QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplemental Research Data Correction Erratum (v19,p875) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Drori, S. Articles by Spiegelman, B. M. Search for Related Content PubMed PubMed Citation Articles by Drori, S. Articles by Spiegelman, B. M. Social Bookmarking                   What's this?

RESEARCH PAPER

Hic-5 regulates an epithelial program mediated by PPAR

¹ Dana-Farber Cancer Institute and the Department of Cell Biology, Harvard Medical School and ² Departments of Medical Oncology and Cancer Biology, Dana-Farber Cancer Institute, and the Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA; ³ Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts 02115, USA

	Abstract

Top Abstract Results Discussion Materials and methods Acknowledgments References

 
PPAR is a dominant regulator of fat cell differentiation. However, this nuclear receptor also plays an important role in the differentiation of intestinal and other epithelial cell types. The mechanism by which PPAR can influence the differentiation of such diverse cell lineages is unknown. We show here that PPAR interacts with Hic-5, a coactivator protein expressed in gut epithelial cells. Hic-5 and PPAR colocalize to the villus epithelium of the small intestine, and their expression during embryonic gut development correlates with the transition from endoderm to a specialized epithelium; expression of both these factors is reduced in tumors. Forced expression of Hic-5 in colon cancer cells enhances the PPAR-mediated induction of several gut epithelial differentiation/maturation markers such as L-FABP, kruppel-like factor 4 (KLF4), and keratin 20. siRNA directed against Hic-5 specifically reduces PPAR-mediated induction of gut epithelial genes in colon cells and in an ex vivo model of embryonic gut differentiation. Finally, forced expression of Hic-5 during 3T3-L1 preadipocyte differentiation inhibits adipogenesis while inducing inappropriate expression of several mRNAs characteristic of gut epithelium in these mesenchymal cells. These results indicate that Hic5 is an important component in determining an epithelial differentiation program induced by PPAR.

[Keywords: Hic-5; ARA55; PPAR; differentiation; selective-coactivator; gut-epithelial]

Received July 16, 2004; revised version accepted December 8, 2004.

Although PPAR is expressed at highest levels in adipose tissues, it is also expressed at significant levels in a number of epithelial cell types, including several that are important in human cancer, such as colon (Mansen et al. 1996; Sarraf et al. 1998; Lefebvre et al. 1999), breast (Mueller et al. 1998), and prostate (Kubota et al. 1998; Mueller et al. 2000). Ligand activation of this receptor in human cancer cells from these tissues resulted in slowing of growth and an increase in cellular maturation (for reviews, see Koeffler 2003; Michalik et al. 2004). For example, PPAR activation in human colon cancer cells reduces cell growth and induces multiple RNAs associated with a more differentiated phenotype, such as keratins 18, 19, and 20 and carcinoembryonic antigen (CEA) family members (Sarraf et al. 1998; Gupta et al. 2001). There is also an alteration in cell morphology, including an increase in cytoplasmic to nuclear ratio, consistent with a more differentiated phenotype (Sarraf et al. 1998).

The notion that PPAR can alter the growth and differentiation state of epithelial cells has led to a number of studies of its potential role in the cause and treatment of cancer. Although agonist treatment of animals suggests possible both pro- or anticancer properties in the colon (Lefebvre et al. 1998; Saez et al. 1998; Sarraf et al. 1998), genetic studies unambiguously demonstrate that PPAR functions as a tumor suppressor in chemically induced colon carcinogenesis in mice (Girnun et al. 2002). Human clinical trials have shown that application of a PPAR ligand to humans with certain forms of liposarcoma yielded impressive histological evidence of tumor differentiation (Demetri et al. 1999); trials of PPAR ligands as a monotherapy in several advanced human malignancies have shown no efficacy (Kulke et al. 2002; Burstein et al. 2003). Taken together, these studies suggest that PPAR may play an important role in the differentiated states of several forms of epithelia, whereas the use of PPAR agonists in the human cancer clinic may be limited to the treatment of early stage disease, or in combination with other drugs.

The apparent ability of PPAR to promote the differentiation and maturation of fat cells of mesenchymal origin and cells of epithelial origin raises interesting issues. Because PPAR and its heterodimeric partner RXR are essentially identical in all of these tissues, this suggests that the execution of radically different genetic programs must occur as a consequence of other molecules that interact, directly or indirectly, with PPAR. Recent data have indicated that coactivator proteins, such as PGC1 or myocardin, can dramatically influence the activity of nuclear receptors and other transcription factors, thereby influencing cell fate (Puigserver et al. 1998; Wang et al. 2001; Yoon et al. 2001; Yoshida et al. 2003). We show here that intestinal epithelial cells contain a coactivator protein, Hic-5, that stimulates PPAR transcriptional activity. Moreover, this factor promotes an epithelial-maturation activity when combined with PPAR in malignantly transformed epithelial cells, and fetal small intestine. Hic-5 can even stimulate expression of PPAR-dependent epithelial genes when expressed in preadipose cells.

	Results

Top Abstract Results Discussion Materials and methods Acknowledgments References

 

Hic-5 binds and coactivates PPAR

We considered the possibility that PPAR may execute a particular program related to gut epithelial biology through the function of a tissue-selective coactivator protein. To identify candidate cofactors of PPAR, we used a yeast two-hybrid assay to screen a human colon cancer cDNA library, using the ligand-binding domain (LBD) of PPAR as bait. Two positive clones, each isolated multiple times, were identified as Hic-5, previously described as a TGF-inducible gene (Shibanuma et al. 1994) and as a LIM-domain coactivator of the androgen receptor in prostate cells, also known as androgen receptor activator 55 (ARA55) (Fujimoto et al. 1999). Hic-5 is known to be expressed in several cell types of epithelial origin: prostate, lung, small intestine, and colon (Shibanuma et al. 1994; Zhang et al. 2000). This pattern of expression along with the ability of Hic-5 to coactivate various nuclear receptors (Fujimoto et al. 1999; Yang et al. 2000) and to be regulated by TGF, an important regulator of the colonic epithelium, prompted our further studies.

The ability of Hic-5 to affect PPAR transcriptional activity was determined using Bosc cells transiently transfected with a luciferase reporter gene linked to a multiple copies of a PPAR response element (DR1). Hic-5 potentiated PPAR activity in a ligand-dependent manner (pioglitazone 1 µM) (Fig. 1A). PGC1, a bona fide coactivator of PPAR (Puigserver et al. 1998), was used as a positive control and showed similar stimulation of PPAR transcriptional activity, although as expected, this activity was largely ligand independent (Fig. 1A).

View larger version (28K): [in this window] [in a new window]   Figure 1. Hic-5 binds and coactivates PPAR in a ligand-depended manner (A) Bosc cells were transiently cotransfected with 50 ng of expression vector for PPAR and 50 ng of a luciferase reporter plasmid with three PPAR response elements (DR1-sites), in the presence of 250 ng Hic-5 or 100 ng PGC1 expression vectors. Empty vector was used as a control. The following day cells were treated with 1 µM pioglitazone (Pio). Cell lysates were then analyzed for relative transcriptional activity, and data were normalized to -gal activity, derived from a cotransfected -gal expressing vector. Values were expressed relative to activation of empty vector and presented as mean ± SE of three different experiments. (B) Coimmunoprecipitation assay. Bosc cells were transfected with either control vector or vector expressing Flag-PPAR. Cell lysates were then incubated with beads bound to an antibody directed against the Flag epitope. The interaction between Flag-PPAR and endogenous Hic-5 was tested in the presence and absence of rosiglitazone. Ectopic expression of gfp-PGC1 was used as a positive control for binding to PPAR. (C) Bosc cells were transiently transfected with various gfp-Hic-5 alleles. The cells were harvested, and extracts were incubated with immobilized GSTPPAR, which was purified from Sf21 baculovirus cells in the presence of PPAR ligand troglitazone (Tro). After washing, bound proteins were eluted and separated by SDS-PAGE, followed by Western blotting using an anti-gfp antibody. (D) In vitro interaction assay. Either GST or GST-PPAR was incubated with 35S-labeled Hic-5 alleles (shown on the right panel). SRC-1 was used as positive control to show in vitro binding to PPAR.

Hic-5 and PPAR are coexpressed in normal and malignant gut development

Gut endoderm develops into a structured epithelium between embryonic days 13 and 15 (E13 and E15) (Simon and Gordon 1995). We investigated whether Hic-5 and PPAR were coexpressed during gut development, using a fetal mouse intestine as a model. After microscopic dissection of intestines at E12.5 to E17.5, RNA was extracted and relative levels of Hic-5, PPAR mRNA, and several other markers of gut development were determined. As shown in Figure 2A, PPAR mRNA increases 2.2-fold between E12.5 and E13.5 and a further fourfold between E13.5 and E15.5. Hic-5 mRNA levels also increase twofold between E12.5 and E13.5 (Fig. 2A). This increase in PPAR and Hic-5 transcripts correlates with the induction of several molecular markers of gut epithelial maturation between E13.5 and E15.5 including keratin 18, keratin 19, and liver fatty-acid-binding protein (L-FABP) (Fig. 2A). To further examine whether PPAR and Hic-5 colocalize in the intestine, paraffinembedded sections of intestine from E13.5, E15.5, E17.5, and 2-wk-old mice were analyzed by immunohistochemistry. As shown in Figure 2B, PPAR protein is localized mainly to the epithelium of embryonic developing gut and in the mucosal lining of the adult villus, with a gradient of increasing expression toward the villus tip (Fig. 2B). Hic-5 shows a similar graded pattern of expression in epithelial cells (Fig. 2A), although its expression is not restricted to the epithelial layer in agreement with previous reports (Fig. 2B; Shibanuma et al. 1994; Yuminamochi et al. 2003). Importantly, these data illustrate that although Hic-5 expression pattern in the gut is broader than the expression of PPAR, PPAR and Hic-5 are colocalized only in the gut epithelial layer and are induced during epithelium development.

View larger version (62K): [in this window] [in a new window]   Figure 2. Hic-5 and PPAR are coexpressed in normal and malignant gut development. (A) Small intestines from mouse embryos were dissected at different stages of embryonic development, and total RNA was extracted. The level of expression of PPAR, Hic-5, keratin 19, keratin 18, and L-FABP was analyzed by using real-time RT-PCR assay. (B) Immunohistochemistry of PPAR and Hic-5 in sections of embryonic and adult mouse small intestines. Brown color indicates positive staining; blue denotes nuclear counter-staining. (C) Mice were treated with AOM as described earlier (Girnun et al. 2002). Colonic tumors and adjacent tissue were harvested, and RNA was extracted. Levels of PPAR, Hic-5, paxillin, keratin 20 KLF4, and PGC1 were analyzed by real-time RT-PCR. The levels of RNA expression were normalized with TBP and are mean±SD. (n = 5). Statistical analysis was done by using Student's t-test. () p < 0.01

Hic-5 enhances PPAR-mediated induction of epithelial genes, in colon cancer cells

A gene-array analysis was used to study the program of endogenous gene expression induced by PPAR in poorly differentiated epithelial cells. Moser human colon cancer cells were treated in the presence and absence of rosiglitazone for 14 h. Genes that showed statistically significant differences in expression between rosiglitazone-treated and -untreated cells were then classified according to their function: epithelial differentiation, cell adhesion, tumor suppressors, oncogenes, lipid metabolisms, and insulin signaling (Supplementary Fig. 1). Since these studies focused on expression of a gut epithelial program, genes classified as related to epithelial-selective functions or cell adhesion were further studied. To determine the role of Hic-5 in the expression of these epithelial cell-selective genes, Moser human colon cancer cells were then retrovirally infected with either a vector expressing Hic-5 or an empty vector, and treated with rosiglitazone; as shown earlier, these cells express relatively high amounts of endogenous PPAR protein (Sarraf et al. 1998). Figure 3A shows that in cells containing only the control vector, rosiglitazone induced several mRNAs characteristic of gut epithelium identified in our gene array analysis, such as keratin 20 (Moll et al. 1993; Gupta et al. 2001) L-FABP (Roth et al. 1990; Gupta et al. 2001), and KLF4 (Chen et al. 2003).

View larger version (23K): [in this window] [in a new window]   Figure 3. Ectopic expression of Hic-5 enhances PPAR-mediated induction of several epithelial markers in colon cancer cells. (A) Human colon cancer cells (Moser) were infected with a control retrovirus or a retrovirus expressing Hic-5. Cells were treated with or without 1 µM rosiglitazone and were then harvested, and RNA was extracted. Levels of mRNA for keratin 20, L-FABP, KLF4, and ADRP were quantified by real-time RT-PCR. Statistical analysis was done using Student's t-test. () p < 0.05; (NS) not significant. (B) Cells were treated as described in A. Levels of mRNA for BPG, CEA, annexin A1, and annexin A2 were determined by real-time RT-PCR analysis. Statistical analysis was done using Student's t-test. (NS) Not significant. (C) Moser cells were infected with either gfp or PGC1 expressing adenovirus. Cells were treated with or without 1 µM rosiglitazone and were then harvested, and RNA was extracted. Levels of mRNA for keratin 20, L-FABP, KLF4, and ADRP were quantified by real-time RT-PCR. Statistical analysis was done using Student's t-test. () p < 0.05; (NS) not significant.

We next asked whether the molecular phenotypes observed were specific for Hic-5 or occurred with the elevation of any PPAR coactivator. We therefore expressed another PPAR coactivator, PGC1 and examined its effects on epithelial gene expression. Interestingly, PGC1 was recently described to be down-regulated in human colon tumors (Feilchenfeldt et al. 2004). Figure 3C shows that PGC1 did not increase the PPAR-mediated induction of epithelial target genes. In fact, PGC1 expression decreased PPAR-mediated induction of keratin 20, and ADRP (Fig. 3C). These data indicate some selectivity for Hic-5 in mediating PPAR induction of epithelial-selective genes.

N terminus of Hic-5 functions as a dominant-negative allele and alters epithelial gene expression

Both the C and N termini of Hic-5 were previously suggested to function as dominant-negative (DN) alleles of Hic-5 (Fujimoto et al. 1999; Kasai et al. 2003). We therefore analyzed the ability of these fragments to alter PPAR-mediated transcription. As shown in Figure 4A, the C-terminal domain (209-444) retains some ability to coactivate PPAR, although the magnitude of this activity is much lower than that seen with full-length Hic-5. In contrast, the N terminus of Hic-5(1-208) lacks the ability to coactivate PPAR (Fig. 4A). To evaluate whether either of these Hic-5-derived proteins might have a suppressive effect on wild-type Hic-5 activity, the full-length protein was expressed with increasing concentration of either the N-terminal or C-terminal halves (Fig. 4B). Coexpression of Hic-5 and its N-terminal fragment resulted in a dose-dependent inhibition of PPAR coactivation (Fig. 4B, lanes 5-10), while this fragment had no significant effect on the ligand activation of PPAR in the absence of added wild-type Hic-5 (Fig. 4A). In contrast, coexpression of full-length Hic-5 with increasing amounts of the C-terminal fragment of Hic-5 resulted in a minor increase in PPAR coactivation (Fig. 4B, lanes 11-16). These results suggest that the N-terminal fragment of Hic-5 can function as a dominant-negative allele and can interfere with Hic-5 coactivation of PPAR. Importantly, this fragment does not appear to interfere significantly with PPAR activity in the absence of wild-type Hic-5.

View larger version (25K): [in this window] [in a new window]   Figure 4. TGF enhances PPAR induction of keratin 20, while dominant-negative Hic-5 blocks it. (A) Cos-1 cells were transiently cotransfected with 50 ng expression vector for PPAR, 50 ng luciferase reporter plasmid with three PPAR response elements (DR1-luc) in the presence of 250 ng gfp-Hic-5, 400 ng gfp-C terminus Hic-5(209-444), gfp-N terminus Hic-5(1-208) expression vectors, or gfp expressing vector as a control. Cells were treated with increasing concentrations of pioglitazone (Pio) as shown. Cell lysate were then analyzed for relative transcriptional activity, as described in Figure 1A. (B) Cos-1 cells were transiently tranfected and treated as described in A. To determine potential dominant-negative activities, Hic-5 expressing vector (100 ng) was cotransfected with increasing amount of either gfp-N terminus Hic-5 or gfp-C terminus Hic-5(100-400 ng). (C) Moser cells transduced with retroviral expressing Hic-5, dominant-negative Hic-5 (gfp-N terminus), or gfp were incubated with or without TGF or PPAR ligand (pioglitazone). After 2 d, total RNA and protein were extracted. (Top) Relative keratin 20 and PPAR mRNA levels were analyzed by real-time RT-PCR. Values of keratin 20 and PPAR were normalized to expression level without treatment in control cells, corrected to -actin and were mean ± SE of three different experiments. (Bottom) Protein lysates were extracted and separated by SDS-PAGE gels. Keratin 20 and GPDH levels were analyzed by Western blot using keratin 20 and GPDH antibodies respectively. GPDH levels are shown as a loading control.

siRNA directed against Hic-5 inhibits PPAR induction of epithelial-selective genes

siRNA directed against Hic-5 were designed and cloned into a retroviral vector. Scrambled oligonucleotides cloned into the same vector were used as a control. Moser cells were infected with both control and siRNA retroviruses. siRNA against Hic-5 was able to downregulate Hic-5 mRNA levels by at least 60% (Fig. 5A) and had an even greater effect on the level of Hic-5 protein (Fig. 5A, inset). Infected cells were treated with or without rosiglitazone. As expected, rosiglitazone was again able to induce L-FABP, KLF4, keratin 20, and ADRP in control cells (Fig. 5B). The expression of the siRNA against Hic-5 did not significantly change the level of these genes in the absence of PPAR agonist (Fig. 5B). However, the ability of the PPAR ligand-mediated induction of each transcript was dramatically decreased in cells expressing the siRNA against Hic-5, while the induction of ADRP was not affected (Fig. 5B). Together, these loss-of function data, using both a dominant-negative allele of Hic-5 and Hic-5-specific siRNA, indicate that Hic-5 and PPAR cooperate specifically to induce multiple genes characteristic of gut epithelial differentiation.

View larger version (22K): [in this window] [in a new window]   Figure 5. siRNA against Hic-5 inhibits PPAR induction of certain epithelial genes. (A) Moser cells infected with siRNA expressing vector (pSuper-neo) directed against Hic-5 (siRNA-5) mRNA or scrambled oligonucleotides as a control. After antibiotic selection cells were treated with and without 1 µM rosiglitazone. (Inset) RNA and proteins were isolated, and Hic-5 mRNA levels were determined by real-time RT-PCR and Western blot analysis. The levels of RNA expression were normalized with TBP and are mean ± SD of three different experiments. Statistical analysis was done using Student's t-test. () p < 0.01. (B) Cells were treated as described in A. RNA was isolated, and levels of mRNA for, KLF4, L-FABP, keratin 20, and ADRP were analyzed by real-time RT-PCR. Statistical analysis was done using Student's t-test. () p < 0.05. (C) Moser cells infected with adenovirus expressing either siRNA directed against PGC1 mRNA (siRNA-) (Koo et al. 2004), or scrambled oligonucleotide as a control. After 48 h cells were harvested, and levels of PGC1 and its target gene ERR were measured by real-time PCR. Analysis was done as described in A. (D) Moser cells were infected as described in C. After infection cells were treated with or without rosiglitazone (1 µM). Levels of KLF4, keratin 20, L-FABP, and ADRP were analyzed as described in B.

Modulation of PPAR and Hic-5 levels in gut development

Mice with total knock out of PPAR are not viable, but heterozygous mice have normal viability (Barak et al. 1999; Kubota et al. 1999). Therefore we used small intestines from PPAR heterozygous and wild-type mice (Akiyama et al. 2002) to examine the role of PPAR in the normal process of gut epithelial differentiation. Figure 6A shows that small intestines harvested from PPAR heterozygous (PPAR^+/-) mice express 50% less mRNA for PPAR compared with small intestines from Ppar wild-type (PPAR^+/-) mice (Fig. 6A, upper left panel). Importantly, small intestines from PPAR^+/- mice also showed 50% reduction in the expression of several mRNA markers of gut epithelium (Fig. 6A): KLF4, keratin 20, and keratin 19 L-FABP, and keratin 18 mRNA levels showed the same trend, albeit, not statistically significant. These results further demonstrate a physiological role for PPAR in gut epithelial differentiation.

View larger version (25K): [in this window] [in a new window]   Figure 6. Modulation of PPAR and Hic-5 expression in epithelial gut development. (A) Small intestines were isolated from PPAR+/+ and PPAR+/- mice (Akiyama et al. 2002). After RNA isolation levels of PPAR keratin 20, keratin 19, keratin 18, KLF4, and L-FABP were determined using real-time RT-PCR analysis, as described in Figure 5 (n = 6). (B) Small-intestines were isolated from mice embryos between E12 and E13 and were injected with either siRNA directed against Hic-5 (siRNA-5-I, siRNA-5-II) or scrambled oligonucleotide expressing vector, followed by electroporation to express the plasmids in the epithelial layer. Embryonic small-intestines were then grown for 2 d ex vivo under differentiation condition, and protein lysates were harvested as described in Tou et al. (2004). Hic-5, keratin 19, and L-FABP protein levels were analyzed by Western blot analysis, using GPDH as loading control. Relative protein expression was determined by using densitometer analysis of the Western blot corrected for GPDH expression levels. This is a representative experiment (n = 5). (C) Embryonic small intestines were isolated and treated as described in B, only Hic-5 expressing vector and an empty vector as a control were used. RNA was isolated, and Hic-5, keratin 20, and KLF4 mRNA levels were analyzed using real-time RT-PCR as described in Figure 5.

We also used the same ex vivo system to study the effects of ectopic Hic-5 expression. Figure 6C shows that Hic-5 mRNA levels were increased threefold in embryonic small intestines injected with Hic-5 expression vector compared with the control injected embryos. This increase caused a threefold increase in keratin 20 levels and a similar trend in KLF4 mRNA levels (Fig. 6C). Taken together, manipulation of Hic-5 expression levels in the fetal mouse intestine resulted in distinct modulation of gut epithelial gene expression. These data strongly suggest a physiological role for Hic-5 in gut epithelial development.

Expression of Hic-5 alters the program of adipocyte differentiation.

Preadipocytes are mesenchymal cells that are ontologically far removed from epithelial cell lineages and are not known to express bona fide markers of epithelial cells. We therefore, examined the effects of forced Hic-5 expression during adipogenesis, a pathway where PPAR plays a dominant role. 3T3-L1 preadipocytes were infected with either an empty retroviral vector or a Hic-5 expressing vector, and the resulting cells were subjected to a standard protocol of differentiation, with or without the addition of a PPAR ligand for 4 d. Figure 7A shows that while endogenous Hic-5 is expressed at detectable levels in 3T3-L1 preadipocytes, these levels decrease during differentiation. Interestingly, forced expression of Hic-5 inhibited morphological aspects of adipogenesis, as shown by phase contrast microscopy and by the accumulation of lipids, as illustrated by Oil-Red-O staining (Fig. 7B). Furthermore, several mRNAs characteristic of adipocyte differentiation, such as aP2, adipsin, and acrp30, were all reduced with forced Hic-5 expression. This effect was even more dramatic when PPAR ligand was added to the differentiation mixture (Fig. 7C). Notably, the cells infected with empty vector do not express several mRNAs characteristic of gut epithelial differentiation, such as keratin 20, L-FABP, and KLF4. Expression of Hic-5 alone mildly increased the levels of these genes. However, when cells expressing Hic-5 were treated with rosiglitazone, robust expression of these epithelial genes was observed (Fig. 7C). These data indicate that Hic-5 and PPAR can collaborate to induce inappropriate expression of genes characteristic of epithelial cells, even in a classical model of adipose differentiation.

View larger version (39K): [in this window] [in a new window]   Figure 7. Ectopic expression of Hic-5 in 3T3-L1 preadipocytes inhibits adipocyte differentiation while inducing markers of epithelial gut differentiation. (A) Hic-5 mRNA levels during 3T3-L1 adipogenesis were analyzed using real-time RT-PCR. (B) 3T3-L1 preadipocytes were infected with either retrovirus expressing Hic-5 or an empty retrovirus. Cells were then differentiated using a standard protocol with addition of rosiglitazone. After 4 d cells were fixed, and phase-contrast pictures were taken (top panel) or stained for lipid droplet formation by Oil-Red-O (bottom panel). (C) 3T3-L1 preadipocytes were treated as described in B. After 4 d of differentiation with or without rosiglitazone, cells were harvested and RNA was extracted. Levels of mRNA for adipocyte markers adipsin, acrp30, and aP2 and gut epithelial markers KLF4 L-FABP and keratin 20 were determined by real-time quantitative RT-PCR. Statistical analysis was done using Student's t-test. () p < 0.05. (D) Illustration of Hic-5 and PPAR combination in epithelial differentiation.

	Discussion

Top Abstract Results Discussion Materials and methods Acknowledgments References

Hic-5, a member of the LIM domain family with striking similarity to paxillin (Shibanuma et al. 1994; Yuminamochi et al. 2003), can shuttle between focal adhesions and the nucleus, suggesting a possible role in relaying cytoplasmic signals into the nucleus (Shibanuma et al. 2003). LIM domain are implicated in various biological processes, including cell lineage determination and organ development (for reviews, see Bach 2000; Gill 2003). Hic-5, in particular, was previously shown to be expressed in epithelial cells (Zhang et al. 2000) and correlated with differentiation processes induced by retinoic acid (Shibanuma and Nose 1998). Indeed, all the genes we examined here that are induced through the combination of Hic-5 and PPAR are related to gut epithelial differentiation. Keratin 20 belongs to the large family of cytokeratins that are expressed almost exclusively in epithelial cells, in this case mainly in mature gastrointestinal epithelium, urothelium, and Merkel cells (Moll et al. 1990); keratin 20 expression is also used to distinguish colon from other cancers in clinical pathology (Nishizuka et al. 2003). L-FABP, a fatty-acid-binding protein, is highly expressed in hepatocytes and enterocytes and is an established marker of intestinal epithelial differentiation and development (Roth et al. 1990). KLF4 is a zinc-finger transcription factor that is enriched in postmitotic gut epithelial cells (Dang et al. 2003; Hinnebusch et al. 2004; Zhao et al. 2004) and required for goblet cell differentiation in the colon (Katz et al. 2002). Importantly, robust induction of these genes required the combined activity of PPAR and Hic-5; expression of Hic-5 alone was not sufficient to activate this program.

Hic-5 has also been implicated in tumorogenesis, although the published literature is ambivalent about its precise role. Hic-5 (ARA55) activates the androgen receptor in prostate cancer and was therefore initially suggested to promote prostate cancer cell growth (Fujimoto et al. 1999; Rahman et al. 2003). However, it was reported more recently that expression of ARA55 (Hic-5), but not that of other androgen receptor coactivators such as ARA54 or ARA70, is decreased in prostate tumors (Mestayer et al. 2003). Consistent with these data, we show in this study that Hic-5 expression is also decreased in AOM-induced colon tumors, further suggesting that Hic-5 is unlikely to be a positive effector of carcinoma progression. The fact that Hic-5 and PPAR collaborate to induce expression of L-FABP, keratin 20, and KLF4, markers of gut maturation, is more consistent with a role for Hic-5 as a tumor suppressor.

Finally, our data suggest a molecular mechanism by which the tumor suppressor activities of TGF and PPAR may be integrated. Hic-5 was originally isolated as a TGF-inducible gene (Shibanuma et al. 1994), and TGF regulates a well-established tumor-suppressor signaling pathway in colon cancer that enhances epithelial differentiation and inhibits cell growth. In advanced stages of colon cancer, aberrant TGF signaling can enhance malignancy, epithelial to mesenchymal transition (EMT), and metastasis (for reviews, see Akhurst and Derynck 2001; Derynck et al. 2001). Genetic studies, including in human colon cancer patients (Sarraf et al. 1999), implicate PPAR as a tumor suppressor. PPAR also regulates cancer cell growth and differentiation in vitro (Sarraf et al. 1998; Gupta et al. 2003). Much like TGF, PPAR has also been shown to be an effective tumor suppressor in the colon early in the process of carcinogenesis. In the presence of damage to the APC gene, PPAR lost its protective function (Girnun et al. 2002) or even enhanced cancer progression (Lefebvre et al. 1998; Saez et al. 1998). It therefore seems likely that induction of Hic-5 by TGF enhances the ability of PPAR to modulate the program of epithelial differentiation in vivo.

Mechanism-based strategies to promote tumor differentiation represent a promising new area in cancer research. The best example of differentiation therapy to date is the use of all-trans-retinoic acid (ATRA) to treat acute promyelocytic leukemia (APL) (Zelent et al. 2001; Fang et al. 2002; Tallman 2004). The ability of ATRA to induce terminal differentiation of malignant myeloid cells to mature neutrophils depends on the expression of a chimeric PML-RAR fusion protein resulting from a t(15;17) chromosomal translocation (Zelent et al. 2001; Fang et al. 2002; Tallman 2004). By analogy, the use of PPAR agonists in differentiation therapy has thus far shown promising results in certain liposarcoma patients, where rosiglitazone induced terminal adipocytic differentiation and tumor regression (Demetri et al. 1999). While clinical trails of PPAR agonists as the sole therapy in advanced epithelial tumors to date have failed to show significant clinical effects (Kulke et al. 2002), new trials will likely combine PPAR ligands with other chemotherapy agents. This study suggests that agonists that enhance interaction of PPAR with Hic-5 could be useful in differentiation therapy of colon and other epithelial cancers.

	Materials and methods

Top Abstract Results Discussion Materials and methods Acknowledgments References

 

Yeast two-hybrid screening

A cDNA encoding amino acids 183-505 of the murine PPAR was cloned in-frame into the GAL4 DNA-binding domain of pGBKT7 (pGBKT7-PPAR, Clontech). An SW480 colon cancer cell cDNA expression library was constructed into the pGAD10 plasmid expressing the GAL4 activation domain (custom generation, Clontech). A yeast two-hybrid screening was performed as described in the Clontech Matchmaker two-hybrid system protocol. Briefly, pGBKT7-PPAR was transformed into Y190 yeast cells by the lithium acetate method and maintained by selection (Y190-PPAR). The pGAD10 plasmid containing the colon cancer cell cDNA library was transformed into Y190-PPAR yeast cells, and positive clones were assayed for -galactosidase activity in a filter assay as described in the Clontech protocol.

Cell culture and transient transfection/reporter assays

293T, Cos-1, Bosc, 3T3-L1, and Moser colon cancer cell lines were cultured in DMEM containing 10% FBS supplemented with penicillin/streptomycin. Sf21 insect cells were cultured in Grace's insect media containing 10% FBS. Bosc or Cos-1 cells were transiently transfected with a PPAR expression vector; a PPAR response element fused to a luciferase construct (DR1-luc); and either a control vector, vectors expressing Hic-5, and/or Hic-5 deletion constructs or PGC-1 expression constructs using Superfect transfection reagent (Qiagen). Cells were treated with various ligand concentrations, and luciferase levels were assayed after 18-24 h of ligand exposure.

Protein interaction analysis

Bosc cells were transfected with a control vector, vectors expressing Flag-PPAR, or gfp-PGC1 as a positive control. Cell lysates were incubated with agarose beads bound to an anti-Flag antibody (Sigma) in the presence or absence of rosiglitazone. Beads were washed and proteins separated on an 8% gel by PAGE and transferred to PVDF. Immunoblotting was performed using antibodies against Hic-5 (BD Transduction Labs), PPAR (Santa Cruz Biotech), or gfp (a kind gift from Dr. Pam Silver, Dana-Farber Cancer Institute, Boston, MA). GST-PPAR was produced using a baculovirus expression system. A cDNA expressing full-length PPAR was cloned into a pAcGHLT baculovirus expression plasmid (Pharmingen). Sf21 cells were infected with the baculovirus expressing GST-PPAR for 2 d and then harvested into lysis buffer. GST-PPAR protein was isolated by incubation with GSH-Sepharose beads and protein quantified by Commassie blue staining. Affinity purification was performed using 1 µg GST-PPAR protein with cell lysate from Bosc cells transfected full-length gfp-Hic-5 or an N-terminal or C-terminal gfp-Hic-5 deletion constructs (gfp-Nterm and gfp-Cterm, respectively) as shown in Figure 1C (a kind gift from Drs. Nose and Shibanuma, Showa University School of Pharmaceutical Sciences, Tokyo, Japan) (Fujita et al. 1998). Briefly, GST-PPAR bound to GSH beads was incubated for 2 h with cell lysates at 25°C. Beads were washed and proteins resolved on an 8% SDS gel. Proteins were transferred to PVDF and immunoblotted using an antibody against gfp (a kind gift from Dr. Pam Silver). Mapping of Hic-5 domains that interact with PPAR was performed by using full-length Hic-5 or Hic-5 deletion constructs, as shown in Figure 1D (a kind gift from Drs. Nose and Shibanuma), or SRC-1 as a positive control. Proteins were translated in vitro in the presence of ³⁵S-methionine according to manufacturer protocol (Promega). GST-PPAR bound to GSH beads was incubated with in vitro translated protein in binding buffer at room temperature for 1 h. Following washing, proteins were eluted and resolved on a 4%-12% gel (Invitrogen) by PAGE. Gels were dried down and exposed to film.

Retroviral preparation and infections

A full-length Hic-5 cDNA was generated by RT-PCR from mouse spleen RNA using Invitrogens TOPO cloning kit. The resulting cDNA was sequenced and subsequently cloned into a pMSCV retroviral vector (Clontech). siRNA against Hic-5 was designed using Oligoengines online design program. The 293T packaging cell line was transfected using Lipofectamine 2000 (Invitrogen). Cells were transfected with expression vectors for vsv and gag-pol and with the following retroviral vectors: pBabe expressing gfp, gfp-Hic-5, or gfp-Nterm, pmscv-puro, pmscv-puro expressing Hic-5, pSuperRetro-neo control, or pSuperRetro-neo siRNA against Hic-5. After 6-12 h, media was replaced, and cells were transferred to 33°C for 24-72 h. Viral supernatants were collected and filtered. Moser or 3T3-L1 cells were plated and infected on the following day with the viral supernatants and 8 µg/mL polybrene. 3T3-L1 cells were subjected to a differentiation protocol when confluent (see below). Moser cells were split 1:3 and selected with either puromycin or neomycin for 1 wk, or sorted for gfp-expressing cells.

Adipocyte differentiation and Oil-Red-O staining

For differentiation assays, confluent 3T3-L1 cells infected with control or Hic-5 expressing retrovirus were treated with or without 1 µM rosiglitazone plus 0.5 mM 3-isobutyl-1-methylxanthine, 5 µg/mL insulin, and 1 µM dexamethasone for 2 d. Cells were then kept in maintenance medium (with or without rosiglitazone) consisting of DMEM with 10% FBS supplemented with penicillin/streptomycin and 5 µg/mL insulin. RNA was harvested and cDNA prepared as described below. Differentiated 3T3-L1 cells were also washed in PBS and then fixed in 10% buffered formalin (Formaldefresh; Fisher). A stock solution of 0.5% Oil-Red-O (Sigma) in isopropanol (w/v) was diluted 60: 40 in water, filtered, and added to fixed cells. Cells were then washed in water and photographed.

Preparation of tissue for RNA and immunohistochemistry

Small intestines were isolated from either E12.5-E17.5 mouse embryos as previously described (Tou et al. 2004), or from adult mice. For immunohistochemistry analysis, isolated small intestine was rinsed with PBS and fixed in 4% paraformaldehyde (EM sciences). Paraffin embedding, sectioning, and immunohistochemistry were performed by the Pathology Core at Brigham and Women's Hospital. A mouse-on-mouse kit was used for Hic-5 antibody staining (Vector labs). Immunohistochemistry was performed using an anti-mouse monoclonal against Hic-5 (1:100) and an anti-rabbit polyclonal antibody against PPAR (1:200). For gene expression analysis, RNA was isolated, cDNA was synthesized, and mRNA levels were analyzed using realtime PCR analysis. For tumor analysis, 8-wk-old C57Bl/6J mice were treated with the colon-specific carcinogen AOM as previously described (Burdette Walter 1970; Girnun et al. 2002). Tumor and adjacent uninvolved tissue were excised and snap frozen in liquid nitrogen. RNA isolation, cDNA synthesis, and real-time PCR analysis of tumor, normal adjacent and embryonic tissues were performed as described below.

Ex vivo embryonic gut development

Day 12.5 fetal mouse intestines were isolated under a dissecting microscope as previously described (Tou et al. 2004). Briefly, to introduce expression vector into explants, 2 µg of control plasmid or plasmid encoding either Hic-5 or siRNA directed against Hic-5 was injected into the intestinal lumen by using a capillary pipette and Nanoject injection device. Explants were bathed in DMEM, placed between platinum electrodes and exposed to three 10-msec pulses of 60 V each by using the BTX830 squarewave pulse electroporator, washed in DMEM, and cultured. After 2 d in culture, explants were harvested in RIPA buffer for protein analysis or subjected to RNA isolation using Trizol.

Transcriptional profiling

Moser human colon cancer cells were treated for 14 h with vehicle control or 1 µM rosiglitazone. Total RNA was prepared using an RNeasy Mini Kit (Qiagen Inc.) and used for global expression analysis. Affymetrix array hybridization and scanning were performed by the Core Facility at Dana-Farber Cancer Institute using human genome U133A chip (Affymetrix). Array data were analyzed with d-CHIP array analysis program (Li and Wong 2001).

Real-time PCR

Total RNA was prepared from cells or tissues using Trizol reagent (Invitrogen) following the manufacturer's instructions. Briefly, chloroform was added to homogenized samples, and the aqueous layer was isolated following centrifugation. RNA was precipitated with isopropanol and washed in 70% ethanol. One microgram of total RNA was DNase treated, and cDNA was synthesized using iScript reverse transcriptase reagent (Bio-Rad). PCR reactions were performed using SYBR Green PCR Master Mix (Bio-Rad). Real-time PCR reactions were carried out using the following conditions: for 2 min at 50°C and for 10 m in at 95°C, followed by 40 cycles of switching between 95°C for 15 sec and 60°C for 1 min (iCycler, Bio-Rad). The cDNA synthesis step included a control reaction without the reverse transcriptase, and the PCR step included a control reaction without the template to rule out contamination and/or genomic amplification. In addition, the TATA-box-binding protein (TBP) gene was used as a control to account for possible variations in initial RNA quantity and efficiency of the cDNA synthesis reaction. Primers for target genes were designed using the PRIMEREXPRESS software (Applied Biosystems). Sequences for real-time PCR primers are available upon request.

	Acknowledgments

Top Abstract Results Discussion Materials and methods Acknowledgments References

 
We thank Drs. Kiyoshi Nose and Motoko Shibanuma for Hic-5 plasmids, Dr. Frank Gonzalez for PPAR targeted mice, Dr. Pam Silver for GFP antibody, and Dr. Marc Montminy for adenovirus expressing siRNA against PGC1. We extend our appreciation to the Dana-Farber Cancer Institute Microarray Core for expert processing of RNA samples, and Flow Cytometry Core for FACS analysis and the Pathology Core at Brigham and Women's Hospital for processing immunohistochemistry of mice samples. We also thank Dr. Sibylle Jaeger and Dr. Christopher Walkey for helpful discussions and careful reading the manuscript, Dr. Melina Fan for real-time PCR data on PGC1, and Adah Levens for administrative assistance. This work was supported by NIH grants RO1DK57670 (B.M.S), DK064685 (G.D.G)., and RO1DK61139 (R.A.S). S.D was supported by a grant from Friends of the Dana-Farber Cancer Institute and by the ADA Mentor based Postdoctoral Fellowship.

	Footnotes

 
Supplemental material is available at http://www.genesdev.org.

Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/gad.1240705.

⁴ These authors contributed equally to this work.

⁵ Corresponding author.
E-MAIL bruce_spiegelman{at}dfci.harvard.edu; FAX (617) 632-5363.

	References

Top Abstract Results Discussion Materials and methods Acknowledgments References

 
Akhurst, R.J. and Derynck, R. 2001. TGF- signaling in cancer: A double-edged sword. Trends Cell. Biol. 11: S44-S51.[Medline]

Akiyama, T.E., Sakai, S., Lambert, G., Nicol, C.J., Matsusue, K., Pimprale, S., Lee, Y.H., Ricote, M., Glass, C.K., Brewer Jr., H.B., et al. 2002. Conditional disruption of the peroxisome proliferator-activated receptor gene in mice results in lowered expression of ABCA1, ABCG1, and apoE in macrophages and reduced cholesterol efflux. Mol. Cell. Biol. 22: 2607-2619.[Abstract/Free Full Text]

Bach, I. 2000. The LIM domain: Regulation by association. Mech. Dev. 91: 5-17.[CrossRef][Medline]

Barak, Y., Nelson, M.C., Ong, E.S., Jones, Y.Z., Ruiz-Lozano, P., Chien, K.R., Koder, A., and Evans, R.M. 1999. PPAR is required for placental, cardiac, and adipose tissue development. Mol. Cell 4: 585-595.[CrossRef][Medline]

Burdette Walter, J. 1970. Carcinoma of the colon and antecedent epithelium. Cancer Res. 30: 253-256.[Free Full Text]

Burstein, H.J., Demetri, G.D., Mueller, E., Sarraf, P., Spiegelman, B.M., and Winer, E.P. 2003. Use of the peroxisome proliferator-activated receptor (PPAR) ligand troglitazone as treatment for refractory breast cancer: A phase II study. Breast Cancer Res. Treat. 79: 391-397.[CrossRef][Medline]

Calnek, D. and Quaroni, A. 1993. Differential localization by in situ hybridization of distinct keratin mRNA species during intestinal epithelial cell development and differentiation. Differentiation 53: 95-104.[CrossRef][M