Smads oppose Hox transcriptional activities

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / y e x c r

Research Article

Smads oppose Hox transcriptional activities Xuelin Li a , Shuyi Nie b , Chenbei Chang b , Tao Qiu a , Xu Cao a,⁎ a

Department of Pathology, University of Alabama at Birmingham, 1670 University Blvd., VH G003, Birmingham, AL 35294-0019, USA Department of Cell Biology, University of Alabama at Birmingham, Birmingham, AL 35294-0005, USA

b

ARTICLE INFORMATION

ABS T R AC T

Article Chronology:

BMPs and Hox proteins play crucial roles in developmental processes. Beyond their mutual

Received 10 August 2005

regulation of gene expression, little is known about the relations between their mechanisms

Revised version received

of actions. Previously, we have shown that Hoxc8 acts as a downstream repressor in the BMP

28 November 2005

signaling pathway. Smad1 and Smad6 interact with Hoxc8 and regulate its repression

Accepted 1 December 2005

activities. The Hox family contains 39 genes divided into 13 paralogs. In this report, we

Available online 10 January 2006

systemically examined the potential functions of all the paralogous Hox proteins as BMP downstream transcription factors. Representative Hox proteins from each paralog were

Keywords:

tested. In the gel-shift assay, we found that Smad1, Smad4, and Smad6 interacted with most

BMP

of the Hox proteins in ways similar to their interactions with Hoxc8. The interactions were

Smad

confirmed in mammalian cells. We also examined the effects of Smads on Hox-induced

Hox

transactivation. Particularly, we determined that for Hoxd10 as a transcriptional activator,

Transcription

both Smad1 and Smad6 opposed its activity. In addition, Smad6 also inhibited Hoxc8- and

Xenopus

Hoxb7-induced osteoprotegerin (OPG) transactivation. Furthermore, Smad1 inhibited

Development

Hoxb4-mediated target gene Irx5 expression during early Xenopus development. Our findings suggest that Hox proteins act as general downstream DNA-binding proteins in

Abbreviations:

BMP signaling cascade and their transcriptional activities are regulated by Smads. © 2005 Elsevier Inc. All rights reserved.

Abd-B; Abdominal-B BMPs; bone morphogenetic proteins GST; glutathione S-transferase Irx5; Xenopus Iroquois 5 OPG; osteoprotegerin OPN; osteopontin PCA; protein-fragment complementation assay TGF-β; transforming growth factor β YFP; yellow fluorescence protein

Introduction Bone morphogenetic proteins (BMPs) are members of the transforming growth factor β (TGF-β) superfamily of signaling molecules that regulate diverse biological events, including

cell growth, differentiation, and apoptosis [1,2]. In particular, these signal molecules regulate a broad range of morphogenetic events during embryonic development, such as neural differentiation, tissue patterning, skeletogenesis, etc. [2]. BMPs trigger cell responses mainly through the Smad pathway,

⁎ Corresponding author. Fax: +1 205 934 1775. E-mail address: [email protected] (X. Cao). 0014-4827/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.yexcr.2005.12.002

E XP E RI ME N TA L CE LL RE S EA RCH 3 12 ( 20 0 6 ) 8 5 4 –86 4

which requires two types of serine/threonine transmembrane receptor kinases and phosphorylation of receptor-regulated Smad proteins (R-Smads, Smad1, 5, and 8) [1,3–7]. Upon phosphorylation, the R-Smads form complexes with the common partner, Smad4, and translocate into the nucleus where the complexes recruit distinct transcription factors and mediate gene transcription in a tissue-specific manner [6,7]. Inhibitory Smads, Smad6 and Smad7, represent another subclass of Smads which block phosphorylation of R-Smads, indicating their negative feedback functions in the signaling [8,9]. While Smad7 is a general inhibitor for the TGF-β superfamily, Smad6 is more specific for the BMP pathway [1,3]. Although the BMP pathway has been well-established, less is known about the downstream transcription factors recruited by BMP signaling [4,6]. Hox transcription factors are key regulators of skeletal patterning during embryonic development [10,11]. The Hox gene family has distinct features such that all the members share a well-conserved homeobox sequence and the gene family has characteristic organization along the genome. In mice and humans, 39 Hox genes have been identified which are arranged into four genomic clusters (Hoxa-d). According to the sequence similarities and positions along the clusters, the 39 genes are divided into 13 paralogs [12–14]. Hox gene mutations lead to structure malformations. For example, homeotic transformations of vertebrae as well as limb skeletal malformations were observed in Hoxa11 mutant mice [15]; mutation of Hoxa13 was found to be the molecular basis for digit arch malformation in Hypodactyly (Hd) mice [16]. Hox proteins and BMPs participate in many common developmental processes during normal embryogenesis, playing either opposite roles or similar roles. First, BMPs are primary factors promoting cartilage and bone formation [2,17–19]. On the other hand, several Hox proteins have been shown to inhibit skeletogenesis. Hoxa2 inhibits cartilage condensation and bone formation in the second branchial arch, likely via excluding Sox9 expression and down-regulating Cbfa1 expression [20]. Hoxd11 and Hoxd13 arrest cartilage growth of tibia and fibula, resulting in decreased bone length [21]. Hoxa13 inhibits the cell proliferation and differentiation of zeugopod cartilage rudiments [22]. Second, BMP-2 [23] as well as 5′ Hoxd [24] genes had been shown to be involved in limb anterior– posterior axis patterning. Third, both BMP-4 and Abd-B-like (paralog 9–13) Hox proteins act as downstream factors of sonic hedgehog signaling and control regionalization of hindgut [25]. All the evidence implies that the functions of BMPs and Hox proteins during development are closely related. Accumulating evidence shows that BMPs can regulate Hox gene expression in many developmental contexts [26–28]. There is also evidence showing that Hox proteins regulate Bmp-4 promoter activities [29]. However, beyond these mutual gene regulations, little is known as to whether their transcriptional mechanisms are intrinsically linked. We have previously identified Hoxc8 as a downstream transcription repressor in BMP-mediated osteopontin (OPN) and osteoprotegerin (OPG) gene expression [30–33]. Other studies further support that osteopontin is a direct target of Hoxc8 in vivo [34]. In response to BMP, Smad1 dislodges Hoxc8 from the promoter and activates both osteopontin and osteoprotegerin gene transcription [30,33]. On the other hand,

855

Smad6 acts as a co-repressor for Hoxc8 and enhances its repressor activity [31]. We also provided evidence that Hoxa9 has similar interactions with Smad1 and Smad6 [30,31]. Similar to our findings, group 13 Hox proteins have been demonstrated to interact with the R-Smads [35]. Also, homeoprotein DLX-1 was found to interact with Smad4 through its homeodomain and inhibit TGF-β/BMP/Activin-induced transactivation [36]. Given that all the Hox proteins share the well-conserved DNA-binding domain known as homeodomain (HD), we wanted to know how universal these Smad–Hox interactions are and whether Smad proteins mediate Hox transcriptional activities as a general rule. For this purpose, we tested 12 Hox proteins, each from a different paralog (except for paralog 8), and demonstrated that most of the Hox proteins tested had similar interactions with Smad1, Smad4, and Smad6. In addition to what we found before for Hoxc8 as a repressor, we determined that when Hoxd10 acts as an activator, both Smad1 and Smad6 opposed its activity. In addition, Smad6 inhibited both Hoxc8- and Hoxb7-induced OPG transactivation. Furthermore, we demonstrated that Smad1 inhibited Hoxb4-induced gene transcription during early Xenopus development. Our data suggest that Hox proteins act as general downstream DNA-binding proteins in BMP signaling pathway and their transcriptional activities are regulated by Smads.

Materials and methods Plasmid construction and cell culture Hoxa1 cDNA was a gift from Dr. V. Zappavigna (University of Modena and Reggio Emilia, Modena, Italy), Hoxa2 cDNA was a gift from Dr. M. Mallo (Max-Planck Institute of Immunobiology, Freiburg, Germany), Hoxb3 and b4 cDNAs were gifts from Dr. K. R. Humphries (British Columbia Cancer Agency, BC, Canada), Hoxa5 cDNA was amplified from human breast cancer cDNA library and confirmed by sequencing, Hoxd10 cDNA was a gift from Dr. M. Featherstone (McGill University, Quebec, Canada), and Hoxa11 cDNA was a gift from Dr. S. S. Potter (Children's Hospital Research Foundation, Cincinnati, OH). Hoxb6, b7, a9, d12, and d13 cDNAs were gifts from Dr. C. Largman (University of California, San Francisco, CA). Smad1 and Smad4 cDNAs were gifts from Dr. R. Derynck (University of California, San Francisco, CA). Smad6 cDNA was a gift from Dr. A. Hemmati–Brivanlou (Rockefeller University, New York, NY). To get GST-fusion constructs, Hoxa1, a2, b3, b4, a5, b7, d10, a11, and d13 cDNAs were cloned into pGEX-5X-1 vector by standard PCR. Hoxb6 and d12 cDNAs were cloned into pGEXKG vector. The generation of GST-Hoxa9 and GST-Smad constructs was described previously [30,31]. The plasmids expressing YFP1 (a.a. 1–158) and YFP2 (a.a. 159–239) were gifts from Dr. S. W. Michnick (University of Montreal, Quebec, Canada) and Odyssey Thera, Inc. (San Ramon, CA). To get Hox– YFP2 constructs, Hoxa1, b4, and d10 cDNAs were cloned inframe to the N-terminus of YFP2. To get YFP1–Smad constructs, Smad1 and Smad6 cDNAs were cloned in-frame to the C-terminus of YFP1 by standard PCR. Expression plasmids for FLAG-tagged Smad1/2/4/6 and HA-tagged Hoxc8 were described previously [30,31]. To get HA-tagged Hoxb7 expression plasmid, Hoxb7 cDNA was cloned into the pCDNA3 vector with

856

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

HA at carboxyl terminal. pSG5-Hoxd10 expression plasmid was obtained from Dr. M. Featherstone. OPG-luciferase (OPGluc) reporter plasmid containing two Hoxc8 binding sites was constructed previously [33]. pTHCR reporter plasmid containing a Hoxd10-responsive element [37] was obtained from Dr. V. Zappavigna. 293T cells, COS-1 cells, and mouse fibroblast NIH3T3 cells were incubated in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C in 5% CO2. Human prostate cancer LNCaP cells were cultured in RPMI-1640 medium with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C in 5% CO2.

Gel-shift assay Gel-shift assays were performed as previously described [38]. The GST-fusion plasmids were transformed into BL21 Escherichia coli. The purification of the GST-fusion proteins was performed as described [39]. The amounts of GST fusion proteins used in the gel-shift assay were as follows: Hoxa1 0.3 μg, Hoxa2 0.1 μg, Hoxb3 50 ng, Hoxb4 0.1 μg, Hoxa5 50 ng, Hoxb6 50 ng, Hoxb7 0.4 μg, Hoxa9 0.1 μg, Hoxd10 0.3 μg, Hoxa11 0.2 μg, Hoxd12 0.5 μg, Hoxd13 0.3 μg, Smad1 3 μg, Smad4 2 μg, and Smad6 3 μg. The osteopontin promoter element containing the Hox binding site (OPN5, 5′-GACATCGTTCATCAGTAATGCTTG-3′) and the Hoxd13 binding element (5′CTGCGATGATTTATGACCGC-3′) were 32P-labeled as probes.

YFP protein-fragment complementation assay (PCA) 293T cells (in 6-well plate, 80% confluence) were transfected with Hox–YFP2 and YFP1–Smad expression plasmids as indicated in Fig. 2A using Tfx-50 reagent according to the manufacturer's instructions (Promega). 48 h after transfection, cells were viewed for green fluorescence under an OLYMPUS IX70 Inverted Research Microscope with objective lens of Hoffman Modulation Contrast®, HMC 20 LWD PL FL, 0.4 NA/1 OPTICS INC at room temperature. Digital pictures were taken with an Olympus 1XTRINOC camera and processed with MagnaFire® SP imaging software (Optronics). To visualize the subcellular localization, COS-1 cells were transfected with the indicated YFP-fusion plasmids (Fig. 2B) using the same method, except using the objective lens of HMC 40 LWD PL FL, 0.6 NA/1.

Transient transfection, luciferase assay and immunoblotting To examine the effects of Smads on Hox protein-induced gene transcription, two reporter plasmids (OPG-luc and pTHCR) were used in the transient transfection assays. For the OPG-luc activity assay, 50 ng OPG-luc and different expression plasmids were transfected into LNCaP cells (in 12-well culture plate, 50% confluence) as indicated in Fig. 3 using Lipofectamine as instructed (Invitrogen). 5 ng pRL-SV40 plasmid expressing Renilla luciferase (Promega) was cotransfected in each well as internal control. Total DNA was kept constant by adding pcDNA3 plasmid. For pTHCR activity assay, NIH3T3 cells (in 24-well culture plate, 50% confluence) were transfected with 25 ng pTHCR reporter plasmid and different expression plasmids as indicated (Fig. 4) using the same method. 5 ng pRL-SV40 plasmid was added in each well.

Cells were harvested 48 h after transfection. Luciferase activity was measured and normalized with Renilla luciferase activity using Dual-luciferase assay kit according to the manufacturer's instructions (Promega). Luciferase values shown in the figures are representative of transfection experiments performed in triplicate in three independent experiments. To determine the expression levels of HA-tagged Hoxb7, c8, and FLAG-tagged Smad6 in Fig. 3, the same transfection was performed in 6-well plate, with the exception that the amounts of the plasmids were doubled. LNCaP cells were lysed in the NP-40 buffer (137 mM NaCl, 2 mM EDTA, 10% glycerol, 1% Nonidet P-40, 20 mM Tris–HCl pH 8.0) 48 h after transfection. Supernatants (50 μg) were boiled and separated by 8.5% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membranes and immunoblotted by 1:1000 anti-FLAG M2 monoclonal antibody (Sigma) or 1:1000 anti-HA monoclonal antibody (Babco). The membranes were washed three times with phosphate-buffered saline containing 0.05% Tween-20 and then incubated with 1:10,000 goat anti-mouse antibody (Bio-Rad). The blots were visualized by the enhanced chemiluminescence (ECL) kit (Amersham).

Embryos, RNA injections, and animal cap assay Xenopus embryos were obtained and staged as described [40]. Hoxb4 mRNA for embryo injection was synthesized in vitro using T7 mMessage mMachine kit (Ambion). The template for Hoxb4 transcription was generated by linearizing the pCIHoxb4 expression plasmid with ClaI. Smad1 and Smad2 mRNAs were synthesized as previously described [41]. The mRNAs were injected into the animal poles of the two-cell stage embryos. Animal caps were removed at late blastula stages and cultured until control embryos reached neurula stages, then total RNA was extracted and subjected to RT-PCR analysis.

RT-PCR assay RT-PCR assay was performed as described previously [42]. Total RNA was extracted from animal caps or whole embryos. Approximately 1 μg total RNA was used for reverse transcription with Taqman kit (Roche). 2 μl of total 30 μl cDNA were used in PCR. Primers used for EF1α and Irx5 amplification were as follows: EF1α-U: 5′-CAGATTGGTGCTGGATATGC-3′ and EF1α-D: 5′ACTGCCTTGATGACTCCTAG-3′; Irx5-U: 5′-ACTCTGGTCCTTGGCAGAGA-3′ and Irx5-D: 5′-AGGGTAAAAGGGGATGCTGT-3′. The PCR conditions were as follows: 25 cycles of 95°C, 30 s; 55°C, 1 min; 72°C, 30 s for EF1α amplification and 27 cycles of 94°C, 30 s; 60°C, 30 s; 72°C, 1 min for Irx5 amplification. 5 μl of total 25 μl PCR products were subjected to 5% PAGE and visualized by autoradiography.

Results Smad1, Smad4, and Smad6 modulate the DNA-binding activities of different paralogous Hox proteins Previously, we demonstrated distinct Smad–Hoxc8 interactions in BMP-mediated gene transcription. Smad1 interacts with Hoxc8 homeodomain (HD) and inhibits its DNA binding

E XP E RI ME N TA L CE LL RE S EA RCH 3 12 ( 20 0 6 ) 8 5 4 –86 4

activity [30,32], while Smad6 forms a heterodimer with Hoxc8 and enhances its binding to DNA [31]. Other studies also show that homeoprotein DLX-1 interacts with Smad4 through its homeodomain [36]. The fact that the homeodomain that mediates Hoxc8–Smad1 interaction is well conserved in the Hox family raises the question as to whether the Smad–Hox interactions are universal for most, if not all, Hox proteins. In vertebrate, the 39 Hox genes are divided into 13 paralogs so that genes within each paralog share maximum sequence homology. To examine our hypothesis, we first chose 12 Hox proteins, each from a different paralog group (except group 8, which has been characterized), to test their interactions with Smads using a gel-shift assay with purified Hox and Smad proteins. As shown in Fig. 1A, 11 Hox proteins bound to the osteopontin promoter element (OPN5), which contains a core TAAT sequence [30], with different affinities. Smad1, Smad4, or Smad6 alone did not bind to the Hox consensus element. When coexpressed, Smad1 inhibited most of the Hox binding to DNA, while Smad4 inhibited all of the Hox binding. Smad6, on the other hand, formed complexes with most of the Hox proteins on the DNA element. These results were similar to what we found previously for Hoxc8 and suggested that Smad–Hox interactions may be a general theme for most Hox paralogs. Since Hoxd13 did not bind OPN5 element (data not shown), we used an alternative DNA target that contains a core TTAT

857

sequence and had been demonstrated to be a selective target for Abd-B-like Hox proteins [43]. Similar to other Hox proteins, Smad1 and Smad4 inhibited Hoxd13 binding to the DNA target (Fig. 1B, lanes 7 and 8), but unlike other Hox proteins, Smad6 did not form a complex with Hoxd13 (Fig. 1B, lane 9). It is possible that complex formation between Smad6 and Hox proteins is dependent on the identity of the DNA target. In our previous studies, the MH1 domain of Smad6 was found to bind on OPN5, and it may contribute to the complex formation of Smad6 and Hoxc8 on OPN5 [44]. So far, we do not have evidence as to whether Smad6 domains bind to a TTAT sequence. Taken together, our results suggest that Smad4 may interact with all, while Smad1 and Smad6 may interact with most of the paralogous Hox proteins and they may potentially act as general cofactors for Hox proteins by modulating their DNA-binding activities.

Smad1 and Smad6 interact with representative Hox proteins in mammalian cells Our gel-shift assays demonstrated that Smads modulated the DNA binding of most of the Hox proteins. To confirm the interactions, we next analyzed the Smad–Hox interactions in 293T cells, using protein-fragment complementation assay (PCA). It had been shown that when the N-terminal and Cterminal YFP fragments were fused with two proteins

Fig. 1 – Smad1, Smad4, and Smad6 alter the DNA-binding activities of representative Hox proteins from different paralogs. (A) The osteopontin Hox binding element (OPN5: 5′-TAGTTAATGACATCGTTCATCAG-3′) was 32P-labeled and used as the probe. The proteins used were purified GST-fusion proteins. 0.3 μg Hoxa1, 0.1 μg Hoxa2, 0.05 μg Hoxb3, 0.1 μg Hoxb4, 0.05 μg Hoxa5, 0.05 μg Hoxb6, 0.4 μg Hoxb7, 0.1 μg Hoxa9, 0.3 μg Hoxd10, 0.2 μg Hoxa11, 0.5 μg Hoxd12 were incubated with the probe alone (lanes 6, 10, 14, 18, 22, 26, 30, 34, 38, 42, and 46), together with 3 μg Smad1 (lanes 7, , 11, 15, 19, 23, 27, 31, 35, 39, 43, and 47), or 2 μg Smad4 (lanes 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, and 48), or 3 μg Smad6 (lanes 9, 13, 17, 21, 25, 29, 33, 37, 41, 45, and 49). (B) Smad1 and Smad4 inhibit Hoxd13 binding to the DNA target. Hoxd13 binding element (5′-CTGCGATGATTTATGACCGC-3′) was used as probe and incubated with 0.3 μg GST–Hoxd13 alone (lane 6) or together with Smads (same amounts as A, lanes 7–9).

858

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

respectively, no fluorescent signals could be detected unless the two proteins interacted to bring the two YFP fragments together to fold in active form and give green signal [45]. In our assays, we chose three distal, nonparalogous Hox proteins (Hoxa1, Hoxb4, and Hoxd10) for their interactions with Smad1 and Smad6 for the following reasons. First, different paralogous Hox proteins define the identities of different domains along the anterior–posterior body axis, and functional redundancy normally occurs between paralogous as well as neighboring nonparalogous Hox proteins, but rarely between distal nonparalogous Hox proteins. The three distal Hox proteins we chose therefore should not share redundant activities. Second, Smad1 inhibited the DNA-binding activities of these three Hox proteins at different levels. The binding of Hoxa1 was remarkably decreased, Hoxd10 binding was moderately decreased, and there was no obvious inhibition of Hoxb4 binding by Smad1 (Fig. 1A, lanes 7, 19, and 39). To test the interactions, we made two sets of expression plasmids. One set of the plasmids express YFP1–Smad (Smad1 and Smad6, respec-

tively), which contain the N-terminal YFP (a.a. 1–158) and the full-length Smad1 or Smad6 separated by a small linker region. Another set of the plasmids express Hox–YFP2 fusion proteins which contain the full-length Hox proteins (Hoxa1, Hoxb4, and Hoxd10, respectively) and the C-terminal YFP (a. a. 159–239). While cells transfected with either YFP1–Smad1 or Hox–YFP2 alone did not have any detectable signal, cotransfection of YFP1–Smad1 and Hox–YFP2 resulted in generation of green fluorescence (Fig. 2A). The data indicate that Smad1 interacts with the three Hox proteins in the cells. Combined with previous results that Smad1 interacts with Hoxc8 and Hoxa9, we propose that Smad1 may interact with most of the Hox proteins directly, though it is also possible that cofactors may exist for a subset of Hox proteins to interact with Smad1. In similar experiments, Smad6 also interacted with the three Hox proteins in 293T cells (Fig. 2A). These results confirmed the Smad–Hox interactions in mammalian cells. To further visualize the subcellular localization of the interactions, we transfected YFP1–Smad and Hox–YFP2

Fig. 2 – Smad1 and Smad6 interact with representative Hox proteins in protein-fragment complementation assays (PCA). (A) 293T cells in a 6-well plate were transfected with Hox–YFP2 and/or YFP1–Smad plasmids as indicated. The amounts of the expression plasmids used for transfection were 1.5 μg each. 48 h after transfection, cells were checked for fluorescent signal under fluorescence microscope at room temperature. (B) COS-1 cells were transfected with indicated expression plasmids. 48 h after transfection, the subcellular protein localization was visualized with higher magnification (see Materials and methods).

E XP E RI ME N TA L CE LL RE S EA RCH 3 12 ( 20 0 6 ) 8 5 4 –86 4

plasmids into COS-1 cells and viewed the signals with an objective lens of higher magnification. Both Smad1 and Smad6 interacted with Hox proteins in the nucleus as well as in the cytoplasm, although the nuclear colocalization was predominant (Fig. 2B). It indicates that Hox proteins may undergo posttranslational modifications in the cytoplasm. One such example is Hoxb6, which is phosphorylated by casein kinase II and PKA [46,47] and is predominantly localized in the cytoplasm of epidermal keratinocytes [47,48].

Smads oppose Hox transcriptional activities Interactions of Smads with most paralogous Hox proteins suggest that Smads may potentially regulate Hox transcriptional activities. We had shown before that Smad1 antagonizes the repressor function of Hoxc8 and activates gene transcription, while Smad6 acts as a co-repressor for Hoxc8 to further silence gene transcription [30–32]. Different Hox proteins act as either transcriptional activators or repressors. Furthermore, the same Hox protein possesses either activation or repression functions, depending on the biological contexts as well as the genes it regulates. One such example is Hoxc8, which mediates a broad range of target gene transcription [34]. Our results indicate that different Hox proteins interact with Smads in similar patterns (Fig. 1). Here we wanted to examine how Smads modulate the activator function of Hox proteins.

859

Besides the target genes regulated by Hoxc8 [34], we noted that Hoxc8 can activate OPG gene transcription when overexpressed in prostate cancer cells (unpublished data), opposite to its roles in osteoblast cells [33]. This observation promoted us to examine the effect of Smad6 on Hoxc8mediated gene activation in this biological context. Interestingly, our results demonstrate that Smad6 blocked Hoxc8-induced OPG transcription dose dependently in LNCaP cells (Fig. 3A). Since Smad6 forms a complex with Hoxc8 on DNA target [31], it is likely that Smad6 switches Hoxc8 from activator to repressor in LNCaP cells. We also examined Hoxb7 (Fig. 3B), which also activated OPG promoter, similar to Hoxc8, likely due to the functional redundancy between these two neighboring Hox proteins. Again, Smad6 exerted inhibitory effect on Hoxb7 activity at a similar dose (Fig. 3B). To further explore the effects of Smads on Hox-mediated gene activation, another reporter system was used. Hoxd10 had been shown to be an activator for Hoxd9 gene transcription; the core element that mediates the transactivation by Hoxd10 had been mapped to a 90 bp Hox cross-talk region (HCR) within the Hoxd9 promoter [37]. We therefore used the luciferase reporter (pTHCR) [37] containing the core element to examine whether the transactivation by Hoxd10 is regulated by Smads. Consistent with the previous findings [37], overexpression of Hoxd10 resulted in 5- to 8-fold increase of the reporter activity (Fig. 4). Importantly, Smad1 blocked the

Fig. 3 – Smad6 blocks Hoxc8- and Hoxb7-induced OPG activation. (A) OPG-luc reporter plasmid (50 ng) was co-transfected with HA-Hoxc8 (100 ng) and FLAG-Smad6 (100 ng and 200 ng) expression plasmids into LNCaP cells. 5 ng pRL-SV40 plasmid expressing Renilla luciferase (Promega) was co-transfected in each well as internal control. Total DNA was kept constant by adding pcDNA3 empty vector. The transfected cells were harvested 48 h later. Reporter activities were measured and normalized to Renilla luciferase levels as internal control. (B) OPG-luc reporter (50 ng) was co-transfected with HA-Hoxb7 (100 ng) and FLAG-Smad6 (200 ng) expression plasmids as indicated. Experiments were repeated three times in triplicate. The expression levels of HA-tagged Hoxb7, c8, and FLAG-tagged Smad6 are shown in the lower panels by Western blot.

860

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

Fig. 4 – Smad1 and Smad6 inhibit Hoxd10 transcriptional activities. (A) Schematic representation of pTHCR reporter. The promoter region contains a 90 bp HCR (Hox-cross talk region) element that is activated by Hoxd10 (Ref. 37). (B) Smad1 inhibits Hoxd10-induced transactivation dose dependently. pTHCR reporter plasmid (25 ng) was co-transfected with Hoxd10 (50 ng), and Smad1 (50 ng, 100 ng, and 150 ng) expression plasmids into NIH3T3 cells. 5 ng pRL-SV40 plasmid expressing Renilla luciferase (Promega) was co-transfected in each well as internal control. Total DNA was kept constant by adding pcDNA3 empty vector. The transfected cells were harvested 48 h later. Reporter activities were measured and normalized to Renilla luciferase levels as internal control. (C) Smad6 inhibits Hoxd10-induced transactivation. NIH3T3 cells were transfected with indicated plasmids. (D) Smad2 does not regulate Hoxd10 transcriptional activity. Experiments were repeated three times in triplicate.

transactivation by Hoxd10 in a dose-dependent manner when coexpressed (Fig. 4B). We also examined the effects of Smad2. However, unlike Smad1, Smad2 did not have effects on Hoxd10-induced transactivation at similar doses (Fig. 4D), suggesting that Smad2 was not involved in regulating Hoxd10 transcriptional activity. Smad6 also inhibited activation by

Hoxd10, similar to Smad1 (Fig. 4C). As our gel-shift results showed that Smad1 reduced Hoxd10 binding to DNA while Smad6 formed a ternary complex with Hoxd10 and DNA, our findings indicate that Smad1 and Smad6 may regulate the activator function of Hoxd10 through different mechanisms. Smad1 inhibited Hoxd10 activity by preventing it from binding

E XP E RI ME N TA L CE LL RE S EA RCH 3 12 ( 20 0 6 ) 8 5 4 –86 4

861

Fig. 5 – Smad1 inhibits Hoxb4 target gene expression during early Xenopus development. (A) Diagram showing animal cap assay carried out in frog embryos. (B) Smad1 inhibits Hoxb4-induced Irx5 gene expression. Embryos were injected with mRNAs encoding Smad1 or Hoxb4 at the two-cell stage. The animal caps were removed at the blastula stage and incubated until control embryos reached neurula stage 19. Total RNA was then extracted and analyzed for Irx5 expression by RT-PCR. EF1α PCR was used as a loading control. Lane 1, uninjected control caps; lanes 2–4, caps injected with Smad1 mRNA (lane 2), Hoxb4 mRNA (lane 3) and combined mRNAs (lane 4); lanes 5 and 6, whole-embryo controls, in the absence (lane 5, −RT) or presence (lane 6) of reverse transcriptase in the RT-PCR reaction. (C) Smad2 is not involved in Hoxb4 target gene expression. Embryos were injected with indicated mRNAs. Then the same procedures were carried out as in (B). The doses of mRNAs used for embryo injection were as follows: Hoxb4 2 ng, Smad1 1 ng, and Smad2 1 ng.

to the DNA target. Smad6, by forming a complex with Hoxd10, could switch it from activator to repressor.

Smad1 modulates Hoxb4 transcriptional activities in vivo Both BMPs and Hox proteins are key factors that mediate many similar morphological events during vertebrate development. The similar expression patterns of BMPs and Hox proteins in many embryonic tissues imply that they may exert functions together to mediate developmental processes. We attempted to investigate whether BMP/Smad signals can regulate Hox target gene transcription in vivo, particularly for Hox proteins as activators. It was noted that Hoxb4 upregulates Irx5 expression during early Xenopus development [49]. Irx family had been shown to be essential for neural plate formation [50,51]. Our aim was to examine whether Smads can modulate Hoxb4-induced Irx5 gene transcription. In the animal cap assay, we injected Hoxb4 mRNA together with or without Smad1 mRNA into the animal poles of frog embryos at the two-cell stage; the animal caps were removed at the blastula stage when endogenous Hoxb4 expression is not initiated. The Irx5 gene expression was examined at the neurula stage by RT-PCR. As shown in Fig. 5B, Irx5 expression was not detected at significant levels in uninjected animal caps (lane 1), which have no endogenous Hoxb4 expression. Overexpressing Hoxb4 clearly induced Irx5 expression. Smad1

significantly blocked the Irx5 induction by Hoxb4 (lane 4). We then compared the effects of Smad1 and Smad2. Unlike Smad1, Smad2 had no obvious effect on Hoxb4-mediated Irx5 upregulation (Fig. 5C, lane 6). These in vivo results suggest that Smad1 inhibits Hoxb4-induced gene transactivation during early development. In addition, these results are consistent with those of the in vitro reporter assays (Fig. 4) and suggest that Smad1 antagonizes the activator function of Hox proteins by dislodging them from DNA targets.

Discussion BMPs are involved in multiple morphological events during embryonic development, including mesoderm patterning, skeletogenesis, limb patterning, and apoptosis [2]. While the BMP/Smad signaling cascade is well characterized, the pathway itself does not explain how BMPs exert such diverse functions. The answer lies in the cross-talk between BMP and other signaling pathways, as well as distinct transcription factors recruited by the BMP pathway. Initially, we identified Hoxc8 as a transcription repressor in BMP-mediated osteopontin and osteoprotegerin gene transcription [30,33]. In this report, we demonstrated that Smad1, Smad4, and Smad6 interact with the representative members of the other 12 Hox paralogs, similar to Hoxc8. More importantly, we determined

862

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

that Smad1 prevented Hox binding to DNA and inhibited transactivation by Hox proteins, and that Smad6 formed DNAbinding complexes with Hox proteins and inhibited Hoxinduced gene transcription. Our results suggest that Hox proteins act as general downstream DNA-binding proteins recruited by BMP/Smad pathway and their transcriptional activities are regulated by Smad proteins. In the gel-shift assay, all of the Hox proteins tested, except Hoxd13, bound to osteopontin promoter element (OPN5) containing TAAT sequence. Smad1, Smad4, and Smad6 interacted with most of the Hox proteins, in ways similar to their interactions with Hoxc8 (Fig. 1A). For Hoxd13, we used alternative DNA target containing a TTAT core element that had been shown to preferentially bind with Abd-B-like Hox proteins [43]. Again, Smad1 and Smad4 inhibited Hoxd13 binding to DNA (Fig. 1B). In our experiments, Smad1/4 always prevented Hox–DNA complex formation. On the other hand, studies on the BMP target gene Msx2 indicate that the presence of both homeoproteins and Smad1/4 on the promoter element is required for the gene activation [52]. It is likely that the Smad binding sites flanking the Hox binding site are critical to determine whether Smad1/4 form complexes with Hox proteins on the DNA elements or prevents them from binding to DNA. To confirm the interactions, we performed PCA assay to test three Hox proteins (Hoxa1, Hoxb4, and Hoxd10) for their interactions with Smad1 in mammalian cells. Although Smad1 inhibits their DNA binding at different levels (Fig. 1A), it interacts with those Hox proteins in mammalian cells similarly (Fig. 2). We propose that Smad1 may interact with most of the Hox proteins directly, though it is also possible that cofactors may exist for a subset of Hox proteins to interact with Smad1. Hox proteins and Smad1 are predominantly colocalized in the nucleus (Fig. 2B). The cytoplasm localization of Hox proteins (Fig. 2B) suggests that they may undergo posttranslational modifications. Smad6 formed complexes with most of the Hox proteins on the OPN5 element (Fig. 1A), but failed to form a complex with Hoxd13 on the TTAT target (Fig. 1B). It is possible that the complex formation between Smad6 and Hox proteins is dependent on the identity of DNA. This situation could be similar to that of Hoxa10–Pbx1a complex, which is only formed on selective DNA sequence [43,53,54]. Our results suggest that Smad1 and Smad6 interact with and modulate the DNA-binding activities of most Hox proteins, if not all. We did not examine other R-Smads. Different R-Smads share highly conserved MH1 and MH2 domains, therefore interactions may also exist between Hox proteins and TGF-β-specific R-Smads as well as Smad5 and Smad8 in BMP pathway. Although Hox proteins are well known as transcription factors, their transcription mechanisms are not well characterized. Cofactors are often required for Hox-mediated transcription. So far the only well-characterized cofactors are nonHox homeoproteins [55]. Pbx proteins interact with Hox proteins (paralog 1–10) and act as partners to offer the DNAbinding specificities for different Hox proteins [53,54]. Meis proteins interact with AbdB-like Hox proteins (paralog 9–13) [56]. Another class of cofactors identified are histone acetyltransferases (HATs) and histone deacetylases (HDACs), which interact with either Hox–Pbx complex or Hox proteins alone [57,58] and mediate transcription by chromatin structure

remodeling [59]. Our results indicate that Smad6 forms heterodimers with Hox proteins and always as Hox corepressor, and that Smad1 inhibits Hox binding to DNA and opposes their transcriptional activities. Specifically, both Smad1 and Smad6 inhibited transactivation by Hoxd10 (Fig. 4). The mechanisms for their actions are likely to be different. Smad1 inhibited Hoxd10 activities mainly by preventing it from binding to the DNA target, whereas Smad6 could switch Hoxd10 from transcription activator to repressor by forming a complex on the DNA element. The inhibitory role of Smad6 on Hox protein-mediated transactivation was further supported by another reporter assay, in which Smad6 blocked both Hoxc8- and Hoxb7-induced OPG promoter activations in LNCaP cells (Fig. 3). Therefore, our results indicate that Smad1 and Smad6 oppose the transactivation induced by Hox proteins. Finally, we investigated the effects of Smads on Hox transcriptional activity during Xenopus development. Again, Hoxb4 examined in the experiments is an activator. We found that Smad1 inhibited Hoxb4-mediated Irx5 gene transcription during early Xenopus development (Fig. 5). The results are consistent with those of the reporter assays (Fig. 4). Although Smad4 showed stronger interaction with Hoxb4 (Fig. 1A, lane 20), we did not find obvious effects of Smad4 on Hoxb4 activities (data not shown). It is likely that Smad4 only serves as cofactor for Smad1–Hox interactions. Without Smad1, Smad4 alone is not sufficient to regulate Hox activities. For both Hoxd10- and Hoxb4-induced transactivation, Smad2 was also tested but showed no obvious effects. It seems that Smad2 does not regulate Hox-mediated gene transcription in these two specific cases (Figs. 4D and 5C). However, in view of so many Hox proteins possessing distinct functions with their specific targets, it is still possible that TGF-β-specific Smads are involved in regulating Hox functions in other biological contexts. Both BMPs and Hox proteins are key regulators during vertebrate development. The facts that many developmental processes require both BMPs and Hox proteins imply the intrinsic functional connections between the two regulators. However, little is known about their relations beyond their mutual regulation of gene expression. In our studies, we characterized the interactions between Hox proteins and BMP-specific Smads and the regulations of Hox transcriptional activities by Smads. We propose the model that Hox proteins act as downstream DNA-binding proteins in the BMP signaling pathway. Their transcriptional activities are regulated by Smads through their physical interactions.

Acknowledgments We thank Dr. S. W. Michnick and Odyssey Thera, Inc., for kindly providing YFP1 and YFP2 expression plasmids for protein-fragment complementation assay. We thank Dr. V. Zappavigna for kindly providing Hoxa1 cDNA and pTHCR reporter, Dr. M. Mallo for Hoxa2 cDNA, Dr. K. R. Humphries for Hoxb3 and b4 cDNAs, Dr. M. Featherstone for Hoxd10 cDNA, Dr. S. S. Potter for Hoxa11 cDNAs, Dr. C. Largman for Hoxb6, b7, a9, d12, and d13 cDNAs, Dr. H. Le Mouellic for Hoxc8 cDNA, Dr. R. Derynck for Smad1, 2, and 4 expression plasmids, and Dr. A. Hemmati-Brivanlou for Smad6 cDNA.

E XP E RI ME N TA L CE LL RE S EA RCH 3 12 ( 20 0 6 ) 8 5 4 –86 4

This work was supported by National Institutes of Health Grant DK57501 (to X.C.).

[22]

REFERENCES [23] [1] J. Massague, TGF-beta signal transduction, Annu. Rev. Biochem. 67 (1998) 753–791. [2] B.L. Hogan, Bone morphogenetic proteins: multifunctional regulators of vertebrate development, Genes Dev. 10 (1996) 1580–1594. [3] S. Itoh, F. Itoh, M.J. Goumans, P. ten Dijke, Signaling of transforming growth factor-beta family members through Smad proteins, Eur. J. Biochem. 267 (2000) 6954–6967. [4] J.L. Wrana, Regulation of Smad activity, Cell 100 (2000) 189–192. [5] A. von Bubnoff, K.W. Cho, Intracellular BMP signaling regulation in vertebrates: pathway or network? Dev. Biol. 239 (2001) 1–14. [6] J. Massague, Y.G. Chen, Controlling TGF-beta signaling, Genes Dev. 14 (2000) 627–644. [7] J. Massague, D. Wotton, Transcriptional control by the TGF-beta/Smad signaling system, EMBO J. 19 (2000) 1745–1754. [8] T. Imamura, M. Takase, A. Nishihara, E. Oeda, J. Hanai, M. Kawabata, K. Miyazono, Smad6 inhibits signalling by the TGF-beta superfamily, Nature 389 (1997) 622–626. [9] A. Nakao, M. Afrakhte, A. Moren, T. Nakayama, J.L. Christian, R. Heuchel, S. Itoh, M. Kawabata, N.E. Heldin, C.H. Heldin, P. ten Dijke, Identification of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling, Nature 389 (1997) 631–635. [10] W. McGinnis, R. Krumlauf, Homeobox genes and axial patterning, Cell 68 (1992) 283–302. [11] R. Krumlauf, Hox genes in vertebrate development, Cell 78 (1994) 191–201. [12] M.P. Scott, Vertebrate homeobox gene nomenclature, Cell 71 (1992) 551–553. [13] M.R. Capecchi, Hox genes and mammalian development, Cold Spring Harbor Symp. Quant. Biol. 62 (1997) 273–281. [14] C. Cillo, M. Cantile, A. Faiella, E. Boncinelli, Homeobox genes in normal and malignant cells, J. Cell. Physiol. 188 (2001) 161–169. [15] K.M. Small, S.S. Potter, Homeotic transformations and limb defects in Hox A11 mutant mice, Genes Dev. 7 (1993) 2318–2328. [16] D.P. Mortlock, L.C. Post, J.W. Innis, The molecular basis of hypodactyly (Hd): a deletion in Hoxa 13 leads to arrest of digital arch formation, Nat. Genet. 13 (1996) 284–289. [17] J.M. Wozney, V. Rosen, A.J. Celeste, L.M. Mitsock, M.J. Whitters, R.W. Kriz, R.M. Hewick, E.A. Wang, Novel regulators of bone formation: molecular clones and activities, Science 242 (1988) 1528–1534. [18] D. Duprez, E.J. Bell, M.K. Richardson, C.W. Archer, L. Wolpert, P.M. Brickell, P.H. Francis-West, Overexpression of BMP-2 and BMP-4 alters the size and shape of developing skeletal elements in the chick limb, Mech. Dev. 57 (1996) 145–157. [19] H. Zou, R. Wieser, J. Massague, L. Niswander, Distinct roles of type I bone morphogenetic protein receptors in the formation and differentiation of cartilage, Genes Dev. 11 (1997) 2191–2203. [20] B. Kanzler, S.J. Kuschert, Y.H. Liu, M. Mallo, Hoxa-2 restricts the chondrogenic domain and inhibits bone formation during development of the branchial area, Development 125 (1998) 2587–2597. [21] D.J. Goff, C.J. Tabin, Analysis of Hoxd-13 and Hoxd-11 misexpression in chick limb buds reveals that Hox genes

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31] [32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

863

affect both bone condensation and growth, Development 124 (1997) 627–636. Y. Yokouchi, S. Nakazato, M. Yamamoto, Y. Goto, T. Kameda, H. Iba, A. Kuroiwa, Misexpression of Hoxa-13 induces cartilage homeotic transformation and changes cell adhesiveness in chick limb buds, Genes Dev. 9 (1995) 2509–2522. P.H. Francis, M.K. Richardson, P.M. Brickell, C. Tickle, Bone morphogenetic proteins and a signalling pathway that controls patterning in the developing chick limb, Development 120 (1994) 209–218. J.C. Izpisua-Belmonte, C. Tickle, P. Dolle, L. Wolpert, D. Duboule, Expression of the homeobox Hox-4 genes and the specification of position in chick wing development, Nature 350 (1991) 585–589. D.J. Roberts, R.L. Johnson, A.C. Burke, C.E. Nelson, B.A. Morgan, C. Tabin, Sonic hedgehog is an endodermal signal inducing Bmp-4 and Hox genes during induction and regionalization of the chick hindgut, Development 121 (1995) 3163–3174. D.M. Duprez, K. Kostakopoulou, P.H. Francis-West, C. Tickle, P.M. Brickell, Activation of Fgf-4 and HoxD gene expression by BMP-2 expressing cells in the developing chick limb, Development 122 (1996) 1821–1828. K. Hashimoto, Y. Yokouchi, M. Yamamoto, A. Kuroiwa, Distinct signaling molecules control Hoxa-11 and Hoxa-13 expression in the muscle precursor and mesenchyme of the chick limb bud, Development 126 (1999) 2771–2783. S.A. Wacker, C.L. McNulty, A.J. Durston, The initiation of Hox gene expression in Xenopus laevis is controlled by Brachyury and BMP-4, Dev. Biol. 266 (2004) 123–137. M. Suzuki, N. Ueno, A. Kuroiwa, Hox proteins functionally cooperate with the GC box-binding protein system through distinct domains, J. Biol. Chem. 278 (2003) 30148–30156. X. Shi, X. Yang, D. Chen, Z. Chang, X. Cao, Smad1 interacts with homeobox DNA-binding proteins in bone morphogenetic protein signaling, J. Biol. Chem. 274 (1999) 13711–13717. S. Bai, X. Shi, X. Yang, X. Cao, Smad6 as a transcriptional corepressor, J. Biol. Chem. 275 (2000) 8267–8270. X. Yang, X. Ji, X. Shi, X. Cao, Smad1 domains interacting with Hoxc-8 induce osteoblast differentiation, J. Biol. Chem. 275 (2000) 1065–1072. M. Wan, X. Shi, X. Feng, X. Cao, Transcriptional mechanisms of bone morphogenetic protein-induced osteoprotegrin gene expression, J. Biol. Chem. 276 (2001) 10119–10125. H. Lei, H. Wang, A.H. Juan, F.H. Ruddle, The identification of Hoxc8 target genes, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 2420–2424. T.M. Williams, M.E. Williams, J.H. Heaton, T.D. Gelehrter, J.W. Innis, Group 13 HOX proteins interact with the MH2 domain of R-Smads and modulate Smad transcriptional activation functions independent of HOX DNA-binding capability, Nucleic Acids Res. 33 (2005) 4475–4484. S. Chiba, K. Takeshita, Y. Imai, K. Kumano, M. Kurokawa, S. Masuda, K. Shimizu, S. Nakamura, F.H. Ruddle, H. Hirai, Homeoprotein DLX-1 interacts with Smad4 and blocks a signaling pathway from activin A in hematopoietic cells, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 15577–15582. V. Zappavigna, A. Renucci, J.C. Izpisua-Belmonte, G. Urier, C. Peschle, D. Duboule, HOX4 genes encode transcription factors with potential auto- and cross-regulatory capacities, EMBO J. 10 (1991) 4177–4187. X. Cao, F.P. Ross, L. Zhang, P.N. MacDonald, J. Chappel, S.L. Teitelbaum, Cloning of the promoter for the avian integrin beta 3 subunit gene and its regulation by 1,25dihydroxyvitamin D3, J. Biol. Chem. 268 (1993) 27371–27380. J.M. Sterner, Y. Murata, H.G. Kim, S.B. Kennett, D.J. Templeton, J.M. Horowitz, Detection of a novel cell cycle-regulated kinase activity that associates with the amino terminus of the

864

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

E XP ER I ME NT A L C EL L RE S EA R CH 3 12 ( 20 0 6 ) 8 5 4 –86 4

retinoblastoma protein in G2/M phases, J. Biol. Chem. 270 (1995) 9281–9288. A.H. Brivanlou, R.M. Harland, Expression of an engrailed-related protein is induced in the anterior neural ectoderm of early Xenopus embryos, Development 106 (1989) 611–617. G. Lagna, A. Hata, A. Hemmati-Brivanlou, J. Massague, Partnership between DPC4 and SMAD proteins in TGF-beta signalling pathways, Nature 383 (1996) 832–836. P.A. Wilson, D.A. Melton, Mesodermal patterning by an inducer gradient depends on secondary cell–cell communication, Curr. Biol. 4 (1994) 676–686. W.F. Shen, S. Rozenfeld, H.J. Lawrence, C. Largman, The Abd-B-like Hox homeodomain proteins can be subdivided by the ability to form complexes with Pbx1a on a novel DNA target, J. Biol. Chem. 272 (1997) 8198–8206. S. Bai, X. Cao, A nuclear antagonistic mechanism of inhibitory Smads in transforming growth factor-beta signaling, J. Biol. Chem. 277 (2002) 4176–4182. I. Remy, A. Montmarquette, S.W. Michnick, PKB/Akt modulates TGF-beta signalling through a direct interaction with Smad3, Nat. Cell Biol. 6 (2004) 358–365. A.A. Fienberg, C. Nordstedt, H.G. Belting, A.J. Czernik, A.C. Nairn, S. Gandy, P. Greengard, F.H. Ruddle, Phylogenetically conserved CK-II phosphorylation site of the murine homeodomain protein Hoxb-6, J. Exp. Zool. 285 (1999) 76–84. W.F. Shen, D. Chrobak, K. Krishnan, H.J. Lawrence, C. Largman, HOXB6 protein is bound to CREB-binding protein and represses globin expression in a DNA binding-dependent, PBX interaction-independent process, J. Biol. Chem. 279 (2004) 39895–39904. L.G. Komuves, W.F. Shen, A. Kwong, E. Stelnicki, S. Rozenfeld, Y. Oda, A. Blink, K. Krishnan, B. Lau, T. Mauro, C. Largman, Changes in HOXB6 homeodomain protein structure and localization during human epidermal development and differentiation, Dev. Dyn. 218 (2000) 636–647. C. Theokli, A.S. Morsi El-Kadi, R. Morgan, TALE class homeodomain gene Irx5 is an immediate downstream target for Hoxb4 transcriptional regulation, Dev. Dyn. 227 (2003) 48–55.

[50] J.L. Gomez-Skarmeta, A. Glavic, E. Calle-Mustienes, J. Modolell, R. Mayor, Xiro, a Xenopus homolog of the Drosophila Iroquois complex genes, controls development at the neural plate, EMBO J. 17 (1998) 181–190. [51] E.J. Bellefroid, A. Kobbe, P. Gruss, T. Pieler, J.B. Gurdon, N. Papalopulu, Xiro3 encodes a Xenopus homolog of the Drosophila Iroquois genes and functions in neural specification, EMBO J. 17 (1998) 191–203. [52] S.M. Brugger, A.E. Merrill, J. Torres-Vazquez, N. Wu, M.C. Ting, J.Y. Cho, S.L. Dobias, S.E. Yi, K. Lyons, J.R. Bell, K. Arora, R. Warrior, R. Maxson, A phylogenetically conserved cisregulatory module in the Msx2 promoter is sufficient for BMP-dependent transcription in murine and Drosophila embryos, Development 131 (2004) 5153–5165. [53] C.P. Chang, W.F. Shen, S. Rozenfeld, H.J. Lawrence, C. Largman, M.L. Cleary, Pbx proteins display hexapeptide-dependent cooperative DNA binding with a subset of Hox proteins, Genes Dev. 9 (1995) 663–674. [54] C.P. Chang, L. Brocchieri, W.F. Shen, C. Largman, M.L. Cleary, Pbx modulation of Hox homeodomain amino-terminal arms establishes different DNA-binding specificities across the Hox locus, Mol. Cell. Biol. 16 (1996) 1734–1745. [55] R.S. Mann, M. Affolter, Hox proteins meet more partners, Curr. Opin. Genet. Dev. 8 (1998) 423–429. [56] W.F. Shen, J.C. Montgomery, S. Rozenfeld, J.J. Moskow, H.J. Lawrence, A.M. Buchberg, C. Largman, AbdB-like Hox proteins stabilize DNA binding by the Meis1 homeodomain proteins, Mol. Cell. Biol. 17 (1997) 6448–6458. [57] M. Saleh, I. Rambaldi, X.J. Yang, M.S. Featherstone, Cell signaling switches HOX–PBX complexes from repressors to activators of transcription mediated by histone deacetylases and histone acetyltransferases, Mol. Cell. Biol. 20 (2000) 8623–8633. [58] A. Chariot, C. van Lint, M. Chapelier, J. Gielen, M.P. Merville, V. Bours, CBP and histone deacetylase inhibition enhance the transactivation potential of the HOXB7 homeodomain-containing protein, Oncogene 18 (1999) 4007–4014. [59] M. Grunstein, Histone acetylation in chromatin structure and transcription, Nature 389 (1997) 349–352.