Inducible and Reversible β-Cell Autoimmunity and Hyperplasia in Transgenic Mice Expressing a Conditional Oncogene

Mice expressing Tag conditionally in β-cells were generated by crossing heterozygous tet-Tag transgenic mice (10) with heterozygous RIP7-rtTA transgenic mice (11). The tet-Tag mice harbor a Tag gene flanked by multiple repeats of tet operator sequences and a minimal promoter (12). They were generated in the C3HeB/FeJ (C3H) strain. In the RIP7-rtTA mice, the reverse tet transactivator (13) was placed under the control of rat insulin II gene regulatory sequences, consisting of 9.5-kb 5′ flanking region and the first intron of the insulin gene (8). This transgenic line was established in the C57BL/6 (B6) × CBA F₂ background and subsequently backcrossed to B6 for at least six generations before being used for the crosses described here. Transmission of both transgenes was monitored by tail DNA polymerase chain reaction analysis, using the sense primer 5′-GACCAGCTACAGTCGGAAACC-3′ and antisense primer 5′-TGCAGTGAGCCAAGATTGTGC-3′ for RIP7-rtTA and the sense primer 5′-GGAATAGTCACCATGAATGAGTACAG-3′ and antisense primer 5′-CATGAACAGACTGTGAGGACTGAG-3′s for tet-Tag. To activate Tag expression, mice in mating cages and double-transgenic mice were treated with 2 mg/ml dox (Sigma) (in 2.5% sucrose) in their drinking water. Untreated double-transgenic mice served as controls. Blood glucose levels were determined in samples obtained from the tail vein using Accutrend strips (Roche).

After the animals were killed, the pancreas was removed, fixed in 4% paraformaldehyde, dehydrated, embedded in paraffin, and sectioned. Sections were stained with hematoxylin and eosin (HE) by standard procedures. For each mouse analyzed, all of the islets in a total of 12 sections (two slides each from the dorsal and ventral pancreatic lobes, each slide with three sections) were examined.

For Tag staining, rehydrated sections were incubated for 20 min in 5% H₂O₂ in methanol, followed by boiling in Antigen Retrieval Citra solution (Biogenex, San Ramon, CA). Cells were permeabilized in 0.25% Nonidet P40 in phosphate-buffered saline (PBS) for 15 min, followed by 30 min blocking in 5% fetal bovine serum (FBS) and 5% BSA in PBS. Tag was detected by an overnight incubation with a mouse monoclonal antibody (Pab101; Santa Cruz Biotechnology) at a dilution of 1:100, followed by biotinated rabbit–anti-mouse and horseradish-peroxidase (HRP)–conjugated streptavidin (both from Vector, Burlingame, CA). For insulin staining, rehydrated sections were blocked as described above and incubated with a guinea pig–anti-insulin serum (1:100; Linco, St. Charles, MO), followed by HRP-conjugated rabbit–anti–guinea pig serum (1:500).

Mice were injected intraperitoneally with 100 μg BrdU/g body wt. The pancreas was removed 6 h later, fixed, and sectioned as detailed above. Tissue sections were immunostained with a BrdU mouse monoclonal antibody as previously described (10).

Apoptotic cells in pancreas sections were detected using the Tumor TACS In situ Apoptosis Detection Kit from R&D Systems (Minneapolis, MN), which uses the terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) approach, followed by counterstaining with 1% methyl green.

To obtain inducible Tag expression in transgenic mice, a mouse line harboring the Tag gene under control of tet operator sequences (tet-Tag [10]) was crossed with a second transgenic line expressing the reverse tet transactivator specifically in pancreatic β-cells under control of an insulin gene regulatory region (RIP7-rtTA [11]). The single transgenic tet-Tag mice, as well as double-transgenic mice not treated with dox, did not express Tag in the islets and had a normal phenotype, as judged by the lack of histological abnormalities in the pancreas by 4 months of age (data not shown). To induce Tag expression, single transgenic mice carrying the two genes were mated, and the mating cages received dox in their drinking water to allow the generation of Tag in β-cells as soon as the RIP is normally activated during pancreas development, around E9.5 (14). Tetracyclines have been shown to cross the placenta as well as accumulate in the milk of lactating females treated with the drugs (15,16). Thus, this treatment allowed continuous exposure of fetuses and newborn mice to dox. Tag expression was detected in cell nuclei in islets of newborn double-transgenic mice (Fig. 1). Mice continuously treated with dox after weaning developed hyperplasia in the majority of the islets by 2 months of age, followed by progression to insulinomas by 3 months of age. The use of the RIP7 promoter, which contains 9.5 kb of the regulatory region of the rat insulin II gene (8), may generate higher levels of Tag expression, compared with the 0.7-kb RIP1 promoter more commonly used (1), thereby resulting in a somewhat faster tumorigenesis process compared with that of RIP-Tag mice. Mice carrying large tumors developed hypoglycemia. Tumors from these mice were used to derive multiple β-cell lines (11).

Expression of Tag depended on continuous treatment with dox. After removal of dox from the drinking water, Tag expression was shut off within 1 week (Fig. 2C). As a result, β-cell proliferation was greatly reduced, as judged by analysis of BrdU incorporation (Fig. 2D). This effect was observed in hyperplastic islets of mice, in which dox treatment was terminated between 4 and 8 weeks of age.

When onset of Tag expression was delayed by initiating the treatment of double-transgenic mice with dox at 3 weeks of age, a complex phenotype developed, involving both β-cell proliferation and insulitis. Within 5 weeks of dox treatment, ∼50% of the islets manifested pronounced hyperplasia (Figs. 3A and 5A). In addition, the majority of the islets showed significant mononuclear cell infiltration, which histologically resembled the insulitis observed in autoimmune diabetes (Figs. 3A, C, and D, 4, and 5A). After Tag shut-off at 8 weeks of age, the infiltrating cells gradually disappeared from the islets. Infiltrates were still present 2 weeks after termination of dox treatment; however, by 4–6 weeks they were cleared from the vast majority of the islets (Figs. 3E and F, 4, and 5C and E).

In addition to the reversibility of insulitis, islet hyperplasia was reversible in all of the islets in 7 of 18 (39%) mice analyzed at 4–8 weeks after dox removal (2 of 4 at 4 weeks, 3 of 10 at 6 weeks, and 2 of 4 at 8 weeks). In these mice, termination of dox treatment resulted in not only a decrease in islet growth but also a reduction in islet size and the restoration of pancreas phenotype back to normal (Figs. 3E and F and 5E). This process involved massive apoptosis of β-cells. As seen in Figs. 5B and 6, hyperplastic islets expressing Tag contained a large number of apoptotic cells, confirming previous findings in RIP-Tag mice, which demonstrated that p53-independent β-cell apoptosis accompanied β-cell transformation (8). Islet cell apoptosis was observed in all dox-treated double-transgenic mice analyzed at 2–3 months of age. Apoptosis persisted in the hyperplastic islets after Tag shutoff, as evidenced by staining of a significant percentage of the nuclei by the TUNEL assay up to 6 weeks after the termination of treatment with dox (Figs. 5D and 6). The apoptotic cells disappeared in individual mice 4–8 weeks after dox removal, concomitantly with the restoration of normal islet size. In contrast, in 10 of 18 (55%) mice analyzed 4–8 weeks after dox removal, the hyperplasia persisted in >30% of the islets, whereas in 1 of 18 (16%) mice, the islets remained mildly hyperplastic. The hyperplastic islets, which persisted 8 weeks after dox removal, did not stain for Tag and did not contain proliferating cells, as judged by analysis of BrdU incorporation (Figs. 7B and C). In addition, these islets did not contain apoptotic cells, as detected by TUNEL analysis (Fig. 7D).

Our results demonstrate that β-cell proliferation during the tumorigenesis process in vivo maintains a dependence on Tag expression. Elimination of Tag by dox removal between 4 and 8 weeks of age results in a great reduction in β-cell proliferation within 1 week (Fig. 2D). These findings indicate that genetic changes that may occur in the transformed cells do not allow Tag-independent proliferation, thereby confirming the findings obtained in the conditionally transformed β-cell lines in tissue culture. In a similar tet-dependent model of Tag expression in salivary gland cells, a tumorigenesis stage has been identified, beyond which the process can progress in the absence of Tag (17). This possibility cannot be excluded in the transgenic mice conditionally expressing Tag in β-cells because no attempt was made to reverse the process at stages later than 8 weeks of age. This is because, unlike the mice with tumors in the salivary gland, the mice carrying insulinomas become hyperinsulinemic and hypoglycemic and die at ∼12 weeks of age. By this time, the pancreas architecture is severely disrupted by the tumors (Fig. 3C). However, the continued dependence of β-cell proliferation on Tag expression in the cell lines derived from the end-stage tumors (10,11) indicates that, unlike the tumorigenesis process in salivary gland cells, the factors activated during the β-cell tumorigenesis process (6,7,8,9) are not sufficient for promoting β-cell proliferation in the absence of the Tag oncoprotein.

The process of Tag-induced islet hyperplasia has been shown by Naik et al. (8) to be accompanied by β-cell apoptosis in RIP-Tag mice, peaking in angiogenic islets at ∼3% of the cells. However, β-cells generated by proliferation outnumber the apoptotic cells, and once solid vascularized tumors develop, apoptosis disappears. In the double-transgenic RIP7-rtTA × tet-Tag mice expressing Tag, a much higher apoptotic index of up to 33% of the islet cells was observed (Figs. 5B and 6). Residual apoptosis of 1% persisted in the hyperplastic islets up to 6 weeks after shut-off of Tag expression (Figs. 5D and 6). In the mice treated with dox between 3 and 8 weeks of age, two distinct groups could be identified. In close to 40% of the mice analyzed, apoptosis led to a reduction in islet size and to the restoration of normal pancreas phenotype within 4–8 weeks after dox removal (Fig. 3E). In the rest of the mice analyzed at these time points, apoptosis subsided before the hyperplastic islet mass was reduced (Fig. 7). These findings suggest the activation of a homeostatic mechanism that operates to regulate the β-cell mass and restore it to the normal range. The efficiency of this mechanism appears to vary between individual mice. It is possible that this mechanism is activated by hypoglycemia or hyperinsulinemia, and variations between individual mice in the amounts of insulin released from hyperplastic islets may determine the extent of the counteracting apoptotic process. Such a mechanism is of great interest, given the role of apoptosis in normal islet remodeling in, for example, newborn mice (18,19) and postpartum females (20).

The delayed onset of Tag expression elicits an autoimmune response against β-cells, confirming the results obtained by Hanahan and colleagues (3,4) in RIP-Tag mice. The findings in the RIP7-rtTA × tet-Tag mice demonstrate that this process is fully reversible within a few weeks of Tag elimination, suggesting that sustained insulitis depends on continuous expression of the target antigen. In addition, these findings suggest that possible β-cell damage resulting from the anti-Tag response does not trigger immune responses to other β-cell antigens, which may be exposed as a result of this damage, since such responses would be expected to persist after Tag expression is shut off.

This work was supported by a grant from the Juvenile Diabetes Research Foundation (to S.E.). We thank Gerhard Christofori for RIP7-rtTA mice and Hasida Orenstien for tissue sectioning.