..............................................................

The transcription factor Eyes absentis a protein tyrosine phosphataseTina L. Tootle1,2*, Serena J. Silver1,2*, Erin L. Davies1*†,Victoria Newman1, Robert R. Latek1, Ishara A. Mills1,2,Jeremy D. Selengut3, Beth E. W. Parlikar1,2 & Ilaria Rebay1,2

1Whitehead Institute for Biomedical Research, Cambridge, Massachusetts 02142,USA2Department of Biology, Massachusetts Institute of Technology, Cambridge,Massachusetts 02142, USA3The Institute for Genomic Research, Rockville, Maryland 20850, USA

* These authors contributed equally to this work

† Present address: Department of Developmental Biology, Stanford University School of Medicine,

Stanford, California 94305, USA

.............................................................................................................................................................................

Post-translational modifications provide sensitive and flexiblemechanisms to dynamically modulate protein function inresponse to specific signalling inputs1. In the case of transcriptionfactors, changes in phosphorylation state can influence proteinstability, conformation, subcellular localization, cofactor inter-actions, transactivation potential and transcriptional output1.Here we show that the evolutionarily conserved transcriptionfactor Eyes absent (Eya)2,3 belongs to the phosphatase subgroupof the haloacid dehalogenase (HAD) superfamily4,5, and proposea function for it as a non-thiol-based protein tyrosine phospha-tase. Experiments performed in cultured Drosophila cells andin vitro indicate that Eyes absent has intrinsic protein tyrosinephosphatase activity and can autocatalytically dephosphorylateitself. Confirming the biological significance of this function,mutations that disrupt the phosphatase active site severelycompromise the ability of Eyes absent to promote eye specifica-tion and development in Drosophila. Given the functionalimportance of phosphorylation-dependent modulation of tran-scription factor activity, this evidence for a nuclear transcrip-tional coactivator with intrinsic phosphatase activity suggests anunanticipated method of fine-tuning transcriptional regulation.

The transcriptional coactivator Eyes absent (Eya) is a member ofan evolutionarily conserved set of nuclear transcription factors andcofactors collectively termed the retinal determination (RD) genenetwork2,3,6. Although RD network members are perhaps bestknown for their roles in eye specification, redeployment of thesegenes, either individually or as a network, contributes to a diversearray of essential developmental processes in all metazoans2,3. Eyafamily members are defined by a conserved ,275-amino-acidmotif, referred to as the Eya domain (ED), that has been shownto bind two other RD members, Sine oculis (So) and Dachshund(Dac)6–8. Together, Eya and So form a potent transcriptionalactivator9; the mechanistic implications of Eya–Dac interactionsare less clear10,11. Emphasizing the functional conservation amongEya homologues, mammalian Eya transgenes can rescue the ‘eyeless’phenotype of Drosophila eya mutations6,12.

We have examined a function for the carboxy-terminal ED of Eyathat was suggested by protein motif searches and structural model-ling studies. These investigations place Eya within the phosphatasesubgroup of the haloacid dehalogenase (HAD) superfamily (Fig. 1aand Supplementary Fig. S1a). HAD family members, found inorganisms ranging from bacteria and archaea to humans, constitutea diverse collection of enzymes that includes dehalogenases,ATPases, phosphonatases, phosphomutases, epoxy hydrolases anda growing number of magnesium-dependent phosphatases4,5,13.Understanding of the in vivo function of HAD family phosphatasesremains extremely limited, particularly in eukaryotic systems.

X-ray crystallography combined with mutagenesis studies ofseveral HAD family proteins has revealed a conserveda/b-hydrolase

fold that unites three non-contiguous sequence motifs to form thecatalytic core of the enzyme13–15. Five conserved residues broughttogether by this tripartite configuration have been shown to beessential for catalysis13–15. Structural modelling studies predict thatthe ED forms a HAD a/b-hydrolase-like fold (Fig. 1b). The catalyticresidues are strikingly conserved in the EDof all Eya proteins (Fig. 1aand Supplementary Fig. S1b). In motif 1 (DXDX(T/V)), theinvariant first aspartic acid serves as the nucleophile in all HADfamily proteins and probably forms a phospho-aspartate intermedi-ate16,17. The second aspartic acid distinguishes the phosphatase/phosphohydrolase subgroup from other branches of the HADsuperfamily4,5,15 and is strictly conserved in all Eya homologues.Motif 2 contains an essential serine/threonine at the end of theb-strand, and motif 3 contributes at least three required residues, alysine and two aspartic acids, the second of which has undergone aconservative substitution to glutamic acid in Eya proteins. Require-ment for the two acidic residues within motif 3 seems to be strictestwithin the phosphatase/phosphohydrolase branch of the HADsuperfamily5. The high degree of conservation of this catalyticquintet (D, S/T, K, D, E) in invertebrate, vertebrate and plant Eyahomologues suggests that Eya belongs to the phosphatase subgroupof the HAD superfamily.To investigate whether Eya has intrinsic phosphatase activity, we

tested the ability of recombinant murine glutathione S-transferase(GST)-tagged ED fusion proteins to dephosphorylate the syntheticsubstrate p-nitrophenyl phosphate (pNPP). We found that Eya canfunction as a phosphatase (Fig. 2), and that mutations altering thepresumptive HAD active-site residues severely compromise activity(Fig. 2a; see Supplementary Information for details). Sensitivity tothe tyrosine phosphatase inhibitor vanadate, but not to inhibitors ofserine/threonine phosphatases (Fig. 2a, b), and a requirement for

Figure 1 Eya is a member of the phosphatase subgroup of the HAD superfamily. a, The

non-contiguous sequences comprising HADmotifs 1, 2 and 3. Dm, D. melanogaster; Mm,

Mus musculus; Hs, Homo sapiens; Dr, Danio rerio; At, Arabidopsis thaliana; Ce,

Caenorhabditis elegans; Bc, Bacillus cereus. Pink residues define the HAD motif; those

mutated in this study are marked with an asterisk. Blue residues are most strongly

conserved among the phosphatase subgroup of the HAD superfamily; green residues are

highly conserved in both ATPases and phosphatases13,28. b, Structural modelling studies

predict a similar active-site configuration for Drosophila Eya (DmEya) and other HAD

proteins. The HAD template backbone is identified with a white ribbon and the Eya model

backbone is rendered with a cyan ribbon. Key active-site residues are highlighted as

sticks, either white for the HAD or yellow for Eya. c, Superimposition of mutant DmEyaHAD

residues on the DmEya model. Alignment of the substitutions (in magenta) and their

wild-type counterparts (in yellow) is shown.

letters to nature

NATURE |VOL 426 | 20 NOVEMBER2003 | www.nature.com/nature 299© 2003 Nature Publishing Group

magnesium are consistent with Eya being a HAD family phospha-tase18. We also tested recombinant Drosophila Eya in these assays,and although its activity was significantly lower, it was also mag-nesium dependent and vanadate sensitive (Fig. 2c). The most likelyexplanation for the weak in vitro activity of the fly ED is that we havenot identified appropriate conditions for purifying correctly folded,fully active protein, although we cannot rule out the possibility thatDrosophila Eya, despite retaining all the conserved residues com-prising the HADmotif (Fig. 1), may have limited ability to functionas a phosphatase. However, the fact that themouse Eya isoform usedin our in vitro assays is able to substitute forDrosophila Eya in vivo12,together with the results of the in vivo experiments described below,leads us to propose that Eya proteins have a conserved phosphatasefunction.To investigate whether Eya has protein phosphatase activity, a

role that has not been definitively demonstrated for any other HADfamily protein15, we tested several phosphotyrosine- or phospho-threonine-containing synthetic peptide substrates. Eya showedrobust activity toward one of the tyrosyl phosphorylated peptides,with a Km significantly lower than that measured using the pNPPsubstrate (Fig. 2d). No measurable activity was detected with thephosphothreonine- or other phosphotyrosine-containing peptides(data not shown; see Methods for details). These results show thatEya has protein tyrosine phosphatase (PTP) activity, but they do notrule out the possibility that Eya could dephosphorylate othersubstrates as well. The fact that not all tyrosyl phosphorylatedpeptides were hydrolysed suggests that Eya has specific sequencepreferences with respect to its putative protein substrates. BecauseHAD family phosphatases use a catalytic aspartate16,17 as thenucleophile rather than the cysteine residue used by standardPTPs19, these results indicate that Eya is the founding member of

a new class of non-thiol-based PTPs.We used the genetically tractable Drosophila system to investigate

the physiological relevance of the putative PTP activity of Eya. Site-directed mutagenesis was used to generate five single and fourdouble mutant combinations of the five HAD active-site residues inDrosophila Eya (Fig. 1c); we refer to them collectively as the EyaHAD

mutants. These mutants were first transfected into Drosophila S2cultured cells. Immunostaining and western blotting analysesrevealed no apparent changes in subcellular localization (data notshown) or expression levels (Supplementary Fig. S2a) of themutants relative to EyaWT.

Eya, like most RD genes, induces formation of eye tissue outsidethe normal eye field when ectopically expressed2,3,6,20. Scoring thepercentage of flies exhibiting ectopic eye formation thereforeprovides a sensitive measurement of Eya activity20. To determinewhether the HAD active-site mutants compromise the ectopic-eye-induction potential of Eya, we generated transgenic lines carryingfull-length EyaHAD mutant expression constructs. All the EyaHAD

mutants showedmarkedly reduced ectopic eye induction relative toEyaWT (Fig. 3a). Protein expression levels from the EyaHAD trans-genes were comparable to those from EyaWT lines (SupplementaryFig. S2b), indicating that the reduction in ectopic-eye-inducingpotential reflects a change in protein activity rather than reducedexpression. Comparable reductions in Eya activity were alsoobserved with EyaHAD transgenes in which two of the five HADactive-site residues were mutated simultaneously (data not shown).

Because theHADmotif active-sitemutants compromise the ability

Figure 2 Eya exhibits phosphatase activity in vitro. a, Kinetic analysis of murine Eya3

GST–ED fusion proteins (GST–MmEya) using the pNPP substrate. The D246N, T420A,

K449Q, D474N and E478Q mutations are analogous to the D493N, S670A, K699Q,

D724N and E728Q mutations described for DmEya. For those enzymes whose activity

was too low to be measured, s indicates a K m significantly higher and an efficiency

(k cat/K m) significantly lower than that measured for D246N. b, In vitro phosphatase

activity of MmEya. c, In vitro phosphatase activity of DmEYA. d, Kinetic analysis of

GSTMmEya using the tyrosyl phosphorylated peptide substrate I(pY)GEF.

Figure 3 EyaHAD mutants have severely reduced activity relative to EyaWT in ectopic-eye-

induction and genetic rescue assays. a, The frequency of ectopic eye induction associated

with expression of Eya transgenes calculated from multiple independent transgenic lines:

EyaWT, 2,465 flies from 8 lines20; EyaD493N, 1,502 flies from 5 lines; EyaS670A, 955 flies

from 3 lines; EyaK699Q, 953 flies from 3 lines; EyaD724N, 265 flies from a single line;

EyaE728Q, 1,239 flies from 4 lines. b, The percentage of eyes from flies of the genotype

eya 2;UAS–eya/dpp–GAL4 exhibiting rescue of the eya 2 ‘eyeless’ phenotype (black bars)

and average size of the rescued tissue relative to a wild-type eye (grey bars) are plotted.

The following transgenic lines were used: EyaWT, 155 flies from two independent lines;

EyaD493N, 124 flies from a single line; EyaS670A, 281 flies from a single line; EyaK699Q, 176

flies from a single line; EyaD493NþS670A, 209 flies from two independent lines;

EyaD493NþD724N, 151 flies from two independent lines. c–f, Scanning electron

micrographs of adult eyes. c, w 1118. d, eya 2. e, eya 2;UAS–eyaWT/dpp–GAL4.

f, eya 2;UAS–eyaHAD/dpp–GAL4; arrow points to a small patch of rescued eye tissue.

letters to nature

NATURE | VOL 426 | 20 NOVEMBER2003 | www.nature.com/nature300 © 2003 Nature Publishing Group

of Eya to induce ectopic eye formation, we asked whether an intactHAD motif is required for normal Eya function during eye develop-ment.We compared the ability of EyaWT versus EyaHAD transgenes tocomplement the eye-specific loss-of-function eya2 allele. Homo-zygous eya2 mutant flies exhibit a completely penetrant ‘eyeless’phenotype, in which the entire eye is missing (Fig. 3c, d). For theseexperiments we define rescue as the ability of a given transgene toproduce recognizable eye tissue within the normal eye field of anadult fly. We also can compare the extent of rescue by estimating thesize of the rescued eye tissue relative to that of a wild-type eye.

Expression of EyaWT transgenes rescues the eya2 ‘eyeless’ pheno-type with complete penetrance in both eyes of each individual fly(Fig. 3b, e; data not shown). In striking contrast, all EyaHAD mutanttransgenes exhibited a significantly reduced frequency and extent ofrescue, with rescue usually occurring in only one of the two eye fieldsof an individual (Fig. 3b, f). For all EyaHAD transgenes tested, evenin cases where rescue efficiency was only twofold to threefold lowerthan that of EyaWT, the size of the rescued eye tissue was alwayssignificantly (fivefold to tenfold) reduced relative to that obtainedwith EyaWT lines (Fig. 3b, e, f).Western blot analyses of eye imaginaldiscs ruled out the possibility that reduced protein expressionmightbe responsible for this result (Supplementary Fig. S2c). In combi-nationwith the ectopic-eye-induction assay data, the results of theserescue experiments argue strongly that the activity of Eya as aputative HAD family phosphatase is required to promote normaleye development in Drosophila.

Because the region of the ED that binds to the RD gene networkprotein So7,12 partially overlaps with motif 1 of the HAD domain(Supplementary Fig. S1b), we checked whether the EyaHAD mis-sense mutations compromise the ability of Eya to interact produc-tively with So. Eya and So interact to form a potent transcriptionalactivator required for eye specification, in which So contributes theDNA-binding domain and Eya provides the transactivation poten-tial7,9,11. Using a transcription assay in Drosophila S2 cultured cells9,we found that the ability of EyaHAD mutant proteins to synergizewith So to activate transcription of a reporter gene is comparable tothat of EyaWT (Fig. 4 and Supplementary Fig. S3). Although wecannot rule out the formal possibility that in vivo the EyaHAD

mutations disrupt interactions with other proteins rather thanblock phosphatase activity, the finding that disruption of theHAD motif active site does not abrogate the ability of Eya tofunction as a transcriptional coactivator in conjunction with Soleads us to propose that Eya proteins have two essential functions: apreviously described role as a transcription factor and a role as aprotein tyrosine phosphatase.

To investigate the intrinsic PTP activity of Eya on physiologically

relevant substrate candidates, we exploited the finding that Eya canbe tyrosine phosphorylated in Drosophila S2 cells (Fig. 5a; seeSupplementary Information for discussion). This allowed us toaffinity purify full-length Eya from these cells and to use it as aprotein substrate in an in vitro phosphatase reaction. Because thephosphotyrosine signal associated with the EyaHADmutant proteinswas consistently elevated relative to EyaWT (Fig. 5a; see Supplemen-tary Information for discussion), EyaHAD protein was used as thesubstrate. We found that incubation of EyaHAD protein withrecombinant murine GST–ED protein strongly reduced the phos-photyrosine signal (Fig. 5b). HAD active-site mutants that exhibitimpaired activity both in vitro and in vivo (Figs 2 and 3) also haveseverely reduced activity in this assay (Fig. 5b). These results showthat Eya can function as a PTP with a full-length endogenousprotein substrate and that such activity depends on an intact HADmotif. Although we do not yet understand the physiologicalrelevance of tyrosine phosphorylation and dephosphorylation ofEya, these results (Fig. 5), together with the previous demonstrationthat Eya is able to self-associate9, suggest that Eya may autocataly-tically dephosphorylate itself.In conclusion, we propose that Eya is the founding member of a

novel class of non-thiol-based PTPs and is, to our knowledge, thefirst example of a transcription factor with intrinsic phosphataseactivity. Further work will be required to understand how tyrosinephosphorylation and dephosphorylation regulate Eya functionin vivo, and what substrates, potentially including Eya itself, maybe regulated by its PTP activity. Elucidation of the biochemicalregulatory mechanisms that coordinate the dual functions of Eya astransactivator and phosphatase during eye specificationwill provideinsights into the function of the RD gene network, and moregenerally establish a new paradigm for transcriptional regulatorystrategies. Although preliminary analyses have not identified otherHAD-motif-containing proteins that are annotated as transcrip-tional regulators (R.R.L. and I.R., unpublished observation), itseems likely that dual-function mechanisms analogous to thatwhich we propose for Eya will prove to be a general strategy forfine-tuning transcriptional output, particularly in highly regulateddevelopmental systems. A

Figure 4 EyaHAD mutations do not disrupt the role of Eya as a transcriptional coactivator

in conjunction with So. The Drosophila cell-culture-based transcription assays were

performed as described9. Lanes: 1, Are-luciferase; 2, EyaWT; 3, EyaD493N; 4, EyaS670A;

5, EyaK699Q; 6, EyaD724N; 7, EyaE728Q; 8, EyaD493NþS670A; 9, EyaD493NþK699Q; 10,

EyaD493NþD724N; 11, EyaD493NþE728Q. See Supplementary Fig. S3 for further details.

Figure 5 Eya has protein tyrosine phosphatase activity. Top panels show immunoblots

probed with anti-phosphotyrosine antibody (anti-P-Y); bottom panels show immunoblots

of the same samples probed with anti-Flag antibody to detect Eya (anti-Eya). a, EyaHAD

proteins exhibit higher levels of tyrosine phosphorylation than EyaWT. Lanes: 1 and 2,

independent transfections of EyaWT; 3, EyaD493NþS670A; 4, EyaD493NþK699Q;

5, EyaD493NþD724N; 6, EyaD493NþE728Q. Fold increase calculated as signal for the EyaHAD

mutant P-Y level relative to the average of the P-Y signal in the EyaWT lanes, and corrected

relative to the strength of the anti-EYA signal. b, Dephosphorylation of DmEya by

recombinant GST–ED. Full-length tyrosine phosphorylated DmEyaD493NþD724N (all lanes)

was immunoprecipitated and incubated with recombinant murine GST–ED, either wild

type (WT) or HAD mutant variants. Lanes: 1 and 2, control; 3 and 4, EyaWT; 5, EyaD246N;

6, EyaT420A; 7, EyaK449Q; 8, EyaD474N; 9, EyaE478Q. The percentage of anti-P-Y signal for

EyaD493NþD724N relative to controls and corrected for relative protein levels is indicated.

Numbers shown are an average from two independent experiments for each GST–ED

tested; results from only one of the two experiments are shown for the GST–ED HAD

mutants.

letters to nature

NATURE |VOL 426 | 20 NOVEMBER2003 | www.nature.com/nature 301© 2003 Nature Publishing Group

MethodsBioinformaticsFor a description of the computational analyses, see Supplementary Information.

Phosphatase assaysPhosphatase assays were performed using GST–ED fusion proteins (purified as describedin Supplementary Information). Assays to measure the enzyme kinetics using thesynthetic substrate p-nitrophenyl phosphate (Sigma) were done in triplicate with sixsubstrate concentrations over six time points. Reactions (80 ml) were performed at 30 8C in200mM PIPES, pH 7.0, 5mM EDTA and 10mM MgCl2, and were quenched by theaddition of 40ml of 10MNaOH. PNPanionwas detected at 405 nm (extinction coefficiente ¼ 1.78 £ 104 lmol21 cm21) using a Tecan GENios plate reader. Reactions werenormalized to buffer-alone controls and the results analysed by Lineweaver–Burk plotusing Microsoft Excel. Synthetic peptide substrates tested were: I(pY)GEF andTSTGPE(pY)EPGENL (Calbiochem); END(pY)INASL, DADE(pY)LIPQQG andRRA(pT)VA (Promega). Reactions (50 ml) were performed at 25 8C in 200mM HEPES,pH 7.0, 10mMMgCl2 and 5mM EDTA, and were quenched with 50 ml of molybdate dyesolution (Promega). Malachite green–ammonium molybdate phosphate complex wasdetected at 595 nm and converted to moles of free phosphate using a phosphateconcentration standard curve. Assays with I(pY)GEF were carried out at five substrateconcentrations over five time points and the results were analysed as described for pNPP.Phosphatase inhibitor cocktail sets I and II (Calbiochem; see Supplementary Informationfor details) were used at a 1:50 dilution in pNPP phosphatase assays. Sodiumorthovanadate was used at 4mM final concentration in pNPP phosphatase assays.

Amino-terminally Flag-epitope-tagged Eya constructs were subcloned into the copperinducible metallothionein promoter vector. Each construct (5 mg of DNA) was transfectedinto S2 cells as previously described21. Following published protocols22–26, cells weretreated with 100 mM NaVO3 and 200 mM H2O2 for 15min before lysis in 100mM NaCl,50mM Tris-HCl, pH 7.5, 2mM EDTA, 2mM EGTA, 1% NP-40, 1mM Na3VO4 and onemini-complete protease inhibitor tablet (Roche) per 10ml. All subsequent solutionsinclude 1mMNa3VO4. Clarified lysates were incubated with 25 ml of anti-Flag M2 agaroseaffinity gel (Sigma) for 1.5 h at 4 8C. Beads were washed twice in lysis buffer and twice in10mM Tris-HCl, pH 7.5, and 150mM NaCl, resuspended in 30 ml of £2 SDS samplebuffer and boiled, and 10 ml were loaded per lane. Western blots were performed asdescribed27 except that blocking and antibody incubations were performed in 1% caseinaccording to Hammarsten (EM Science). Antibodies: guinea-pig anti-Eya 1:16,000; rabbitanti-phosphotyrosine 1:400 (0.21mg21ml21; Upstate); HRP-conjugated goat anti-guinea-pig and anti-rabbit 1:5,000 (Jackson ImmunoResearch). Determination of the foldincrease in phosphotyrosine signal relative to Eya protein amounts was performed usingNIH Image software; samples analysed in this way were run together on the same gel.

To obtain sufficient tyrosine phosphorylatedDrosophila Eya to use as a substrate in thein vitro phosphatase assay, a stable cell line expressing Flag-tagged EyaD493NþD724N wasgenerated. Cells (500 ml) were immunoprecipitated as described above, except that 1mMNa3VO4 was omitted from the wash buffer. The washed immunoprecipitates wereincubated in phosphatase assay reaction buffer (as described above but without pNPP),either with GST agarose or with 100mg GST–ED proteins for 1 h at 30 8C, processed forwestern blotting and analysed as described above.

Molecular biology and geneticsSite-directed mutagenesis, subcloning, generation of transgenics, crosses, ectopic eyescoring, calculation of per cent ectopic eye induction and scanning electron microscopywere performed as previously described20,21. Fly crosses were performed at 25 8C with theexception of the genetic rescue assays, which were performed at 20 8C.

Received 3 June; accepted 22 September 2003; doi:10.1038/nature02097.

1. Hunter, T. Signaling—2000 and beyond. Cell 100, 113–127 (2000).

2. Treisman, J. E. A conserved blueprint for the eye? Bioessays 21, 843–850 (1999).

3. Wawersik, S &Maas, R. L. Vertebrate eye development as modeled inDrosophila.Hum. Mol. Genet. 9,

917–925 (2000).

4. Collet, J. F., van Schaftingen, E. & Stroobant, V. A new family of phosphotransferases related to P-type

ATPases. Trends Biochem. Sci. 23, 284 (1998).

5. Thaller,M. C., Schippa, S. & Rossolini, G.M. Conserved sequencemotifs among bacterial, eukaryotic,

and archaeal phosphatases that define a new phosphohydrolase superfamily. Protein Sci. 7, 1647–1652

(1998).

6. Bonini, N. M., Bui, Q. T., Gray-Board, G. L. & Warrick, J. M. The Drosophila eyes absent gene directs

ectopic eye formation in a pathway conserved between flies and vertebrates. Development 124,

4819–4826 (1997).

7. Pignoni, F. et al.The eye-specification proteins So and Eya form a complex and regulatemultiple steps

in Drosophila eye development. Cell 91, 881–891 (1997).

8. Chen, R., Amoui, M., Zhang, Z. & Mardon, G. Dachshund and Eyes absent proteins form a complex

and function synergistically to induce ectopic eye development inDrosophila.Cell 91, 893–903 (1997).

9. Silver, S. J., Davies, E. L., Doyon, L. & Rebay, I. A functional dissection of Eyes absent reveals newmodes

of regulation within the retinal determination gene network.Mol. Cell. Biol. 23, 5989–5999 (2003).

10. Kim, S. S. et al. Structure of the retinal determination protein Dachshund reveals a DNA binding

motif. Structure (Camb.) 10, 787–795 (2002).

11. Ikeda, K., Watanabe, Y., Ohto, H. & Kawakami, K. Molecular interaction and synergistic activation of

a promoter by Six, Eya, and Dach proteins mediated through CREB binding protein.Mol. Cell. Biol.

22, 6759–6766 (2002).

12. Bui, Q. T., Zimmerman, J. E., Liu, H. & Bonini, N. M. Molecular analysis of Drosophila eyes absent

mutants reveals features of the conserved Eya domain. Genetics 155, 709–720 (2000).

13. Collet, J. F., Stroobant, V. &Van Schaftingen, E.Mechanistic studies of phosphoserine phosphatase, an

enzyme related to P-type ATPases. J. Biol. Chem. 274, 33985–33990 (1999).

14. Aravind, L., Galperin, M. Y. & Koonin, E. V. The catalytic domain of the P-type ATPase has the

haloacid dehalogenase fold. Trends Biochem. Sci. 23, 127–129 (1998).

15. Selengut, J. D. Mdp-1 is a new and distinct member of the haloacid dehalogenase family of aspartate-

dependent phosphohydrolases. Biochemistry 40, 12704–12711 (2001).

16. Ridder, I. & Dijkstra, B. Identification of theMg2þ -binding site in the P-type ATPase and phosphatase

members of the HAD (haloacid dehalogenase) superfamily by structural similarity to the response

regulator protein CheY. Biochem. J. 339, 223–226 (1999).

17. Cho, H. et al. Beryllium fluoride acts as a phosphate analog in proteins phosphorylated on aspartate:

structure of a beryllium fluoride complex with phosphoserine phosphatase. Proc. Natl Acad. Sci. USA

98, 8525–8530 (2001).

18. Selengut, J. D. & Levine, R. L. MDP-1: A novel eukaryotic magnesium-dependent phosphatase.

Biochemistry 39, 8315–8324 (2000).

19. Andersen, J. N. et al. Structural and evolutionary relationships among protein tyrosine phosphatase

domains. Mol. Cell. Biol. 21, 7117–7136 (2001).

20. Hsiao, F. C., Williams, A., Davies, E. L. & Rebay, I. Eyes absent mediates cross-talk between retinal

determination genes and the receptor tyrosine kinase signaling pathway. Dev. Cell 1, 51–61 (2001).

21. Tootle, T. L., Lee, P. S. & Rebay, I. CRM1-mediated nuclear export and regulated activity of the

receptor tyrosine kinase antagonist YAN require specific interactions with MAE. Development 130,

845–857 (2003).

22. Cohen, J., Altaratz, H., Zick, Y., Klingmuller, U. & Neumann, D. Phosphorylation of erythropoietin

receptors in the endoplasmic reticulum by pervanadate-mediated inhibition of tyrosine

phosphatases. Biochem. J. 327, 391–397 (1997).

23. Huyer, G. et al. Mechanism of inhibition of protein-tyrosine phosphatases by vanadate and

pervanadate. J. Biol. Chem. 272, 843–851 (1997).

24. Imbert, V. et al. Induction of tyrosine phosphorylation and T-cell activation by vanadate peroxide, an

inhibitor of protein tyrosine phosphatases. Biochem. J. 297, 163–173 (1994).

25. Ruff, S. J., Chen, K. & Cohen, S. Peroxovanadate induces tyrosine phosphorylation of multiple

signaling proteins in mouse liver and kidney. J. Biol. Chem. 272, 1263–1267 (1997).

26. Scanga, S. E. et al. The conserved PI3K/PTEN/Akt signaling pathway regulates both cell size and

survival in Drosophila. Oncogene 19, 3971–3977 (2000).

27. O’Neill, E. M., Rebay, I., Tjian, R. & Rubin, G.M. The activities of two Ets-related transcription factors

required forDrosophila eye development are modulated by the Ras/MAPK pathway. Cell 78, 137–147

(1994).

28. Aravind, L. & Koonin, E. V. The HD domain defines a new superfamily of metal-dependent

phosphohydrolases. Trends Biochem. Sci. 23, 469–472 (1998).

Supplementary Information accompanies the paper on www.nature.com/nature.

Acknowledgements We thank M. Voas, D. Doroquez, L. Doyon and B. Chaffee for technical

assistance, J. Flynn, S. Neher and S. Flaugh for advice, reagents and technical help, D. Maas for

murine Eya complementary DNAs, the Rebay lab for advice and discussions, F. Pignoni for the

eya2 stocks used in the rescue assay, and R. Hegde for sharing unpublished information. We are

grateful for advice, comments and encouragement from R. Fehon, G. Fink, T. Orr-Weaver,

S. Shenolikar, F. Solomon and J. York. T.L.T., S.J.S., I.A.M. and B.E.W.P. are supported by the

Ludwig Foundation, the Howard Hughes Medical Institute, the Packard Foundation and the

National Science Foundation, respectively. I.R. is supported by the National Eye Institute and the

Burroughs Wellcome Fund.

Competing interests statement The authors declare that they have no competing financial

interests.

Correspondence and requests for materials should be addressed to I.R. (rebay@wi.mit.edu).

..............................................................

FKF1 is essential forphotoperiodic-specificlight signalling in ArabidopsisTakato Imaizumi1, Hien G. Tran1, Trevor E. Swartz2, Winslow R. Briggs2

& Steve A. Kay1

1Department of Cell Biology, The Scripps Research Institute, 10550 North TorreyPines Road, La Jolla, California 92037, USA2Department of Plant Biology, Carnegie Institution of Washington, 260 PanamaStreet, Stanford, California 94305, USA.............................................................................................................................................................................

Adaptation to seasonal change is a crucial component of anorganism’s survival strategy. To monitor seasonal variation,organisms have developed the capacity to measure day length(photoperiodism). Day-length assessment involves the photo-periodic control of flowering in Arabidopsis thaliana, wherebythe coincidence of light and high expression of CONSTANS (CO)

letters to nature

NATURE | VOL 426 | 20 NOVEMBER2003 | www.nature.com/nature302 © 2003 Nature Publishing Group