Impaired di¡erentiation of endocrine and exocrine cells of the pancreasin transgenic mouse expressing the truncated type II activin receptor

Shuichi Shiozaki a, Tomoko Tajima a, You-Qing Zhang a, Megumi Furukawa a,Yoichi Nakazato b, Itaru Kojima a;*

a Department of Cell Biology, Institute for Molecular and Cellular Regulation, Gunma University, Maebashi 371, Japanb First Department of Pathology, Gunma University School of Medicine, Maebashi 371, Japan

Received 28 September 1998; received in revised form 22 January 1999; accepted 19 February 1999

Abstract

Activin A is expressed in endocrine precursor cells of the fetal pancreatic anlage. To determine the physiologicalsignificance of activins in the pancreas, a transgenic mouse line expressing the truncated type II activin receptor under thecontrol of L-actin promoter was developed. Histological analyses of the pancreas revealed that the pancreatic islets of thetransgenic mouse were small in size and were located mainly along the pancreatic ducts. Immunoreactive insulin was detectedin islets, some acinar cells, and in some epithelial cells in the duct. In addition, there were abnormal endocrine cells outsidethe islets. The shape and the size of the endocrine cells varied and some of them were larger than islets. These cells expressedimmunoreactive insulin and glucagon. In the exocrine portion, there were morphologically abnormal exocrine cells, whichdid not form a typical acinar structure. The cells lacked spatial polarity characteristics of acinar cells but expressedimmunoreactive amylase, which was distributed diffusely in the cytoplasm. Plasma glucose concentration was normal in thetransgenic mouse before and after the administration of glucose. The insulin content of the pancreas in transgenic andnormal mice was nearly identical. These results suggest that activins or related ligands regulate the differentiation of thepancreatic endocrine and exocrine cells. ß 1999 Elsevier Science B.V. All rights reserved.

Keywords: Islet ; L-Cell ; Insulin; Glucagon; Amylase

1. Introduction

The members of the transforming growth factor(TGF)-L supergene family regulate cell proliferation,di¡erentiation, morphogenesis and tissue remodeling(for reviews see [1^3]). Activins belong to this super-family and elicit diverse e¡ects in various cell systems[4,5]. In amphibians, activins and related factors are

thought to be a regulator of mesoderm induction [6^10]. They induce mesoderm of a dorsoanterior naturein Xenopus explants. They also modulate the forma-tion of neural tissue [11,12]. Despite the importanceof activins in the regulation of development in am-phibians, the role of activins in the development ofmammalians is less clear. In the mouse embryonicstem cells, activin A is shown to induce the forma-tion of dorsoanterior-like mesoderm [13]. Mutantmice lacking genes encoding activin subunits weredeveloped. Activin LA-de¢cient mice develop toterm but die within 24 h of birth. They lack whiskersand lower incisors and have abnormalities in cranio-

0167-4889 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 7 - 4 8 8 9 ( 9 9 ) 0 0 0 2 2 - 1

Abbreviations: tActRII, truncated type II activin receptors;BMP, bone morphogenic protein; OP, osteogenic protein

* Corresponding author. Fax: +81-27-220-8893;E-mail : ikojima@akagi.sb.gunma-u.ac.jp

BBAMCR 14455 21-4-99

Biochimica et Biophysica Acta 1450 (1999) 1^11

facial development [14]. Mice lacking the activin LB

subunit have an apparent failure of eyelid fusion [15].Mice de¢cient in both activin LA and LB subunitsshowed the defects of both individual mutants [15].On the other hand, mutant mice lacking the type IIactivin receptor have suppressed follicle-stimulatinghormone secretion and defects in reproductive per-formance [16]. Mutant mice lacking the type IIB ac-tivin receptor have severe anomalies in the cardiovas-cular system [17]. Given that activins are expressed invarious organs [18] and are considered to regulatevarious cellular functions, abnormalities found inthese mutant mice were fewer than expected. Thismay be due at least partly to the existence of multiplesubunits of activin [19^21], multiple subtypes of ac-tivin receptors [22^24] and other possible reasons in-cluding maternal delivery of activins.

In the pancreas, activin A is expressed in endocrinecells of the islets [25] and is expressed during devel-opment [26]. In particular, activin A is expressed inthe endocrine precursor cells of the pancreatic anlage[26]. Since activin A is capable of converting amy-lase-secreting pancreatic AR42J cells into endocrinecells [27], it is possible that activin A modulates thedi¡erentiation of pancreatic endocrine cells in vivo.To clarify the role of activins in the pancreas, wegenerated transgenic mice expressing truncated typeII activin receptors (tActRII). The tActRII acts as adominantly negative form of the activin receptor sys-tem [11,28] and blocks the action of activins andrelated factors. These transgenic mice had variousabnormalities of the pancreas.

2. Materials and methods

2.1. Transgene construct

A 0.5 kb fragment of tActRII cDNA was linked tocytomegalovirus enhancer and chicken-L-actin pro-moter (Fig. 1). A 2.9 kb fragment that containedrabbit L-globin gene, SV40 polyA, was isolatedfrom plasmid vector by digestion with SalI andPstI site.

2.2. PCR analysis

The presence of transgene in the founders and

their progeny was con¢rmed by polymerase chainreaction (PCR) typing of tail DNA using primersspeci¢c for human type II activin receptor sequence5P-TTCTCCGGCGCCTCGGGAAA-3P and rabbitL-actin sequence 5P-GTGGTATTTGTGAGCCAG-GG. For PCR, 50 Wl of the reaction mixture (con-taining PCR bu¡er, 1.5 mM MgCl2, 0.2 mM dNTPmix, 0.5 WM each of the two primers, 2.5 U Taqpolymerase) was added to the microfuge tube, andthen subjected to 30 PCR cycles. In each cycle, sam-ples were heated at 94³C for 45 s to denature DNA,cooled to 68³C for 45 s to anneal the primers andheated at 72³C for 45 s to activate the polymerase.The reaction mixture was then electrophoresed in2.0% agarose gel. The band of transgene was de-tected at 550 base pairs.

2.3. Generation and screening of transgenic mice

B6C3F1 hybrid (females; approximately 8 weeksof age), C57BL/6N (males; approximately 8 weeksof age), ICR (females; approximately 8 weeks ofage) mice were obtained from Charles River Japan(Tokyo, Japan). Embryos at the 1-cell stage wereisolated from super ovulated B6C3F1 females matedto C57BL/6N males. Transgenic mice were generatedby pronuclear microinjection of the DNA constructat a concentration of approximately 2 ng/Wl in PBSat pH 7.4. Embryos surviving the injection were re-implanted into pseudopregnant ICR foster mothersand allowed to develop to term. After the mice hadbeen weaned, total genomic DNA was isolated froma part of the tail to test for the presence of the trans-gene construct by Southern blot hybridization andby PCR analysis.

2.4. Northern blot analysis

Total RNA was obtained by the guanidinium thio-cyanate method. RNA yields were estimated fromthe absorbance at 260 nm. Ten Wg of RNA wasfractionated on 1.0% agarose formamide gels andblotted onto nylon membrane. The membrane wasprehybridized at 42³C for 2 h in a medium(5USSC, 1.0% SDS, 5UDenhardt's solution, 50%formamide and 0.5 mg/ml of denatured salmonsperm DNA) before adding the 32P-labeled tActRIIcDNA probe. The membrane was hybridized at 42³C

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^112

overnight and washed in 2USSC, 0.1% SDS solutionfor 10 min, then exposed to X-ray overnight.

2.5. Histochemistry and immunocytochemistry

Tissues were ¢xed in 4% (wt/vol) paraformalde-hyde or 10% (v/v) formalin. Standard techniqueswere used for embedding, sectioning and stainingof tissues. Immunocytochemistry was done as de-scribed previously [25]. Antibodies against insulinand glucagon were provided by Dr. Wakabayashiof Biosignal Research Center, Gunma University.Anti-activin A antibody, provided by Dr. Y. Eto ofthe Central Research Laboratory, Ajinomoto Inc.(Kawasaki, Japan), cross-reacts with both LA dimerand LA monomer but not with either inhibin A orfollistatin-bound activin A [25]. Antibody againstCD68 was purchased from Santa Cruz Biotechnol-ogy (Santa Cruz, CA, USA) and anti-myoglobinantibody was from Daco (Glostrup, Denmark). Tocompare the size of islets in normal and transgenicpancreas, we measured the diameter of islets. Whenthe islet was oval, the mean of the major and minoraxes was calculated. We also measured the distancebetween the margin of an islet and its nearest duc-tule. The area of islets and exocrine tissue was meas-ured by using a computer program `Photoshop'.

2.6. Glucose tolerance test

After overnight fasting, 1 g/kg of glucose dissolvedin physiological saline was administered intraperito-neally. At 0, 30, 60, 90 and 120 min after the admin-

istration of glucose, blood glucose concentration wasmeasured by the glucose oxides method. Plasma in-sulin concentration was measured as described pre-viously [27]. Statistical analysis was done by Stu-dent's t-test.

3. Results

A transgene was constructed by fusing a kinase-deleted human ActRII cDNA downstream of thehuman L-actin promoter (LA-tActRII) (Fig. 1). Wehave previously shown that mutated human ActRIIlacking kinase activity blocks activin receptor signal-ing by forming non-functional oligomers with wild-type mouse ActRI that binds the same ligand [28].The L-actin promoter was chosen because it directs ahigh level of transgene [29]. Expression of this trans-gene should inhibit activin receptor signaling.

Of 47 potential transgenic founder mice born, sev-en carried the transgene. All of them were born aliveand appeared relatively well. They were phenotypi-cally normal except that many of them lackedwhiskers and reached adulthood. Five of them werefemale and two were male. Female founders wereinfertile. Two male founders were fertile and twotransgenic lines were generated (S19 and S2). In theS19 mouse line, the intact LA-tActRII transgene wasstably integrated at a single site containing three cop-ies per haploid genome and transmitted in typicalMendelian fashion. The S19 line of LA-tActRIItransgenic mice actively expressed human tActRIIRNA. Total RNA was isolated from representative

Fig. 1. Transgene construct. Schematic presentation of the LA-tActRII is shown. The tActRII was produced by PCR, which retainedthe extracellular domain and transmembrane domain but deleted most of the intracellular domain.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^11 3

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^114

tissues of the transgenic line and analyzed by North-ern blotting using a human ActRII speci¢c probe.The transgene was transcribed in a wide variety oftissues including the pancreas (data not shown).

The S19 transgenic mice were apparently normalexcept that approximately 50% of them lackedwhiskers. They grew normally but female mice wereinfertile although histological abnormality was notfound in the ovary. Histological analyses were donein the transgenic mice aging 2 to 5 months. The sizeand the wet weight of the pancreas in the transgenicmice were nearly identical to those in normal mice.Histological analyses of the pancreas of the trans-genic mice revealed that most of the pancreatic isletswere small and located in the vicinity of the pancre-atic ducts (Fig. 2A). In normal mice, the mean diam-eter of islets was 235 Wm (n = 495 from 10 mice) andthe diameter was less than 100 Wm in only 11% of theislets (Fig. 3). In the transgenic mice, the mean diam-eter of islets was 76 Wm (n = 426 from 8 mice). In64% of the islets, the diameter was less than 100

Wm (Fig. 3). In normal mice, the distance from mar-gin of islet to the nearest ductule was less than 200Wm in 32% of the islets whereas, in the transgenicmice, it was less than 200 Wm in 87% of the islets.In some islets of the transgenic mice, the distributionof the islet cells and staining of nuclei were irregular(Fig. 2B). This was partly due to the inclusion of aductule in the islet. In¢ltration of mononuclear cellswas occasionally found in connective tissues sur-rounding the pancreatic duct (Fig. 2C). These cellswere lymphocytes since they expressed CD3, amarker of T lymphocytes (data not shown). Immu-nocytochemistry using anti-insulin antibody showedthat, as in islets of normal mice, the positive immu-noreactivity was found in the core of small pancre-atic islets (Fig. 4A). In addition, it was detected insome portions of the small acini which are locatednear the pancreatic duct. Immunoreactive insulin wasalso found in some cells of the pancreatic duct (Fig.4A). Similarly, immunoreactive glucagon (Fig. 4B)was observed in the ductal epithelial cells. Notethat cells expressing both glucagon and insulin werenot observed in the pancreatic duct. Interestingly,cells morphologically di¡erent from either acinar orislet cells were observed among acini (Fig. 5A,B).The shapes of the cells were quite variable. Somewere polygonal whereas others extended processesand connected each other. Under light microscopy,some of them appeared to be syncytial multi-nucleated cells (Fig. 5A). Numerous number of nu-clei were distributed irregularly and were often clus-tered in the cells. The sizes of the cells were quitevariable and some of them were greater than thatof islets in the transgenic mice. These cells were pos-itively stained with anti-insulin antibody (Fig. 6A).Immunoreactive glucagon was also detected in thecells (Fig. 6B). Hence, the multinucleated cells weredouble-positive cells expressing insulin and glucagon.Immunoreactivities of insulin and glucagon were de-tected throughout the cell body and were particularlystrong in some compartments. These multinucleatedcells were not macrophages since they did not express

Fig. 3. Comparison of the size of islets in normal and transgen-ic mice. The diameter of islets was measured in normal (E) andtransgenic (F) mice. The numbers of islets examined in normaland transgenic mice were 495 and 426, respectively.

Fig. 2. Histological analysis of the pancreas of the transgenic mouse. (A) Hematoxylin^eosin (HpE)-stained section of pancreas from4-month-old transgenic (TG) and normal (N) mice (100U). (B) HpE-stained section of pancreas from 4-month-old transgenic (TG)and normal (N) mice (400U). Arrows indicate irregular distribution of islet cells. (C) HpE-stained section of pancreas from a4-month-old transgenic mouse. Note that in¢ltration of mononuclear cells were observed in connective tissues surrounding the pan-creatic duct (arrow).6

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^11 5

Fig. 4. Immunohistochemistry of the pancreas of the transgenic mouse. (A) Immunohistochemistry of a section of pancreas from4-month-old transgenic (TG) and normal (N) mice using anti-insulin antibody (100U). In the transgenic mouse, insulin-positive cellswere observed in the core of the small islets. In addition, immunoreactive insulin was observed in some acinar cells (arrowhead) andin epithelial cells of the pancreatic duct (arrow). (B) Immunohistochemistry of a section of pancreatic duct from 4-month-old trans-genic (TG) and normal (N) mice using anti-glucagon antibody (400U). Note that immunoreactive glucagon was detected in the ductalcells of the transgenic but not the normal mouse.

Table 1Changes in plasma glucose concentration before and after intraperitoneal administration of glucose

Plasma glucose concentration (mM)

0 30 60 90 120 min

Normal (n = 6) 5.3 þ 0.6 8.3 þ 0.3 6.4 þ 0.3 6.9 þ 0.4 6.0 þ 0.6Transgenic (n = 5) 5.2 þ 0.9 8.1 þ 0.6 7.5 þ 0.3 7.2 þ 0.4 5.5 þ 0.5

Glucose (1 g/kg) was injected intraperitoneally and the plasma glucose concentration was measured at various times after the adminis-tration of glucose. Values are the means þ S.E. for indicated numbers.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^116

CD68, a marker antigen of macrophage. They didnot express myoglobin, a marker of muscle cells,either. Some of the abnormal endocrine cells alsoexpressed immunoreactive activin A (data notshown). Additionally, cells containing immunoreac-tive glucagon and activin A were also observed in thecore of some islets (Fig. 7). The shape of the cellswere irregular and they formed clusters. In the exo-crine portion of the pancreas of the transgenicmouse, most of the acini were morphologically nor-mal. However, some of the exocrine portion wasmorphologically abnormal. The area of abnormalexocrine cells was 20.9 þ 4.8% (mean þ S.E., n = 10)of the exocrine tissue. The shapes and the positionsof the cells were irregular and an acinar structure

was not organized. In normal acinar cells, zymogengranules were observed in the center of the acinusand basophilic ergastoplasm was distributed in thebasolateral side of the acinus (Fig. 8A). In the ab-normal portion, the ergastoplasm was not observed(Fig. 8C). Although the morphology was di¡erentfrom normal acinar cells, they were positively stainedwith anti-amylase antibody (Fig. 8D). In normal aci-ni, immunoreactive amylase distributed in the apicalside of acini (Fig. 8B) whereas in abnormal exocrinecells, immunoreactive amylase distributed di¡usely(Fig. 8D). Despite of the histological abnormalitiesin pancreatic islets, the urinary sugar was negative.The plasma glucose concentrations before and afteran intraperitoneal administration of glucose in thetransgenic mouse was nearly identical to that of nor-

Fig. 5. Histological analysis of the atypical endocrine cells inthe pancreas of the transgenic mouse. HpE-stained section ofpancreas from a 4-month-old transgenic mouse. Note that mul-tinuclear cells with various shapes were observed (arrows)(A,B).

Fig. 6. Immunohistochemistry of the atypical endocrine cells.Immunohistochemistry of adjacent sections of pancreas from a4-month-old transgenic mouse using anti-insulin (A) and anti-glucagon (B) antibodies.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^11 7

Fig. 7. Immunohistochemistry of the islet of the transgenic mouse. Immunohistochemistry of an islet from 4-month-old transgenic(TG) and normal (N) mice using anti-glucagon (Glu) and anti-activin A (Act) antibodies.

Fig. 8. Immunohistochemistry of the pancreas of normal and transgenic mice. (A) HpE-stained section of pancreas of a 4-month-oldnormal mouse (400U). Basophilic ergastoplasm was observed in the basolateral side of the acinar cells. (B) Immunohistochemistry ofpancreas of a normal mouse using anti-amylase antibody (400U). Amylase immunoreactivity was found in the luminal side of the aci-nus. (C) HpE-stained section of pancreas of the transgenic mouse (400U). Basophilic ergastoplasm was not observed. (D) Immuno-histochemistry of pancreas of a 4-month-old transgenic mouse using anti-amylase antibody. Amylase immunoreactivity was distributeddi¡usely in the cytoplasm.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^118

mal mouse (Table 1). The insulin response to glucosein the transgenic mice was identical to that in normalmice (Table 2). The insulin content of the pancreas inthe transgenic mice was identical to that in normalmice (data not shown).

4. Discussion

In the present study, we established a transgenicmouse line expressing tActRII under the control ofthe L-actin promoter. The tActRII blocks the signal-ing pathway activated by activins and related ligandsin a dominantly negative fashion [11,28]. The L-actinpromoter enables the expression of the transgene ef-fectively in various tissues [29]. Yet, the expressionlevel of the transgene may vary depending upon thecell types. Therefore, the blockade of the e¡ect ofactivins may not be complete in some types of cells.This may explain why some of the abnormalities de-scribed in the mutant mice lacking L-subunits of ac-tivin [14,15] were not observed in our transgenicmice. In this regard, we have probably underesti-mated the abnormalities caused by the blockade ofthe activin action. Despite these limitations, thepresent results provide some important informationconcerning the physiological signi¢cance of activinsand related ligands in the development and di¡eren-tiation of the pancreatic cells.

The transgenic mice expressing tActRII had vari-ous abnormalities in pancreatic endocrine cells. Mostof the pancreatic islets were small in size and manyof them located along the pancreatic ducts. In addi-tion, insulin-producing cells were found in the pan-creatic duct and small acini located by the duct.

These observations were similar to the histological¢ndings found in regenerating pancreas and suggestthat neogenesis of endocrine cells took place in thetransgenic mice. Presumably, di¡erentiation and/ormaturation of the islet endocrine cells may havebeen attenuated in these mice and neogenesis of en-docrine cells was promoted as a compensatory mech-anism. Migration of the islets may have also beeninhibited to some extent since many islets were lo-cated by the ducts. These results were in agreementwith a recent report by Yamaoka et al. [30]. Theyshowed that in transgenic mice expressing tActRIIunder the control of the insulin promoter, pancreaticislets were severely hypoplastic. In their transgenicmice, glucose tolerance was impaired and the re-sponse of insulin was signi¢cantly lower than thatin control mice. In contrast, glucose tolerance wasalmost normal in our transgenic mice. The di¡erencemay be due at least partly to the di¡erence in theexpression levels of tActRII in L-cells since Yamaokaet al. [30] used powerful insulin promoter and thecopy number was greater.

The most intriguing observation obtained in thepresent study was that there were yet undescribedabnormal endocrine cells in the pancreas. The shapesof the endocrine cells were quite variable and theirsizes were sometimes greater than those of islets.Morphologically, some of them appeared to be syn-cytium and numerous nuclei were randomly clusteredin the cytoplasm. But since we did not examine themby electron microscopy, we are not totally certainwhether or not these cells are de¢nitely syncytialcells. In any event, these cells expressed both gluca-gon and insulin and some of them also expressedactivin A. In the fetal mouse pancreas, endocrineprecursor cells ¢rst appeared on embryonic day 9(E9) [31]. These precursors express both glucagonand insulin [31^33]. We reported that in the fetalrat pancreas, endocrine precursor cells ¢rst expressedactivin A [26]. Subsequently, cells expressing activinA, insulin and glucagon were observed [26]. Hence,the double- or triple-positive endocrine cells found inthe transgenic mice resemble the endocrine precursorcells in the fetal pancreas. At present, the exactmechanism for the formation of large and possiblymultinuclear cells is unknown. In this regard, our(M.F., I.K.) unpublished observation indicated thatmultinuclear cells extending multiple long processes

Table 2Changes in plasma insulin concentration before and after intra-peritoneal administration of glucose

Plasma insulin concentration(pM)

0 30 min

Normal (n = 5) 126 þ 28 402 þ 39Transgenic (n = 5) 103 þ 14 372 þ 45

Glucose was administered as for Table 1 and the plasma insulinconcentration was measured before and after 30 min of the ad-ministration of glucose. Values are the mean þ S.E. for indicatednumber.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^11 9

were observed in regenerating rat pancreas after 90%pancreatectomy. Quite interestingly, these cells ex-pressed immunoreactive insulin, glucagon and activinA. The multinuclear cells found in the transgenicmouse may be similar to the cells found in regener-ating pancreas. Cells expressing insulin, glucagonand activin A were also observed in the core of theislets in the transgenic mice. In the fetal rat pancreas,the triple-positive cells were found in the islets onbetween E13.5 to E15 and they eventually disap-peared [26]. The existence of the triple-positive cellsin the adult islets of the transgenic mice stronglysuggested that di¡erentiation of endocrine cells wasattenuated in the transgenic mice. Activin A mayhave modulated di¡erentiation of the endocrine cellsin vivo as an autocrine factor [26]. It should be notedthat atypical endocrine cells were not observed in thetransgenic mice described by Yamaoka et al. [30].The di¡erence may be due to the di¡erent promotersused in two studies. We used the L-actin promoterwhile Yamaoka et al. [30] used the insulin promoter.It is expected that tActRII is expressed in endocrineprecursor cells in our transgenic mice whereas in thetransgenic mice developed by Yamaoka et al. [30],tActRII may be expressed after endocrine cells arecommitted to produce insulin.

In addition to the double- or triple-positive cells,another type of abnormal cells were observed in theexocrine portion of the transgenic mouse. They ex-pressed amylase and, in fact, morphologically normalacini were occasionally found among the clusters ofthe abnormal cells. Although these cells expressedamylase, they did not have the characteristics of po-larized acinar cells. Di¡erentiation into acinar cellswas inhibited to some extent in these cells. Miralleset al. [34] showed that coculture of embryonic mes-enchyme is inhibitory for the formation of endocrinecells in E12.5 rat pancreatic rudiment in culture. In-terestingly, the addition of follistatin, an inhibitor ofthe activin action, decreases the number of endocrinecells and increases the number of amylase-containingcells [34]. Their results suggest that blocking the ac-tion of endogenous activin or related ligand leads tothe formation of exocrine cells. Thus, it would beexpected that cells tend to convert to exocrine lineageif the activin action is blocked. The amylase-positivecells found in the transgenic mouse might have beentransitional cells during the course of exocrine di¡er-

entiation. We have reported recently that activin Aconverts amylase-secreting pancreatic AR42J cellsinto endocrine cells [27]. The results suggest that ac-tivin A mediates the commitment of amphicrineAR42J cells to di¡erentiate into endocrine cells. Inthe transgenic mouse, neogenesis of endocrine cellstook place. This observation, however, does not ex-clude the possibility of the involvement of activin Ain the endocrine determination. If the commitment isregulated by multiple factors including activin A,then the inhibition of commitment in the transgenicmouse would be only partial. Our unpublished ob-servation indicates that TGF-L and bone morpho-genic protein-4 (BMP-4) partially reproduces the ef-fect of activin A under some conditions in AR42Jcells. In this regard, addition of TGF-L to the cultureof pancreatic rudiment inhibits the development ofacinar tissue and promotes the development of endo-crine cells [35]. It is thus possible that endocrine de-termination is regulated by multiple members of theTGF-L superfamily. As the inhibition of the activinaction by follistatin leads to the formation of exo-crine cells [34], the appearance of abnormally di¡er-entiated exocrine cells suggests that di¡erentiation ofthe precursor cells into endocrine lineage is attenu-ated partially in the transgenic mouse. Bottinger etal. [36] developed transgenic mice expressing trun-cated type II TGF-L receptor. In these mice, severeabnormalities were observed in exocrine portion ofthe pancreas including ductular transformation, neo-angiogenesis, ¢brosis and in¢ltration of macro-phages. In addition, many of acinar cells expressedproliferating cell nuclear antigen and the number ofapoptotic cells was increased. Normal di¡erentiationand/or maintenance of di¡erentiated phenotype weredisrupted in these mice. Since abnormalities found inthese mice and transgenic mice expressing tActRIIare di¡erent, TGF-L and activin A exert di¡erente¡ects in the development of pancreatic exocrine tis-sue.

The tActRII blocks the e¡ect of activins in a dom-inantly negative fashion. In addition, the tActRIIprobably inhibits the action of osteogenic protein-1/BMP-7 (OP-1/BMP-7) since OP-1/BMP-7 also bindsto the type II activin receptor [37]. The mRNA forOP-1/BMP-7 is expressed in the pancreas althoughthe exact site of expression is not yet known [38].Therefore, we cannot rule out the possibility that

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^1110

the abnormalities found in the transgenic mice weredue at least partly to the inhibition of OP-1 action.In any event, the present results suggest that activinsor related ligands may play an important role in thedi¡erentiation of both pancreatic endocrine and exo-crine cells.

Acknowledgements

The present study was supported by Grants-in-Aidfor Scienti¢c Research from the Ministry of Educa-tion, Science, Sports and Culture of Japan, andgrants from the Japanese Pancreatic Foundation,the Japanese Diabetes Foundation, the NovartisFoundation of Japan, the Uehara Foundation, theLife Science Foundation, and the YamanouchiFoundation for Research on Metabolic Disorders.

References

[1] J. Massague, A. Hata, F. Liu, Trends Cell. Biol. 7 (1993)187^192.

[2] J.A. Barnard, R.M. Lyons, H.A. Moses, Biochim. Biophys.Acta 1032 (1990) 79^87.

[3] W.A. Border, N.A. Noble, N. Engl. J. Med. 331 (1994)1286^1292.

[4] S.Y. Ying, Endocrine Rev. 9 (1988) 267^293.[5] L.V. DePaolo, T.A. Bicsak, G.F. Erickson, S. Shimasaki, N.

Ling, N. Proc. Soc. Exp. Biol. Med. 198 (1991) 500^512.[6] J.C. Smith, B.M.J. Price, K.V. Nimmen, D. Huylebroeck,

Nature 345 (1990) 729^731.[7] G. Thomsen, T. Woolf, M. Whitman, S. Sokol, V. Vaughan,

W. Vale, D.A. Melton, Cell 63 (1990) 485^493.[8] A. Hemmati-Brivanlou, D.A. Melton, Nature 359 (1992)

609^614.[9] R.A. Cornell, D. Kimelman, Development 120 (1994) 453^

462.[10] C. LaBonne, M. Whitman, Development 120 (1994) 463^

472.[11] A. Hemmati-Brivanlou, D.A. Melton, Cell 77 (1994) 273^

281.[12] A. Hemmati-Brivanlou, D.A. Melton, Cell 77 (1994) 283^

295.[13] B.M. Johansson, M.V. Wiles, Mol. Cell. Biol. 15 (1995) 141^

151.[14] M.M. Matzuk, T.R. Kumar, A. Vassalli, J.R. Bickenbach,

D.R. Roop, R. Taenisch, A. Bradley, Nature 374 (1995)354^356.

[15] A. Vassalli, M.M. Matzuk, H.A.R. Gardner, K.F. Lee, R.Jaenisch, Genes Dev. 8 (1994) 414^427.

[16] M.M. Matzuk, T.R. Kumar, A. Bradley, Nature 23 (1995)356^359.

[17] S.P. Oh, E. Li, Genes Dev. 11 (1997) 1812^1826.[18] H. Meunier, C. Rivier, R.M. Evans, W. Vale, Proc. Natl.

Acad. Sci. USA 85 (1998) 247^251.[19] S. Oda, S. Nishimatsu, K. Murakami, N. Ueno, Biochem.

Biophys. Res. Commun. 210 (1995) 581^588.[20] G. Hotten, H. Neidhardt, C. Schneider, J. Pohl, Biochem.

Biophys. Res. Commun. 206 (1995) 608^613.[21] J. Fang, W. Yin, E. Smiley, S.Q. Wang, J. Bonadio, Bio-

chem. Biophys. Res. Commun. 228 (1996) 669^674.[22] L.S. Mathews, W. Vale, Cell 65 (1991) 973^982.[23] L. Attisano, J.L. Wranna, S. Cheifetz, J. Massague, Cell 68

(1992) 97^108.[24] P. ten Dike, H. Ichijo, P. Franzen, P. Schulz, J. Saras, J.

Toyoshima, C.H. Heldin, K. Miyazono, Oncogene 8 (1993)2879^2887.

[25] H. Yasuda, K. Inoue, H. Shibata, T. Takeuchi, E. Eto, Y.Hasegawa, N. Sekine, Y. Totsuka, T. Mine, E. Ogata, I.Kojima, Endocrinology 133 (1993) 624^630.

[26] M. Furukawa, Y. Eto, I. Kojima, Endocrine J. 42 (1995) 63^68.

[27] H. Mashima, H. Ohnishi, K. Wakabayashi, J. Miyagawa, T.Hanafusa, M. Seno, H. Yamada, I. Kojima, J. Clin. Invest.97 (1996) 1647^1654.

[28] Y.Q. Zhang, H. Mashima, M. Kanzaki, H. Shibata, I. Ko-jima, Hepatology 25 (1997) 1370^1375.

[29] M. Ishii, F. Tashiro, S. Hagiwara, T. Toyoshima, C. Hashi-moto, I. Takei, K. Yamamura, J. Miyazaki, Endocrine J. 41,((suppl)) (1994) S9^S16.

[30] T. Yamaoka, C. Idehara, M. Yano, T. Matsushita, T. Ya-mada, S. Ii, M. Moritani, J. Hata, H. Sugino, S. Noji, M.Itakura, J. Clin. Invest. 102 (1998) 294^301.

[31] G.K. Gittes, W.J. Rutter, Proc. Natl. Acad. Sci. USA 89(1992) 1128^1132.

[32] G. Teitelman, J.K. Lee, Development 121 (1987) 454^466.[33] G. Teitelman, S. Alpert, J.M. Polak, A. Martinez, D. Hana-

han, Development 118 (1993) 1031^1039.[34] F. Miralles, P. Czernichow, R. Soharfmann, Development

125 (1998) 1017^1024.[35] F. Sanvito, P.L. Herrera, J. Huarte, A. Nicholos, R. Mon-

tesano, L. Orci, J.D. Vassalli, Development 120 (1994) 3451^3462.

[36] E.P. Bottinger, J.L. Jakubczak, I.S.D. Roberts, M. Mumy,P. Hemmati, K. Bagnall, G. Merlino, L. Wake¢eld, EMBOJ. 16 (1997) 2621^2633.

[37] H. Yamashita, P. ten Dike, D. Huylebroeck, T.K. Sampath,M. Andries, J.C. Smith, C.H. Heldin, K. Miyazono, J. Cell.Biol. 130 (1995) 217^226.

[38] S. Vukicevic, V. Latin, P. Chen, R. Batorsky, A.H. Reddi,T.K. Sampath, Biochem. Biophys. Res. Commun. 198 (1994)693^700.

BBAMCR 14455 21-4-99

S. Shiozaki et al. / Biochimica et Biophysica Acta 1450 (1999) 1^11 11