodels by specifically inhibiting oneand only one event of the cell cycle.If an event is dependent upon the prioroccurrence of an earlier event, a spe-cific block of the earlier event shouldprevent the occurrence of the laterevent. If, on the other hand, the twoevents are independent of one another,then a specific block of the earlierevent should not prevent the occurrenceof the subsequent event. Indeed, studiesemploying inhibitors that act specifical-ly on one event of the cycle, such asDNA synthesis or mitosis, have alreadyprovided some information on the in-terdependence of cell cycle events.However, the temperature-sensitive cdcmutants of S. cerevisiae permit moredetailed conclusions, both because ofthe greater number of specific cell cycleblocks in a single organism and be-cause of the greater assurance that a

single gene defect directly affects oneand only one event in the cell cycle.

Mutations Affecting the Cell Cycle

Cell division cycle mutants of S.cerevisiae were detected among a col-lection of temperature-sensitive mu-tants by looking for mutants in whichdevelopment was arrested at the re-strictive temperature at a specific stagein the cell cycle, as evidenced by thecellular and nuclear morphology (6).The phenotype of each mutant class isdescribed in Table 1 by the sequenceof events that occurs in a cell when itis shifted from the permissive temper-ature to the restrictive temperature atthe beginning of the cell division cycle(that is, at cell separation, see CS inFig. 1). The initial defect in a mutantis defined as the first cell cycle event(among those which can presently bemonitored) that fails to take place atthe restrictive temperature. The eventsfor which initial defects have beenfound in mutants include the initiationof DNA synthesis, bud emergence,DNA synthesis, medial nuclear division,late nuclear division, cytokinesis, andcell separation. Information on the in-terdependence of steps in the cell di-vision cycle is obtained by observingwhich events in the first cell cycle atthe restrictive temperature occur or donot occur after arrest at the initial de-fect.

47

on

Sep

tem

ber

17, 2

009

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

Two dependent pathways in the cycle.The model of the cell division cyclepresented in Fig. 3 can be derived fromthe phenotypes of the mutants (Table1 ) by the following reasoning. First, letus compare the phenotypes of these mu-tants with the predictions of the de-pendent pathway model. Working back-wards through the cell cycle we seethat this model is adequate for the se-quence: cell separation, cytokinesis,late nuclear division, medial nucleardivision, DNA synthesis, and the ini-tiation of DNA synthesis. A mutantwith an initial defect in any one of thesesix processes fails to complete any ofthe other events in this group whichnornmally occurs later in the cycle. Thesimplest explanation of these observa-tions is that these six events comprisea dependent pathway in which thecompletion of each event is a necessaryprerequisite for the occurrence of theimmediate succeeding event (Fig. 3).

In contrast, although bud emergenceand DNA synthesis normally occur atabout the same time in the cell cycle,they must be on separate pathways(Fig. 3) since they are independent ofone another. Mutants defective in theinitiation of DNA synthesis (cdc 4 andcdc 7) or in DNA synthesis (cdc 8

CS

CK

il

mND 23

Fig. 3. The circuitry of the 3cycle. Events connected by an a

proposed to be related such thatevent is dependent for its occurrethe prior completion of theevent. The abbreviations are thein Fig. 1. Numbers refer to cdlc Efare required for progress fromto the next; HU and TR refer tosynthesis inhibitors hydroxyureamon, respectively; MF refers to tlfactor, (y factor.

and cc/c 21 ) undergo bud enand mutants defective in bud et(cf/c 24) undergo DNA qFurthermore, inhibitors that blcsynthesis (hydroxyurea and tido not inhibit bud emergence

Although we do not havewith initial defects in nuclea

Table 1. Summary of mutant phenotypes. Cells were shifted from 230 to 36°C atof cell separation. Abbreviations are as in Fig. 1. A minus sign indicates that an evenoccur, a plus indicates that the event occurs once, and a double plus indicates thatoccurs more than once.

.d. Initial Events completed at restrictive temperaturedefect BE iDS DS NM mND IND CK CS

2824478

21269

1316172023141531011

StartBEiDSiDSDSDSmNDmNDmNDmNDmNDmNDmNDmNDINDINDCKCKCKCS

++++

++ ?_ +

±+ +±

± + - +± ±+

++ + +

± ±+

+ +1+ + +

± ++ ++ ±+ +± +

± ±+ ±+ +± ±± ±

+ + + + ±± ± + + +++ ++ ++ ++ ++ ++++ ++ ++ ++ +± ++++ ++ ++ ++ ±++++ ++ ++ d-+ ++ ++ ++

* AlthouLgh muLtations in 32 cdc genes have been discovered. only 19 of these genes are infor consideration in developing a model of the cell cycle. Most of those not included wbecause they progress through several cycles at the restrictive temlperatuLre before devarrested and this prevents an analysis of DNA synthesis during their terminal cycle. The1 (19) was excltuded because macromolecule synthesis, as sell as btud emiiergence, is rapiin this mutant at the resti-ictive temliperatuLre, and we suLspect that this inhibition of growthe occuri ence of sonic events which arc not noi mllally dependent tupon buLd emei-gence, bildependent on growth.

48

tion, it is apparent that this event, likethe bud emergence event, occurs in all

24 mutants defective in the initiation of8 \ DNA synthesis and in DNA synthesisHU BE (Table 1). Nuclear migration also oc-TR I curs when DNA synthesis is inhibited

z2 / with hydroxyurea or trenimon (7, 8).9 / Nuclear migration is therefore inde-

pendent of initiation of DNA synthesisNM and DNA synthesis. Furthermore, since

the nucleus normally migrates into theneck between the bud and parent cell,

yeast cell it seems reasonable to suppose that nu-arrow are clear migration is dependent upon budthe distal emergence. We propose, therefore, thatnce upon nuclear migration is an event on theproximalsame as same pathway as bud emergence and

genes that subsequent to it on this pathway (Fig.one event 3).the DNA Finally, medial and late nuclear di-and treni- vision are completed in the ;mutant de-hie mating fective in bud emergence (cdc 24) but

neither cytokinesis nor cell separationoccurs in this mutant. These observa-

nergence, tions suggest that the separate pathwaymergence that leads to bud emergence and nu-synthesis. clear migration joins the first pathway)ck DNA at the event of cytokinesis (Fig. 3).renimon) Thus, cytokinesis and cell separation(7). are dependent upon bud emergence as

mutants well as upon nuclear division.ir migra- A commnon step controls both path-

ways. Although bud emergence is notnecessary for the initiation of DNA

t the time synthesis, and vice versa, the productit does not of gene cdc 28 is required for boththe event processes (Table 1). Furthermore, the

mating factor produced by cells ofmating type a (a factor) blocks both

Reference bud emergence and the initiation ofDNA synthesis in cells of mating type

(18) a (9, 10). One hypothesis to explain(2) these observations is that the two path-(16,18) ways leading, respectively, to bud(18) emergence and to initiation of DNA(16,18) synthesis diverge from a common(18) pathway, and that both the cdc 28(17) gene product and the a factor sensitive(17) step are elements of this common path-(17) way.(2) A prediction of this hypothesis is(2) that the a factor sensitive step and(2) the step mediated by the cdc 28 gene(2) product should precede and be required(17) for the cdc 4 and cdc 7 mediated steps(17) that lead to the initiation of DNA syn-(19) thesis and for the cdc 24 mediated(19) step that leads to bud emergence. Both(19) of these predictions have been confirmed

(Il). We assume, therefore, that theicluer lefto ctc 28 gene product and the a factorelopment is sensitive step mediate some early eventimiutant cdcIdly arresteci r events In the cell cycle that areith prevents necessary prerequisites for both of theItewhich are

dlependent pathways described above.

SCIENCE, VOL. 183

on

Sep

tem

ber

17, 2

009

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

We shall term this event "start" (Fig.3). In principle, completion of the"start" event can be monitored by theacquisition of insensitivity to a factorin haploids of mating type a or by theacquisition of insensitivity to temper-ature in a cdc 28 mutant, althoughthis is not possible in all experimentalsituations.

Several observations suggest thatstart" is in fact the beginning of the

yeast cell cycle. First, stationary phasepopulations obtained by limiting anyone of several nutrients (glucose, am-monia, sulfate, phosphate) consist al-most exclusively of cells which are ar-rested at a point in the cell cycle aftercell separation, but prior to bud emer-gence and the initiation of DNA syn-thesis (12). Stationary phase cells ofmating type a do not undergo budemergence after inoculation into freshmedium in the presence of a factor. Itappears, therefore, that as yeast cellsexhaust their nutrients they finish cellcycles and become arrested prior to"start" in the cell cycle.

Similarly, when cultures are grownwith limited glucose in a chemostatthere is a striking correlation betweenthe generation time and the proportionof unbudded cells in the population (13).The increase in the proportion of un-budded cells as the generation time in-creases suggests that the unbudded cellsdelay the "start" of new cycles untilsome requirement for growth or forthe accumulation of energy reserves(14) has been met, and that, as ex-pected, the time necessary to meet thisrequirement is a function of the rateof supply of glucose.

Finally, passing "start" in the cellcycle appears to represent a point ofcommitment to division, as opposed tomating, for haploid cells. If a cell ofmating type a is beyond "start" in thecell cycle at the time of exposure to afactor, it proceeds through the cell

Fig. 4. Time-lapse photographs of diploidstrains homozygous for two cdc mutations.Cells were grown at the permissive tem-perature (23°C) and shifted onto agarplates at the restrictive temperature (36'C).The cells were photographed at the timeof the temperature shift, and at successiveintervals while the plate was maintainedat 36°C. (a) cdc 4 cdc 24 (strain RD314,182-1-1, 2) at time 0; (b) cdc 4 cdc 24after 6 hours at 36°C; (c) cdc 4 cdc 8(strain RD314, 198, 3) at time 0; (d)cdc 4 cdc 8 after 6 hours at 36'C; (e)cdc 4 cdc 13 (strain RD314, 428, 3) attime 0; (f) cdc 4 cdc 13 after 7 hours at36'C; (g) cdc 4 (strain 314D5) at time0; (h) cdc 4 after 11 hours at 36°C.

11 JANUARY 1974

cycle to cell separation at a normalrate, and then both daughter cells be-come arrested at "start" (10). Further-more, cdc mutants arrested at variouspositions in the cycle are unable tomate with cells of opposite matingtype, with two exceptions: mutants thatare arrested at "start" (cdc 28), andmutants that repeatedly pass through.'start' at the restrictive temperature(as evidenced by the attainment of amultinucleate state, Table 1) appearto mate relatively well (15).

IO"4

-**

...:

44"

#4

Events Necessary for "Start"

Let us consider what events of onecell cycle must be completed in orderto permit the "start" event of the nextcell cycle. Since in these experimentswe could not monitor "start" directly,our conclusions regarding this eventmust be considered tentative. However,it appears from the following observa-tions that an initial defect in cell sepa-ration, cytokinesis, or bud emergencedoes not prevent a cell from tLindergo-

*N

> @2

* *4i.

S44*eS

4: :4,

V* !* *: O

4 * *:,.#

or

..'.

J0;

b4A

4,4* f s,, .) I j#E a

E

0., ~. I"I.NV#

Is

41a

C 2.0; t .'.-t ,"'.

'4'

.44S

*44 *,4.A

4444.

*44

04.)

,i- A..I

O-

f:

A&

I"*

*5

4 e4 *

'-'4

.

*'

:*

46

49

..' 4 -..

4..,w-, *..4.

... p. .*

9a;*s7

N.41

I0 o

n S

epte

mbe

r 17

, 200

9 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

ing the "start" event of subsequent cellcycles.

First, although mutants defectivein cell separation have not been ex-tensively studied, they have been iso-lated from mutagenized cultures afterselecting for large cell aggregates byfiltration through nylon mesh (8). Adefect in cell separation does notappear to be lethal, and cells can gothrough an indefinite number of cellcycles despite a failure to complete thisevent.

Second, mutants defective in cyto-kinesis (cdc 3, cdc 10, cdc 11) undergomultiple rounds of bud emergence, ini-tiation of DNA synthesis, DNA syn-thesis, and nuclear division, frequentlyattaining an octanucleate stage. Thesemutants do not continue to go throughcell cycles indefinitely, and the reasonfor their eventual cessation of develop-ment is unknown, although it is thecase that many of the cells lyse afterextended incubation at the restrictivetemperature.

Finally, mutants defective in budemergence frequently undergo addi-tional nuclear cycles in that about 50percent of the cells in a diploid strainhomozygous for the cdc 24 lesion be-come tetranucleate at the restrictivetemperature. Haploid cdc 24 mutantsusually stop development at the bi-nucleate stage, and rarely become tetra-nucleate, but an analysis of DNA syn-thesis in the haploid cells at the restric-tive temperature suggests that many ofthe cells synthesize a second round ofDNA (8).These observations suggest that the

completion of the events bud emer-gence, cytokinesis, and cell separation,which comprise one of the two de-pendent pathways in the cycle, is not anecessary condition for the "start"event in a second cycle (Fig. 3). Al-though a mutant defective in nuclearmigration has not been found we an-ticipate that arrest at this event willalso permit the start of subsequentcycles.

With one exception, none of the mu-tants blocked in the pathway from ini-tiation of DNA synthesis to late nu-clear division show evidence of goingthrough additional cell cycles afterbeing arrested at the sites of their ini-tial defects. We interpret this result tomean that it is necessary to completethese events in order to undergo the"start" event in a subsequent cell cycle(Fig. 3).

50

A Timer Controls Bud Emergence

Mutants defective in the cdc 4 gene(required for initiation of DNA syn-thesis) are exceptional in that theycontinue bud emergence for multiplecycles at the restrictive temperature,attaining as many as five buds on asingle mononucleate cell (16). Thesesuccessive cycles of budding continuedespite the fact that the initiation ofDNA synthesis, DNA synthesis, nu-clear division, cytokinesis, and cell sep-aration are not occurring. Furthermore,

40

30

20

10

0

E

40

30

20

10

0

0 0 .2 .4 .6 .8 1.0

length bud / length parent cell

Fig. 5. Correlation of phenotype at 36'Cwith bud size at time of temperature shiftfor double mutant strains. Cells from a

large number of photographs like thosepresented in Fig. 4 were measured todetermine the ratio of the length of thebud to the length of the parent cell at thetime of the shift, and were scored forwhether they developed a morphologycharacteristic of cells with the cdc 8 or

cdc 13 mutation (a single round bud, solidbars), or a morphology characteristic ofthe cdc 4 mutation (one to five elongatedbuds on a single cell, open bars). (Top)The results for cdc 4 cdc 8 (strain RD314,198, 3); (bottom) the results for cdc 4cdc 13 (strain RD314, 428, 3).

the time interval between successivebudding events in cdc 4 mutants at therestrictive temperature maintains aperiodicity of about one cell cycle time.This observation suggests that sometype of intracellular timer initiates thesuccessive cycles of budding, and thatthis timer can run independently ofmany of the cell cycle events.The unusual behavior of mutants de-

fective in cdc 4, and the surprisingconclusion that their phenotype sug-gests, prompted us to consider the pos-sibility that this phenotype might notbe reflecting normal control mech-anisms, but might be -a result of anartifact. For example, the putative budson cdc 4 mutants might not be the re-sult of normal bud emergence events,but might be caused by some unrelatedmorphologic alteration. Alternatively,they might be the result of normal budemergence events, but these eventsmight be activated in an anomolousway by the abnormal cdc 4 gene prod-uct. Although we cannot completelyrule out the hypothesis of artifact inthe behavior of cdc 4 mutants, theproperties of a few double cdc mutantsdo eliminate some possible sources oferror. Double mutant strains contain-ing a defect in the initiation of DNAsynthesis (cdc 4) as one mutation, anda defect in bud emergence (cdc 24),DNA synithesis (cdc 8), or medial nu-clear division (cdc 13) as the secondmutation, were constructed and exam-ined by time-l#pse photomicroscopy(Fig. 4). A diploid strain carrying onlythe homozygous cdc 4 mutation isshown in Fig. 4, g and h, for compari-son.The double mutant strain defective

in cdc 4 and cdc 24 does not undergomultiple rounds of bud emergence atthe restrictive temperature (Fig. 4, aand b). This result indicates that mu-tants defective in cdc 4 require a func- -tional cdc 24 gene product in order todisplay the phenotype of repeated budemergence and this -phenotype is not,therefore, unrelated to the normal bud-ding process.The double mutant strains harboring

lesions in cdc 4 and cdc 8, or in cdc 4and cdc 13, exhibit an unusual patternof development at the restrictive tem-perature (Fig. 4, c and d and e and f)(16). The result is striking in that thepopulations of cells from both doublemutant strains behave heterogeneously.Some cells continue periodic bud emer-gence (characteristic of a defect in cdc

SCIENCE, VOL. 183

=41

on

Sep

tem

ber

17, 2

009

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

4 alone), and other cells terminate de-velopment with a single large bud oneach parent cell (characteristic of adefect in cdc 8 or cdc 13 alone). Fur-thermore, the phenotype that a particu-lar cell exhibits is correlated with theposition of that cell in the cell divisioncycle at the time of the shift to the re-strictive temperature. This correlationis evident in Fig. 4, c and d and e and f,and is recorded for a larger number of

-cells in Fig. 5. In both double mutantstrains most of the cells that do notcontinue bud emergence were either un--budded or had small buds, while mostof those that do continue bud emer-gence were unbudded or had large buds.These observations are interpreted tomean that the former class of cells arethose that block at the cdc 8 or cdc13 mediated processes, DNA synthesis,and medial nuclear division, respec-tively, while the latter class of cells arethose that block at the cdc 4 mediatedprocess, initiation of DNA synthesis.This interpretation is consistent withour previous determinations of the timeof function of these gene products (16,17). We may conclude, therefore, thatcontinued bud emergence in a mutantstrain is not due to the lesion in genecdc 4 per se, but is merely a result ofthe cell's position in the cell divisioncycle at the time it is arrested.These results seem to us to be best

interpreted by the hypothesis of a timerthat controls bud emergence and thatcan express itself at only one discretestage in the cell cycle, the stage ofarrest in the cdc 4 mutant. The roleof this timer in the normal cell cycle,and, in particular, its relation to the"start" event, are at present unclear.The action of the timer might be aprerequisite for, be dependent upon,or be part of the "start" event.

Implications of the Model

Let us return now to the questionwe posed at the outset: How are the-events of the cell cycle coordinatedso that their sequence remains invari-ant? The phenotypes of the cdc mu-

tants suggest that the following eventsare ordered in a single dependent path-way: "start," initiation of DNA syn-thesis, DNA synthesis, medial nucleardivision, late nuclear division, cyto-kinesis, and cell separation. Hence, thetemporal sequence of these events iseasily accounted for by the fact thatno event in this pathway can occurwithout the prior occurrence of all pre-ceding events. A second dependentpathway is comprised of the events"start," bud emergence, nuclear migra-tion, cytokinesis, and cell separation.Thus, the temporal sequence of thesefive events is also assured. Furthermore,the integration of the two pathways isaccomplished by the facts that bothdiverge from a common event, "start,"and that both converge on a commonevent, cytokinesis.

Although evidence was found for theexistence of a timer that controls budemergence, there is no indication thatthis timer plays any role in coordinat-ing different events of the cell cycle.It is conceivable that the timer servesto phase bud emergence with respect tothe events of the DNA synthesis andnuclear division pathway, but it seemsto us that the joint dependence of budemergence and initiation of DNA syn-thesis on "start" is sufficient to explainthe coordination between the two path-ways. Although the function of thetimer in the cell cycle is unknown, wefavor the idea that the timer is eitherphasing successive "start" events, per-haps by monitoring cell growth, or isphasing successive bud emergenceevents in order to limit the cell to onesuch event per cycle. A variation ofthe dependent pathway model appearsto be sufficient, therefore, to accountfor the coordination of cell cycleevents, and it does not appear to benecessary to invoke the model of in-dependent pathways with a central tim-ing mechanism.

Applicability to other organismns. Theevents that comprise the cell divisioncycle have their origin in a distant evo-lutionary past common to all eukaryoticorganisms. The complexity of thisprocess suggests that a high degree of

conservation of its basic elements mightbe expected. In this context, it is in-teresting to note that the only eventsof the S. cerevisiae cell cycle that arenot common to most eukaryotes, budemergence and nuclear migration, areon a separate pathway from the otherevents, as if they were appendagesadded to the basic plan. We would notbe surprised, therefore, if in mosteukaryotes an event, "start," activatesand acts as a point of commitment forthe dependent pathway of events lead-ing from the initiation of DNA syn-thesis, to DNA synthesis, to successivestages of nuclear division, and finallyculminating in cytokinesis and, whereapplicable, cell wall separation. Fur-thermore, the completion of some stagesof nuclear division, but not cytokinesisor cell separation, may in general benecessary in one cell cycle for the"start" of the next cell cycle.

References and Notes

1. D. H. Williamson, in Cell Synchrony, I. L.Cameron and G. M. Padilla, Eds. (AcademicPress, New York, 1966), p. 81; L. H. Hartwell,Annu. Rev. Genet. 4, 373 (1970).

2. L. H. Hartwell, R. K. Mortimer, J. Culotti,M. Ctulotti, Genetics 74, 267 (1973).

3. J. M. Mitchison, The Biology of the Cell Cy-cle (Cambridge Univ. Press, New York, 1972),p. 244

4. E. Zeuthen and N. E. Williams, in NucleicAcid Metabolism, Cell Differentiation, andCancer Growth, E. V. Cowdry and S. Seno,Eds. (Pergamon, Oxford, 1969), p. 203.

5. D. Prescott, Recent Results Cancer Res. 17,79 (1969); H. Halvorson, J. Gorman, P.Tauro, R. Epstein, M. LaBerge, Fed. Proc.23, 1002 (1964).

6. L. H. Hartwell, J. Culotti, B. Reid, Proc.Natl. Acad. Sci. U.S.A. 66, 352 (1970).

7. M. L. Slater, J. Bacteriol. 113, 263 (1973);L. Jaenicke, K. Scholz, M. Donike, Euir. J.Biochem. 13, 137 (1970).

8. L. H. Hartwell, unpublished results.9. E. Throm and W. Duntze, J. Bacteriol. 104,

1388 (1970).10. E. Biicking-Throm, W. Duntze, L. H. Hart-

well, T. Manney, Exp. Cell Res. 76, 99 (1973);L. H. Hartwell, ibid., p. 111.

11. L. Hereford and L. H. Hartwell, in prepa-ration.

12. D. H. Williamson and A. W. Scopes, Exp.Cell Res. 20, 338 (1960); J. R. Pringle, R. J.Maddox, L. H. Hartwell, in preparation.

13. H. K. von Meyenberg, Pathol. Microbiol. 31,117 (1968); C. Beck and H. K. von Meyen-berg, J. Bacteriol. 96, 479 (1968).

14. M. T. Kiienzi and A. Fiechter, Arch. Mikro-biol. 64, 396 (1969); ibid. 84, 254 (1972).

15. B. J. Reid, in preparation.16. L. H. Hartwell, J. Mol. Biol. 59, 183 (1971).17. J. Culotti and L. H. Hartwell, Exp. Cell Res.

67, 389 (1971).18. L. H. Hartwell, J. Bacteriol. 115, 966 (1973).19. , Exp. Cell Res. 69, 265 (1971).20. Supported by research grant 6M17709 from

the Institute of General Medical Sciences toJ.C.. J.R.P. and B.J.R.

11 JANUARY 1974 Sl

on

Sep

tem

ber

17, 2

009

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om