LETTERS

A spatial gradient coordinates cell size and mitoticentry in fission yeastJames B. Moseley1, Adeline Mayeux2, Anne Paoletti2 & Paul Nurse1

Many eukaryotic cell types undergo size-dependent cell cycle tran-sitions controlled by the ubiquitous cyclin-dependent kinase Cdk1(refs 1–4). The proteins that control Cdk1 activity are welldescribed but their links with mechanisms monitoring cell sizeremain elusive. In the fission yeast Schizosaccharomyces pombe,cells enter mitosis and divide at a defined and reproducible sizeowing to the regulated activity of Cdk1 (refs 2, 3). Here we showthat the cell polarity protein kinase Pom1, which localizes to cellends5, regulates a signalling network that contributes to the con-trol of mitotic entry. This network is located at cortical nodes inthe middle of interphase cells, and these nodes contain the Cdk1inhibitor Wee1, the Wee1-inhibitory kinases Cdr1 (also known asNim1) and Cdr2, and the anillin-like protein Mid1. Cdr2 estab-lishes the hierarchical localization of other proteins in the nodes,and receives negative regulatory signals from Pom1. Pom1 forms apolar gradient extending from the cell ends towards the cell middleand acts as a dose-dependent inhibitor of mitotic entry, workingthrough the Cdr2 pathway. As cells elongate, Pom1 levels decreaseat the cell middle, leading to mitotic entry. We propose that thePom1 polar gradient and the medial cortical nodes generateinformation about cell size and coordinate this with mitotic entryby regulating Cdk1 through Pom1, Cdr2, Cdr1 and Wee1.

Recurring cycles of cellular growth and division present a simplequestion: how do cells link these processes and divide only uponreaching an appropriate size? Genetic screens have identified com-ponents in the fission yeast controlling cell size at division6,7. Thecentral factor Cdk1 drives entry into mitosis by phosphorylating awide range of targets8. In small cells that are in early G2, Cdk1 is keptinactive by the kinase Wee1, which directly phosphorylates and inhi-bits Cdk1. As cell size increases during G2, the phosphatase Cdc25removes this inhibitory phosphorylation to activate Cdk1 and allowmitotic entry. Thus, the mitotic entry control system coordinates abalance of Wee1 and Cdc25 activity with changes in cell size togenerate a reproducible cell size at division. Although several factorsregulating Wee1 and Cdc25 have been identified3, the links betweencell size and the Wee1-Cdc25-Cdk1 module have remained elusive.

We noted that the protein kinase Cdr2, which negatively regulatesWee1, is localized to a band of cortical nodes in the middle of inter-phase cells9. After mitotic entry, cytokinesis factors such as myosin IIare recruited to similar nodes, which subsequently condense to formthe cytokinetic ring10. The presence of Cdr2 in these medial nodesduring G2, when the inhibition of Wee1 contributes to the G2–Mtransition, suggested that the nodes might act as potential sites of cellcycle regulation. To investigate this possibility further, we first ana-lysed the composition of medial nodes. We found that the previouslyuncharacterized protein Blt1 (also known as SPBC1A4.05), identifiedusing a proteomic approach (Supplementary Fig. 1), co-localized withCdr2 in the medial interphase nodes, together with Mid1 (Fig. 1a, b),which was previously shown to localize to similar interphase

structures11. Furthermore, physical interactions between Blt1–Mid1,Blt1–Cdr2 and Cdr2–Mid1 were detected by co-immunoprecipitation(Fig. 1d). The Cdr2-related kinase Cdr1, which directly phosphorylatesand inhibits Wee1 in vitro12–14, also co-localized with Blt1 in the nodes(Fig. 1c and Supplementary Fig. 2a). The presence of two componentsof the mitotic control system, Cdr1 and Cdr2, in nodes during G2supports the possibility that these structures might function to regulatemitotic entry. To investigate the recruitment of these proteins to theinterphase nodes, we monitored green fluorescent protein (GFP)-tagged node proteins in cells deleted for the other components, andfound that Cdr2 localized to interphase nodes in mid1D, blt1D andcdr1D cells, but was required for all other proteins to reach the inter-phase cortical nodes (Fig. 2a and Supplementary Fig. 3a). We deter-mined a hierarchy for recruitment of these proteins to interphasenodes by examining all combinations of GFP-tagged proteins andmutants, including the additional interphase node proteins Klp8and Gef2 (Supplementary Figs 2, 3 and data not shown). Theseresults indicate that medial cortical nodes are formed by the ordered,Cdr2-dependent assembly of multiple interacting proteins duringinterphase.

The presence of the mitotic regulators Cdr1 and Cdr2 in the inter-phase nodes prompted us to examine whether Blt1 also regulatesmitotic entry. blt1D cells had an increased length at division, consist-ent with a delay in mitotic entry. This phenotype was synthetic with atemperature-sensitive cdc25ts mutation (Supplementary Table 1),but not with cdr1D and cdr2Dmutants, which act through Wee1 (refs15–17). These genetic data are consistent both with the localizationhierarchy described earlier and with a pathway in which Cdr2 func-tions upstream of Cdr1 and Blt1 to promote mitosis through thenegative regulation of Wee1 (Fig. 2b).

To determine whether these nodes might act as sites of Wee1regulation, we next examined the cellular distribution of Wee1.GFP–Wee1 under the control of the low-level P81nmt1 promoterwas found in three cellular structures: the medial cortical nodes, thenucleus and the spindle pole body (Fig. 2c). GFP–Wee1 co-localizedwith Blt1–mCherry at the medial cortical nodes (Fig. 2d), and withthe spindle pole body marker Sid4–mCherry (Supplementary Fig. 4).To confirm this localization pattern for endogenously expressedWee1, we integrated a triple-GFP tag at the genomic wee11 locus.Although present at low levels, Wee1–33GFP was found in the med-ial cortical nodes and the nucleus (Fig. 2e). Wee1 localization to themedial cortical nodes depended on Cdr2 but not on Cdr1 or Blt1(Supplementary Fig. 5), reinforcing the importance of Cdr2 in gen-erating these structures. Moreover, a Cdr2(E177A) mutant, whichlocalizes properly but is predicted to abolish Cdr2 kinase activity9,disrupted the recruitment of Wee1 to the medial cortex and delayedentry into mitosis (Supplementary Fig. 6), indicating a role for Cdr2kinase activity in signalling to Wee1. Because Wee1 is present in thesame medial cortical nodes as its inhibitory network, we propose that

1The Rockefeller University, New York, New York 10065, USA. 2Institut Curie, Centre de Recherche and CNRS UMR144, Paris 75248 cedex 05, France.

Vol 459 | 11 June 2009 | doi:10.1038/nature08074

857 Macmillan Publishers Limited. All rights reserved©2009

the interphase nodes control the onset of mitosis through Cdr2-dependent negative regulation of Wee1.

Placement of Cdr2 at the top of the Wee1 regulatory networksuggests that mechanisms regulating Cdr2 could act as inputs tothe mitotic entry control system. We considered that cell polarityfactors positioned at the cell tip might provide spatial cues to limitCdr2 distribution to the cell middle. To test this possibility, cellpolarity factor mutants (for3D, mod5D, orb2ts, orb6ts, pom1D, tea1Dand tea4D) were screened for defects in Cdr2 localization. Thisdemonstrated a highly specific role for the protein kinase Pom1, aconserved member of the dual-specificity tyrosine-phosphorylationregulated kinase (DYRK) family of kinases5. Cdr2 was restricted tothe cell middle in wild-type cells, but localized throughout half of thecell in a pom1D mutant (Fig. 3a). pom1D cells grow in a monopolarmanner, and the Cdr2 nodes extended from the cell middle out to thenon-growing cell end. This strong defect in Cdr2 localization was not

observed in other monopolar or cell polarity mutants(Supplementary Fig. 7), although a more minor defect was foundin tea1D and tea4D mutants, which fail to localize Pom1 to cellends5,18. Interphase node components downstream of Cdr2 also loca-lized to the cell end in pom1D mutants (Supplementary Fig. 8; previ-ously shown for Mid1 (refs 19, 20)).

These results indicate that Pom1 provides inhibitory signals thatconfine Cdr2 to the cell middle. Pom1-dependent signals control Cdr2phosphorylation, as shown by the retardation of Cdr2–haemagglutinin(HA) migration in protein gels owing to phosphorylation afteroverexpression of the Pom1 protein kinase (Fig. 3b). Consistently,

a

c

d

b

Max

pro

ject

Sin

gle

Max

pro

ject

Sin

gle

Max

pro

ject

Sin

gle

MergeBlt1–mCherryCdr2–mEGFP

MergeBlt1–mCherryMid1–mEGFP

MergeBlt1–mCherryCdr1–3×GFP

Mid1

Blt1–GFP

Anti-GFP IP

Input

Anti-GFP IP

Input

Anti-GFP IP

Input

Blt1–3×HA

Cdr2–GFP

Mid1–13×Myc

Cdr2–GFP– ++ +

– ++ +

– ++ +

Anti-Mid1 probe Anti-HA probe Anti-Myc probe

Figure 1 | Interphase medial nodes contain multiple factors that physicallyinteract. a–c, Co-localization of Blt1–mCherry and indicated GFP-taggedproteins. Top row images are inverted maximum projections; cells areoutlined in white in the merged images. Regions boxed in red are magnifiedin the bottom rows, which show inverted single focal planes. Max project,maximum projection; mEGFP, monomeric enhanced GFP. Scale bars,3.5 mm. d, Co-immunoprecipitation of the indicated node proteins from ananti-GFP immunoprecipitation (IP) from native cell extracts. Samples wereprobed with the indicated antibodies.

a

Maximum projection Top focal plane Medial focal planec

d GFP–Wee1

Blt1–mCherry Merge

GFP–Wee1

Blt1–mCherry Merge

bMid1–mEGFP Cdr1–3×GFP Blt1–mEGFP

cdr2

+cd

r2Δ

GFP

–Wee

1

e Max project Top plane Medial plane

Wee

1–3×

GFP

Wee

13×

GFP

?

Cdk1-P(off)

Cdk1

Cdr2

Wee1

Cdr1 Blt1

Cdc25(on)

Figure 2 | Interphase nodes are organized by Cdr2 and contain Wee1.a, Localization of Mid1–mEGFP, Cdr1–33GFP and Blt1–mEGFP in wild-type (cdr21) and cdr2D cells. Cells are outlined in black, and images areinverted maximum projections. Scale bar, 3.5 mm. b, Pathway for control ofmitotic entry by interphase node proteins on the basis of genetic epistasisand localization analysis. Dashed line indicates potential for directregulation of Wee1 by Cdr2 (ref 16). Blt1 functions downstream of bothCdr1 and Cdr2, but requires only Cdr2 for proper localization. Themolecular function of Blt1 remains unclear, as denoted by the questionmark. c, Localization of GFP–Wee1 from S. pombe strain JM551. Cells areoutlined in the left panel. Scale bar, 5mm. d, Co-localization of GFP–Wee1and Blt1–mCherry; images are inverted single focal planes. Left panels showa single cell outlined in merged image. The region boxed in red is magnifiedin the right panels. e, Localization of Wee1–33GFP expressed from theendogenous promoter and chromosomal locus. Scale bar, 3.5mm.

LETTERS NATURE | Vol 459 | 11 June 2009

858 Macmillan Publishers Limited. All rights reserved©2009

Cdr2–HA migrated faster after the deletion of pom11 (Fig. 3b andSupplementary Fig. 9). These results indicate that Pom1 may actdirectly on Cdr2 in cells, although an indirect effect through otherprotein kinases remains possible. To investigate potential functionalroles for this regulation, we examined the length of pom1D cells atdivision. These divided at a smaller size than wild-type cells, indicatingpremature entry into mitosis (Fig. 3c). pom1D suppressed the elon-gated cell-length phenotype of a cdc25ts mutant, but had little effectwhen combined with cdr2D and cdr1D. Conversely, increased pom11

expression generated a delay in mitotic entry that was syntheticwith cdc25ts but not with cdr2D, cdr1D or wee1ts. This mitotic delaycorrelated with increased phosphorylation of Cdk1 on Tyr 15(Supplementary Fig. 10), the target of Wee1 kinase2. These data indi-cate that Pom1 functions in a dose-dependent manner to delay entryinto mitosis through negative regulation of the Cdr2-Cdr1-Wee1pathway. The pom1D size phenotype is not as severe as wee11 deletion,indicating that further Wee1 regulatory mechanisms are likely to beoperating. We conclude that Pom1 is a potential functional linkbetween polarized cell growth and mitotic entry by regulating thesetwo processes.

Pom1 forms a polar gradient that peaks at cell ends19. Using quant-itative microscopy on cells expressing Pom1–GFP, we examined thechanges in Pom1 distribution as cells elongate and progress throughG2 (Fig. 4a). Fluorescence-intensity measurements revealed the pres-ence of Pom1 in the middle of small cells. As cell length increased,Pom1 levels in the cell middle showed a progressive decrease (Fig. 4b);this size-dependent decrease was confirmed in a larger number ofcells (Supplementary Fig. 12). Thus, Cdr2 localized at the cell middleencounters progressively lower levels of inhibitory Pom1 as the cellelongates and progresses through G2. Given these results, we proposea model for how the relationship between Cdr2 medial nodes and aPom1 inhibitory gradient might relay information about cell size to

regulate mitotic entry (Fig. 4c). In the model, small cells in early G2contain sufficient levels of Pom1 at the cell middle to negativelyregulate Cdr2 function and prevent entry into mitosis. After polar-ized growth during G2, Pom1 levels at the cell centre decrease, allow-ing Cdr2 nodes to promote mitosis by inhibiting Wee1. In support ofcell size-dependent regulation by Pom1, Cdr2–HA from small cells(wee1ts) migrates slower in protein gels than Cdr2–HA from largecells (cdc25ts) (Supplementary Fig. 9c).

This model predicts that cell cycle regulation at medial nodesresponds to changes in Pom1 localization. To test this prediction,we first examined tea1D cells, in which Pom1 is not concentrated atcell ends but does retain full kinase activity21. This altered distri-bution within the cell should allow Pom1 more access to Cdr2 atthe cell middle. Consistent with the model, tea1D cells showed a delayin mitotic entry, and this phenotype was additive with the cdc25ts

mutation but not with the cdr2D or pom1D mutants (SupplementaryTable 2). As a more direct test, we ectopically localized Pom1 kinaseactivity throughout the cortex by fusing the Pom1 carboxy-terminalkinase domain (Pom1C) to a plasma membrane targeting domain(PMT). This chimaera was efficiently targeted throughout the cortexand disrupted Cdr2 localization in the cell middle (Fig. 4d).

Cdr2–mEGFP

Cdr2–HA

Blankophor

pom

1+p

om1Δ

a c

b 1 2 3 4 5

Strain Length at division (µm)

WT 14.3 ± 0.9

pom1Δ 11.7 ± 2.0

cdc25ts 21.0 ± 2.4

pom1 cdc25ts 17.0 ± 1.6

cdr2Δ 20.4 ± 1.3

19.6 ± 1.6

cdr1Δ 19.5 ± 1.3

19.2 ± 1.4

P3nmt1-pom1+ 17.7 ± 1.4

P3nmt1-pom1+ cdc25ts 33.9 ± 5.4

P3nmt1-pom1+ cdr2Δ 19.4 ± 2.0

P3nmt1-pom1+ cdr1Δ 19.4 ± 1.6

wee1ts 6.8 ± 0.9

P3nmt1-pom1+ wee1ts 6.8 ± 1.1

pom1 cdr2Δ

pom1Δ cdr1Δ

1: pom1Δ2: Wild type

3: pom1 OE (~×3)4: pom1 OE (~×20)

5: pom1 OE (~×20) + λ-PPase

Δ

Δ

Figure 3 | Pom1 regulates Cdr2 and mitotic entry. a, Localization ofCdr2–mEGFP in wild-type (pom11) and pom1D cells that are also stainedwith Blankophor, a cell wall dye that brightly stains growing cell ends.Images are inverted maximum projections. Scale bar, 3.5 mm. b, Migration ofCdr2–33HA from wild-type, pom1D and pom1-overexpressing (OE) cells.Whole-cell extracts were separated by SDS–PAGE, transferred tonitrocellulose, and probed with an anti-HA antibody. Levels of pom11

overexpression were quantified by western blot (Supplementary Fig. 11a, b).l-PPase, l-phosphatase. c, Length of dividing, septated cells of the indicatedgenotypes (mean 6 s.d.; n . 50 for each value). wee1-50 strains were shiftedto 36.5 uC for 6 h before measurements; all other strains were measured at25 uC. P3nmt1-pom11 cells were grown in the presence of thiamine, whichincreases Pom1 levels approximately threefold compared to endogenouslevels (Supplementary Fig. 11a).

Early G2 Late G2

Pom1Cdr2 signalling (off)

Cdr2 signalling (on)

a b

c

cdr2

-GFP

cdr2

cdc2

5ts

Growth

d

PMT–mCherry

Pom1C–mCherry

Chimaera-mCherry

Cdr2–GFPmCherry eConstruct Length at division (µm)

None 14.8 ± 0.9

PMT 14.7 ± 1.1

Pom1C 15.9 ± 1.2

Chimaera 20.3 ± 1.2

None 20.5 ± 1.3

PMT 20.3 ± 1.4

Pom1C 20.7 ± 1.6

Chimaera 20.5 ± 1.2

None 21.0 ± 2.4

PMT 20.7 ± 2.3

Pom1C 28.3 ± 3.5

Chimaera 38.2 ± 6.7

7.6 µm

8.7 µm

9.8 µm

10.7 µm

11.5 µm

7.6 µm

8.7 µm

9.8 µm

10.7 µm

11.5 µm

400

300

200

100

00 2 4 6 8 10 12 14

Fluo

resc

ence

(a.u

.)

Distance (µm)

Pom1–GFP

Δ

Cellmiddle

Figure 4 | Pom1 levels at the cell centre control Cdr2-dependent entry intomitosis. a, Localization of Pom1–GFP in cells of indicated size. Images areinverted sum projections. b, Fluorescence intensity quantifications for cellsfrom a. a.u., arbitrary units. c, Model for regulation of mitotic entry at Cdr2nodes by Pom1 gradient. d, Left: localization of PMT, Pom1C and chimaera(PMT–Pom1C fusion). Constructs are tagged with mCherry. Right:localization of Cdr2–GFP in cells expressing the indicated constructs.Images are single focal planes, and the same cells are shown for mCherry andCdr2–GFP images. Scale bar, 3.5 mm. e, Length of dividing, septated cellsexpressing the indicated constructs in cdr2-GFP, cdr2D or cdc25-22background, as indicated (mean 6 s.d.; n . 50 for each value).

NATURE | Vol 459 | 11 June 2009 LETTERS

859 Macmillan Publishers Limited. All rights reserved©2009

Expression of the chimaera generated a delay in mitotic entry thatwas equivalent to the deletion of cdr21, whereas the chimaera did notaffect cdr2D cells (Fig. 4e). Furthermore, expression of the chimaerain cdc25ts cells led to a strong mitotic delay. We conclude that thelevels of Pom1 at the cell middle regulate entry into mitosis throughthe negative regulation of the Cdr2 pathway.

By exploiting the simple geometry and genetic tractability of fis-sion yeast, we have revealed a regulatory system coordinating cell sizeand division by means of a polar gradient that inhibits the function ofcortical structures in the cell middle. A central component of thismechanism is the conserved kinase Pom1, which acts as a molecularlink between cell growth and mitotic entry through a Cdr2-Cdr1-Wee1-Cdk1 pathway. We propose that the Pom1 polar gradientprovides information about cell size and geometry for the Cdk1mitotic regulatory system. An important next step will be to definethe biochemical nature of interactions within this signalling network,which will lead to the generation of quantitative models for cell cycleregulation at interphase nodes. Such models have already providedinsight into the mechanism of cytokinetic ring assembly by post-mitotic cortical nodes22.

The ability of diffusible morphogen gradients to coordinate sizeand geometry has been much studied in multicellular developmentalsystems. However, recent work has demonstrated that gradients alsooperate within cells to control a growing number of processes,including GTPase signalling23, mitotic spindle assembly24, and themitotic entry system that we have described here. Such intracellulargradients operate on a smaller spatial scale, but are able to generateinformation about cell size and shape23. Thus, the use of gradients togenerate spatial information represents a widely conserved mech-anism operating at a range of scales in biology.

METHODS SUMMARYStandard S. pombe media and methods were used25. Strains are listed in

Supplementary Table 3. Gene deletion and tagging were performed by PCR

and homologous recombination26. Cells for length measurements were stained

with Blankophor (MP Biochemicals), imaged in Metamorph (MDS Analytical

Technologies) using an epifluorescence microscope (Axioplan 2, Carl Zeiss, Inc.)

equipped with a CoolSNAP HQ camera (Roper Scientific), and analysed inImageJ (National Institutes of Health). More than 50 septated cells were mea-

sured for each strain.

For protein localization, cells were imaged on agar pads or in liquid media

under a coverslip with 0.4 or 0.5mm step size using a DeltaVision Spectris

(Applied Precision) comprised of an Olympus IX71 wide-field inverted fluor-

escence microscope, an Olympus UPlanSApo 3100, numerical aperture 1.4, oil

immersion objective, and a Photometrics CoolSNAP HQ camera (Roper

Scientific). Images were captured and processed by iterative constrained decon-

volution using SoftWoRx (Applied Precision), and analysed in ImageJ. For

quantitative Pom1 localization, Pom1–GFP cells (JM414) were imaged in 14

Z-sections of 0.4 mm each, and a sum projection of each single cell was com-

pressed into a one-dimensional line along the long axis of the cell using ImageJ.

The fluorescence intensity profile of this line was plotted, and the background

fluorescence from cells not expressing GFP was subtracted. The minimum fluor-

escence value in the centre of each cell was identified from these background-

subtracted profiles to generate the mean values in Supplementary Fig. 12.

For co-immunoprecipitation, cells were washed in IP buffer (20 mM HEPES,

pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.2% NP-40, 1 mM PMSF and Completeprotease inhibitor (Roche)) and lysed by bead-beating. Extracts were cleared by

centrifugation at 16,000g and incubated with Dynabeads coated with anti-GFP27.

Precipitates were washed in IP buffer, separated by SDS–PAGE, and analysed by

western blotting. Whole-cell lysates were prepared as described28.

Full Methods and any associated references are available in the online version ofthe paper at www.nature.com/nature.

Received 31 January; accepted 21 April 2009.Published online 27 May 2009.

1. Mitchison, J. M. Growth during the cell cycle. Int. Rev. Cytol. 226, 165–258 (2003).2. Jorgensen, P. & Tyers, M. How cells coordinate growth and division. Curr. Biol. 14,

R1014–R1027 (2004).

3. Rupes, I. Checking cell size in yeast. Trends Genet. 18, 479–485 (2002).4. Dolznig, H., Grebien, F., Sauer, T., Beug, H. & Mullner, E. W. Evidence for a size-

sensing mechanism in animal cells. Nature Cell Biol. 6, 899–905 (2004).5. Bahler, J. & Pringle, J. R. Pom1p, a fission yeast protein kinase that provides

positional information for both polarized growth and cytokinesis. Genes Dev. 12,1356–1370 (1998).

6. Nurse, P., Thuriaux, P. & Nasmyth, K. Genetic control of the cell division cycle inthe fission yeast Schizosaccharomyces pombe. Mol. Gen. Genet. 146, 167–178(1976).

7. Nurse, P. Genetic control of cell size at cell division in yeast. Nature 256, 547–551(1975).

8. Ubersax, J. A. et al. Targets of the cyclin-dependent kinase Cdk1. Nature 425,859–864 (2003).

9. Morrell, J. L., Nichols, C. B. & Gould, K. L. The GIN4 family kinase, Cdr2p, actsindependently of septins in fission yeast. J. Cell Sci. 117, 5293–5302 (2004).

10. Wu, J. Q., Kuhn, J. R., Kovar, D. R. & Pollard, T. D. Spatial and temporal pathway forassembly and constriction of the contractile ring in fission yeast cytokinesis. Dev.Cell 5, 723–734 (2003).

11. Paoletti, A. & Chang, F. Analysis of mid1p, a protein required for placement of thecell division site, reveals a link between the nucleus and the cell surface in fissionyeast. Mol. Biol. Cell 11, 2757–2773 (2000).

12. Wu, L. & Russell, P. Nim1 kinase promotes mitosis by inactivating Wee1 tyrosinekinase. Nature 363, 738–741 (1993).

13. Parker, L. L., Walter, S. A., Young, P. G. & Piwnica-Worms, H. Phosphorylation andinactivation of the mitotic inhibitor Wee1 by the nim1/cdr1 kinase. Nature 363,736–738 (1993).

14. Coleman, T. R., Tang, Z. & Dunphy, W. G. Negative regulation of the wee1 proteinkinase by direct action of the nim1/cdr1 mitotic inducer. Cell 72, 919–929 (1993).

15. Breeding, C. S. et al. The cdr21 gene encodes a regulator of G2/M progression andcytokinesis in Schizosaccharomyces pombe. Mol. Biol. Cell 9, 3399–3415 (1998).

16. Kanoh, J. & Russell, P. The protein kinase Cdr2, related to Nim1/Cdr1 mitoticinducer, regulates the onset of mitosis in fission yeast. Mol. Biol. Cell 9, 3321–3334(1998).

17. Russell, P. & Nurse, P. The mitotic inducer nim11 functions in a regulatory networkof protein kinase homologs controlling the initiation of mitosis. Cell 49, 569–576(1987).

18. Martin, S. G., McDonald, W. H., Yates, J. R. & Chang, F. Tea4p links microtubuleplus ends with the formin for3p in the establishment of cell polarity. Dev. Cell 8,479–491 (2005).

19. Padte, N. N., Martin, S. G., Howard, M. & Chang, F. The cell-end factor pom1pinhibits mid1p in specification of the cell division plane in fission yeast. Curr. Biol.16, 2480–2487 (2006).

20. Celton-Morizur, S., Racine, V., Sibarita, J. B. & Paoletti, A. Pom1 kinase linksdivision plane position to cell polarity by regulating Mid1p cortical distribution. J.Cell Sci. 119, 4710–4718 (2006).

21. Bahler, J. & Nurse, P. Fission yeast Pom1p kinase activity is cell cycle regulated andessential for cellular symmetry during growth and division. EMBO J. 20,1064–1073 (2001).

22. Vavylonis, D., Wu, J. Q., Hao, S., O’Shaughnessy, B. & Pollard, T. D. Assemblymechanism of the contractile ring for cytokinesis by fission yeast. Science 319,97–100 (2008).

23. Meyers, J., Craig, J. & Odde, D. J. Potential for control of signaling pathways via cellsize and shape. Curr. Biol. 16, 1685–1693 (2006).

24. Bastiaens, P., Caudron, M., Niethammer, P. & Karsenti, E. Gradients in the self-organization of the mitotic spindle. Trends Cell Biol. 16, 125–134 (2006).

25. Moreno, S., Klar, A. & Nurse, P. Molecular genetic analysis of fission yeastSchizosaccharomyces pombe. Methods Enzymol. 194, 795–823 (1991).

26. Bahler, J. et al. Heterologous modules for efficient and versatile PCR-based genetargeting in Schizosaccharomyces pombe. Yeast 14, 943–951 (1998).

27. Cristea, I. M., Williams, R., Chait, B. T. & Rout, M. P. Fluorescent proteins asproteomic probes. Mol. Cell. Proteomics 4, 1933–1941 (2005).

28. Sreenivasan, A. & Kellogg, D. The Elm1 kinase functions in a mitotic signalingnetwork in budding yeast. Mol. Cell. Biol. 19, 7983–7994 (1999).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements We thank members of the Nurse laboratory for discussionsand critical reading of the manuscript, A. Puszynska for technical assistance, TheRockefeller University Bio-Imaging Resource Center for assistance withmicroscopy, The Rockefeller University Proteomics Resource Center for massspectrometry, K. Gould, J.-Q. Wu, T. Pollard, J. Bahler, F. Chang and K. Sawin foryeast strains, plasmids and antibodies, and S. Martin for sharing unpublished dataand for discussions. J.B.M. was supported by a postdoctoral fellowship from theAmerican Cancer Society (PF-07-129-01-MBC); A.P. by funding from ANR(BLAN06-3_135468), ARC (no. 4934) and FRM (INE20071110973); and P.N. bythe Breast Cancer Research Foundation and The Rockefeller University.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. Correspondence and requests for materials should beaddressed to J.B.M. (jmoseley@rockefeller.edu).

LETTERS NATURE | Vol 459 | 11 June 2009

860 Macmillan Publishers Limited. All rights reserved©2009

METHODSYeast methods. Yeast strains were generated by tetrad dissection, when applic-

able. For cell-length measurements, cells were grown at 25 uC in EMM media

lacking supplements. Cells measured for Supplementary Fig. 6d were grown in

EMM14S media. Strains measured in Fig. 3c were: JM366, JM185, JM14, JM618,

JM554, JM616, JM555, JM492, JM717, JM787, JM805, JM887, JM228 and

JM786. Strains measured in Fig. 4e were: AP2458, AP2542, AP2541, AP2540,

JM554, AP2545, AP2544, AP2543, JM14, AP2587, AP2468 and AP2588. Strains

measured in Supplementary Fig. 6d were: JM823, JM822 and JM826. Strains

measured in Supplementary Table 1 were: JM366, JM206, JM14, JM357, JM554,JM888, JM555, JM889 and JM625. Strains measured in Supplementary Table 2

were: JM366, JM885, JM14, JM619, JM554, JM614 and JM886. Statistical ana-

lyses of genetic interactions are listed in Supplementary Table 4 (a5 0.05). The

only GFP-tagged strains for which we noticed an altered cell length at division

were cdr1-3GFP (11.4 6 1.3 mm) and wee1-3GFP (16.5 6 1.6mm); these pheno-

types suggest increased activity of Cdr1 and Wee1, respectively, in these strains.

P81nmt1-GFP-wee11 strains were grown in the presence of thiamine, and

expression was induced by removal of thiamine for 20 h before imaging. PMT

encodes amino acids 500–920 of Mid1 and localizes to the cortex independently

of Cdr2 (Supplementary Fig. 13c). Pom1C encodes amino acids 591–1087 of

Pom1. PMT, Pom1C and the PMT–Pom1C chimaera were subcloned into the

previously described pAP140 (ref. 29), modified to contain the mCherry coding

sequence (from pKS390, a gift from K. Sawin). To generate pJM353, the cdr21

promoter, cdr21 coding sequence, GFP coding sequence and cdr21 terminator

were PCR-amplified and subcloned into the integration plasmid pJK148. The

mutation E177A was introduced by site-directed mutagenesis of pJM353 using

Quikchange IIXL kit (Stratagene) to generate pJM385. For pJM394, the cdr21

locus including promoter (2800) and terminator (11,000) was PCR-amplifiedand subcloned into integration plasmid pJM391 (gift from D. Coudreuse). The

mutation E177A was introduced by site-directed mutagenesis of pJM394 using

Quikchange IIXL kit (Stratagene) to generate pJM397.

Biochemistry. For mass spectrometry, protein complexes were isolated from

strains JM10 and JM151 using described methods27. Polypeptides were sepa-

rated by SDS–PAGE followed by Coomassie staining. The gel bands were

excised, diced and treated with 10 mM dithiothreitol in 0.1 M ammonium

bicarbonate for protein reduction. Free cysteine residues were alkylated with

freshly made 55 mM iodoacetamide in 0.1 M ammonium bicarbonate.

Tryptic digestion was started with the addition of 25 ng ml21 Sequence

Grade Modified Trypsin (Promega) in ammonium bicarbonate buffer. The

protein was digested for at least 16 h at 37 uC. The resultant tryptic peptide

mixture was separated by gradient elution with the Dionex nano-HPLC

system that was interfaced with a Thermo LTQ mass spectrometer (MS).

Automated data-dependent acquisition was used to acquire MS/MS data. The

acquired MS/MS spectra were converted to a MASCOT acceptable format

and searched by the Mascot search engine to identify proteins from primary

sequence databases.

Western blots were probed with antibodies anti-HA (F-7, Santa Cruz

Biotechnology, Inc.), anti-Myc (9E10, Santa Cruz Biotechnology, Inc.), anti-

GFP (generated as described27), anti-Mid1 (ref. 29), anti-Cdc2 (phospho-

Tyr15) (9111, Cell Signaling Technology), anti-PSTAIRE (sc-53, Santa Cruz

Biotechnology, Inc.), and anti-Pom1 (gift from J. Bahler). Quantification of

Pom1 levels is shown in Supplementary Fig. 11. Band intensities of scanned blots

were quantified using ImageJ. The integrated intensity of a fixed area was mea-

sured, and background levels were subtracted.

29. Celton-Morizur, S., Bordes, N., Fraisier, V., Tran, P. T. & Paoletti, A. C-terminalanchoring of mid1p to membranes stabilizes cytokinetic ring position in earlymitosis in fission yeast. Mol. Cell. Biol. 24, 10621–10635 (2004).

doi:10.1038/nature08074

Macmillan Publishers Limited. All rights reserved©2009