Conditional Gene Targeting in Mouse Pancreatic β-Cells: Analysis of Ectopic Cre Transgene Expression in the Brain

Transgenic Cre and R26R reporter mouse lines used in this study are listed in Table 1. Experimental animals were generated by crossing Tg(Ins2-cre)^25Mgn (termed RIP-Cre^Mgn) (11), Tg(Ins2-cre)^1Herr (termed RIP-Cre^Herr) (10), Tg(Ins2-creEsr1)^1Dam (termed RIP-Cre/ERT) (12), Tg(Pdx1-cre)^89.1Dam (termed Pdx1-Cre^Dam) (13), Tg(Ipf1-cre)^1Tuv (termed Pdx1-Cre^Tuv) (14), Tg(Pdx1-cre/ERT)^1Mga (termed Pdx1^AI-III-Cre/ERT) (15), or Tg(Ins1-cre/ERT)^1Lphi (termed MIP-Cre/ERT) (Tamarina et al., unpublished data) transgenic lines with a reporter strain expressing either lacZ Gt(ROSA)26Sor^tm1Sor (termed R26R^wt/lacZ) (17,19) or enhanced YFP Gt(ROSA)26Sor^tm1(EYFP)Cos (termed R26R^wt/YFP) (18). Both R26R reporter strains on C57BL/6 background were obtained from Jackson Laboratories (Bar Harbor, ME). Pdx1^tm1Cvw mice (16) on B6D2 F1 background were obtained from Dr. C.V. Wright (Vanderbilt University). Complete details of the sources for all mouse strains used in this study are listed in supplementary Table 1 (available at http://diabetes.diabetesjournals.org/cgi/content/full/db10-0624/DC1). For timed pregnancies, noon on the day of the vaginal plug was considered embryonic day 0.5 (e0.5). All animal studies were approved by the Institutional Animal Care and Use Committees at the relevant institutions.

Primary antibodies included guinea pig anti-porcine insulin IgG (1:500; Dako, Carpinteria, CA), guinea pig anti-insulin antibody (1:1,000; Millipore, Billerica, MA), rabbit anti–β-gal IgG (1:5,000; MP Biomedicals, Solon, OH), goat anti–β-gal IgG (1:1,000; Biogenesis Ltd, Poole, UK), rabbit anti-STAT3 phosphorylation (pSTAT3) IgG (1:1,000; Cell Signaling Technologies, Beverly, MA), rabbit anti-orexin IgG (1:2,000; Calbiochem, EMD Biosciences/Merck, Darmstadt, Germany), and rabbit anti-Cre antibody (1:1,000, cat. #69050; EMD Biosciences, San Diego, CA). Fluorescent-labeled secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA) and Invitrogen (Carlsbad, CA). Recombinant mouse leptin was obtained from the National Hormone and Peptide Program (Los Angeles, CA).

Over a 5-day period, mice were injected subcutaneously or intraperitoneally with 3 doses of 1–8 mg tamoxifen (Sigma, T5648) freshly dissolved in corn oil (Sigma, C8267) at 10 mg/ml, 20 mg/ml, or corn oil vehicle. The subcutaneous injection site was sealed with a drop of Vetbond tissue adhesive (3M). Following tamoxifen administration, the mice were housed individually for 5–10 days before being analyzed for Cre-recombinase–mediated activity.

β-Gal activity was detected by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) staining as described previously (20) with slight modifications. Briefly, pancreata and brains were dissected in ice-cold 10 mmol/l PBS and fixed in freshly prepared 1–2% paraformaldehyde for either 2–4 h at room temperature or overnight at 4°C. Brains (2-mm slices) and pancreata were permeabilized for 5 h in 2 mmol/l MgCl₂, 0.01% sodium deoxycholate, 0.02% NP-40, 10 mmol/l PBS, and then stained overnight in the dark in 2 mmol/l MgCl₂, 5 mmol/l potassium ferricyanide, 5 mmol/l potassium ferrocyanide, 1 mg/ml X-gal, 0.01% sodium deoxycholate, 0.02% NP-40, 10 mmol/l PBS pH 7.4 at ambient temperature or 37°C. Tissues were washed in PBS, postfixed in 4% paraformaldehyde for 1 h, washed in PBS, and placed into 70% ethanol prior to whole mount imaging. For YFP detection, embryos were dissected at e15.5 and imaged in whole mount.

Mice were injected intraperitoneally with leptin (5 mg/kg) or vehicle (PBS) and then rested for 2 h prior to perfusion.

Immunodetection of pancreatic Cre expression was performed in 5-μm paraffin sections prepared from paraformaldehyde-fixed pancreata of RIP-Cre^Mgn, RIP-Cre/ERT, Pdx1^AI-III-Cre/ERT, and MIP-Cre/ERT mice. Transgenic lines expressing Cre/ERT received the third dose of tamoxifen on the day prior to being killed. After antigen retrieval, sections were incubated with primary antibodies to Cre and insulin (Millipore). For immunodetection of β-gal expression in the brain, mice were perfusion-fixed, and brains were removed and postfixed overnight as described previously (21). Following cryoprotection, brains were sectioned into 30-μm coronal slices, collected in four consecutive series, and stored at −20°C. Sections were then incubated with primary antibodies to pSTAT3, β-gal (Biogenesis), or orexin overnight at 4°C. Immunolabeling was visualized with appropriate fluorescent-labeled secondary antibodies. Digital images were acquired by confocal microscopy. One-way ANOVA analysis was used to compare the percent of β-cells that express Cre in the islets of the different transgenic lines.

Islets (22) and hypothalamus were isolated from adult Tg(Pdx1-cre)^89.1Dam (13) mice and their controls. Total cellular RNA was isolated using the RNAqueous Small Scale Phenol-Free Total RNA isolation kit (Ambion, Austin, TX), and trace contaminating DNA was removed with the TURBO DNA-free kit (Ambion). High-quality RNA had a 28S–to–18S ratio from 1.2 to 2.0 and an RNA integrity number from 8.2 to 8.9. Single-stranded cDNA was generated by reverse transcription from 180-ng total RNA using the Superscript III First Strand Synthesis kit (Invitrogen). cDNA (40 ng/reaction) was analyzed by quantitative RT-PCR using the ABI Prism 7900 Sequence Detection System and POWER SYBR Green Master Mix (Applied Biosystems, Foster City, CA). Samples were analyzed in duplicates, and relative cDNA levels were determined by comparing cycle threshold values of Cre cDNA to Hypoxanthine phosphoribosyl transferase (HPRT) cDNA. Samples with cycle threshold values greater than 40 were considered to have undetectable amounts of template. Primer sequences were the following; Cre (5′TGCAACGAGTGATGAGGTTC3′ and 5′GCAAACGGACAGAAGCATTT3′), HPRT (5′TACGAGGAGTCCTGTTGATGTTGC3′ and 5′GGGACGCAGCAACTGACATTTCTA3′), and Pdx1 (5′CTGAGGGACAAAGATGCAGA3′ and 5′TTCTAATTCAGGGCGTTGTG3′). One-way ANOVA with Newman-Keuls multiple comparison tests were used to compare outcomes in mice of different genotypes. Data were expressed as mean ± SE.

Using the R26R reporter line, the RIP-Cre^Mgn line (11) was previously shown to have robust Cre-mediated recombination within the β-cells and the ventral brain during development (9). To investigate whether Cre-mediated recombination occurred within the brain of other transgenic Cre lines using the rat Ins2 or Pdx1 promoter (Table 1), these mouse strains were crossed with the R26R reporter strain and analyzed for β-gal activity in whole mount brain slices (Figs. 1 and 2). No X-gal staining was detected in the brain or pancreas from control R26R^wt/lacZ littermates indicating that β-gal is not expressed in the absence of Cre activity (Fig. 1,B and F and supplementary Fig. 1). In RIP-Cre^Mgn;R26R^wt/lacZ mice, widespread X-gal staining was detected in most brain areas with robust expression in the mid-brain and ventral regions, which was consistent with previous reports (9) (Fig. 1,C and supplementary Fig. 2). In the brains of RIP-Cre^Herr;R26R^wt/lacZ mice, X-gal staining was less widespread and had a more punctate pattern without any obvious regionalization (Fig. 1,D and supplementary Fig. 3). The brains of RIP-Cre/ERT;R26R^wt/lacZ mice with Cre activity induced by three 2-mg doses of tamoxifen revealed a diffuse intermediate pattern of X-gal staining that was more extensive than in RIP-Cre^Herr;R26R^wt/lacZ mice but less than in RIP-Cre^Mgn;R26R^wt/lacZ mice (Fig. 1,E and supplementary Fig. 4). All three transgenic lines, RIP-Cre^Mgn, RIP-Cre^Herr, and RIP-Cre/ERT, showed a high level of recombination in pancreatic islets (Fig. 1 G–I and supplementary Figs. 2–4).

Cre-mediated recombination within the brain of Pdx1-Cre transgenic lines has not been examined, but ectopic recombination was reported in the pharyngeal region of Pdx1-Cre^Dam;R26R^wt/lacZ embryos (23). Unlike the widespread recombination in brains from RIP-Cre transgenic lines, X-gal staining in Pdx1-Cre^Dam;R26R^wt/lacZ brains (Fig. 2,B and supplementary Fig. 5) and Pdx1-Cre^Tuv;R26R^wt/lacZ brains (Fig. 2,C and supplementary Fig. 6) was localized primarily to the hypothalamus and brain stem. Analysis of Cre mRNA by quantitative RT-PCR in the Pdx1-Cre^Dam line confirmed expression in the hypothalamus with levels of hypothalamic expression 12.6-fold lower than in islets (supplemental Fig. 5). Pdx1^AI-III-Cre/ERT transgenic mice express the tamoxifen-inducible Cre (15). Injection of a single dose of tamoxifen into pregnant females (2 mg/40 g body weight) at e16.5 did not result in recombination in the brains of Pdx1^AI-III-Cre/ERT;R26R^wt/lacZ embryos dissected at e20.5 (15). In adult Pdx1^AI-III-Cre/ERT;R26R^wt/lacZ mice injected with three 1-mg doses of tamoxifen, recombination was detected mainly in the hypothalamus (supplementary Fig. 7, left panel). However, three 8-mg doses of tamoxifen induced much broader recombination throughout the brain, suggesting that the extent of recombination in the adult brain is dependent upon the tamoxifen dose (Fig. 2 D and supplementary Fig. 7, right panel). These data suggest that the Pdx1^AI-III-Cre/ERT transgene is expressed in the adult brain but not in the e16.5 brain, although it is possible that higher tamoxifen levels may be needed to induce Cre-mediated recombination within the embryonic brain.

To further examine the timing of Cre expression in the brains of the Pdx1-Cre lines expressing constitutively active Cre, we studied the Pdx1-Cre^Tuv transgenic line crossed into either R26R^lacZ/lacZ or R26R^YFP/YFP reporter mice and analyzed embryos at e15.5 (Fig. 2 J–K and supplementary Fig. 8). Both reporter strains demonstrated Cre activity in the brain stem and ventral region of the developing brain that gives rise to the hypothalamus, indicating that functional Cre protein is expressed in the ventral region of the Pdx1-Cre^Tuv brain prior to e15.5.

To determine whether Cre activity in the hypothalamus of the three different Pdx1-Cre transgenes was due to previously unrecognized endogenous Pdx1 expression, a mouse line with a lacZ reporter cassette in the Pdx1 locus was examined (16). Both adult and embryonic (e15.5) Pdx1^wt/lacZ (Fig. 2 E and L and supplementary Fig. 9) brains were negative for X-gal staining. Furthermore, expression of the endogenous Pdx1 gene was undetectable in the hypothalamus by real-time RT-PCR (data not shown) indicating that Pdx1-Cre^Dam, Pdx1-Cre^Tuv, and Pdx1^AI-III-Cre/ERT transgenes are ectopically expressed in the brain.

Detection of Cre-mediated recombination in the hypothalamus of Pdx1-Cre^Dam;R26R^wt/lacZ mice (Fig. 2,B, supplementary Fig. 5, and supplementary Fig. 10) raised the possibility that Cre protein may be expressed in neurons involved in the regulation of energy and glucose homeostasis. To determine the extent of Cre-mediated recombination within these specific neuronal populations, β-gal positive cells in brain sections from leptin-treated Pdx1-Cre^Dam;R26R^wt/lacZ mice were co-localized with orexin and leptin-induced pSTAT3, respectively. In the lateral hypothalamus, β-gal protein was expressed in a complex pattern that partially overlapped with both the orexin-expressing and LepRb-expressing neuronal populations (Fig. 3), although significant populations of β-gal positive cells did not overlap with the neuronal cell population in either the lateral hypothalamus or in other hypothalamic regions including the arcuate nucleus. Nonetheless, these data clearly illustrate that the Pdx1-Cre^Dam line induces Cre-mediated recombination in subpopulations of hypothalamic neurons involved in energy expenditure and glucose metabolism.

A new transgenic line, MIP-Cre/ERT, which employs an 8.5-kb fragment of the mouse Ins1 promoter has been recently developed to express the tamoxifen-inducible Cre/ERT in β-cells (Tamarina et al., unpublished data) (Table 1). Following three doses of 2-mg tamoxifen, strong β-gal activity was detected in the pancreatic islets of adult MIP-Cre/ERT;R26R^wt/lacZ mice but not in their R26R^wt/lacZ littermates (Fig. 4). By contrast, no β-gal activity was detected in any region of the brain from MIP-Cre/ERT;R26R^wt/lacZ mice (Fig. 4). Furthermore, when the tamoxifen dose was increased to three doses of 8-mg, β-gal activity was not detected within the brain (data not shown). Cre expression efficiency in the β-cells, as determined by immunohistochemistry, was similar in MIP-Cre/ERT (89.0 ± 8.0%), RIP-Cre (85.4 ± 5.5%), RIP-Cre/ERT (88.9 ± 5.8%), and Pdx1-Cre/ERT (92.1 ± 6.6%) mice (supplemental Fig. 11).

The studies examining the in vivo role of genes associated with biological processes in the pancreas and β-cells have relied largely upon fragments of the rat Ins2 gene promoter or the Pdx1 gene promoter (3,8,,,,–13,16). Although insulin secretion from β-cells plays an important role in glucose control, many other tissues including the brain are intimately involved in the regulation of glucose metabolism. Previous studies have demonstrated that the 668 bp rat Ins2 promoter fragment drives Cre-recombinase expression within the central nervous system of the mouse transgenic line, RIP-Cre^Mgn [Tg(Ins2-cre)^25Mgn] (9,24,25). In this study, we examined whether Cre-mediated recombination occurred in the brain of six mouse transgenic lines that have been extensively used to express Cre specifically within the pancreas or islet β-cells (Table 1). Analysis of Cre-mediated recombination using the R26R reporter strain demonstrated that all six transgenic lines expressed Cre recombinase to varying extents within the brain, raising the possibility that alterations of gene expression in the brain may complicate the analysis and that the observed phenotype may not be solely due to changes in the β-cell. This possibility was highlighted by a recent study that used the RIP-Cre^Mgn line to selectively delete the Stat3 gene in the β-cells (24). STAT3-deficient mice displayed increased food intake, obesity, and leptin resistance; physiological effects that the authors attributed to STAT3 deficiency in the brain leading to impaired leptin signaling.

There are no previous studies reporting Cre-mediated recombination in the brain with Pdx1-Cre lines. Our findings indicate that the Pdx1-Cre^Dam line causes recombination in a subset of hypothalamic neurons involved in energy and nutrient homeostasis. The similarity of β-gal–expression patterns in the hypothalamus with the other two Pdx1-Cre transgenic lines suggest that Cre-mediated recombination in these lines may also affect similar neuronal subpopulations. The lack of β-gal activity in the brains of Pdx1^wt/lacZ mice indicates that the Cre expression in the brain of the Pdx1-Cre^Dam, Pdx1-Cre^Tuv, and Pdx1^AI-III-Cre/ERT mice is not a reflection of endogenous Pdx1 gene expression. A likely explanation for this spurious expression is the removal of these Pdx1 gene promoter fragments from their endogenous gene context. Furthermore, the similar recombination pattern generated with Pdx1-Cre lines makes it unlikely that this brain expression is a result of neighboring sequences at the sites of integration, which are almost certainly different for each line. While a lacZ knock-in reporter to analyze endogenous Ins2 expression is currently not available, a recent study demonstrated expression of the Ins2 gene but not the Ins1 gene in the mouse hypothalamus (26). Thus, in contrast to the infidelity of Cre transgene expression found with the Pdx1 promoter, Cre expression observed using the Ins2 promoter may reflect, in part, endogenous promoter activity. It is not known whether the ectopic expression of the insulin or Pdx-1 transgenes occurs during the embryonic period, adult periods, or in both periods.

Several caveats should be considered in interpreting our results. First, we did not examine all currently available insulin, PDX-1, or other gene promoters used to direct Cre expression to the pancreas or β-cell. Thus, it is essential that investigators examine brain expression in any Cre line thought to be pancreas- or β-cell–specific. Second, our results should not be interpreted to indicate specific expression (or lack of expression) in any brain region or nuclei as we did not perform detailed mapping of brain regions following lacZ staining. We did note regions with strong X-gal staining, but this should not be taken as evidence that other areas do not express Cre, and it is possible that a more isolated or diffuse expression in other brain regions may also lead to Cre activity. More detailed work is needed to identify which brain regions are positive or negative for Cre activity. Third, whether Cre expression leads to excision of a floxed DNA fragment is an incompletely understood process that depends on both Cre expression and the floxed allele. We mostly used a single reporter line, and we do not know if the results would differ with other lines that express other reporters such as alkaline phosphatase. In fact, we predict that some reporter lines will not show the same Cre activity we observed given that a range of sensitivity of floxed alleles to Cre-mediated recombination is likely (with the R26R^wt/lacZ line being more Cre-sensitive) (27). This possibility further complicates interpretation of studies using Cre to inactivate a gene of interest. Thus, we urge caution in extrapolating that the lack of Cre-mediated recombination with a certain reporter gene predicts a lack of Cre-mediated recombination of a gene of interest in the brain. Based on the current study, it is clear that the R26R^wt/lacZ floxed allele is susceptible to Cre-mediated recombination in several brain regions, including orexin-positive and leptin-responsive neuronal populations. Finally, the experimental intent for using Cre transgene must be considered. If lineage tracing is the goal, is it preferable to use a “sensitive” or “insensitive” reporter? If gene inactivation is the goal of Cre-mediated recombination, then whether other tissues or cells endogenously express the gene of interest becomes a critical factor. If the gene of interest is expressed in places other than the pancreas or β-cell (especially in the brain where a large number of genes are known to be expressed in both tissues), then the current finding of Cre-mediated recombination in brain regions involved in glucose homeostasis, appetite, weight, and energy expenditure make attributions of the phenotype to the β-cell more difficult.

The above analysis clearly illustrates that new transgenic lines are needed to ensure fidelity of conditional Cre expression in islet β-cells. In transgenic Tg(Ins1-EGFP)^1Hara mice (28), an 8.5-kb fragment of the mouse Ins1 gene promoter was successfully used to express enhanced green fluorescent protein (eGFP) specifically within islet β-cells in the absence of eGFP expression in other tissues. This same Ins1 promoter fragment was used in the MIP-Cre/ERT transgenic line to direct Cre/ERT gene expression in β-cells (Tamarina et al., unpublished data). The improved fidelity of Cre expression observed in the MIP-Cre/ERT line is likely due, in part, to the additional regulatory elements within the larger promoter fragment employed and because the mouse Ins1 gene is not expressed in the hypothalamus (26). Thus, the MIP-Cre/ERT mice appear to represent a transgenic line to express Cre efficiently and specifically in islet β-cells.

In conclusion, this study reveals that the current transgenic lines utilizing Ins2 and Pdx1 promoter fragments target Cre expression not only to the islet β-cells, but also to the brain. While not invalidating the use of these lines, our data indicate that studies conducted using these Cre transgenic mice should be interpreted carefully to assess whether manipulation of the target gene within the brain could contribute to the observed phenotype. The lack of Cre-mediated recombination in the brain of MIP-Cre/ERT;R26R^wt/lacZ mice suggests that the newly developed MIP-Cre/ERT line is currently the only available β-cell–specific Cre line. As with all newly developed Cre transgenic lines, caution must also be exhibited when using this line until its potential as a β-cell–specific Cre line has been validated through further experimental analysis.

This work was supported by the following grants: National Institutes of Health (NIH) DK63363, Juvenile Diabetes Research Foundation (JRDF) 1-2008-473, American Diabetes Association (ADA) 7-09-BS-41, University of Michigan Diabetes Research and Training Center Pilot & Feasibility (DRTC P&F) P60 DK020572, and the D.R.E.A.M. Foundation (to P.J.D.); ADA Junior Faculty Award and University of Chicago DRTC P&F grant P60 DK20595 (to B.W.); NIH DK057768 and American Heart Association Established Investigator Award (to M.G.M.); NIH DK074966 (to M.W.R.); NIH DK065131, DK071052, and JRDF 1-2007-548 (to M.G.); NIH HD36720 and JDRF 1-2006-219 and Vanderbilt DRTC P&F grant P60 DK020593 (to P.A.L.); NIH T32DK07563 (to J.P.); NIH DK073716 and JDRF 1-2008-147 (to E.B.); NIH DK69603, DK66636, DK072473, a Merit Review Award from the U.S. Department of Veterans Affairs Research Service, and JDRF 1-2008-512 (to A.C.P.). Core facilities used in this study were supported by the National Institute of Diabetes and Digestive and Kidney Diseases DRTCs from the University of Chicago (P60 DK020595), University of Michigan (P60 DK020572), Vanderbilt University (P60 DK020593), and the Vanderbilt Mouse Metabolic Phenotyping Center (DK59637). The schematics of the mouse brain shown in the figures are from the Allen Brain Atlas Resources (http://www.brain-map.org/) (29).

No potential conflicts of interest relevant to this article were reported.

B.W., M.B., A.C.P., and P.J.D. conceived the study. B.W., M.B., P.A.L., M.G., M.G.M., A.C.P., and P.J.D. designed the experimental approach. B.W., M.B., W.Y., D.M.O., J.L.P., R.B.R., L.M.D., A.S., and L.E. researched data. E.B.-M., M.G.M., and M.G. contributed to the analysis and interpretation of data. N.A.T., L.H.P., and M.W.R. provided novel reagents. B.W., M.B., M.G., A.C.P. and P.J.D. wrote the manuscript.