Remodeling the exocrine pancreas at metamorphosisin Xenopus laevisSandeep Mukhi*, Jinzhe Mao†, and Donald D. Brown*‡

*Department of Embryology, Carnegie Institution, 3520 San Martin Drive, Baltimore, MD 21218; and †Center for Neuroscience Research,Children’s National Medical Center, Washington, DC 20010

Contributed by Donald D. Brown, April 11, 2008 (sent for review January 8, 2008)

At metamorphosis the Xenopus laevis tadpole exocrine pancreasremodels in two stages. At the climax of metamorphosis thyroidhormone (TH) induces dedifferentiation of the entire exocrinepancreas to a progenitor state. The organ shrinks to 20% of its size,and �40% of its cells die. The acinar cells lose their zymogengranules and �75% of their RNA. The mRNAs that encode exocrine-specific proteins (including the transcription factor Ptf1a) undergoalmost complete extinction at climax, whereas PDX-1, Notch-1, andHes-1, genes implicated in differentiation of the progenitor cells,are activated. At the end of spontaneous metamorphosis when theendogenous TH has reached a low level, the pancreas begins toredifferentiate. Exogenous TH induces the dedifferentiation phasebut not the redifferentation phase. The tadpole pancreas lacks themature ductal system that is found in adult vertebrate pancreases,including the frog. Exocrine pancreases of transgenic tadpolesexpressing a dominant negative form of the TH receptor controlledby the elastase promoter are resistant to TH. They do not shrinkwhen subjected to TH. Their acinar cells do not dedifferentiate atclimax, nor do they down-regulate exocrine-specific genes oractivate Notch-1 and Hes-1. Even 2 months after metamorphosisthese frogs have not developed a mature ductal system and theacinar cells are abnormally arranged. The TH-dependent dediffer-entiation of the tadpole acinar cells at climax is a necessary step inthe formation of a mature frog pancreas.

dedifferentiation � thyroid hormone � Thyroid hormone receptor dominantnegative � gene expression

The concentration of thyroid hormone (TH) controls amphib-ian metamorphosis. The TH level increases steadily in

tadpoles as they proceed to metamorphosis and then decreasesto a base line level after metamorphic climax (1). Metamorphosisin Xenopus laevis involves the remodeling of various organsincluding skin, brain, intestine, liver, and pancreas (2). AddingTH to the rearing water induces organ remodeling, and anti-thyroid chemicals like methimazole or perchlorate prevent re-modeling. The essential role of thyroid receptors (TRs) has beendemonstrated for diverse metamorphic programs by the inhib-itory properties of a dominant negative form of the TR (TRDN)on the metamorphosis of various cell types and organs (3–7).

The tadpole exocrine pancreas consists of acinar cells withwell developed zymogen granules. Exocrine-specific proteinshave been detected as early as stage NF41 (8). The size of thepancreas steadily increases as the tadpole grows up to the climaxof metamorphosis, and then at this stage the pancreas loses�80% of its volume (9). This pancreatic ‘‘regression’’ at climaxis marked by histolysis and cell death (2, 10, 11) and alsodown-regulation (‘‘extinction’’) of exocrine-specific mRNAs andtheir proteins (11, 12).

We report here that the regression of the tadpole exocrinepancreas involves a TH-controlled dedifferentiation of the aci-nar cells to a progenitor state. These cells then redifferentiate inthe frog to form a typical vertebrate pancreas. A major differ-ence between the tadpole and frog pancreas is the almostcomplete absence of a recognizable ductal system in the tadpolepancreas. We present evidence that the dedifferentiation of the

acinar cells during metamorphic climax is essential for theformation subsequently of a typical adult exocrine pancreas inthe frog with a complete set of ducts.

ResultsMorphological Changes of the Pancreas During Spontaneous Meta-morphosis. Transgenic tadpoles with the rat elastase promoterdriving GFP (pElastase-GFP) were used to observe the changeof shape and position of the pancreas during metamorphosis(Fig. 1). The stages of metamorphosis are divided into premeta-morphic (up to NF55; Fig. 1 A), prometamorphic (NF56–58; Fig.1B) climax (NF59–65; Fig. 1C) and froglet (NF66; Fig. 1D). Thepancreas grows proportionately with the size of the animal to theend of prometamorphosis (NF58; Fig. 1 E and F). The tadpolepancreas is a large flaccid organ that is curled ventrally aroundthe U-shaped duodenum and touches the posterior end of thestomach. During climax the pancreas shrinks dramatically (Fig.1G), reaching �20% of its original volume by NF63. At the endof metamorphosis the frog pancreas is f lat and firm having lostits f laccid consistency (Fig. 1H). Throughout premetamorphosisthe pancreas is filled with acinar cells that contain characteristiceosinophilic zymogen granules (Fig. 1I). When the tadpolereaches NF58/59, the onset of climax, the acinar cells start to losetheir zymogen granules (Fig. 1J). At climax (NF63) most of theircytoplasm is lost and no acinar features remain. Many cellscontain cytoplasmic vacuoles (Fig. 1K). Cell debris that is visibleby electron microscopy and apoptotic bodies confirmed byTUNEL assay indicate a significant amount of cell death atclimax [supporting information (SI) Fig. S1]. From NF58 toNF62 the total DNA content of the pancreas drops by �40%(Fig. S2), reflecting the significant cell death that has beenreported (10, 11). However, at the same time the total RNA inthe remaining cells drops 75% (Fig. S2) as the acinar cells losetheir ribosome-rich cytoplasm. Throughout climax the dediffer-entiated exocrine cells have prominent nuclei and nucleoli.Later, we will demonstrate that these cells have changed theirgene expression profile. At the end of metamorphosis (NF66)regular acinar structures are still not visible (Fig. 1L). They donot reappear until �3–4 weeks after stage NF66.

Dedifferentiation and Redifferentiation of the Pancreas During Meta-morphosis. We examined the expression pattern of exocrine-specific genes by in situ hybridization, real-time RT-PCR, andimmunohistochemistry at different stages of metamorphosis.Differentiated exocrine pancreatic cells in tadpole and adult arerich in exocrine-specific product such as amylase, mRNA (Fig. 2A and D), and protein (Fig. 2 E and H). Dedifferentiation of theexocrine pancreas is characterized by the loss (extinction) of

Author contributions: S.M., J.M., and D.D.B. designed research; S.M. and J.M. performedresearch; S.M. and D.D.B. analyzed data; and S.M. and D.D.B. wrote the paper.

The authors declare no conflict of interest.

‡To whom correspondence should be addressed. E-mail: brown@ciwemb.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0803569105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

8962–8967 � PNAS � July 1, 2008 � vol. 105 � no. 26 www.pnas.org�cgi�doi�10.1073�pnas.0803569105

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

exocrine-specific mRNAs at the climax of metamorphosis (Fig.2B). Down-regulation of amylase mRNA was also confirmed byreal-time RT-PCR (Fig. S3). Reappearance of exocrine mRNAsbegins in 1-week-old froglets (data not shown) and graduallyincreases. Even though the amylase mRNA is lost at climax theamylase protein (leftover protein) remains throughout climax inall of the exocrine cells, confirming that the origin of theundifferentiated cells at climax is the previously differentiatedtadpole acinar cells (Fig. 2 F and G). Ptf1a, a transcription factor

that is required to express the exocrine-specific genes (13),follows the same down-regulation kinetics as amylase (Fig. 2 Iand J), but its mRNA reappears at the end of metamorphosisbefore the up-regulation of amylase mRNA begins (Fig. 2K).

The transcription factor PDX-1 is expressed only in �-cells inthe adult mammalian pancreas (14). However, both exocrine andendocrine progenitor cells express PDX-1 (15). In the tadpoleand frog pancreas PDX-1 mRNA (Fig. 3 A and C) is expressedexclusively in islet cells. At climax PDX-1 expression is activatedin the dedifferentiated exocrine cells (Fig. 3B), providing evi-dence for the progenitor state of the pancreas exocrine cells atclimax. Up-regulation of PDX-1 mRNA was also confirmed byreal-time RT-PCR (Fig. S3).

Notch signaling in the mammalian pancreas has been impli-cated in exocrine cell fate and differentiation (16–18). ActivatedNotch-1 induces Hes-1 expression, and Hes-1 can bind to andinactivate Ptf1a (19), thereby leading to the inhibition of exo-crine-specific gene expression and conceivably in the case ofamphibian metamorphosis the dedifferentiation of the exocrinepancreas. Notch-1 and Hes-1 mRNAs are up-regulated duringmetamorphic climax (Fig. 3 E and H and Fig. S3) and thendecrease as the pancreas redifferentiates into mature pancreas(Fig. 3 F and I and Fig. S3). TH can dedifferentiate the exocrinepancreas in 48 h of 3,5,3�-L-triiodothyronine (T3) induction bydown-regulation of exocrine mRNAs (described below), how-ever, Notch-1 and Hes-1 mRNA are unchanged (Fig. S3).Therefore the activation of the Notch-1 and Hes-1 genes is toolate a TH response to account for the first steps of TH-induceddedifferentiation.

Role of TH in Pancreatic Remodeling. We assessed the role of THin pancreas remodeling by using two methods. First, we addedexogenous TH to the rearing water of premetamorphic tadpolesto induce metamorphic changes. Second, we prepared transgenicanimals in which the rat elastase promoter controls a dominantnegative form of TR� fused to GFP at its carboxyl terminus(pElastase-TRDN-GFP). Previous experiments demonstratedthat expression of this competitive inhibitor blocks a variety ofTH-induced programs (3). The protein encoded by the TRDN-

Fig. 1. Morphological and histological changes of the X. laevis pancreas atdifferent stages expressing GFP driven by the rat elastase promoter. (A–D)Stages of X. laevis metamorphosis. (E–H) Abdominal view of the pancreasobserved by fluorescent microscope. (I–L) H&E-stained sections of pancreasesat the corresponding stages. Expression of the elastase-GFP transgene had noeffect on normal development and remodeling of pancreas. (Scale bars: A–D,4 mm; E–H, 1 mm; I–L, 40 �m.)

A B C D

E F G H

I J K L

Fig. 2. Change in gene expression in the exocrine pancreas during meta-morphosis. (A–D and I–L) In situ hybridization with amylase (A–D) and Ptf1a(I–L) probes. Every mRNA encoding a ‘‘terminally differentiated’’ exocrineprotein that we have tested is extinguished at metamorphic climax (NF62) tothe same extent as these 2 mRNAs (see discussion). (E–H) Although amylasemRNA is extinguished at climax, amylase protein was still detected by immu-nohistochemistry, confirming the dedifferentiation stages of previously dif-ferentiated exocrine cells. (Scale bar: 40 �m.)

A B C

D E F

G H I

Fig. 3. Change in gene expression in the exocrine pancreas during meta-morphosis. (A–C) In situ hybridization with PDX-1 probes. PDX-1 is expressedexclusively in islet cells in the tadpole (A) and frog pancreases (C), and at climaxexpression is extended to the dedifferentiated exocrine cells (B). (D–I) Notch-1(D–F) and Hes-1 (G–I) genes are activated at climax, although low level ofexpression of Notch-1 and Hes-1 has been observed at premetamorphic andadult frogs. (Scale bar: 40 �m.)

Mukhi et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8963

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

GFP transgene is concentrated in the nucleus because of anuclear localization signal within the TR. The TRDN-GFPtransgene is expressed only in exocrine cells and not in insulin-immunoreactive cells (Fig. S4).

A WT premetamorphic pancreas (Fig. 4A) when induced byT3 resembles a normal pancreas at the climax of spontaneousmetamorphosis. It shrinks dramatically in size (Fig. 4B), and theacinar cells lose their zymogen granules, and much of theircytoplasmic volume (Fig. 4E). Transgenic tadpoles expressingthe TRDN-GFP transgene in their exocrine pancreas are pro-tected from T3-induced alteration (Fig. 4 C and F). In situhybridization of the T3-exposed premetamorphic control ani-mals demonstrated that T3 induces down-regulation of exocrine-specific markers such as amylase after 4 days of exposure (Fig.4H). This induced down-regulation of exocrine mRNAs is com-parable with that of spontaneous metamorphosis at stage NF63(Fig. 2B). However, animals expressing the TRDN-GFP trans-gene failed to down-regulate amylase (Fig. 4I). In some of thetransgenic animals protection was not homogenous throughoutthe pancreas, and patches of cells were not protected against theaction of T3. Loss of acinar integrity, loss of zymogen granules,and disappearance of exocrine mRNAs in T3-induced premeta-morphic tadpoles along with the protective nature of the TRDNprotein against exogenous T3 provides additional proof thatTH controls the process of dedifferentiation of the exocrinepancreas.

The TRDN Transgene Prevents Remodeling of the Exocrine PancreasDuring Spontaneous Metamorphosis. Transgenic animals wereraised through spontaneous metamorphosis. The expression ofthe TRDN transgene did not affect either tadpole growth or theirtime to metamorphosis. All external signs of metamorphosiswere normal in the transgenic animals. Internal organs like theintestine that do not express the transgene underwent normalremodeling at climax. However, the pancreas did not change itsshape during metamorphosis nor did it shrink at climax. UnlikeWT (Fig. 5A), the TRDN-transgenic pancreases did not losetheir zymogen granules at NF66 (Fig. 5E) nor did they undergo

the RNA changes that characterize the normal pancreas atclimax. Amylase mRNA was down-regulated in WT froglets(NF66; Fig. 5B) and not in TRDN-transgenic froglets of thesame stage (Fig. 5F), and Notch-1 and Hes-1 mRNAs were notactivated at climax in these transgenic animals (Fig. 5 G and H).Their protection was not always complete. Some transgenicanimals had small islands of cells that responded to spontaneousmetamorphosis (data not shown). The growth rates of transgenicand nontransgenic sibling frogs as measured by weighing theanimals twice a week for 2 months were the same (data notshown). However, the pancreases in 2-month-old transgenicfrogs were abnormal. The acinar cells were disorganized (Fig.6C), and there were regions of the pancreas that were vacuolatedand others with tightly packed nuclei and without proper acini(Fig. S5) that no longer expressed the transgene.

Pancreatic Duct Maturation During Spontaneous Metamorphosis.H&E-stained sections identified the pancreatic ducts by theirmorphology (20) and were confirmed by antibody stainingagainst carbonic anhydrase II (CAII) (data not shown), whichstains all types of ductal cells (21). Like other vertebrates themature frog pancreas has a complex ductal system consisting ofintercalated, lobular, main ducts (Fig. 6B) and extra-pancreaticducts that connect with the bile duct and attach to the intestine(Fig. S6). The premetamorphic pancreas has simple ducts witha single cuboidal epithelial cell lining (Fig. 6A). These structuresare visible at all stages of metamorphosis sparsely distributed inthe exocrine pancreas. By the end of metamorphosis largerlobular and collecting ducts have begun to form. We havequantified the presence of ductal components at different de-velopmental stages by counting the number of duct cells. Thereis a large increase in the number of duct cells in the postmeta-morphic frog pancreas compared with the premetamorphicpancreas (Fig. 6D). The intercalated ducts that connect the acinito the lobular duct were not detected in the premetamorphicpancreas but appear in postmetamorphosis along with theredifferentiation of the acinar cells (4-week-old froglets; data notshown). The extra-pancreatic ducts that connect the pancreas tothe intestine are present at all stages of development. However,these larger ducts also undergo maturation. They have a singlelayer of epithelial cells in the tadpole with very little collagen orsmooth muscle matrix (Fig. S6). The surrounding matricesappear after metamorphosis. In the frog these ducts are largerin diameter and have a thick collagen matrix surrounding them.

A B C

D E F

G H I

Fig. 4. The TRDN-transgene prevents TH-induced exocrine dedifferentia-tion. (A, D, and G) WT control NF54 tadpole. (B, E, and H) NF54 tadpole after4 days of 10 nM T3 exposure. (C, F, and I), pElastase-TRDN-GFP NF54 transgenictadpole after 4 days of 10 nM T3 exposure. (A–C) Abdominal view of thepancreas. The thin white line outlines each pancreas. (D–F) H&E-stainedsections. (G–I) In situ hybridization for amylase. Exogenous TH induces thededifferentiation of the exocrine pancreas, and the TRDN transgene preventsthe dedifferentiation process. (Scale bars: A–C, 1 mm; D–I, 40 �m.)

A B C D

E F G H

Fig. 5. The TRDN transgene prevents dedifferentiation of acinar cells in spon-taneous metamorphosis. NF66 WT (A–D) and NF66 pElastase-TRDN (E–H) pan-creas sections. (A and E) Stained with H&E. (B–D and F–H) In situ hybridization. (Band F) Amylase. (C and G) Notch-1. (D and H) Hes-1. (Scale bar: 40 �m.)

8964 � www.pnas.org�cgi�doi�10.1073�pnas.0803569105 Mukhi et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

The pancreatic ductal system in a 2-month-old TRDN frog wasunderdeveloped compared with the control sibling (Fig. 6). Nointercalated ducts were observed and the lobular ducts werehighly disorganized and were scarce (Fig. 6C and Fig. S5).However, the maturation of extra-pancreatic ducts was notaffected in the TRDN frogs (Fig. S6), as the transgene was neverexpressed in those duct cells.

DiscussionRemodeling Begins with a TH-Dependent Dedifferentiation of theExocrine Pancreas. The premetamorphic tadpole pancreas is alarge flaccid structure with well developed acini (Fig. 1) that arerich in zymogen granules and contain high levels of the proteinsthat characterize vertebrate exocrine pancreases. The dramaticreduction in pancreas volume that occurs at metamorphic climaxor can be induced by exogenous TH has been described for Hyla(22), Bufo (10), Xenopus (22), and Rana (23). Organ shrinkageat climax has been estimated to be between 70% and 90%. Janes(22) concluded that the pancreas undergoes ‘‘partial degenera-tion’’ during metamorphosis followed by ‘‘regeneration.’’ Cellu-lar ‘‘necrosis’’ and ‘‘dehydration’’ have been held responsible forthe decrease in pancreatic mass (11, 22, 23). Apoptosis ofpancreatic cells is elevated at climax (10). We have also observedincreased apoptosis and the accumulation of cellular debris atthe climax of metamorphosis by electron micrography (Fig. S1).The cell loss in the X. laevis pancreas during metamorphosismeasured by the total DNA content of the pancreas was �40%(Fig. S2). However, this amount of cell death accounts for justpart of the organ shrinkage. After just 4 days of TH induction orat the climax of metamorphosis, all of the exocrine cells lost theirzymogen granules, �70% of their total RNA, and most of theircytoplasmic volume (Fig. 4). Every terminally differentiatedpancreas-specific gene whose mRNA we measured was down-regulated at climax. The list includes carboxypeptidase A and B,

� amylase, lipase, chymotrypsin, trypsin, zymogen granule mem-brane protein, DNase I, elastase I, and protein disulfide isomer-ase (data not shown). Ptf1a is a key transcription factor that isrequired for the expression of exocrine-specific genes in X. laevis(24, 25) as it is in mammals (26). TH-induced down-regulationof Ptf1a mRNA is not rapid enough to cause the drop of theexocrine-specific mRNAs (data not shown). However, this geneis reactivated before the pancreatic enzyme genes at the end ofmetamorphosis (Fig. 2) and might play an important role in theredifferentiation of the exocrine pancreas.

PDX-1, Notch-1, and Hes-1 are expressed in exocrine progen-itor cells (18, 27, 28). These three genes are up-regulated atclimax (Fig. 3). PDX-1 is expressed presumably exclusively in theislet cells in the differentiated tadpole and frog pancreas (Fig. 3)as it is in the mature mammalian pancreas (14). However, PDX-1expression is activated in progenitor exocrine cells (15, 29), andwe show that there is a similar pervasive activation of PDX-1 inthe dedifferentiated exocrine cells at climax (Fig. 3). These samecells up-regulate Notch-1 and Hes-1 (Fig. 3) at climax. Blockingthe TH response by overexpressing a TRDN inhibits thededifferentiation of the pancreas at climax of spontaneousmetamorphosis (Fig. 5) and that induced by exogenous TH (Fig.4). At the end of metamorphosis these transgenic animals stillhave a tadpole-like pancreas with acinar cells containing zymo-gen granules. The exocrine-specific mRNA, amylase, is notdown-regulated at climax nor do these transgenic pancreasesup-regulate Notch-1 or Hes-1 mRNAs at climax (Fig. 5).

The final redifferentiation phase of remodeling occurs whenthe endogenous TH has returned to a baseline level after theclimax of metamorphosis (1), giving us reason to believe that THis not involved in the redifferentiation of the pancreas aftermetamorphosis in the frog.

Plasticity of the Pancreas. The adult mammalian exocrine pancreasexhibits plasticity and has been shown to dedifferentiate underseveral conditions. Duct ligation is an established technique tocause tissue damage, and the acini become metaplastic andtransdifferentiate into tubular duct-like structures (30). Matureacinar cells lose their identity and subsequently dedifferentiateand redifferentiate in chronic pancreatitis (31) upon exposure tothe pharmacological agent, cerulein (32, 33) and when placedinto cell culture (34).

In view of the mammalian experiments we considered whetherinjury to the pancreatic duct at climax in X. laevis might play arole in TH-induced dedifferentiation. The extensive intestinalremodeling that occurs at the same time (35) might secondarilydamage the pancreatic duct. However, the elastase TRDNtransgenic animals undergo normal or exogenous TH-inducedintestinal remodeling while their pancreases are inhibited fromdedifferentiation (Fig. 4). Therefore, in Xenopus, the dediffer-entiation process is an organ autonomous event mediated di-rectly by TH. Differentiation of the exocrine pancreas follows awell studied sequence of gene activation events in progenitorcells, including Notch signaling activation (27, 29). Redifferen-tiation in spontaneous metamorphosis is marked by the activa-tion of the Notch-1 pathway at about the same time as PDX-1.The period of exocrine expression of PDX-1 ends at aroundNF66, whereas the Notch pathway remains activated for at least4 weeks after metamorphosis. Notch pathway activation has alsobeen observed after duct ligation (36) and after caeruleinexposure (33) in mammalian models. Notch-1 activation isknown to induce dedifferentiation of the pancreas presumably byinducing Hes-1 that binds to the critical exocrine transcriptionfactor Ptf1a (19). However, TH induction does not activateNotch-1 or Hes-1 in the first 48 h when down-regulation ofexocrine-specific mRNAs has begun. Possibly the low level ofNotch-1 protein in tadpoles is activated by protease cleavage (37)and that initiates dedifferentiation.

A B

C D

Fig. 6. Differences in ductal systems between tadpole and frog. (A) Premeta-morphic tadpoles have very few ducts compared with (B) adult pancreas. Thecomplex ductal system arises after metamorphosis. (C) Sections of a 2-month-old pElastase-TRDN-GFP frog showing reduced number of ducts. (D) All typesof duct cells in control animals (n � 5 per group) were counted in a unit areafrom H&E-stained sections of NF54 tadpoles (PreMM), 2 month-old frog (2M),and an adult frog. (Scale bar: 40 �m.)

Mukhi et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8965

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

Redifferentation of acinar cells into endocrine and ductal cellshas been reported by several authors (34, 36, 38). In a separatestudy, we have shown that the dedifferentiated exocrine cells donot give rise to insulin-producing cells during the redifferenti-ation phase (S.M. and D.D.B., unpublished data). The TRDNtransgenic frogs have a greatly reduced ductal system with noevidence of intercalated duct as late as 2 months after meta-morphosis, further indicating that the intercalated ducts origi-nate from dedifferentiated acinar cells.

The Difference Between the Tadpole and the Frog Pancreas. The moststriking feature of the tadpole pancreas is its underdevelopedduct system. The only readily identifiable ducts are an occa-sional simple tube that consists of a single layer of smallcuboidal cells without supporting matrix (Fig. 6). These smallducts do not change their size or abundance during metamor-phosis. In the tadpole the major pancreatic duct that joins thebile duct and then attaches to the duodenum has a simpleepithelial wall that is one cell thick with no collagen supportmatrix (Fig. S6). All sizes of ducts are plentiful in the frogpancreas, and the larger ones have a thick collagen matrix (Fig.6). The tadpole acinar cells synthesize a high level of the samedigestive enzymes as the frog acinar cells, but it is not clear howthe tadpole exocrine products are collected for delivery intothe intestine considering the paucity of ducts. One explanationfor TH’s role in maturation of the pancreas at metamorphosisis to induce the exocrine cells, which are almost exclusivelyacinar in the tadpole, to revert to a progenitor state that thenredifferentiates into a more traditional pancreas containingnot only acini but a complete ductal system. The animalsexpressing the TRDN transgene cannot dedifferentiate theirpancreas at climax nor form a normal duct system even 2months after metamorphosis (Fig. 6 and Fig. S5). Theseanimals grow normally to sexual maturity.

These studies with spontaneous metamorphosis and TH in-duction in WT and transgenic animals confirm the essential roleof TH in pancreatic dedifferentiation at metamorphic climaxand the importance of this temporary return to a progenitorstate to form a mature vertebrate pancreas after metamorphosis.The proteins that are expressed in the tadpole acinar cells do notsurvive long enough to be lineage tracers for the new ductalsystem in the frog. Yet all facts indicate that these duct cells arederived from previously differentiated tadpole acinar cells. Aschematic representation of exocrine pancreas architecture andthe gene expression changes with development is illustrated inthe Fig. 7. We report that pancreatic dedifferentiation andredifferentiation as a normal part of an animal’s developmentalprogram and life cycle. The fact that dedifferentiation of thetadpole pancreas is controlled completely by a simple hormoneincreases its value as a model system for studying pancreasdevelopment.

Materials and MethodsPlasmid Constructs and Transgenesis. Transgenic Xenopus were prepared byusing a 205-bp rat elastase promoter (39, 40) driving either GFP (a gift fromM. Horb, Clinical Research Institute of Montreal, Canada) or a dominantnegative (TRDN) form of the TR� fused to GFP (5). TRDN is the codingsequence of X. laevis TR� lacking 36 bp from its C terminus fused by a 15-bplinker to GFP. This TRDN-flex-GFP was prepared from the constructpCS2�[teto]TRDN-GFP3 (5) by PCR using a primer that incorporates anEcoRI site at the 5� end and a SmaI site at the 3� end. The PCR product waspurified, cut with EcoRI and SmaI, and ligated into linearized pElastase-GFPthat had been cleaved with the same restriction enzymes. Both pElastase-GFP and pElastase-TRDN-GFP were linearized with NotI for transgenesis.

Transgenic animals were prepared by restriction enzyme-mediated in-tegration nuclear transplantation (41) with some modifications (42). Thelinearized pElastase-TRDN-GFP was always coinjected with a linearized�-crystaline-GFP construct (43). Successful transgenesis was confirmed fromtail tip DNA by PCR. Only animals with visible GFP expression in their

pancreas confirmed by fluorescent stereomicroscopy and through immu-nohistochemistry (described below) of cryosections with a GFP antibodywere used for experiments. The promoter drives expression only in theacinar cells of NF54 tadpoles (S.M. and D.D.B., unpublished work). This istrue for both the GFP and the TRDN-GFP reporters. The former GFP proteinis localized in the cytoplasm, and the latter fused protein is in the nucleus.In these experiments we used F1 progeny of pElastase-GFP males and F0

pElastase-TRDN-GFP animals.

Raising and Treatment of Premetamorphic Tadpoles. Tadpoles were raised in aflow-through system at 22°C with a 12-h light-dark cycle. X. laevis tadpoleswere staged according to Nieuwkoop and Faber (44). Tadpoles and frogs wereanesthetized with ice-cold MS-222 (3-aminobenzoic acid ethyl ester; Sigma)before their pancreases were isolated for direct visualization with a fluores-cent microscope (Leica) or for tissue fixation. In TH-induced metamorphosisstudy premetamorphic (NF54) tadpoles (at �5 animal per liter; n � 5/6 pergroup) were reared in 4 liters of 0.1� MMR (10 mM NaCl, 0.2 mM KCl, 0.1 mMMgCl2, 0.2 mM CaCl2, and 0.5 mM Hepes, pH 7.5). A final concentration of 10nM T3 was added to the rearing water, and duration of exposure was for 4days. The water with T3 was changed every 2 days.

Histology, Immunohistochemistry, and In Situ Hybridization. Pancreases werefixed in Bouin’s fixative, embedded in paraffin, sectioned, and stained withH&E. For immunohistochemistry samples were fixed in 4% paraformaldehyde,embedded in OCT compound, cryosectioned, and immunostained as de-scribed (6). Primary antibodies used for immunohistochemistry were guineapig anti-insulin (Linco Research) at 1:500 dilution, rabbit anti-GFP (Torrey PinesBiolabs) at 1:500 dilution, CAII (Chemicon International) at 1:200 dilution andmouse anti-amylase (Santa Cruz Biotechnology) at 1:50 dilution. Amylasestaining was done in paraffin-embedded sections, and antigen-unmaskingtechniques was used according to the manufacturer’s protocol (Vector Labo-ratories). Fluorescence-conjugated secondary antibodies (Molecular Probes)were used at a dilution of 1:400. Nuclei were counterstained with 0.5 �g/mlDAPI.

Intercalated and lobular ducts (both intralobular and interlobular ducts)were identified by their characteristic morphology (20) from the H&E-stained sections and confirmed by immunostaining with the antibodyagainst CAII (data not shown) (21). Nuclei for each kind of duct werecounted by digital micrography from premetamorphic tadpoles, 2-month-old control, and adult frog pancreases (n � 5 per group). The area of eachsection was estimated by digital imaging using ImageJ software (NationalInstitutes of Health). The number of each ductal cell type is expressed persquare millimeter of pancreas.

Total pancreatic RNA and DNA were isolated and quantified at differentdevelopmental stages after extraction with TRIzol (Invitrogen). In situhybridization was performed on 4% paraformaldehyde fixed and cryosec-tioned tissues according to the protocol established in our laboratory (6).The digoxigenin-labeled antisense probes were prepared for amylase

Fig. 7. Schematic representation of morphological and gene expressionchanges in the exocrine pancreas during X. laevis metamorphosis. The adultfrog pancreas has all of the types of ducts that have been described for adultvertebrates. Tadpoles lack the intercalated ducts. At the climax of metamor-phosis, TH induces dedifferentiation of the exocrine pancreas to a progenitorstate that has extinguished the mRNAs encoding terminally differentiatedproteins (represented by amylase). PDX-1, Notch-1, and Hes-1 are up-regulated. Late in climax Ptf1a is reactivated followed by the differentiationof acinar and duct cells. Abbreviations: ac-acinar; ld-lobular duct, icd-intercalated duct. Figure is not drawn to scale.

8966 � www.pnas.org�cgi�doi�10.1073�pnas.0803569105 Mukhi et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

(GenBank accession no. BF048382), Ptf1a (GenBank accession no.AY372268), Notch-1 (GenBank accession no. M33874), Hes-1 (GenBankaccession no. AF383160), and PDX-1 (GenBank accession no. X16849) asdescribed (45). Images were taken with a Nikon Eclipse E800 microscopeand SPOT RT digital camera (Diagnostic Instruments). In each group, 5–10animals were used for in situ hybridization, immunostaining, and histol-ogy. Identical results were obtained for samples from the same stage.

Real-time RT-PCR was used to quantify mRNA in both spontaneous andT3-induced pancreatic samples (SI Text).

ACKNOWLEDGMENTS. We thank Drs. Marko Horb, Steven Leach, and TomasPieler for theirvaluableadviceandgiftsofplasmidsandRejeanneJusteforexperttechnical assistance. The research was funded by grants from the National Insti-tutes of Health and the G. Harold and Leila Y. Mathers Charitable Trust.

1. Leloup J, Buscaglia M (1977) Triiodothyronine: Metamorphosis hormone of the am-phibians. C R Acad Sci 284:2261–2263.

2. Dodd MHI, Dodd JM (1976) The Biology of Metamorphosis (Academic, New York).3. Schreiber AM, Das B, Huang H, Marsh-Armstrong N, Brown DD (2001) Diverse devel-

opmental programs of Xenopus laevis metamorphosis are inhibited by a dominantnegative thyroid hormone receptor. Proc Natl Acad Sci USA 98:10739–10744.

4. Marsh-Armstrong N, Cai L, Brown DD (2004) Thyroid hormone controls the develop-ment of connections between the spinal cord and limbs during Xenopus laevis meta-morphosis. Proc Natl Acad Sci USA 101:165–170.

5. Das B, Schreiber AM, Huang H, Brown DD (2002) Multiple thyroid hormone-inducedmuscle growth and death programs during metamorphosis in Xenopus laevis. ProcNatl Acad Sci USA 99:12230–12235.

6. Cai L, Brown DD (2004) Expression of type II iodothyronine deiodinase marks the timethat a tissue responds to thyroid hormone-induced metamorphosis in Xenopus laevis.Dev Biol 266:87–95.

7. Brown DD, et al. (2005) Thyroid hormone controls multiple independent programsrequired for limb development in Xenopus laevis metamorphosis. Proc Natl Acad SciUSA 102:12455–12458.

8. Horb ME, Slack JM (2002) Expression of amylase and other pancreatic genes in Xeno-pus. Mech Dev 113:153–157.

9. Bollin E, Jr, Carlson CA, Kim KH (1973) Anuran pancreas development during thyroxine-induced metamorphosis of Rana catesbeiana: RNA metabolism during the regressivephase of pancreas development. Dev Biol 31:185–191.

10. Milano EG, Chimenti C (1995) Morphogenesis of the pancreas of Bufo bufo duringmetamorphosis. Gen Comp Endocrinol 97:239–249.

11. Leone F, Lambert-Gardini S, Sartori C, Scapin S (1976) Ultrastructural analysis of somefunctional aspects of Xenopus laevis pancreas during development and metamorpho-sis. J Embryol Exp Morphol 36:711–724.

12. Shi YB, Brown DD (1990) Developmental and thyroid hormone-dependent regulationof pancreatic genes in Xenopus laevis. Genes Dev 4:1107–1113.

13. Lin JW, et al. (2004) Differential requirement for ptf1a in endocrine and exocrinelineages of developing zebrafish pancreas. Dev Biol 270:474–486.

14. Guz Y, et al. (1995) Expression of murine STF-1, a putative insulin gene transcriptionfactor, in �-cells of pancreas, duodenal epithelium, and pancreatic exocrine andendocrine progenitors during ontogeny. Development 121:11–18.

15. Habener JF, Stoffers DA (1998) A newly discovered role of transcription factors involvedin pancreas development and the pathogenesis of diabetes mellitus. Proc Assoc AmPhysicians 110:12–21.

16. Murtaugh LC, Stanger BZ, Kwan KM, Melton DA (2003) Notch signaling controlsmultiple steps of pancreatic differentiation. Proc Natl Acad Sci USA 100:14920–14925.

17. Hald J, et al. (2003) Activated Notch1 prevents differentiation of pancreatic acinar cellsand attenuate endocrine development. Dev Biol 260:426–437.

18. Esni F, et al. (2004) Notch inhibits Ptf1 function and acinar cell differentiation indeveloping mouse and zebrafish pancreas. Development 131:4213–4224.

19. Ghosh B, Leach SD (2006) Interactions between Hairy/Enhancer of Split-related pro-teins and the pancreatic transcription factor Ptf1–p48 modulate function of the PTF1transcriptional complex. Biochem J 393:679–685.

20. Egerbacher M, Bock P (1997) Morphology of the pancreatic duct system in mammals.Microsc Res Tech 37:407–417.

21. Lohr M, et al. (2001) Immortalized bovine pancreatic duct cells become tumorigenicafter transfection with mutant k-ras. Virchows Arch 438:581–590.

22. Janes RG (1937) Studies on the amphibian digestive system II. Comparative histology ofthe pancreas, following early larval development, in certain species of anura. J Mor-phol 61:581–611.

23. Race JJ, Robinson C, Terry RJ (1966) The influence of thyroxine on the normal devel-opment of pancreas in Rana pipens larvae. J Exp Zool 162:181–192.

24. Jarikji ZH, et al. (2007) Differential ability of Ptf1a and Ptf1a-VP16 to convert stomach,duodenum, and liver to pancreas. Dev Biol 304:786–799.

25. Afelik S, Chen Y, Pieler T (2006) Combined ectopic expression of Pdx1 and Ptf1a/p48results in the stable conversion of posterior endoderm into endocrine and exocrinepancreatic tissue. Genes Dev 20:1441–1446.

26. Cockell M, Stevenson BJ, Strubin M, Hagenbuchle O, Wellauer PK (1989) Identificationof a cell-specific DNA-binding activity that interacts with a transcriptional activator ofgenes expressed in the acinar pancreas. Mol Cell Biol 9:2464–2476.

27. Jensen J, et al. (2000) Control of endodermal endocrine development by Hes-1. NatGenet 24:36–44.

28. Habener JF, Kemp DM, Thomas MK (2005) Minireview: Transcriptional regulation inpancreatic development. Endocrinology 146:1025–1034.

29. Apelqvist A, et al. (1999) Notch signaling controls pancreatic cell differentiation.Nature 400:877–881.

30. Wang RN, Kloppel G, Bouwens L (1995) Duct- to islet-cell differentiation and isletgrowth in the pancreas of duct-ligated adult rats. Diabetologia 38:1405–1411.

31. Tezel E, et al. (2004) REG I as a marker for human pancreatic acinoductular cells.Hepatogastroenterology 51:91–96.

32. Strobel O, et al. (2007) �-Cell transdifferentiation does not contribute to preneoplastic/metaplastic ductal lesions of the pancreas by genetic lineage tracing in vivo. Proc NatlAcad Sci USA 104:4419–4424.

33. Jensen JN, et al. (2005) Recapitulation of elements of embryonic development in adultmouse pancreatic regeneration. Gastroenterology 128:728–741.

34. Means AL, et al. (2005) Pancreatic epithelial plasticity mediated by acinar cell trans-differentiation and generation of nestin-positive intermediates. Development132:3767–3776.

35. Schreiber AM, Cai L, Brown DD (2005) Remodeling of the intestine during metamor-phosis of Xenopus laevis. Proc Natl Acad Sci USA 102:3720–3725.

36. Rooman I, et al. (2006) Expression of the Notch signaling pathway and effect onexocrine cell proliferation in adult rat pancreas. Am J Pathol 169:1206–1214.

37. Sawey ET, Johnson JA, Crawford HC (2007) Matrix metalloproteinase 7 controls pan-creatic acinar cell transdifferentiation by activating the Notch signaling pathway. ProcNatl Acad Sci USA 104:19327–19332.

38. Lardon J, Bouwens L (2005) Metaplasia in the pancreas. Differentiation 73:278–286.39. Beck CW, Slack JM (1999) Gut-specific expression using mammalian promoters in

transgenic Xenopus laevis. Mech Dev 88:221–227.40. Kruse F, Rose SD, Swift GH, Hammer RE, MacDonald RJ (1993) An endocrine-specific

element is an integral component of an exocrine-specific pancreatic enhancer. GenesDev 7:774–786.

41. Kroll KL, Amaya E (1996) Transgenic Xenopus embryos from sperm nuclear transplan-tations reveal FGF signaling requirements during gastrulation. Development122:3173–3183.

42. Huang H, Marsh-Armstrong N, Brown DD (1999) Metamorphosis is inhibited in trans-genic Xenopus laevis tadpoles that overexpress type III deiodinase. Proc Natl Acad SciUSA 96:962–967.

43. Smolich BD, Tarkington SK, Saha MS, Stathakis DG, Grainger RM (1993) Characteriza-tion of Xenopus laevis �-crystallin-encoding genes. Gene 128:189–195.

44. Nieuwkoop PD, Faber J (1956) Normal tables of Xenopus laevis (Daudin) (NorthHolland, Amsterdam).

45. Berry DL, Schwartzman RA, Brown DD (1998) The expression pattern of thyroidhormone response genes in the tadpole tail identifies multiple resorption programs.Dev Biol 203:12–23.

Mukhi et al. PNAS � July 1, 2008 � vol. 105 � no. 26 � 8967

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021