THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 15, Issue of May 25, pp. 10705-10708,1992 Printed in U.S.A.

Glucagon Gene 5”Flanking Sequences Direct Expression of Simian Virus 40 Large T Antigen to the Intestine, Producing Carcinoma of the Large Bowel in Transgenic Mice*

(Received for publication, November 27, 1991)

Ying C. Lee, Sylvia L. Asa, and Daniel J. Drucker From the Departments of Medicine, Clinical Biochemistry, Genetics, and Pathology and the Banting and Best Diabetes Centre, University of Toronto, Toronto, Ontario M5G 2C4, Canada

Glucagon and the glucagon-like peptides play impor- tant roles in the regulation of glucose homeostasis. Previous studies have demonstrated that -1300 base pairs of rat glucagon gene 5’-flanking sequences direct transgene expression to the pancreas and brain, but not to the intestine, of transgenic mice. These obser- vations suggested that different tissue-specific en- hancer elements mediate activation of glucagon gene transcription in the pancreas and intestine. We have now generated mice that express SV40 large T antigen under the control of -2000 base pairs of glucagon gene 5’-flanking sequences. Transgene expression was ob- served in the brain and pancreas in association with the development of pancreatic endocrine tumors. In contrast to the mice described previously, we also de- tected transgene expression throughout the gastroin- testinal tract in endocrine cells of the stomach and small and large intestine. Focal areas of enteroendo- crine cell hyperplasia in the large bowel invariably progressed to invasive and metastasizing plurihor- monal endocrine carcinoma, which was clinically and pathologically evident by 4 weeks of age. In contrast, transgene expression in the small bowel and stomach was not associated with progression to either hyper- plasia or carcinoma. The results of these studies pro- vide functional evidence for the existence of an up- stream cis-acting regulatory domain that directs glu- cagon gene transcription to the endocrine cells of the intestine in transgenic mice.

The glucagon gene is expressed in the A cells of the endo- crine pancreas, in the L cells of the intestine, and in the central nervous system, predominantly in brainstem neurons (1, 2). Glucagon and the glucagon-like peptides are encoded within a common proglucagon precursor that undergoes tis- sue-specific post-translational processing to liberate a differ- ent profile of proglucagon-derived peptides in the pancreas, intestine, and brain (3). These peptides have important bio- logical functions in vivo (4); glucagon secretion from the pancreas modulates hepatic gluconeogenesis and glycogeno- lysis, and glucagon-like peptide I (GLP-I)’ from the intestine mediates glucose-dependent insulin secretion from the pan- creatic B cell. The specific function(s) of the proglucagon-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are: GLP-I, glucagon-like peptide I (7- 37); bp, base pair(s); PCR, polymerase chain reaction; ACTH, adre- nocorticotropic hormone.

derived peptides in the brain have not yet been elucidated. The sequences of proglucagon cDNAs isolated from differ-

ent tissues (1, 2, 5-7) and the results of ribonuclease protec- tion studies (1, 2, 8, 9) have demonstrated that proglucagon mRNA transcripts are identical in the pancreas, intestine, and brain. A single promoter mediates transcription initiation from an identical transcription start site in the pancreas, intestine, and brain (10); and the results of gene transfer studies have identified specific cis-acting glucagon gene en- hancer and promoter sequences that direct glucagon gene transcription preferentially in islet cell lines (11, 12). In contrast, little is known about the control of glucagon gene expression in the intestine. Fetal rat primary intestinal cell cultures have been utilized for the study of intestinal glucagon gene expression in vitro (l), but molecular factors or cis-acting domains that control glucagon gene transcription in the in- testine have not been identified. Furthermore, intestinal glu- cagon-producing cell lines suitable for gene transfer studies have not yet been isolated.

Previous studies analyzing the expression of a glucagon- SV40 large T antigen transgene containing -1300 bp of rat glucagon 5”flanking sequences demonstrated expression in the brain and pancreas, but not in the gastrointestinal tract, of transgenic mice (13). These observations implied that DNA regulatory sequences mediating islet and neural specific glu- cagon gene expression were different from those necessary to direct transgene expression to the intestine. To understand the molecular basis for glucagon gene expression in the intes- tine, we have generated transgenic mice containing additional glucagon gene 5”flanking sequences upstream of the SV40 large T antigen. These mice express the transgene in the small and large intestine and develop endocrine carcinoma of the large bowel. These observations provide important evi- dence for the localization of a tissue-specific element capable of directing glucagon gene expression to the intestine in vivo.

EXPERIMENTAL PROCEDURES

Generation of Transgenic Mice-For vector codstruction, a 2.1- kilobase rat glucagon gene EcoRI/AccI fragment was ligated to a Bgn/ BamHI fragment of SV40 T antigen coding sequences to produce the GLUTag transgene. This fusion gene contains -2.0 kilobases of rat glucagon gene sequences, including 5‘-flanking sequences, and 58 bp of exon 1 (14) and contains -700 additional nucleotides of 5’-flanking sequence upstream of the KpnI site that delineates the 5’-boundary of the GLUTag-SV40 transgene described previously (13). The KpnI site is located at approximately -1300 bp (and not at -850 bp as described (13)) according to analysis of the glucagon gene in our laboratory. Transgenic mice were generated on an outbred CD-1 background as previously described (15). DNA prepared from mouse tails was used for Southern blot analysis to identify transgenic progeny.

Expression Analysis-For immunohistological and RNA analyses,

10705

10706 Glucagon Gene Expression a minimum of eight mice from each different age group were analyzed individually. RNA was isolated from tissues by the acid/phenol pre- cipitation method (16), and Northern blot analysis was carried out as previously described (2). The polymerase chain reaction (PCR) was utilized for the identification of specific mRNA transcripts as previously described (17). The primers used to identify mouse glucagon mRNAs were 5’-TTCACCAGTGACTACAGCAAA-3’ and 5’-GGTT-TGAATCAGCCAGTTGAT-3’, which resulted in the gen- eration of a 307-bp glucagon cDNA fragment. The primers used to generate a 219-bp SV40 T antigen cDNA fragment were 5’-AGAG- GAATCTTTGCAGCTAA and 5’-TGCATCCCAGAAGCCTCCAA.

Morphological Methods-Fresh tissue was fixed in 10% phosphate- buffered formalin and embedded in paraffin for histological exami- nation. Sections 5 pm thick were stained with hematoxylin and eosin. To localize endocrine peptides, the avidin-biotin-peroxidase complex technique was applied to paraffin sections using primary antiserum at different dilutions as previously described (18). The duration of exposure to primary antiserum was 24 h at 4 “C. To localize SV40 large T antigen, fresh tissue was snap-frozen in liquid nitrogen; frozen sections 5 pm thick were fixed in acetone, and the indirect immuno- peroxidase technique was performed with monoclonal antibody MAB- 419 at a dilution of 1:20, with incubation for 1 h at room temperature. For both methods of immunolocalization, the reaction was visualized using 3,3’-diaminobenzidine and hydrogen peroxide. Negative con- trols were performed by replacing the primary antibody with nonim- mune rabbit serum or normal mouse ascites fluid.

RESULTS

Three different transgenic founders were obtained that expressed the GLUTag transgene. The three established lines demonstrated identical tissue-specific patterns of GLUTag gene expression as assessed by Northern blot, PCR, and immunohistochemical analyses. Transgene expression was detected in the brain, pancreas, and intestine of all three lines; the detailed analysis described below was carried out on the line designated GLUTag-Y. Transgenic GLUTag-Y mice were visibly distinguishable from their nontransgenic littermates by 2-3 weeks of age. The transgenic mice were smaller than their wild-type littermates by 3 weeks of age and weighed 40- 50% less than the nontransgenic littermates by 4-6 weeks of age. The mice became gradually wasted and died 4-12 weeks after birth.

To determine the distribution and tissue specificity of transgene expression, RNA was prepared from various tissues and analyzed for expression of T antigen mRNA transcripts by Northern blotting. RNA prepared from day 1 mice con- tained T antigen mRNA transcripts in the pancreas and intestine, but not in the stomach, even with longer exposures of the same blot (Fig. 1, upper). In contrast, by day 10, T antigen mRNA transcripts were detectable in the stomach and small bowel; the relative amounts of T antigen mRNA transcripts were reduced in the pancreas and comparatively more abundant in the large bowel (Fig. 1, upper). Transgene expression in the brain was analyzed using reversed-tran- scribed RNA and PCR in view of the extremely low levels of glucagon (2) and glucagon-SV40 transgene (13) expression in the brain. T antigen mRNA transcripts were detected in RNA from the whole brain as well as in RNA from the brainstem and hypothalamus (Fig. 1, lower). Aliquots of the identical reverse-transcription reactions were also analyzed using glu- cagon-specific primers. In contrast to the widespread distri- bution of T antigen expression, glucagon mRNA transcripts were restricted in expression to the brainstem, in agreement with the results of previous studies (2).

Autopsies of the transgenic mice consistently revealed pa- thology in the large intestine. By 4-8 weeks of age, marked dilatation of the cecum was associated with a firm thickened segment of the large bowel extending from the cecum to the rectum (Fig. 2). To trace the temporal development of this lesion, mice were examined at different ages. In fetuses at day

DAY 1 DAY 10

B B S H B B S H

wt 19

G * a W t 18

FIG. 1. Upper, Northern blot analysis of T antigen mRNA tran- scripts from different tissues of day 1 and 10 transgenic animals. The blot depicts 10 pg of total cellular RNA from the stomach (ST), pancreas (PAN), total intestine (ZNT), small bowel (SB) , or large bowel (LB). The arrowhead indicates the migration position of the two T antigen mRNA transcripts. Equal amounts of RNA were loaded in each lane as assessed by ethidium bromide staining of the gel. The blot was exposed for 3 days. Lower, PCR analysis of T antigen and glucagon mRNA transcripts in the brain. Ten pg of total cellular RNA from the brainstem (BS) , hypothalamus (H), or brain dissected free of the hypothalamus and brainstem ( B ) was reverse-transcribed; and one-twentieth of the reaction was used in a PCR reaction with primers specific for either SV40 T antigen (Tag) or glucagon ( G ) . RNA was prepared from nontransgenic littermates (wt) and trans- genic animals (tg) 5 weeks of age. Aliquots (YIo) of the PCR reactions were analyzed by gel electrophoresis and transferred to a nylon membrane, and the blots were hybridized with a probe for T antigen (upper panel) or glucagon (lower panel).

19 of gestation, no obvious pathology was detected in any tissue with hematoxylin and eosin. The number of GLP-I- immunoreactive cells in the large bowel was slightly increased at day 19 (Fig. 3, upper and center). These cells were found individually or in very small groups and were confined to the epithelium of the mucosal crypts. In contrast, the pancreases and small bowels of fetal day 19 mice showed no histochemical evidence of hyperplasia of glucagon-containing cells using either the GLP-I or glucagon antisera (data not shown). The distribution and number of the other pancreatic endocrine cell types was entirely normal, with the exception of the pancreatic polypeptide cell, which was not consistently de- tected. Animals from l to 21 days following birth demon- strated progressive development of epithelial cell proliferation in the lamina propria of the large bowel. Immunohistochemis- try demonstrated GLP-I-immunoreactive cells both within the mucosal crypts and infiltrating into the lamina propria. These cells exhibited strong nuclear staining for SV40 T antigen (Fig. 3, lower). In the stomach and small bowel, GLP- I- and SV40 T antigen-immunoreactive cells were detected, but no areas of cellular hyperplasia were observed (data not shown). Pancreases from animals 1-3 weeks of age contained a number of irregular islets with large cells harboring pleo- morphic nuclei and clear cytoplasm at the periphery (Fig. 4, upper). The large pleomorphic cells had strong nuclear posi- tivity for SV40 T antigen (Fig. 4, lower), and occasional cells were immunopositive for glucagon or GLP-I (data not shown). The central cytologically normal cells within these islets contained insulin and somatostatin in the usual distribution,

Glucagon Gene Expression 10707

FIG. 4. Upper, at 3 weeks of age, a transgenic mouse pancreas contains clusters of abnormal cells a t the periphery of an islet; these cells contain pleomorphic nuclei and abundant chromophobic cyto- plasm (hematoxylin and eosin stains; magnification X 200). Lower, the large cells at the periphery of the islet contain focal nuclear immunopositivity for SV40 large T antigen (indirect immunoperoxi- dase technique; magnification X 360).

FIG. 2. Terminal ileum, cecum, and large bowel from 8- week-old nontransgenic control (upper) and transgenic litter- mate (lower). The thickened segment of the large bowel consists almost exclusively of endocrine carcinoma; the cecum and appendix are dilated. The ruler scale is in centimeters.

FIG. 3. At day 19 of gestation (upper), the large bowel of a trans- genic fetus contains more GLP-I-immunoreactive cells than that of a nontransgenic littermate (center). Three days postnatally (lower), the large bowel of a transgenic mouse contains clusters of cells that stain for SV40 T antigen both within crypts and infiltrating the lamina propria. Upper and center, avidin-biotin-peroxidase complex technique; lower, indirect immunoperoxidase technique. Magnifica- tion X 220.

but no pancreatic polypeptide was detected in the islets of transgenic mice after birth.

In transgenic mice 4 weeks of age or older, the large bowel was increased in thickness, with nests and solid sheets of epithelial cells filling the lamina propria, infiltrating into and

. i

FIG. 5. Lower, by 4 weeks of age, the large bowel contains foci of invasive endocrine carcinoma that penetrate through the muscular wall of the bowel (hematoxylin and eosin stains; magnification X 70). Upper, the nuclei of the invasive carcinoma are strongly positive for SV40 large T antigen (indirect immunoperoxidase technique; mag- nification X 180).

through the muscularis propria (Fig. 5) and beyond the serosa of the bowel. Occasional animals had lymph node metastases involving pericolonic and para-aortic lymph nodes. The tumor was composed of epithelial cells with pleomorphic nuclei often harboring multiple nucleoli and occasionally containing intra- nuclear eosinophilic inclusions; numerous tumor cells exhib- ited strong nuclear positivity for SV40 T antigen (Fig. 5). Tumor cells were positive for neuron-specific enolase and synaptophysin; stains for chromogranin were negative (data not shown). The tumor cells contained moderate cytoplasmic immunoreactivity for glucagon and were strongly positive for GLP-I and peptide YY. In a few animals, the tumors stained positively in a focal distribution for cholecystokinin, and scattered tumor cells were positive for the a-subunit of gly- coprotein hormones. The tumors were entirely negative for insulin, somatostatin, pancreatic polypeptide, vasoactive in- testinal peptide, serotonin, gastrin, calcitonin, bombesin,

10708 Glucagon Gene Expression

ACTH, @-endorphin, corticotropin-releasing hormone, growth hormone releasing hormone, and carcinoembryonic antigen.

In the pancreas, by 4 weeks of age, many islets were large and displayed disrupted architecture (data not shown). The large islets contained two distinct cell populations. There were central compressed clusters of bland epithelial cells with vesicular nuclei and eosinophilic cytoplasm. The periphery of the islets was composed of numerous pleomorphic cells with hyperchromatic nuclei that often harbored intranuclear eosin- ophilic inclusions and stained for SV40 T antigen; their cytoplasm was abundant and chromophobic, and the cell borders were indistinct. Occasional large cells contained cy- toplasmic GLP-I immunopositivity. The small clusters of residual islet cells contained insulin and somatostatin but no reactivity for SV40 T antigen, pancreatic polypeptide, gluca- gon, or GLP-I. Although in the largest islets the pleomorphic cells comprised the majority of the islet parenchyma, there was no evidence of infiltration into the surrounding exocrine pancreas.

DISCUSSION

The results of previous studies (13) have suggested that tissue-specific enhancer elements may regulate the expression of the glucagon gene in uiuo. The observations described here provide evidence for a glucagon gene intestinal specific cis- acting element that resides between -1300 and 2000 bp rela- tive to the start of transcription. This element directs trans- gene expression to endocrine cells in the stomach as well as to the small and large intestine of transgenic mice. In contrast, GLU2-Tag mice transgenic for a glucagon-SV40 transgene containing 1300 bp of rat glucagon gene 5"flanking sequences did not express the transgene in the stomach or intestine in any of the lines studied (13). The pancreatic pathology in our GLUTag-Y mice also differs significantly from that described previously in the GLU2-Tag transgenic line. GLUTag-Y transgenic mice contained two distinct populations of islet cells that were distinguishable by 3 weeks of age; large hyper- plastic islets were consistently detected by 4 weeks, and progression to tumor formation was observed from 4 to 8 weeks of age. In contrast, proliferation of A cells in the GLU2- Tag mice was only observed at 5 months of age (13); and by 9-12 months of age, only a few islets had progressed to become solid tumors. In contrast, RIP1-Tag mice transgenic for an insulin-SV40 transgene developed islet cell hyperplasia and tumor formation by 8 weeks of age (19). These differences in the development of hyperplasia and tumor formation are not explained simply by differences in the developmental onset of transgene expression in the pancreas in that GLUTag-Y, GLU2-Tag, and RIP1-Tag2 mice all contained SV40 T anti- gen-reactive cells in fetal and neonatal pancreases. It is pos- sible that the DNA sequences upstream of -1300 bp direct higher level expression of the transgene in the endocrine pancreas, accelerating the onset of histologically detectable abnormalities in the pancreatic islets of GLUTag-Y mice. The relative differences in the transcriptional properties of the -2000 uersus -1300 bp glucagon gene 5'-flanking sequences should be testable in uitro using gene transfer experiments and islet cell lines.

The localization of GLUTag transgene expression in the brain differed in part from the localization of endogenous mouse glucagon gene mRNA transcripts. As described in the rat (2, 20), the mouse glucagon gene also appears to be expressed predominantly in the brainstem in both transgenic animals and nontransgenic littermates. In contrast, T antigen

mRNA transcripts were detected in RNAs from the hypo- thalamus, brainstem, and total brain dissected free of the hypothalamus and brainstem. The more widespread expres- sion of the GLUTag transgene compared with the endogenous glucagon gene agrees with previous reports of a discrepancy between brain regions displaying immunoreactivity for T antigen and the glucagon-like peptides (13). One potential explanation for these results invokes the possibility that glucagon gene sequences (either additional 5'-flanking or downstream sequences) not present in the transgenes studied to date are required for the correct region-specific restriction of glucagon gene expression in the brain. Alternatively, but less likely, the glucagon gene sequences that mediate correct brain-specific expression in the rat may not be identical to the corresponding sequences utilized for brain-specific expres- sion in the mouse. Furthermore, it is possible that the reporter used (SV40 T antigen) may lead to generalized activation of glucagon gene expression in the brain.

GLUTag-Y transgenic mice consistently developed endo- crine cell hyperplasia progressing to carcinoma in the large intestine. Although transgene expression was also detected in endocrine cells throughout the small intestine and stomach, no endocrine cell hyperplasia was detected in these two tis- sues. A number of different mechanisms may potentially account for this observation, including differing levels of transgene expression and variation in the time of onset of transgene expression in endocrine cells of the small and large intestine. The relative degree of transgene expression alone cannot explain the tissue-specific differences since levels of T antigen mRNA transcripts were higher in the pancreas than in the intestine during the first few days of life; however, hyperplasia developed more rapidly in the large intestine than in the pancreas. The mechanisms underlying the tissue-spe- cific differences in susceptibility to neoplastic transformation remain unknown. The availability of glucagon-producing in- testinal tumors should permit the isolation of intestinal cell lines that, when propagated in uitro, will be useful for the identification of the cis- and tram-acting factors important for the molecular control of glucagon gene expression in the intestine.

REFERENCES 1. Drucker, D. J., and Brubaker, P. L. (1989) Proc. Natl. Acad. Sci. U. S. A.

2. Drucker, D. J., and Asa, S. (1988) J. Biol. Chern. 263,13475-13478 3. Mojsov, S., Heinrich, G., Wilson, I. B., Ravazzola, M., Orci, L., and

4. Unger, R. H. (1985) Diabetolo ia 28,574-578 5. Be!!,?. I., Santerre, R. F., ancfMullenbach, G. T. (1983) Nature 302,716-

86,3953-3957

Habener, J. F. (1986) J. Biol. Chern. 261,11880-11889

6. Lo ez, L. C., Frazier, M. L., Su, C. J., Kumar, A,, and Saunders, G. F.

7. Heinrich, G., Gros, P., Lund, P. K., Bentley, R. C., and Habener, J. F.

8. Novak, U., Wilks, A., Buell, G., and McEwen, S. (1987) Eur. J. Biochern.

9. Seino, S, Welsh, M., Bell, G. I., Chan, S. J., and Steiner, D. F. (1986) FEES

10. Lee. Y. C.. Brubaker, P. L.. and Drucker, D. J. (1990) Endocrinology 127 ,

' I 18

k983) Proc. Natl. Acad. Sci. U. S. A. 80,5485-5489

(1984) Endocrinology 115,2176-2181

164,553-558

Lett. 203,25-30

11.

12.

13.

14.

15.

16. 17. 18.

19. 20.

Philippe, J., Drucker, D. J., Knepel, W., Jepeal, L., Misulovin, z., and

Drucker, D. J., Philippe, J., Jepeal, L., and Habener, J. F. (1987) J. Bioi.

Efrat. S.. Teitelman. G.. Anwar, M., Ruaeiero, D., and Hanahan, D. (1988)

2217-2222

Habener, J. F. (1988) Mol. Cell. Biol. 8,4877-4888

Chern. 262,15659-15665

Philippe, J., Drucker, D. J., Knepel, W., Jepeal, L., Misulovin, z., and

Drucker, D. J., Philippe, J., Jepeal, L., and Habener, J. F. (1987) J. Bioi.

Efrat. S.. Teitelman. G.. Anwar, M., Ruaeiero, D., and Hanahan, D. (1988)

2217-2222

Habener, J. F. (1988) Mol. Cell. Biol. 8,4877-4888

Chern. 262,15659-15665

Neuron 1,605-6i3 Heinrich, G., Gros, P., and Habener, J. F. (1984) J. Biol. Chern. 259 ,

14082-14087 Ho an, B., Constantini, F., and Lacy, E. (1986) Manipulating the Mouse

lfrnbryo: A Laboratory Manual, p 1633-1641, Cold Sprlng Harbor

Chomczynski, P., anZSaccb1, N. (1987) Anal. Chern. 162,156-159, Laboratory, Cold S ring Harbor, N!

Dong, J., Asa, S. L., and Drucker, D. J. (1991) Mol. Endocnnol. 6, In press Asa, S. L., Singer, W., Kovacs, K., Horvath, E., Murray, D., Colapinto, N.,

Hanahan D. (1985) Nature 315 , 115-122 Han, V. K. M., Hynes, M. A,, Jin, C., Towle, A. C., Lauder, J. M., and

"

and Thorner, M. 0. (1987) Acta Endocrinol. 115,331-337

Lund, P. K. (1986) J. Neuroscr. Res. 16, 97-107