letter - Salk Institute

9 downloads 0 Views 3MB Size Report
precedes the diminution of FGF4 and FGF8 transcripts indicates that the arrest in limb outgrowth observed after RLIM overex- pression is due primarily to ...
letter

© 1999 Nature America Inc. • http://genetics.nature.com

RLIM inhibits functional activity of LIM homeodomain transcription factors via recruitment of the histone deacetylase complex

LIM domains1 are required for both inhibitory effects on LIM homeodomain transcription factors and synergistic transcriptional activation events1–4. The inhibitory actions of the LIM domain can often be overcome by the LIM co-regulator known as CLIM2, LDB1 and NLI (referred to hereafter as CLIM2; refs 2–4). The association of the CLIM cofactors with LIM domains does not, however, improve the DNA-binding ability of LIM homeodomain proteins4,5, suggesting the action of a LIM-associated inhibitor factor. Here we present evidence that LIM domains are capable of binding a novel RING-H2 zinc-finger protein, Rlim (for RING finger LIM

aa m Rlim c RLIM

m Rlim c RLIM

. . . . . . . . MENSDSNDKGS-DQSAAQRRSQMDRLDREEAFYQFVNNLSEEDYRLMRDNNLLGTPGESTEEELLRRLQQIKEGPPPQSPDENRAGESSD ||.|||.|||. |||.|||:||.|||||||||||||||||||||||||||||||||||.|||||||||:|:|||||.|..||||::||:: MESSDSSDKGNIDQSEAQRQSQLDRLDREEAFYQFVNNLSEEDYRLMRDNNLLGTPGEITEEELLRRLHQVKEGPPQQNSDENRGAESTE . . . . . . . . DVTNSDSIIDWLNSVRQTGNTTRSGQRGNQSWRAVSRTNPNSGDFRFSLEINVNRNNGSQTSENESEPSTRRLSVENMESSSQRQMENSA ||.|:|||||||||||||||||||||||||||||||||||||||||||||||||||||. ..|.|.|||... :.|:.|. || | . DVSNGDSIIDWLNSVRQTGNTTRSGQRGNQSWRAVSRTNPNSGDFRFSLEINVNRNNGNTNPETENEPSAEPSGGEDLEN-SQSDSEIPR

domain-binding protein), which acts as a negative co-regulator via the recruitment of the Sin3A/histone deacetylase corepressor complex. A corepressor function of RLIM is also suggested by in vivo studies of chick wing development. Overexpression of the gene Rnf12, encoding Rlim, results in phenotypes similar to those observed after inhibition of the LIM homeodomain factor LHX2, which is required for the formation of distal structures along the proximodistal axis, or by overexpression of dominant-negative CLIM1. We conclude that Rlim is a novel corepressor that recruits histone deacetylase-containing complexes to the LIM domain.

89

cc

dd

α-Clim α-Lhx3 Lhx3 Clim1 Rlim*

90

179

– + – – +

– + + – +

+ – – + +

+ αHA Rlim (HA) Lhx3 LIM (myc) Lmo2 (myc)

+ – – –

+ + –

+ + – +

179

Rlim

bb

consensus

C

C H

H

C

523

ff

GST Zyx LIM

GST Pax LIM

GST CRPLIM

+ + +

input 20% GST GST Isl1 LIM GST Lhx2

GST Lhx3 ∆LIM

GST Lhx3 LIM

GST Lhx3

+ + –

gg

Clim1

600 593

C

– + –

53O

ALKTCSVCITEYTE----GNKLRKLPCSHEFHVHCIDRWLSE------NSTCPICRRAVL EYDVCAICLDEYED----GDKLRILPCSHAYHCKCVDPWLTKT-----KKTCPVCKQKVV DSDCCAICIEAYKP----TDTIRILPCKHEFHKNCIDPWLIE------HRTCPMCKLDVL EDATCAICLDNLQN(20)GTTVIVMPCKHRFHYFCLTLWLEA------QQTCPTCRQKVK MELTCGLCGESIGDQ---NSQLQALPCSHLFHLKCLQTDG--------NRGCPNCKRSSV IGEKCLICEESISSTFT-GEKVVESTCSHTSHYNCYLMLFETLYFQGKFPECKICGEKSK PGKSCDECGKFLQ-----IKKFIVFPCGHCFHWNCIIRVIL(27)NIIVEKCGLCSDINI KNQTCFMCRLTL------DIPVVFFKCGHIYHQHCLNEEEDTLESERKLFKCPKCLVDLE C

Clim1 Lhx3 Rlim

Rlim

448

RING-H2 finger Rlim CRZF goliath CELG PSMP FAR1 PEP3 PEP5

GST Lmk LIM

c RLIM

GST Lmo2

m Rlim

. . . RING-H2 finger . . NLAMRSFGENDALKTCSVCITEYTEGNKLRKLPCSHEFHVHCIDRWLSENSTCPICRRAVLSSGNRESVV |||||.|||||||||||||||||||||||||||||||:|||||||||||||||||||||||.|||||||| NLAMRNFGETDALKTCSVCITEYTEGNKLRKLPCSHEYHVHCIDRWLSENSTCPICRRAVLASGNRESVV

441

input 10%

c RLIM

. . . . . . . . SGGGNSSGSSSSSSPSPSSSGESSESSSKMFEGSSEGGSSGPSRRDGRH-RAPVTFDESGSLPFLSLAQFFLLNEDDEDQPRGLTKEQID :|||||. | |||| |:|. ||||:::.||:||: |:.|||:||||||||||||||||||||:|||||||||||| -----GSGSSSNESTD-VSSGEVFEGSN-------EGGSTSGARREGRNTRGSVTFEESGSLPFLSLAQFFLLNEDDDDQPRGLTKEQID

359

Rlim

GST

m Rlim

351

Rlim

GST Rlim (403-600)

c RLIM

Rlim

GST Rlim (208-423)

m Rlim

. . . . . . . . ERGGFRRTFSRSERAGVRTYVSTIRIPIRRILNTGLSETTSVAIQTMLRQIMYGFGELSYFMYSDSDSEPSASVSSRNVERVESRNGRGS |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||.:||:. ...||:. |..|| ERGGFRRTFSRSERAGVRTYVSTIRIPIRRILNTGLSETTSVAIQTMLRQIMYGFGELSYFMYSDSDADPSGPTPNQNVDASEPQNG---

ee input 20%

c RLIM

269

GST Rlim (1-207)

m Rlim

. . . . . . . . LRQQISGPELL--GRGLFAASGSRNPSQGTSSSDTGSNSESSGSGQRPPTIVLDLQVRRVRPGEYRQRDSIASRTRSRSQAPNNTVTYES |||:. | |: .||.| : .|:.:||:|:: |||.:.||||||||||||||||||||||||||||.|||||||:||||||||| LRQHAVGTEIPSENAVLFSALETGPVPQAAGSSETNGASESAAPGQRPPTIVLDLQVRRVRPGEYRQRDSIANRTRSRSQTPNNTVTYES

GST

c RLIM

263

GST

m Rlim

. . . NLS . . . . . SESASARPSRAERNSTEAVTEVPTTRAQRRARSRSPEHRRTRARAERSVSPLQPTSEIPRRA------PTLEQSSENEPEGSSRTRHHVT |||:|.| . .||.::| :|| :. |:||||||||||:|||||| :|| ||:.|.||.|||. |.::|. ||:|||||||:||| SESPSVRQPGSERSTSEELTEEASPRGQRRARSRSPEQRRTRARTDRSRSPINPVSEAPRRSHHNTSSQTFDHSAVNEAEGSSRTRQHVT

input 10%

© 1999 Nature America Inc. • http://genetics.nature.com

Ingolf Bach1,7, Concepción Rodriguez-Esteban2, Catherine Carrière1, Anil Bhushan3, Anna Krones1, David W. Rose4,5, Christopher K. Glass6, Bogi Andersen5, Juan Carlos Izpisúa Belmonte2 & Michael G. Rosenfeld1

Lhx3/Clim1

Clim1 Lhx3

C

Fig. 1 Rlim, a novel LIM domain-binding RING zinc-finger protein. a, Amino acid sequences of mouse and chicken Rlim (m Rlim and c RLIM, respectively) are 75% identical. A putative nuclear localization signal (NLS) is shaded and the RING-H2 zinc finger is boxed. b, Sequence comparison of RING-H2 zinc-finger domains of Rlim, CRFZ, goliath, CELG, PSMP, FAR1, PEP3 and PEP5. c, Co-immunoprecipitation of full-length [35S]Met-labelled Rlim (asterisk) with bacterially expressed Lhx3 and Clim1 using anti-Lhx3 (αLhx3) and anti-Clim (αClim) antibodies. The first lane corresponds to 10% input. d, Co-immunoprecipitations of nuclear extracts from CV1 cells co-transfected with HA-tagged Rlim and Myc-tagged LIM domains of Lhx3 or Lmo2 expression plasmids with the anti-HA (αHA) antibody. The western blot was visualized using an anti-Myc antibody. e, GST protein interaction assays between [35S]-labelled Rlim protein and bacterially expressed GST fusion proteins: full-length Lhx3 (GST-Lhx3), Lhx3 mutant proteins containing only the LIM domain (GST-Lhx3LIM) or no LIM domain (GST-Lhx3∆LIM); Lhx2 (GST-Lhx2); Lmo2 (GST-Lmo2); and different LIM domains from Isl-1 (GST-Isl1LIM), LIM kinase 1 (GST-Lmk1LIM), CRP (GST-CRPLIM), paxillin (GST-PaxLIM) and zyxin (GST-ZyxLIM). f, GST protein interaction assays between [35S]labelled Clim1 and bacterially expressed regions of Rlim fused to GST. g, Inhibition by Rlim of Lhx3/Clim1 interactions in an EMSA. A [32P]-labelled Lhx3/Clim1/DNA complex supershift on an Lhx3 DNA binding site was inhibited by addition of 2×Rlim, and completely inhibited with 5×Rlim (data not shown).

1Howard Hughes Medical Institute, Eukaryotic Regulatory Biology Program, University of California, San Diego, School of Medicine, 9500 Gilman Drive, La Jolla, California 92093-0648, USA. 2Gene Expression Laboratory, 3The Clayton Foundation, The Salk Institute for Biological Sciences, 10010 North Torrey Pines Road, La Jolla, California 92186, USA. 4Whittier Diabetes Program, 5Division of Endocrinology and Metabolism, 6Division of Cellular and Molecular Medicine, University of California, San Diego, School of Medicine, 9500 Gilman Drive, La Jolla, California 92093-0648, USA. 7Center for Molecular Neurobiology, University of Hamburg, Martinistr. 85, 20246 Hamburg, Germany. Correspondence should be addressed to I.B. (e-mail: [email protected]), M.G.R. (e-mail: [email protected]) or J.C.I.B. ([email protected]).

394

nature genetics • volume 22 • august 1999

© 1999 Nature America Inc. • http://genetics.nature.com

© 1999 Nature America Inc. • http://genetics.nature.com

We screened a mouse embryonic pituitary expression library with the LIM domain of the LIM homeodomain factor Lhx3 (also known as PLim) and obtained cDNAs encoding a novel protein (which we called Rlim) that shares homology with RING-H2 finger proteins6, including the cytotactin/tenascininducible chicken RING finger protein CRZF (Fig. 1a,b; ref. 7). We confirmed by co-immunoprecipitation experiments that Rlim bound the LIM domains of both Lhx3 and Lmo2 (Fig. 1c,d). In GST protein interaction assays, Rlim bound to the LIM domains of Lhx3 (on either finger motif), Isl1, Lhx2, Lmo2 and the LIM kinase 1 (known as Lmk1 and Kiz1), but only weakly or not at all to the LIM domains of the cytoplasmic LIM proteins CRP, enigma, paxillin and zyxin (Fig. 1e, and data not shown). This binding specificity is similar to that observed for Clim2 (refs 3,4,8), suggesting that Rlim binds to most or all of the nuclear LIM proteins. Furthermore, Rlim interacted with the Clim cofactors in vitro (Fig. 1c), with two separable Clim interaction domains defined on Rlim (aa 1–207, 206–423; Fig. 1f). Addition of Rlim to Clim1 and Lhx3 inhibited Lhx3/Clim1/DNA ternary complex formation in electrophoretic mobility shift assays (EMSA; Fig. 1g). These data suggest that Rlim is capable of modulating Clim actions on LIM homeodomain factors at the level of Clim binding or at the level of its function. The highest levels of Ldb1 (encoding Clim2) and Ldb2 (encoding Clim1) mRNA expression overlapped with regions of expression of several LIM homeodomain proteins4,8,9. In situ hybridization revealed that Rnf12 (encoding Rlim) and Ldb1 mRNAs were ubiquitously expressed at embryonic day (E) 7.0 (data not shown) and E7.5, whereas Ldb2 transcripts were mainly detected in the neuroectoderm, with highest expression in the future headfold (Fig. 2a). Rnf12 and Ldb1 mRNA expression patterns were similar in all embryonic regions examined at E11.5, E15.5 and E17 (Fig. 2, and data not shown). Ldb2 was expressed in a subset of Ldb1- and Rnf12-expressing regions in the nervous system (Fig. 2b,c), whereas Ldb2 expression appeared highest in regions with lower Rnf12 expression in many non-neuronal tissues (that is, nasal epithelium (Fig. 2d), submandibular gland (Fig. 2e) and gut (Fig. 2f)), and lowest in regions of higher Rnf12 expression (for example skin; Fig. 2b,e,g). A haemagglutinin (HA) epitope-tagged Rlim was localized to the nucleus of transfected COS-7 cells (Fig. 3a), as are Clim proteins. The distinct but overlapping tissue and cellular distributions of mRNAs encoding Rlim and Clim cofactors throughout mouse embryogenesis are consistent with the possibility that both of these cofactors may function to differentially regulate the activity of LIM homeodomain proteins. To explore potential inhibitory functions of Rlim, we carried out transient transfection assays using the promoters of genes encoding the pituitary trophic hormones prolactin (encoded by PRL), thyroid-stimulating hormone β-subunit (TSH-β, encoded by TSHB) and glycoprotein hormone α-subunit (α-GSU, encoded by CGA), which are synergistically activated by Lhx3 acting with other classes of transcription factors4,10. For all three transcription units, co-transfection with Rlim inhibited the transcriptional activity of Lhx3 (Fig. 3b,c), consistent with the actions of other RING zinc-finger proteins that act as transcriptional repressors11,12. This inhibitory action was specific and conferred by the LIM domain because co-transfections of Rlim with Pit1 or Pitx1 exhibited little or no inhibitory effect, and LIM domaindeleted Lhx3 (Lhx3∆LIM) was not inhibited by Rlim (Fig. 3c). Indeed, a Gal/LIM domain fusion protein imparts repression on an upstream activating sequence (UAS)-dependent reporter in a single-cell nuclear microinjection assay (data not shown). Furthermore, the observed synergistic actions of Lhx3 with Pit1 or nature genetics • volume 22 • august 1999

Rnf12

Ldb2

letter Ldb1

aa

b b

cc

dd

ee

ff

gg

Fig. 2 Comparison of Rnf12 (encoding Rlim), Ldb2 (encoding Clim1) and Ldb1 (encoding Clim2) expression patterns during mouse embryogenesis by in situ hybridizations on serial sections at E7.5 (a) and E15.5 (b–g). Sections are of the entire embryo (a), forebrain (b), neural tube (c), nasal epithelia (d), submandibular gland (e), intestine (f) and limb (g).

with Pitx1 and Clim were blocked in the presence of Rlim (Fig. 3b,c). Similarly, Rlim also inhibits activity of other LIMhomeodomain factors (Fig. 3d). The inhibitory action of Rlim raised the question as to whether Rlim and Clim exhibit a functional balance analogous to that suggested between coactivators such as CBP/P300 or SRC-1 and corepressors such as Ncor or Smrt (ref. 13). A Gal4/Rlim fusion protein was able to act as a repressor on a UAS/tk promoter using single-cell nuclear microinjection assays. Specific purified antimouse Sin3A/B IgGs and anti-HDAC2 IgGs (ref. 14), which have no effect on control transcription units (Fig. 3e), blocked the ability of Rlim to act as a repressor; in contrast, IgGs against the SAP30 component of the Sin3A/HDAC complex (ref. 15) failed to alter the efficiency of Rlim (Fig. 3e). Finally, an anti-HA antibody can co-immunoprecipitate mSin3A with HA epitopetagged Rlim (Fig. 3e), suggesting that Rlim binding to the LIM domain may mediate recruitment of corepressor complexes15, analogous to the roles of Ncor/Smrt on specific nuclear receptors14,16. As LIM homeodomain factors such as LHX2 have been shown to regulate development17–19, we investigated the potential bio395

letter

© 1999 Nature America Inc. • http://genetics.nature.com

TSHB

fold activation

50

150 40

100

20

Pit1 Lhx3 Rlim

GAS

10

+––+–++ –+––+++ – –+++–+

6xSp1 70

70

60

60

60

50

50

40

40

30

30

20

20

10

10

0

0

αmSin3A/B – + – – αHDAC2 – –+– αSAP30 – ––+

0

Pit1 + – – + – + + Lhx3 – + – – + + + Rlim – – + + + – +

3xUAS/tk

70

% blue cells

% blue cells

© 1999 Nature America Inc. • http://genetics.nature.com

e

30

50 0

c

10 0

– + – – Gal/Rlim – + + + + – – + – αSin3A/B – – + – – – – – + αHDAC-2 – – – + – αSAP30 – – – – +

60 50 40 30 20 10

Lhx3 Pitx1 Rlim Clim1 Lhx3∆LIM

0

400 300 200 100

+ – –– + –– +++ – + –– – –+ + + + – – –+ + ++ + – + – – –– – –– – ++ – – +– – +– – – –

CT

Lhx2 Rlim

0

–+–+ – –++

αmSin3A IgG

mSin3A

kD 150

Gal/Rlim

30 20

CGA 500

80 70

f

50 40

d

CGA

60

light units

PRL 200

fold activation

b

a

Gal/Rlim +αmSin3A/B Gal/Rlim +αHDAC2

Rlim(HA) Sin3A

αHA αSin3A

– + + –

+ + + –

– + – +

Fig. 3 Rlim repressor function. a, Nuclear staining of COS-7 cells transfected with an HA-tagged Rlim expression plasmid. b, Co-transfections of the PRL promoter/enhancer and the TSHB promoter with Rlim, Lhx3 and Pit1 expression. c, Co-transfection of the CGA promoter with Rlim, Lhx3, Lhx3∆LIM, Pitx1 or Clim1 expression plasmids. d, Co-transfection of the CGA promoter with Lhx2 and Rlim expression plasmids. e, Rlim is a corepressor that requires mSin3A/B and HDAC2 for function. 3xUAS/tk/LacZ, 6xSp1, p36/LacZ or GAS/LacZ reporters14 were microinjected into nuclei of quiescent cells and the effect of co-injecting specific, purified IgGs against mSin3A/B, HDAC2 or SAP30 were evaluated for ability to block Gal/Rlim-dependent repression. Photomicrographs show lacZ and rhodamine (αIgG) staining of representative fields (CT, control). f, Rlim can interact with Sin3A in vivo. Co-immunoprecipitation of Sin3A from CV1 cells nuclear extracts transfected with Sin3A or HA-tagged Rlim (Rlim(HA)) expression plasmids with anti-Sin3A or anti-HA antibodies. The western blot was developed with specific anti-Sin3A IgGs. Similar interaction was obtained using anti-Sin3A IgGs to immunoprecipitate Rlim (data not shown).

logical role of RLIM in chick limb development. We first isolated a chicken RNF12 cDNA that encoded a protein with 75% amino acid sequence identity with mouse Rlim (Fig. 1a). RNF12, LHX2, CLIM1 and CLIM2 mRNAs were first detected at stages 15–16 (data not shown) in the presumptive chick limb mesoderm (Fig. 4a, stage 18). As limb outgrowth proceeded, we detected RNF12, LHX2, CLIM1 and CLIM2 in the entire limb bud, with a higher concentration in the distal mesoderm throughout limb development until the latest stage examined (stage 28, data not shown). We observed similar expression patterns for both forelimbs and hindlimbs (data not shown). Expression in the overlying ectoderm or apical ectodermal ridge (AER) was not observed. To overexpress RLIM throughout the entire limb bud, we constructed replication-competent retroviral vectors encoding fulllength RLIM. Limb buds infected by injection of RNF12 in chick limb primordia at stages 8–12 lacked a defined AER and exhibited an arrest in limb outgrowth when analysed after 60–72 hours (data not shown). When allowed to develop further, 28% of the injected embryos had alterations of limb outgrowth (Fig. 4b). Approximately 16% of the altered limbs were severely truncated, lacking all digits as well as radius and ulna. In other embryos (22%), the zeugopodal segment (radius and ulna) was affected. The most frequent phenotype (62%) was a reduction in size and number of digits. Overall, the phenotypes observed resembled the alterations obtained after repressing LHX2 function during chick limb outgrowth by expressing an ENG repression domain (encoded by Drosophila melanogaster engrailed)/LHX2 homeodomain fusion protein (ENGRD/LHX2-HD), with which perturbations in all segments 396

around the proximodistal axis were observed19 (Fig. 5a). We used in situ hybridization to examine expression of genes involved in the outgrowth and patterning of the limb at different times after virus injection. FGF4 and FGF8 are expressed in the AER, and can substitute for the growth function of the AER (refs 20–22). FGF4 and FGF8 transcripts were reduced or absent in the AER 55–60 hours after infection with the RNF12 virus (Fig. 5b, and data not shown), and were preceded by downregulation of SHH in the underlying mesenchyme (Fig. 5b). SHH has been shown to act in a feedback loop with FGF4 that is required for the normal outgrowth of the limb along its proximodistal and anterior-posterior axes23,24. The fact that downregulation of SHH precedes the diminution of FGF4 and FGF8 transcripts indicates that the arrest in limb outgrowth observed after RLIM overexpression is due primarily to changes in mesodermal gene expression, and thus only indirectly to changes in the AER. Expression of the LIM homeodomain gene LMX1, which is required for the establishment of the dorsoventral limb axis17,18, was not altered (Fig. 5b). LHX2 transcripts were reduced, probably due to lack of the cells that normally express the gene (Fig. 5b). The reduction or absence of expression of genes required for distal outgrowth of the limb, without alteration of dorsoventral markers, correlates with phenotypes observed at later stages (Fig. 4b). Recent in vivo experiments in Xenopus laevis demonstrated that full-length CLIM2 protein is required for synergistic gene activation by Lhx1, and that amino-terminal or carboxy-terminal deletions in CLIM2 abolish this synergism3,25. To compare these results with RNF12 phenotypes, we designed a dominantnegative CLIM (DN-CLIM) that contains nuclear localization nature genetics • volume 22 • august 1999

letter

© 1999 Nature America Inc. • http://genetics.nature.com

Methods Cloning of mouse and chicken cDNAs and in situ hybridizations. We carried out protein-protein interaction screens of a mouse λgt11 embryonic pituitary library using a radioactively labelled Lhx3LIM-GST fusion as described4. We obtained chicken RNF12, CLIM1 and CLIM2 cDNAs by low-stringency screenings of a chicken limb cDNA library with mouse Rnf12, Ldb1 and Ldb2 probes. We carried out in situ hybridization of sagittal sections of mouse embryos as described4 using a mouse Rlim cDNA linearized with SalI as template for the generation of a [35S]-radiolabelled antisense probe of 705 nt. The probes for Ldb2 and Ldb1 have been described4. We performed whole-mount in situ hybridization on chicken embryos as described19. Chicken CLIM1, CLIM2 and RNF12 cDNAs were linearized with Xho1, Nco1 and Stu1 and used as template for generation of [35S]-radiolabelled antisense probes of 522, 742 and 979 nt, respectively. Probes for FGF4 (ref. 27), FGF8 and SHH (ref. 22), LHX2 (ref. 19) and LMX1 have been described17.

nature genetics • volume 22 • august 1999

a CLIM1

CLIM2

RNF12

22J

18J

LHX2

26J

b

b WT

signal and LIM interaction domain of CLIM1 (aa 225–341). The functionality of DN-CLIM was demonstrated (using bacterially expressed protein) by its ability to efficiently compete with full-length CLIMs for binding to the LIM domain of Lhx3 (Fig. 5c, and data not shown). Injection of a retrovirus expressing DN-CLIM resulted in variable limb phenotypes (probably reflecting variable infection efficiencies), but these alterations were restricted to the proximodistal limb axis, as in the cases of RLIM and ENG-RD/LHX2-HD misexpression (Fig. 5a,d). Reduction in the number and size of digits was the most frequent phenotype (48% of the affected embryos). We observed alterations or absence of radius and ulna in 25% of the affected embryos (Fig. 5d). These perturbations of limb outgrowth were also preceded by alterations in the AER and reduced limb buds (data not shown). In contrast, infection with retroviral vectors expressing full-length CLIM1, CLIM2 or LHX2 did not cause a phenotype, consistent with the normal expression of these genes in the distal limb (data not shown) and indicating that they are not limiting. Downregulation of SHH occurs with similar kinetics in the DN-CLIM and ENG-RD/LHX2-HD experiments as in the RLIM experiments (data not shown), consistent with them belonging to a common pathway. The identification of LIM domain-binding RLIM and CLIM cofactors that are expressed in distinct, but partially overlapping, manners during development suggests that their alternate or combinatorial recruitment may control the transcriptional activity of LIM homeodomain proteins. In conferring a transcriptional repressor role via the LIM domain, RLIM appears to act through a HDAC complex and, with the CLIMs, provides components for the negative and positive functions mediated by nuclear LIM proteins. A developmental connection between CLIM and LIM homeodomain proteins has been established, but mutations of the Drosophila LDB1 homologue, CHIP, result in an early segmentation phenotype well before the expression of any known LIM homeodomain factor26. Indeed, CLIM proteins bind other classes of transcription factors in addition to LIM homeodomain factors4. By analogy, RLIM is also likely to exert effects based on interactions with additional classes of transcription factors.

a

humerus radius

II III IV

ulna

humerus radius

humerus

RLIM

© 1999 Nature America Inc. • http://genetics.nature.com

Fig. 4 RNF12 regulates chick limb development. a, Spatiotemporal pattern of expression of chicken RNF12 and CLIM1, CLIM2 and LHX2 during chick limb bud development at stages 18, 22 and 26. b, Overexpression of RLIM perturbs limb outgrowth. Left, whole-mount chick embryos 7/8 days after infecting the wing primordia at stages 8–12 with a retroviral vector encoding full-length RLIM. Right, the same embryos after cartilage staining. Top, non-infected chick wing. The most severe phenotypes are shown in the second row, where the ulna and the entire distal part of the wing are missing. Less severe reductions are shown in the bottom two rows with absence of only the most posterior digit (third row), or reduction of the ulna and absence of digits (fourth row).

II

radius

III ulna humerus radius

ulna

Protein-protein interaction assays, transfections and DNA binding assays. In vitro co-immunoprecipitations were carried out as described4. We obtained radiolabelled proteins by in vitro transcription/translation with a kit (Promega) and [35S]–methionine (NEN Life Science). We used specific anti-HA and anti-Myc monoclonal antibodies (Boehringer and Berkeley Antibody) and specific anti-Lhx3 and anti-Clim1 rabbit antisera4. For in vitro GST protein-protein interaction assays10, we cloned cDNAs encoding Lhx3LIM and Lhx3∆LIM into pGEX vector; GST-Lhx3, GST-Clim1 and -Clim2 fusion constructs have been described4,10. GST fusion constructs of the cytoplasmic LIM-only proteins CRP, paxillin, zyxin and the LIM-kinase LIM domain were generously provided by G.N. Gill. The GST-mouse Rlim fusions were cloned in pGEX-KGK vector by a PCR-based approach: Rnf12 (1–207) using forward primer (F) 5´–TCTCCCCATGGAGAACTCAGATTCTAAC–3´ and reverse primer (R) 5´–CCTCTGAGCTCAGG TGGTCGGCACTTCTGTTAC–3´ (NcoI/SacI); Rnf12 (208–423) F, 5´–TCCTCCCATGGGGAGG CAAGAAGCCGGAGCC–3´, and R, 5´–CTTCTCTCGAGTCAGGCACTAGGCTCTGAATCACTG–3´ (NcoI/XhoI); and Rnf12 (403–600) F, 5´–CCTTCCCATGGCAGGCTTTGGTGAGCTAAGCTAC–3´, and R, 5´–TTCT TC TCGAGTCACACAACACTTTCTCTGTTCCC–3´ (NcoI/XhoI). To produce DN-CLIM protein, CLIM1 (aa 225–341) was amplified with specific primers as mentioned above and inserted into pGEX-KGK. Full-length Rnf12 was cloned in pGEX-KGK by direct insertion. We carried out protein-mediated EMSA with bacterially expressed proteins as described4. The oligonucleotide used in these experiments encompasses the Lhx3 recognition site on the CGA promoter10. For cell culture and transient transfections of CV1 and COS-7 cells10 we used reporter plasmids αGSU-luciferase, TSH-β-luciferase and prolactin-luciferase, and the expression plasmids cytomegalovirus (CMV) promoter-driven Lhx3,

397

letter

© 1999 Nature America Inc. • http://genetics.nature.com

b

a WT

humerus radius

II III

ulna

IV

ENG-RD/LHX2-HD

humerus

RLIM injected SHH+FGF4

RLIM injected

FGF8

RLIM injected

RLIM injected

LHX2

radius

radius II

LMX1

III ulna

dd

c

humerus

+ – + – + + + + – – + – + + + + + +

III ulna

IV

LHX3/CLIM1 FL LHX3/DN-CLIM LHX3 radius

DN-CLIM

© 1999 Nature America Inc. • http://genetics.nature.com

CLIM1 FL DN-CLIM Lhx3

radius*

ulna

Fig. 5 Overexpression of dominantnegative CLIM1 or LHX2 perturbs limb outgrowth. a, Blocking of LHX2 function by an ENG-RD/LHX2-HD protein causes arrest in limb outgrowth. Top, control limb. Remaining panels, chick limbs infected at stages 10–12 with the ENG-RD/LHX2-HD retroviral vector19. Second row, the ulna, as well as digits, are missing. Third row, alteration at only the most distal level is shown. b, Overexpression of RLIM perturbs gene expression during chick limb development. Injected limb buds are on the left. Top right, stage 23 embryo, SHH transcripts are absent (arrowhead) and FGF4 transcripts are still present (arrow) in the injected limb bud. Top left, stage 24 embryo, FGF8 expression is abolished in the injected limb bud. Bottom left, stage 25 embryo, LMX1 expression appears normal in the reduced limb bud (left arrow). Bottom right, LHX2 transcripts are still present in the infected limb bud (arrow). c, EMSA shows that DN-CLIM competes with and replaces, with increasing amounts (2×, 5×), fulllength CLIM1 in a CLIM1/LHX3/DNA ternary complex. d, Overexpression of DN-CLIM perturbs limb outgrowth. Top row, the radius and digit II are missing. Second row, the ulna and all digits are missing. The radius is absent in the third row and a severe reduction of the radius and absence of all digits (a remnant of digit IV is still present) is observed in the fourth row.

radius* ulna

Lhx3∆LIM, Pit1 (ref. 10), Ptx1, Clim1 and Clim2 (ref. 4). For the Rlim expression plasmids, we cloned NotI-digested fragments encoding fulllength mouse Rnf12 cDNA into pcDNA3 (Invitrogen) and inserted the N-terminal HA epitope-tagged Rlim expression vector using the Quick Change Site-Directed Mutagenesis kit (Stratagene): F, 5´–GATCACCAAGATGGAGTATCCATATGACGTACCAGACTATGCAAACTCAGATTCTAACG–3´; the reverse primer contained the complementary sequence. We constructed Myc-tagged expression plasmids of Lhx3LIM and Lmo2 by introducing the PCR products (F) 5´–CCTTCCCATGGATGAGGTGCTGCAGATACCC–3´ and (R) 5´–CCCTTGTCGACTTGGTCCA CTCGTAGATGTCTTG–3´ into the CS2MT vector. Nuclear extraction from CV1 cells co-transfected with HA epitope-tagged Rlim and Sin3A CMV expression vectors14, in vivo co-immunoprecipitation and westernblot analysis were carried out as described14 . Production of retrovirus, chicken infection and histology. The viruses used for chicken infections contained the constructs ENG-RD/LHX2-HD (ref. 19), DN-CLIM and full-length chicken RNF12. cDNAs encoding DNCLIM and full-length RNF12 were cloned into the shuttle vector Slax-13 and then introduced into the retroviral vector RCAS(BP)A (ref. 28). Virus preparations and injections of chicken embryos (MacIntyre Poultry or SPAFAS) were made in the wing and leg primordia at stages 8–12 as described28. They were then returned to the incubator at 37 °C and fixed at different time points for either in situ hybridization or phenotypic analysis. The percentage of injected embryos showing an abnormal phenotype (which varied with the extent of the infection) was as follows: ENGRD/LHX2-HD, 19%; DN-CLIM, 23%; and Rnf12, 25%. For cartilage visualization, we fixed embryos in trichloroacetic acid, stained them with 0.1% Alcian green and dehydrated/cleared in methylsalycilate.

398

Nuclear microinjection assays. We performed microinjection analysis essentially as described14. Before the injection, Rat-1 fibroblasts were rendered quiescent by incubation in serum-free medium for 24–36 h. We injected plasmids (CMV-Gal/Rlim and a Gal/tk-LacZ reporter) into the nuclei of cells at a concentration of 100 µg/ml. Anti-Ncor, anti-Rpd3 or anti-Sin3A/B IgGs were co-injected with rabbit pre-immune IgGs, allowing the unambiguous identification of the injected cells. We detected βgalactosidase activity by incubation with X-gal. Injected cells were identified by staining with tetramethylrhodamine-conjugated donkey anti-rabbit IgGs. Injected cells were counted as those exhibiting either nuclear rhodamine fluorescence or blue-X-gal staining or both. More than 200 hundred cells were microinjected and counted for each data point. GenBank accession numbers. Mouse Rnf12, AF069992; chicken RNF12, AF069993; chicken CLIM1, AF069990; chicken CLIM2, AF069991. Acknowledgements

We thank C. Kintner for initial X. laevis mRNA injections; O. Bernard for the Kiz-1 expression plasmid; C. Nelson for cell culture; P. Meyers for assistance in figure preparation; M. Fisher for assistance in manuscript preparation; E. De Robertis, E. Osmundson, M. Ruchhoef, S. O’Connell, M. Wegner and R. Stan for providing reagents and advice; and A. Ryan and H. Ostendorff for comments on the manuscript. C.C. was a fellow of the LALOR Foundation. This work is supported by NIH grants to B.A. (AR044882 and AR02080), J.C.I.B. and M.G.R.; by the Irving Weinstein Foundation to B.A.; and by a G. Harold and L.Y. Mathers Charitable Foundation grant to J.C.I.B.

Received 29 October 1998; accepted 28 June 1999.

nature genetics • volume 22 • august 1999

© 1999 Nature America Inc. • http://genetics.nature.com

1. 2.

3. 4.

5. 6. 7.

8.

9.

10.

11.

© 1999 Nature America Inc. • http://genetics.nature.com

12.

13.

14.

Dawid, I.B., Breen, J.J. & Toyama, R. LIM domains: multiple roles as adaptors and functional modifiers in protein interactions. Trends Genet. 14, 156–162 (1998). Taira, M., Otani, H., Saint-Jeannet, J.-P. & Dawid, I.B. Role of the LIM class homeodomain protein Xlim-1 in neural and muscle induction by the Spemann organizer in Xenopus. Nature 372, 677–679 (1994). Agulnick, A.D. et al. Interactions of the LIM-domain-binding factor Ldb1 with LIM homeodomain proteins. Nature 384, 270–272 (1996). Bach, I., Carrière, C., Ostendorff, H.P., Andersen, B. & Rosenfeld, M.G. A family of LIM domain interacting cofactors confer transcriptional synergism between LIM and Otx homeoproteins. Genes Dev. 11, 1370–1380 (1997). Jurata, L.W. & Gill, G.N. Functional analysis of the nuclear LIM domain interactor NLI. Mol. Cell. Biol. 17, 5688–5698 (1997). Freemont, P.S. The RING finger. A novel protein sequence motif related to the zinc finger. Ann. N. Y. Acad. Sci. 684, 174–192 (1993). Tranque, P., Crossin, K.L., Cirelli, C., Edelman, G.M. & Mauro, V.P. Identification and characterization of a RING zinc finger gene (C-RZF) expressed in chicken embryo cells. Proc. Natl Acad. Sci. USA 93, 3105–3109 (1996). Jurata, L.W., Kenny, D.A. & Gill, G.N. Nuclear LIM interactor, a rhombotin and LIM homeodomain interacting protein, is expressed early in neuronal development. Proc. Natl Acad. Sci. USA 93, 11693–11698 (1996). Toyama, R., Kobayashi, M., Tomita, T. & Dawid, I.B. Expression of LIM-domain binding protein (ldb) genes during zebrafish embryogenesis. Mech. Dev. 71, 197–200 (1998). Bach, I. et al. P-Lim, a LIM homeodomain factor, is expressed during pituitary organ and cell commitment and synergizes with Pit-1. Proc. Natl Acad. Sci. USA 92, 2720–2724 (1995). Schoorlemmer J. et al. Ring 1A is a transcriptional repressor that interacts with the polycomb-M33 protein and is expressed at rhombomere boundaries in the mouse hindbrain. EMBO J. 19, 5930–5942 (1997). Moosmann, P., Georgiev, O., Le Douarin, B., Bourquin, J.P. & Schaffner W. Transcriptional repression by RING finger protein TIF β that interacts with the KRAB repressor domain of KOX1. Nucleic Acids Res. 24, 4859–4867 (1996). Torchia, J., Glass, C.K. & Rosenfeld, M.G. Co-activators and co-repressors in the integration of transcriptional responses. Curr. Opin. Cell Biol. 10, 373–383 (1998). Heinzel, T. et al. A complex containing N-CoR, mSin3 and histone deacetylase mediates transcriptional repression. Nature 387, 43–48 (1997).

nature genetics • volume 22 • august 1999

letter

15. Laherty, C.D. et al. SAP30, a component of the mSin3 corepressor complex involved in N-CoR-mediated repression by specific transcription factors. Mol. Cell 2, 33–42 (1998). 16. Nagy, L. et al. Nuclear receptor repression mediated by a complex containing SMRT, mSin3A, and histone deacetylase. Cell 89, 373–380 (1997). 17. Vogel, A., Rodriguez, C., Warnken, W. & Izpisúa Belmonte, J.C. Dorsal cell fate specified by chick Lmx1 during vertebrate limb development. Nature 378, 716–720 (1995). 18. Riddle, R. et al. Induction of the LIM homeobox gene Lmx-1 by Wnt7a establishes dorsoventral pattern in the vertebrate limb. Cell 83, 631–640 (1995). 19. Rodriguez-Esteban, C. et al. Lhx2, a vertebrate homologue of apterous, regulates vertebrate limb outgrowth. Development 125, 3925–3934 (1998). 20. Mahmood, R. et al. A role for FGF8 in the initiation and maintenance of vertebrate limb bud outgrowth. Curr. Biol. 5, 797–806 (1995). 21. Crossley, P.H., Minowada, G., MacArthur, C.A. & Martin, G.R. Roles for FGF-8 in the induction, initiation and maintenance of chick limb development. Cell 84, 127–136 (1996). 22. Vogel, A., Rodriguez, C. & Izpisúa Belmonte, J.C. Involvement of FGF-8 in initiation, outgrowth and patterning of the vertebrate limb. Development 122, 1737–1750 (1996). 23. Laufer, E., Nelson, C.E., Johnson, R.L., Morgan, B.A. & Tabin, C. Sonic hedgehog and Fgf-4 act through a signaling cascade and feedback loop to integrate growth and patterning of the developing limb bud. Cell 79, 993–1003 (1994). 24. Chiang, C. et al. Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function. Nature 383, 407–413 (1996). 25. Breen, J.J., Agulnick, A.D., Westphal, H. & Dawid, I. Interactions between LIM domains and the LIM domain-binding protein Ldb1. J. Biol. Chem. 273, 4712–4717 (1998). 26. Morcillo, P., Rosen, C., Baylies, M.K. & Dorsett, D. Chip, a widely expressed chromosomal protein required for segmentation and activity of a remote wing margin enhancer in Drosophila. Genes Dev. 11, 2729–2740 (1997). 27. Niswander, L., Tickle, C., Vogel, A., Booth, I. & Martin, G.R. FGF-4 replaces the apical ectodermal ridge and directs outgrowth and patterning of the limb. Cell 75, 579–587 (1993). 28. Morgan, B.A., Izpisúa Belmonte, J.C., Duboule, D. & Tabin, C.J. Targeted misexpression of Hox-4.6 in the avian limb bud causes apparent homeotic transformations. Nature 358, 236–239 (1992).

399