Dev Genes Evol (2006) 216: 109–118DOI 10.1007/s00427-005-0037-4

ORIGINAL ARTICLE

Masayuki Ikuzawa . Katsuhiko Shimizu .Shigeki Yasumasu . Ichiro Iuchi . Yun-Bo Shi .Atsuko Ishizuya-Oka

Thyroid hormone-induced expression of a bZip-containing

transcription factor activates epithelial cell proliferation

during Xenopus larval-to-adult intestinal remodeling

Received: 14 April 2005 / Accepted: 29 September 2005 / Published online: 15 November 2005# Springer-Verlag 2005

Abstract In the intestine during amphibian metamorpho-sis, stem cells appear, actively proliferate, and differentiateinto an adult epithelium analogous to the mammaliancounterpart. To clarify the molecular mechanisms regulat-ing this process, we focused on a bZip-containing tran-scription factor (TH/bZip). We previously isolated TH/bZip from the Xenopus intestine as one of the candidategenes involved in adult epithelial development. Northernblot and in situ hybridization analyses showed that thetransient and region-dependent expression of TH/bZipmRNA correlates well with the growth of adult epithelialprimordia originating from the stem cells throughout theXenopus intestine. To investigate its role in the adultepithelial development, we established an in vitro genetransfer system by using electroporation and organ culturetechniques, and we overexpressed TH/bZip in the epithe-lium of Xenopus tadpole intestines. In the presence ofthyroid hormone (TH) where the adult epithelial primordiaappeared after 3 days of cultivation, overexpression of TH/bZip significantly increased their proliferating activity. On

the other hand, in the absence of TH where the epitheliumremained as larval-type without any metamorphic changes,ectopic expression of TH/bZip significantly increased theproliferating activity of the larval epithelium but had noeffects on its differentiated state. These results indicate thatTH/bZip functions as a growth activator during amphibianintestinal remodeling, although TH/bZip expression in theepithelium alone is not sufficient for inducing the stemcells.

Keywords Intestinal remodeling . TH/bZip .Electroporation . Organ culture . Xenopus

Introduction

During amphibian metamorphosis, the intestinal epitheliumdramatically changes from the larval to adult form to adaptto the terrestrial carnivorous life. Most of the larvalepithelium degenerates by apoptosis (Ishizuya-Oka andUeda 1996). The primordia of the adult epithelium appearin the basal region of the larval epithelium as isletsconsisting of a small number of undifferentiated cells.These cells express Musashi-1 (Ishizuya-Oka et al. 2003),which is proposed to be a marker for mammalian intestinalstem cells and their immediate descendents (Booth andPotten 2000; Potten et al. 2003). They actively proliferateand finally differentiate into absorptive epithelial, goblet andendocrine cells similar to the mammalian ones (Hourdryand Dauca 1977; McAvoy and Dixon 1977, 1978; Shi andIshizuya-Oka 1996). Thus, the adult epithelial primordiainclude multipotential stem cells analogous to those in themammalian intestine. The amphibian intestine duringmetamorphosis provides a unique model system for thestudy of the regeneration of the epithelium originatingfrom the stem cells.

It is well known that amphibian metamorphosis is con-trolled essentially by a single hormone—the thyroid hor-mone (TH) (Dodd and Dodd 1976; Kikuyama et al. 1993;Shi 1999). To understand its molecular mechanisms, anumber of TH response genes have been isolated from the

Communicated by R.P. Elinson

M. Ikuzawa . S. Yasumasu . I. IuchiLife Science Institute, Sophia University, Chiyoda-ku,Tokyo, 102-8554, Japan

K. ShimizuYoshizato Project, CLUSTER,Hiroshima Prefectural Institute of IndustrialScience and Technology, Higashi-Hiroshima,Hiroshima, 739-0046, Japan

Y.-B. ShiLaboratory of Gene Regulation and Development,National Institute of Child Healthand Human Development,Bethesda, MD 20892-5431, USA

A. Ishizuya-Oka (*)Department of Biology, Nippon Medical School,Kawasaki, Kanagawa 211-0063, Japane-mail: a-oka@nms.ac.jpTel.: +81-44-7333580Fax: +81-44-7221231

tail (Wang and Brown 1993), the hindlimb (Buckbinderand Brown 1992), the brain (Denver et al. 1997), and theintestine (Shi and Brown 1993; Amano and Yoshizato1998) of Xenopus laevis tadpoles. However, only limiteddata are available regarding the functions of TH responsegenes in the larval-to-adult remodeling of these organs. Forexample, in the intestine, a TH response gene encoding thematrix metalloproteinase, stromelysin-3 plays importantroles in the apoptosis of the larval epithelium (Shi andIshizuya-Oka 2001). However, little is known about genesinvolved in the development of the adult epithelium.

We have previously discussed that the adult epitheliumcan develop in the presence of TH in organ cultures of theanterior part of the intestine isolated from X. laevis premeta-morphic tadpoles (Ishizuya-Oka and Shimozawa 1992),suggesting that all genes necessary for adult epithelial de-velopment are expressed organ-autonomously. To identifysuch genes, we have isolated 16 cDNA clones from culturedanterior intestines (Shimizu et al. 2002). Among them, wefocused on two clones whose expression was most dra-matically up- and then down-regulated (more than 20-fold)during intestinal remodeling. One of them was previouslyidentified as Xtld, a Xenopus homolog of DrosophilaTolloid closely related to bone morphogenetic protein-1(BMP-1) and was proposed to act on adult epithelial de-velopment by regulating BMP-4 (Shimizu et al. 2002).Here, we identified the other clone as a TH/bZip, a memberof the basic leucine zipper family of transcription factorsthat was originally isolated from the X. laevis tail (Wangand Brown 1993) and the intestine (Shi and Brown 1993)as a TH response gene. During amphibian metamorphosis,TH/bZip is also expressed in the head and the hindlimb(Denver et al. 1997; Berry et al. 1998a), suggesting itsimportant roles common in all of these organs. AlthoughTH/bZip shares homology with the mammalian E4BP4gene encoding a bZip transcriptional repressor (Cowellet al. 1992), its function is expected to be different fromthe mammalian E4BP4. This is because the repressiondomain of E4BP4 (Cowell and Hurst 1994) is almost ab-sent in the TH/bZip protein (Brown et al. 1996).

To investigate the possible roles of TH/bZip during theintestinal remodeling, we first examined a relationshipbetween the expression of TH/bZip mRNA and the adultepithelial development, which progresses from the anteriorto the posterior parts of the X. laevis small intestine(Ishizuya-Oka et al. 1997a). In addition, we examined theexpression profile of TH/bZip mRNA in the stomach,another major digestive organ. The larval epithelium in thestomach consists of the surface epithelium and glands.During metamorphosis, the primordia of the adult epithe-lium appear in the basal region of the larval glands andreplace the degenerating larval epithelial cells similar tothose in the intestine (Ishizuya-Oka et al. 1998). In bothorgans, we found that TH/bZip expression closely cor-relates with the growth of the adult epithelial primordia,especially with active cell proliferation. Furthermore, wedeveloped a gene transfer system by using electroporation,which has been recently used for the study of avian em-bryonic gut (Sakamoto et al. 2000; Yasugi and Nakamura

2000). By combining this with the organ culture techniquesthat we established earlier (Ishizuya-Oka and Shimozawa1991), we demonstrated for the first time that TH/bZipfunctions to promote the proliferation of adult epithelialcells during amphibian intestinal remodeling.

Materials and methods

Animals

Tadpoles of the South African clawed frog (X. laevis) atstages 57–66 (Nieuwkoop and Faber 1967) were purchasedfrom local suppliers and used throughout the experiments.

Cloning of full-length TH/bZip cDNA

To obtain the full-length sequence of a 786-bp cDNAfragment isolated from the anterior part of the TH-treatedsmall intestines (IA009; Shimizu et al. 2002), we screeneda cDNA library constructed from the cDNA of guts atstages 61–62 by using a λ Zap II cDNA library kit(Stratagene). Since the 3.3-kb cDNA obtained did notcontain any protein-coding region, we performed 5′ rapidamplification of cDNA ends-polymerase chain reaction(RACE-PCR). The cDNA was synthesized from the RNAof the guts at stages 61–62 by a Marathon cDNA Am-plification kit (Clontech). The first PCR was performedusing an adapter primer 1 and a gene specific primer(5′- ACCTCCAGTGTGTTCAAGCTAGTGTTTCCC-3′).The nested PCR was performed using an adapter primer 2and a nested gene specific primer (5′-TACACACACCCTCCATCTCACAGGAGG-3′). After screening and performingthe 5′ RACE-PCR, a full-length cDNA (4,549 bp) was ob-tained. The cDNA consisted of a 5′-untranslated region(189 bp), an open reading frame (1,005 bp) encoding 335amino acids, and a long 3′-untranslated region (3,355 bp).The nucleotide sequence of the coding region shares 99.6%homology with that of TH/bZip B (Brown et al. 1996;Accession no. U41859) whose nucleotide sequence lacks a3-kb 3′-untranslated region. We then determined the full-length of the cDNA encoding the TH/bZip B protein.

Northern blot analysis

The total RNAwas extracted from the stomach, the anterior(from the bile duct junction to the end of the typhlosole) andthe posterior parts (about two spirals just before the largeintestine) of the small intestine, and the large intestine oftadpoles at various stages by using TRIZOL solution(Invitrogen). The total RNAwas also extracted from wholesmall intestine of stage 57 tadpoles treated with 5 nM 3,3′,5-triiodo-L-thyronine (T3) for up to 4 days. Digoxigenin(DIG)-labeled DNA probe was prepared from IA009cDNA in pCR2.1 by polymerase chain reaction with TaqDNA polymerase (QIAGEN) and DIG-11-deoxyuridinetriphosphate (Roche Diagnostics). Northern blotting was

110

performed with Northern Starter Kit (Roche Diagnostics)according to the manufacturer’s protocol. In short, 1 μg ofthe total RNA was separated in a 1% agarose gel byelectrophoresis, transferred onto positively charged nylonmembranes, and hybridized with the DIG-labeled probe byusing a Max-Gly Northern Blotting Kit (Ambion). Todetect signals, the membranes were incubated with alkalinephosphatase-conjugated anti-DIG antibody and subse-quently with CDP-Star. Chemiluminescence images wereobtained through Lumi-Imager (Roche Diagnostics). Tonormalize the amount of RNA loaded, the hybridizedprobes were stripped off the membranes and rehybridizedwith elongation factor (EF) 1α cDNA probe.

In situ hybridization

The procedures for in situ hybridization were the same asdescribed previously (Ishizuya-Oka et al. 1994). In brief,DIG-labeled antisense and sense probes were preparedaccording to the instruction of a DIG RNA labeling kit(Roche Diagnostics) by using IA009 cDNA in pCRII.Tissue fragments isolated from the stomach, the anteriorand posterior parts of the small intestine, and the largeintestine were fixed with 4% paraformaldehyde in phos-phate buffered saline (pH 7.4) at 4°C for 4 h. They werethen frozen on dry ice and cut at 7 μm. Hybridization buffercontaining DIG-labeled antisense RNA probe (100 ng/ml)was applied to the sections. After hybridization at 42°C for16 h, the hybridized DIG-probes were immunologicallydetected with a DIG-probe Detection Kit (Roche Diagnos-tics) and examined microscopically. As controls, somesections were hybridized with DIG-labeled sense RNAprobe (100 ng/ml). All control sections showed only thebackground levels of the signals (data not shown).

Construction of plasmids expressing TH/bZip

To construct an expression vector for the TH/bZip, theprotein-coding region of TH/bZip was amplified by PCRusing the following primers: forward primer comprising aSacI restriction site and the first ATG codon at 5′-end,5′ -ATATGAGCTCATGAACGGCCCTGTTCCTTGCCCAAC-3′, and reverse primer comprising a XmaI restrictionsite and a stop codon at 5′-end, 5′-ATATCCCGGGTCATCATTCATGACCCACGCGGGTCACTACC-3′. ThePCR product was subcloned into a PGM-T easy vector andthe insert was sequenced to ensure that there would be noerrors introduced by Tag polymerase. The plasmid wasdigested with SacI and XmaI enzyme, and the insert wascloned into the corresponding sites of a pIRES2-EGFP(enhanced green fluorescent protein) vector (Clontech).The resultant construct containing a cytomegalovirus(CMV) promoter was a bicistronic message as shown inFig. 1 and named ZCG. pIRES2-EGFP was used as acontrol EGFP-expressing plasmid.

Transfection of plasmids into intestinal epitheliumby electroporation and organ culture

Tissue fragments were isolated from the anterior part of thesmall intestine of X. laevis tadpoles at stages 57–58 (beforemetamorphic climax) and were cut open lengthwise. Theapparatus for electroporation was set up as shown in Fig. 2,modified after the gene transfer method as described inavian embryonic guts (Sakamoto et al. 2000). The isolatedfragments were put into a small chamber (1 mm in height,5 mm in width, and 2 mm in length) fixed in a glass dishwith their epithelial side toward the cathode of electrodes.The chamber was filled with tissue-equivalent (TE) con-taining plasmids ZCG or pIRES2-EGFP at the concentra-tion of 0.1–0.5 μg/μl. For the maximum efficiency oftransfection, six 50-ms pulses at 12 V were generated by anelectroporator (CUY21E; Nepa gene). The transfectedintestines were cultured as described (Ishizuya-Oka andShimozawa 1991). Briefly, they were placed on membranefilters (type HAWP; Millipore) laid on steel grids andincubated at 26°C in 60% Leibovitz-15 medium (Gibco-BRL) supplemented with 100 U/ml of penicillin, 100 μg/ml of streptomycin, and 10% charcoal-treated fetal bovineserum (CTS medium) (Gibco-BRL). To induce metamor-phosis, T3, insulin, and hydrocortisone (Sigma) were addedto the CTS medium at 10 nM, 5 μg/ml, and 0.5 μg/ml,respectively. With or without T3, the expression of EGFPbecame detectable on the first day of cultivation and at-tained its maximal level on the third day, and then graduallydecreased. Thus, we cultured the explants for 3 days.

Immunohistochemistry

Some of the explants were fixed with 4% paraformalde-hyde and frozen at −80°C. The sections cut at 7 μm wereobserved directly by fluorescence microscopy to detectEGFP expression. To identify differentiated intestinal ab-sorptive cells, some of the sections were incubated with therabbit anti-Xenopus intestinal fatty acid binding protein(IFABP) antiserum (diluted 1:500; Ishizuya-Oka et al.1997a) for 1 h and with Alexa Fluor 488-conjugated anti-rabbit IgG (1:500; Molecular Probes) for 1 h at roomtemperature. The sections were observed by fluorescencemicroscopy.

Other explants were fixed with 95% ethanol at 4°C for4 h, embedded in paraffin, and then cut at 5 μm. To identifyproliferating cells, some sections were incubated with the

Fig. 1 Schematic diagram of ZCG construct used for transfection.This construct contains the coding region of TH/bZip under thecontrol of CMV promoter and also that of EGFP for identification oftransfected cell. The resultant TH/bZip protein is translated sepa-rately from EGFP protein and not tagged

111

mouse anti-PCNA monoclonal antibody (diluted 1:100;Novacastra) for 1 h at room temperature and visualizedwith streptavidin–biotin–peroxidase complex (Nichirei)and DAB/H2O2 as described previously (Ishizuya-Okaet al. 2001). The labeling indices were calculated as theratio of labeled nuclei to the total nuclei in >5 sectionsrandomly selected from each explant. More than threeexplants were examined for each experimental point. Theresults were statistically analyzed by the analysis of vari-ance (ANOVA). If the difference was significant based onthe ANOVA analysis, the difference between the controland each group was further analyzed by Student’s t test.

To detect transfected cells, sections of ethanol-fixed ex-plants were incubated with the rabbit anti-GFP antibody(diluted 1:500;MBL) and visualized as described above. Toexamine the relationship between EGFP expression afterTH/bZip expression and cell proliferation, some of thesections were incubated with the anti-PCNA antibody,visualized with streptavidin–biotin–alkaline phosphatasecomplex (Nichirei) and a new fuchsin kit (Nichirei), andthen used for GFP immunostaining. Other sections weredouble-immunostained with a mixture containing the rabbitanti-GFP antibody (diluted 1:500; MBL) and the mouseanti-PCNA antibody at 4°C overnight, and then with amixture containing Alexa Fluor 568-conjugated anti-rabbitIgG (1:500; Molecular Probes) and Alexa Fluor 488-conjugated anti-mouse IgG (1:500; Molecular Probes). Thesections were then observed by fluorescence microscopy.

To distinguish adult epithelial primordia from larvalepithelial cells during metamorphosis, some of the sectionsprepared for immunohistochemistry or in situ hybridizationwere stained with methyl green-pyronin Y (Muto) for 5 minas described (Ishizuya-Oka and Ueda 1996). The adultepithelial primordia are intensely stained with red becauseof their RNA-rich cytoplasm, whereas the staining intensityof the remaining larval epithelial cells becomes weaker asthey undergo apoptosis both in vivo (Ishizuya-Oka andUeda 1996) and in vitro (Ishizuya-Oka et al. 1997a).

Results

TH/bZip expression is transiently up-regulatedby TH throughout the digestive tract

As intestinal remodeling proceeds in an anterior-to-poste-rior direction (Ishizuya-Oka et al. 1997a), we examinedwhether there were any regional differences in TH/bZipexpression along the anteroposterior axis of the intestine.The levels of TH/bZip mRNA were analyzed by Northernblot hybridization in several parts of the intestine and thestomach during spontaneous metamorphosis. There was nosignificant expression until stage 57, it became detectabletoward the beginning of the metamorphic climax, andreached the highest level around stage 61 (Fig. 3a). At thisstage the level of TH/bZip mRNA in the anterior part of the

Fig. 2 Electroporation to trans-fect DNA into the epithelium ofthe X. laevis small intestine andcultivation of the transfectedintestine. A tubular fragmentwas isolated from the anteriorpart of the tadpole small intes-tine, cut lengthwise, and put in achamber filled with DNA. Aftertransfection, the fragment wascultured in the medium

Fig. 3 Northern blot analyses of TH/bZip mRNA levels. a Devel-opmental expression of TH/bZip mRNA in several parts of the X.laevis digestive tract during metamorphosis. F froglet. b Up-regulation of TH/bZip mRNA expression in the small intestine of

stage 57-tadpoles treated with 5 nM T3 for 0 to 4 days. TH/bZipmRNA becomes detectable on day 2. The probe recognizes thetranscripts with apparent length of 4.5 kb (arrowheads). Hybridiza-tion with EF1α serves as a control

112

small intestine was higher than in its posterior part,coinciding with the higher proliferating activity of epithe-lial cells in the anterior part at stage 61 (Ishizuya-Oka et al.1997a). Thereafter, the TH/bZip mRNA level decreasedmore rapidly in the stomach and the anterior part of thesmall intestine than in its posterior part. The TH/bZipmRNA level at stage 65 was lower in the anterior part ofthe small intestine than in its posterior part, coinciding withthe lower proliferating activity of epithelial cells in theanterior part at this stage (Ishizuya-Oka et al. 1997a). Afterthe completion of metamorphosis, few transcripts wereobserved throughout the digestive tract.

To confirm that the expression of TH/bZip mRNA is up-regulated by TH in vivo, stage 57 tadpoles were treatedwith TH for 0–4 days. TH/bZip mRNA level was up-regulated in the whole small intestine around day 2 after thestart of TH treatment (Fig. 3b), similar to earlier studies ofTH/bZip A gene, which showed low levels of up-regu-

lation after 1 day but high levels after 2 days (Brown et al.1996; Ishizuya-Oka et al. 1997b).

Transient expression of TH/bZip mRNAspatiotemporally correlates with growthof adult epithelial primordia

To determine the spatiotemporal relationships between TH/bZip expression and adult epithelium development moreprecisely, in situ hybridization and immunohistochemicalanalyses were performed in both the intestine and thestomach during spontaneous metamorphosis.

No specific signals were detected in the entire intestinethroughout the larval period, which is in agreement with theNorthern blot data (Fig. 3). By stage 60, when primordia ofthe adult epithelium appear throughout the intestine, cellsexpressing TH/bZip mRNA became detectable and in-creased in number toward stage 61. The adult epithelial

Fig. 4 Correlation between ex-pression of TH/bZip mRNA andadult epithelial development inthe Xenopus intestine at stage61. Cross sections were hybrid-ized with antisense (c, f, h) orsense (d) probes, stained withmethyl green-pyronin Y (a, e, g),and immunostained with theanti-PCNA antibody (b). a–dAnterior small intestine. Pri-mordia of the adult epithelium(arrowheads) are intenselystained with methyl green-pyro-nin Y while the staining intensityof larval cells (LE) becomesweaker. The adult primordiaprotrude into the connective tis-sue (CT) by active proliferation(b) and show strong signals forTH/bZip (c). Positive cells arealso detected in the connectivetissue and muscles (M). Controlsection hybridized with the senseprobe shows no specific signal(d). e, f Posterior small intestine.g, h Large intestine. The adultprimordia are detectable as isletsstained intensely with methylgreen-pyronin Y in the basalregion of the larval epithelium(arrowheads) as in the anteriorintestine. Note that the moreposterior the intestine is, thesmaller the adult epithelialprimordia are, and the fewerpositive cells are there. Scalebars, 20 μm

113

primordia in the anterior part of the small intestine aredetectable as islets consisting of undifferentiated cellsstained strongly with methyl green-pyronin Y between thelarval epithelium and the connective tissue (Fig. 4a). Here,the adult epithelial primordia rapidly protruded into theconnective tissue by active cell proliferation (Fig. 4b).These primordia were strongly positive for TH/bZip(Fig. 4c). Specific signals were also detected in the sur-rounding connective tissue and muscles, which began toproliferate as the adult epithelium (Fig. 4b; Ishizuya-Okaand Ueda 1996), but not in the larval epithelium (Fig. 4c).The latter became weakly stained with methyl green-py-ronin Y (Fig. 4a) and often possessed yellowish lysosomesas it degenerated. At this stage in the posterior part of thesmall intestine (Fig. 4e,f) and the large intestine (Fig. 4g,h),the adult epithelial primordia were smaller in size. Theexpression of TH/bZip mRNA was localized in the regioncorresponding to the primordia that were detectable assmall islets between the lysosome-rich larval epitheliumand the connective tissue. Overall, there were fewer TH/bZip-expressing cells in the posterior intestine than in theanterior part at stage 61. Thereafter, as the adult epitheliumcompletely replaced the larval epithelium in an anterior-to-posterior direction, the expression of TH/bZip mRNAwas

rapidly down-regulated and became undetectable at the endof metamorphosis throughout the intestine.

Similarly, in the stomach, no positive cells were detectedthroughout the larval period (Fig. 5a). The larval epithe-lium in the stomach consists of the surface epitheliumfacing the lumen and glands surrounded by a thin layer ofconnective tissue. Around stage 60, when primordia of theadult epithelium became detectable as islets stained strong-ly with methyl green-pyronin Y in the basal region of thelarval glands (Fig. 5b Ishizuya-Oka et al. 1998), they beganto proliferate actively (Fig. 5c). Strong signals for TH/bZipmRNA was localized in the region corresponding to theprimordia (Fig. 5d). Signals were also weakly detectedin the surrounding tissues, which began to proliferate(Fig. 5c). The degenerating larval epithelium showed verylittle signal. When the adult epithelium was replacing thelarval epithelium from the bottom of the glands toward thelumen (Fig. 5e), signals for TH/bZip remained detectable inthe region corresponding to the adult epithelium (Fig. 5f).After stage 62, TH/bZip expression was rapidly down-regulated throughout the stomach.

All control hybridizations with the sense probe gaveonly background levels of signals (Figs. 4d and 5g). Thus,the expression profile of TH/bZip mRNA correlates well

Fig. 5 Expression of TH/bZipmRNA in the Xenopus stomachduring metamorphosis. Crosssections were hybridized withantisense (a, d, f) or sense (g)probes, stained with methylgreen-pyronin Y (b, e), andimmunostained with anti-PCNAantibody (c). a Stage 57. Thelarval epithelium consists of thesurface epithelium (LEs) andglands (LEg). b–d Stage 60.Primordia of the adult epitheli-um (arrowheads) are detectableas islets stained strongly withmethyl green-pyronin Y in thebasal region of the larval glands(b). They actively proliferate (c)and are positive for TH/bZip(d). Signals are also detectedweakly in the connective tissue(CT) and muscles (M). e–gStage 61. The adult epithelium(AE), which is replacing thelarval epithelium (LE) from thebottom toward the lumen (e),remains positive (f). Controlsection hybridized with thesense probe shows no specificsignals (g). Scale bars, 20 μm

114

with the development of the adult epithelium, especiallyits active cell proliferation in both the intestine and thestomach.

Expression of TH/bZip activates cell proliferationin the epithelium

We previously established the intestinal organ culture sys-tem. Differentiated larval epithelium of the small intestineisolated from X. laevis tadpoles at stage 57 can be main-tained as larval-type for at least 10 days in the absence ofTH (Ishizuya-Oka and Shimozawa 1991). At this stage, thedifferentiated larval epithelium continues to express IFABP,a marker for absorptive epithelial cells (Ishizuya-Oka et al.1997a). Both the connective tissue and muscles remainundeveloped as a larval-type in the absence of TH. By usingthis culture system, we examined whether the ectopic ex-pression of TH/bZip could affect the larval epithelium. Theanterior part of the small intestine at stage 57 was trans-fected with plasmids expressing both TH/bZip and EGFPproteins (ZCG) or EGFP protein alone (pIRES2-EGFP) byelectroporation and cultured in the absence of TH. In theintestines transfected with any plasmid, EGFP expressionbecame detectable after day 1 and attained its maximal levelaround day 3 when cells positive for EGFP were detecteduniformly in a wide area of the epithelium. EGFP was notdetected in other tissues except for a few connective tissuecells close to the epithelium (Fig. 6a). Almost all of theepithelial cells in the transfected intestine were positive forEGFP (Fig. 7a), whereas no positive cells were detected inthe untransfected intestine (Fig. 7c). It should be noted inexplants transfected with ZCG that the number of pro-liferating epithelial cells positive for PCNA was larger(Fig. 7e) than those in control explants transfected with

pIRES2-EGFP (Figs. 7d and 8). In the ZCG-transfectedexplants, the double-immunofluorescence labeling withanti-GFP and anti-PCNA antibodies directly indicated thatepithelial cells expressing the transgene have a high pro-liferation activity (Figs. 6c and 7b). The layer of theepithelium became thicker (Fig. 7b,e) than that in controlexplants (Fig. 7d), possibly due to the increase of thenumber of epithelial cells by more active proliferation.Nevertheless, the epithelium remained as a single layer andpositive for IFABP (Fig. 6b). No undifferentiated cells suchas adult epithelial primordia were detected in any explantthroughout the cultivation.

To investigate whether TH/bZip also affects the prolif-eration of the adult epithelial primordia, the anterior part ofthe small intestine at stage 58 was transfected with ZCG,cultured in the presence of TH, and examined by PCNAimmunohistochemistry. In the presence of TH, the larval-to-adult epithelial replacement was induced by TH in vitroas in vivo, although it was not done so in the absence of TH(Ishizuya-Oka and Shimozawa 1991). In the controlexplants transfected with pIRES2-EGFP as in the un-transfected explants, primordia of the adult epitheliumappeared around day 3 as islets consisting of undiffer-entiated cells strongly stained with methyl green-pyronin Y(Fig. 7f), whereas the degenerating larval epithelium wasmore weakly stained. Both the connective tissue and mus-cles increased in cell number in the presence of TH. Inexplants transfected with ZCG where TH/bZip was over-expressed in the epithelium, adult epithelial cells stronglystained with methyl green-pyronin Y became more nu-merous (Fig. 7h) by increased active proliferation (Fig. 7i)than those in the control explants (Figs. 7f and 8). Theexpression of EGFP was somewhat uneven in the presenceof TH (Fig. 6d) compared with that in the absence ofTH (Fig. 6c), probably due to reduced expression in the

Fig. 6 Intestines transfected with ZCG and cultured for 3 days. a, bFrozen sections of the intestines cultured in the absence of TH. Thelarval epithelium (LE) expresses EGFP (green) and remains positivefor IFABP (red). The connective tissue (CT) is negative except for afew cells close to the epithelium (arrows). c, d Paraffin sections ofthe cultured intestines, double-immunostained with anti-EGFP

(green) and anti-PCNA antibodies (red). Regardless of the absence(c) or presence of TH (d), epithelial nuclei positive for PCNA(arrows) are localized in cells positive for EGFP. In the presence ofTH (d), nuclei positive for PCNA are also detected in the developingconnective tissue. Scale bars, 20 μm

115

degenerating larval epithelium. Nevertheless, cells express-ing the transgene showed a high activity of proliferationsimilar to that in the absence of TH (Fig. 6d).

Discussion

TH-up-regulated TH/bZip expression correlateswith growth of adult epithelial primordia throughoutthe stomach and intestine

TH/bZip gene was originally isolated as a TH responsegene in the X. laevis tail (Wang and Brown 1993) and theintestine (Shi and Brown 1993). Because of the tetraploidyin the X. laevis genome, there are duplicated copies of thisgene, TH/bZip A and B, which contain TH response ele-ments (TREs) in their promoters (Furlow and Brown 1999).In this paper, we identified TH/bZip B as a candidate for agene involved in adult epithelial development during in-testinal remodeling (Shimizu et al. 2002). The expressionprofile of TH/bZip B in the present study agrees with that of

Fig. 7 Effects of TH/bZip ex-pression on cell proliferation inintestines cultured for 3 days.Cross sections were stained withmethyl green-pyronin Y (f, h),immunostained with anti-GFP(a, c) and anti-PCNA antibody(d, e, g, i), and double immu-nostained with the both anti-PCNA (red) and anti-GFP(brown) antibodies (b). a–e Inthe absence of TH. Almost allof larval epithelial cells (LE)transfected with ZCG expressEGFP (a) and actively prolifer-ate (b), whereas there are nocells positive for EGFP in theuntransfected intestine (c). Theepithelium in explants transfec-ted with ZCG becomes a highercolumnar (b, e) than that incontrol explants transfected withpIRES2-EGFP (d). CT, connec-tive tissue; M, muscles. f–i Inthe presence of TH. Adult epi-thelial cells stained intenselywith methyl green-pyronin Y(AE, arrowheads) in explantstransfected with ZCG are morenumerous by more active pro-liferation (h, i) than those incontrol explants (f, g). Arrowsin b and e indicate labelednuclei. Scale bars, 20 μm

Fig. 8 TH/bZip expression promotes cell proliferation in theepithelium of X. laevis intestines cultured for 3 days regardless ofthe presence of TH. Each value represents the mean+SD of morethan three explants. **P<0.001, *P<0.05 compared with controlexplants expressing EGFP alone

116

TH/bZip A, which we previously reported in the anteriorpart of the small intestine (Ishizuya-Oka et al. 1997b),suggesting their common roles in the intestinal remodeling.The present in situ hybridization analysis shows that re-gional differences in the amount of TH/bZip B mRNAalong the anteroposterior axis of the small intestine coincidewith those in the growth of adult epithelial primordia,which progresses in an anterior-to-posterior direction(Ishizuya-Oka et al. 1997a). Furthermore, in the stomachwhere primordia of the adult epithelium appear only at thebottom of larval glands (Ishizuya-Oka et al. 1998), TH/bZipB mRNA became detectable as soon as the primordiaappeared. These results suggest the possibility that TH/bZipis involved in appearance and/or growth of the adult epi-thelial primordia throughout the stomach and intestine ofthe amphibian.

During metamorphosis of the amphibian, TH/bZip isexpressed in developing organs such as the head and thehindlimb (Denver et al. 1997; Berry et al. 1998a), and inproliferating areas such as the spinal cord of the resorbingtail (Berry et al. 1998b). These profiles suggest that TH/bZip expression may ubiquitously activate cell prolifera-tion during amphibian organ remodeling. However, itsfunction has not yet been experimentally demonstrated inany organ.

Expression of TH/bZip activates cell proliferationin both the differentiated larval epitheliumand undifferentiated adult epithelial primordia

For functional analysis of TH/bZip, we developed an invitro system by using electroporation and organ culturetechniques. Because tubular fragments of the tadpole in-testines were slit open lengthwise with the epitheliumoutside in this system, exogenous plasmid DNA couldeasily access the epithelium of the intestines just like that inthe avian embryonic gut (Yasugi and Nakamura 2000). Inaddition, because the tadpole epithelium is a single layer,the whole epithelial layer toward the cathode of electrodescan be uniformly transfected. Thus, electroporation provedto be an easy and efficient way to introduce DNA into theepithelium of the amphibian postembryonic intestine. Al-though the transcription of foreign DNA gradually ceasesafter 3 days of transfection in vitro, this period is sufficientfor studying functions of numerous early TH responsegenes which were previously isolated from the X. laevisintestine (Shi and Brown 1993).

We previously reported the accuracy of PCNA immu-nohistochemistry for detecting proliferating cells (Ishizuya-Oka et al. 2001, 2003), since this method gives the samelabeling pattern and almost the same numbers of labelednuclei as BrdU immunohistochemistry during Xenopusintestinal remodeling (Ishizuya-Oka and Ueda 1996;Ishizuya-Oka et al. 1997a). Using PCNA immunohisto-chemistry and the in vitro system, we have shown that TH/bZip functions as a growth activator, regardless of thepresence of TH. In the presence of TH, where numerous TH

response genes other than TH/bZip are expressed (Shi andIshizuya-Oka 1996), primordia of the adult epitheliumincluding multipotential stem cells analogous to those inthe mammalian intestine become detectable (Ishizuya-Okaet al. 2003). The remaining epithelial cells undergo ap-optosis. Overexpression of TH/bZip increased the prolif-erating activity of the adult epithelial primordia. In theabsence of the TH where the epithelium remains as alarval-type (Ishizuya-Oka and Shimozawa 1991), the ec-topic expression of TH/bZip also increased the number ofproliferating cells in the larval epithelium. In this case, theproliferating activity of the epithelium in the transfectedintestine was much higher than that of the epithelium in theuntransfected intestine during TH-induced metamorphosisin vitro (Fig. 8). This may be due to a greater number ofcells expressing the transgene in the transfected intestine(Fig. 7a) than during TH-induced metamorphosis. In thelatter case, TH/bZip is endogenously expressed in the adultepithelial primordia but not in numerous larval epithelialcells. It is noteworthy that even though the ectopic ex-pression of TH/bZip up-regulates cell proliferation of thelarval epithelium extensively, adult epithelial primordia didnot appear. This suggests that TH/bZip expression in thelarval epithelium alone is not sufficient for inducing un-differentiated stem cells.

During metamorphosis of the amphibian, the expressionof TH/bZip is transiently up-regulated in the connectivetissue and muscles when these tissues begin to develop byactive cell proliferation in both the intestine (Shi andIshizuya-Oka 1996) and the stomach (Ishizuya-Oka et al.1998). It seems likely that TH/bZip promotes cell prolif-eration in developing tissues, independent of cell-types. Itis worth studying the potential roles of TH/bZip in cellproliferation in other tissues and/or organs. Alternatively,there remains a possibility that the expression of TH/bZipin the connective tissue plays some role in adult epithelialdevelopment. Our previous culture study indicated thatadult epithelial development requires the presence of con-nective tissue (Ishizuya-Oka and Shimozawa 1992). In thisstudy, it was discovered that TH/bZip expression in theepithelium alone could not induce adult epithelial primor-dia. Taken together, it is possible that some target genes ofTH/bZip in the connective tissue are involved in theappearance of adult stem cells. It would be interesting inthe future to investigate the effects of connective tissue-specific expression of TH/bZip on the adult epithelialdevelopment.

In conclusion, we have established an in vitro genetransfer system for functional analysis of TH responsegenes. Using this system, we have shown that at least oneof the functions of TH-induced transcription factor TH/bZip is to stimulate cell proliferation of the adult epitheliumduring the larval-to-adult intestinal remodeling. This invitro system combined with the use of transgenic frogs formetamorphic studies (Damjanovski et al. 2002; Fu et al.2002; Buchholz et al. 2003) should provide a powerful toolfor dissecting the molecular mechanisms underlying am-phibian intestinal remodeling.

117

Acknowledgements We would like to thank Prof. K. Yoshizato forgenerously supplying the IA009 cDNA fragment and for hisencouragement during the course of this work. This work wassupported by JSPS Grants-in-Aid for Scientific Research (C) (to A.I.-O.) and the Intramural Research Program of the National Instituteof Child Health and Human Development, NIH, (to Y.-B. S.).

References

Amano T, Yoshizato K (1998) Isolation of genes involved inintestinal remodeling during anuran metamorphosis. WoundRepair Regen 6:302–313

Berry DL, Rose CS, Remo BF, Brown DD (1998a) The expressionpattern of thyroid hormone response genes in remodelingtadpole tissue defines distinct growth and resorption geneexpression programs. Dev Biol 203:24–35

Berry DL, Schwartzman RA, Brown DD (1998b) The expressionpattern of thyroid hormone response genes in the tadpole tailidentifies multiple resorption programs. Dev Biol 203:12–29

Booth C, Potten CS (2000) Gut instincts: thoughts on intestinalepithelial stem cells. J Clin Invest 105:1493–1499

Brown DD, Wang Z, Furlow JD, Kanamori A, Schwartzman RA,Remo BF, Pinder A (1996) The thyroid hormone-induced tailresorption program during Xenopus laevis metamorphosis. ProcNatl Acad Sci USA 93:1924–1929

Buchholz DR, Hsia S-CV, Fu L, Shi Y-B (2003) A dominant-negative thyroid hormone receptor blocks amphibian metamor-phosis by retaining corepressors at target genes. Mol Cell Biol23:6750–6758

Buckbinder L, Brown DD (1992) Thyroid hormone-induced geneexpression changes in the developing frog limb. J Biol Chem267:25786–25791

Cowell IG, Hurst HC (1994) Transcriptional repression by thehuman TH/bZip factor E4BP4: definition of a minimalrepression domain. Nucleic Acids Res 22:59–65

Cowell IG, Skinner A, Hurst HC (1992) Transcriptional repressionby a novel member of the TH/bZip family of transcriptionfactors. Mol Cell Biol 12:3070–3077

Damjanovski S, Sachs LM, Shi Y-B (2002) Function of thyroidhormone receptors during amphibian development. MethodsMol Biol 202:153–176

Denver RJ, Pavgi S, Shi Y-B (1997) Thyroid hormone-dependentgene expression program for Xenopus neural development.J Biol Chem 272:8179–8188

Dodd MHI, Dodd JM (1976) The biology of metamorphosis. In:Lofts B (ed) Physiology of the Amphibia. Academic, NewYork, pp 467–599

Fu L, Buchholtz D, Shi Y-B (2002) Novel double promoterapproach for identification of transgenic animals: a tool for invivo analysis of gene function and development of gene-basedtherapies. Mol Reprod Dev 62:470–476

Furlow JD, Brown DD (1999) In vitro and in vivo analysis of theregulation of a transcription factor gene by thyroid hormoneduring Xenopus laevis metamorphosis. Mol Endocrinol13:2076–2089

Hourdry J, Dauca M (1977) Cytological and cytochemical changesin the intestinal epithelium during anuran metamorphosis. IntRev Cytol Suppl 5:337–385

Ishizuya-Oka A, Shimozawa A (1991) Induction of metamorphosisby thyroid hormone in anuran small intestine culturedorganotypically in vitro. In Vitro Cell Dev Biol 27A:853–857

Ishizuya-Oka A, Shimozawa A (1992) Connective tissue is involvedin adult epithelial development of the small intestine duringanuran metamorphosis in vitro. Roux’s Arch Dev Biol 201:322–329

Ishizuya-Oka A, Shimozawa A, Takeda H, Shi Y-B (1994) Cell-specific and spatio-temporal expression of intestinal fatty acid-binding protein gene during metamorphosis. Roux’s Arch DevBiol 204:150–155

Ishizuya-Oka A, Ueda S (1996) Apoptosis and cell proliferation inthe Xenopus small intestine during metamorphosis. Cell TissueRes 286:467–476

Ishizuya-Oka A, Ueda S, Damjanovski S, Li Q, Liang VC-T, ShiY-B (1997a) Anteroposterior gradient of epithelial transforma-tion during amphibian intestinal remodeling: immunohisto-chemical detection of intestinal fatty acid-binding protein. DevBiol 192:149–161

Ishizuya-Oka A, Ueda S, Shi Y-B (1997b) Temporal and spatialregulation of a putative transcriptional repressor implicates it asplaying a role in thyroid hormone-dependent organ transfor-mation. Dev Genet 20:329–337

Ishizuya-Oka A, Inokuchi T, Ueda S (1998) Thyroid hormone-induced apoptosis of larval cells and differentiation of pepsi-nogen-producing cells in the stomach of Xenopus laevis invitro. Differentiation 63:59–68

Ishizuya-Oka A, Ueda S, Inokuchi T, Amano T, Damjanovski S,Stolow M, Shi Y-B (2001) Thyroid hormone-induced expres-sion of Sonic hedgehog correlates with adult epithelialdevelopment during remodeling of the Xenopus stomach andintestine. Differentiation 69:27–37

Ishizuya-Oka A, Shimizu S, Sakakibara S, Okano H, Ueda S (2003)Thyroid hormone-up-regulated expression of Musashi-1 isspecific for progenitor cells of the adult epithelium duringamphibian gastrointestinal remodeling. J Cell Sci 116: 3157–3164

Kikuyama S, Kawamura K, Tanaka S, Yamamoto K (1993) Aspectsof amphibian metamorphosis: hormonal control. Int Rev Cytol145: 105–148

McAvoy JW, Dixon KE (1977) Cell proliferation and renewal in thesmall intestinal epithelium of metamorphosing adult Xenopuslaevis. J Exp Zool 202:129–138

McAvoy JW, Dixon KE (1978) Cell specialization in the smallintestinal epithelium of adult Xenopus laevis. J Anat 125:155–169

Nieuwkoop PD, Faber J (1967) Normal table of Xenopus laevis(Darwin). North-Holland, Amsterdam

Potten CS, Booth C, Tudor GL, Booth D, Brady G, Hurley P,Ashton G, Clarke R, Sakakibara S, Okano H (2003) Identifi-cation of a putative intestinal stem cell and early lineagemarker; Musashi-1. Differentiation 71:28–41

Sakamoto N, Fukuda K, Watanuki K, Sakai D, Komano T, ScottingPJ, Yasugi S (2000) Role for cGATA-5 in transcriptionalregulation of the embryonic chicken pepsinogen gene byepithelial–mesenchymal interactions in the developing chickenstomach. Dev Biol 223:103–113

Shi Y-B (1999) Amphibian metamorphosis: from morphology tomolecular biology. Wiley-Liss, New York

Shi Y-B, Brown DD (1993) The earliest changes in gene expressionin tadpole intestine induced by thyroid hormone. J Biol Chem268:20312–20317

Shi Y-B, Ishizuya-Oka (1996) Biphasic intestinal development inamphibians: embryogenesis and remodeling during metamor-phosis. Curr Top Dev Biol 32:205–235

Shi Y-B, Ishizuya-Oka (2001) Thyroid hormone regulation ofapoptotic tissue remodeling: implications from molecularanalysis of amphibian metamorphosis. Prog Nucleic Acid ResMol Biol 65:53–100

Shimizu K, Ishizuya-Oka A, Amano T, Yoshizato K, Ueda S (2002)Isolation of connective-tissue-specific genes involved inXenopus intestinal remodeling: thyroid hormone up-regulatesTolloid/BMP-1 expression. Dev Genes Evol 212:357–364

Wang Z, Brown DD (1993) Thyroid-hormone-induced gene ex-pression program for amphibian tail resorption. J Biol Chem268:16270–16278

Yasugi S, Nakamura H (2000) Gene transfer into chicken embryosas an effective system of analysis in developmental biology.Develop Growth Differ 42:195–197

118