A dual color, genetically engineered mouse model for multi-spectral imaging of the pancreatic microenvironment - PMC

Abstract

INTRODUCTION

The use of fluorescent proteins for imaging of biological processes has yielded important information in many fields of research. The development of genetically engineered mice that express fluorescent proteins either constitutively or in a tissue-specific manner has been an important advance, allowing for the imaging of in vivo processes that range from normal vertebrate development to human cancer biology.^1–4

Conditional gene targeting using site-specific recombinase systems (Cre/loxP, Flp/FRT) is a powerful genetic technology being applied to the study of gene function.⁵ Primarily used for tissue-specific loss of function (gene knockout) studies in mice, conditional gene targeting technology has also been used to control the temporospatial expression of fluorescent protein-encoding reporter transgenes.^6,7 These Cre reporter mouse models are useful for characterizing and validating newly-generated Cre recombinase-expressing transgenic lines prior to beginning tissue-specific gene knockout studies. As fluorescence imaging technologies advance, allowing for improved multi-spectral imaging of biological processes, applications for the simultaneous use of multiple fluorophores in in vivo studies expand.^8,9

Recently, a conditional fluorescent reporter mouse model (mT/mG) was developed in which transgene expression of tdTomato (a red fluorescent protein) converts to the expression of GFP (enhanced green fluorescent protein) following Cre recombinase-mediated intramolecular rearrangement of the fluorescent protein-encoding transgene.¹⁰ We wished to explore and validate the use of mT/mG mice for fluorescence imaging studies of pancreas biology. Pdx1-Cre transgenic mice have been used extensively to induce genetic recombination within the murine pancreas. PDX-1, also known as IPF-1, IDX-1, and STF-1, was originally identified as a transcription factor that regulates expression of insulin and other hormones produced by islet cells of the adult pancreas.^11–13 PDX-1 is also a homeodomain protein expressed specifically during early stages of mouse embryonic development in tissues destined to become the pancreas, antral stomach, and rostral duodenum. PDX-1 is sometimes referred to as a master regulator of vertebrate pancreas development.¹⁴ PDX-1-expressing cells of embryonic mice give rise to all three pancreatic epithelial cell types: exocrine acinar cells, endocrine islet cells, and duct cells.¹⁵ In adult mice PDX-1 continues to be expressed at high levels by insulin-producing, beta cells of the pancreatic islets.¹¹ Pdx1-Cre mice express Cre recombinase within the developing foregut-derived pancreas and duodenum during embryogenesis,^15,16 recapitulating the expression pattern of the endogenous pdx1 gene.

To generate mice in which the pancreatic parenchyma, expressing membrane-targeted GFP, is highlighted against the membrane-targeted red fluorescent backlight of stromal and non-epithelial-derived pancreatic tissues, we crossed Pdx1-Cre transgenic mice with mT/mG Cre reporter mice. To assess the suitability of this mouse model for imaging studies of the pancreas, we imaged offspring of Pdx1-Cre and mT/mG intercrosses at both the macroscopic and microscopic levels using bright light and fluorescence imaging modalities.

MATERIALS AND METHODS

RESULTS

Crosses of Pdx1-Cre mice with mTmG mice yielded litters with the expected ratios of genotypes. Mice obtained from these crosses fell into three phenotypic groups: mTmG-negative (either Pdx1-Cre -positive or Pdx1-Cre -negative, but non-fluorescent and therefore termed +/+ for imaging purposes); mTmG-positive, Pdx1-Cre -negative (mTmG); and mTmG-positive, Pdx1-Cre -positive (Pdx1-Cre/mTmG) double transgenic mice.

Pdx1-Cre/mTmG mice showed no apparent defects, bred normally, and had normal life expectancies (>1 year). For imaging studies, same-sex adult littermates from each of the three phenotypic groups were imaged simultaneously. Non-fluorescent littermates were imaged alongside their fluorescent (mTmG-positive) siblings to establish optimal exposure times and to control for auto-fluorescence, particularly of the gastrointestinal tract. Three 6-week old female littermates, imaged using an Olympus OV100 small animal imaging system (Figure 1A) and a CRi Maestro^™ EX in vivo imaging system (Figure 1B) and either brightfield illumination or fluorescence (tdTomato or GFP) illumination settings, demonstrated bright, generalized expression of tdTomato and tissue-specific expression of GFP. Figure 1C shows genotyping data for the mice in Figure 1. In mTmG-positive mice, all tissues examined expressed the mTmG transgene and demonstrated bright tdTomato fluorescence in the absence of Cre expression (Figures 1 and 2 and Supplemental Figure 1). As expected, the Pdx1-Cre transgene showed tissue-specific expression. Imaging of all major organs from Pdx1-Cre/mTmG, mice demonstrated that expression of GFP was limited to the pancreas and the proximal small intestine (Figures 1 and 2, Pdx1-Cre/mTmG genotype). GFP expression in the proximal small intestine (duodenum) tapered off distally (Figure 2B). Images obtained from Pdx1-Cre/mTmG mice ranging in age from early post-natal life to late adulthood have established that the fluorescent phenotypes described above are established during embryogenesis and are stably maintained throughout murine adult life (data not shown).

The conversion of red (mT) to green (mG) fluorescence is extensive but not complete in Pdx1-Cre/mTmG pancreata, resulting in a mosaic pattern of fluorescence within the pancreatic parenchyma that varied from individual to individual. Pancreatic mosaicism of mT and mG expression is best demonstrated macroscopically in Figure 2C (Littermate 2 versus Littermate 1, both PdxCre⁺, mTmG/+) and microscopically in Figure 3B. Figure 2C Littermate 2 was the most extreme example of genetic mosaicism seen in our studies. Most of the Pdx1-Cre/mTmG mice we examined more closely resembled the Pdx1-Cre/mTmG mouse shown in Figure 1 and Littermate 1 in Figure 2C, both of which demonstrate extensive pancreatic expression of mG. Confocal imaging of mTmG-positive pancreata demonstrated bright membrane staining and clear delineation of cell borders, especially in acinar cells (Figure 3). Islet cells of Pdx1-Cre/mTmG mice also demonstrated GFP expression. The intensity of membrane staining in islet cells was significantly less than in the surrounding acinar cells (Figure 3A). Small pancreatic ducts and individual pancreatic ductal cells were difficult to discriminate from the brightly fluorescent acinar cells in these unstained cryosections.

The vascular scaffolding, the most prominent component of the stromal matrix in normal pancreas, demonstrated bright red fluorescence in Pdx1-Cre/mTmG pancreata (Figure 3A, tdTomato panel). Although incomplete on a tissue basis, we found that conversion of mT to mG fluorescence appeared complete within individual pancreatic cells of Pdx1-Cre/mTmG mice. Co-expression of red and green fluorescence, which would appear as yellow fluorescence in images, was not seen within individual cells (Figure 3B).

To assess for unanticipated toxic effects of fluorescent protein (mT or mG) expression on the pancreas, we performed histological studies on pancreata from fluorescent (Pdx1-Cre/mTmG and mTmG) experimental mice and from non-fluorescent (mTmG-negative) control mice. The histological appearance of exocrine (acini and ducts) and endocrine (islets) elements was unremarkable in H&E-stained sections taken from Pdx1-Cre/mTmG (Figure 4) and mTmG (data not shown) mice and was comparable to the H&E findings seen in non-fluorescent, control pancreata.

DISCUSSION

Applications for the in vivo use of expressed fluorescent reporters to address biological questions are expanding rapidly, mirroring ongoing technological advancements in imaging and an ever increasing number and diversity of genetically-modified organisms available for use in these studies.^19,20 Fluorescent reporter mice designed for use in Cre/loxP conditional gene targeting studies are an example.^21–23 In the present study we used mTmG (mT/mG) fluorescent reporter mice and Pdx1-Cre transgenic mice to generate a dual-color mouse model suitable for use in studies of pancreas biology using fluorescence imaging technology.

There are now many Cre recombinase-expressing lines available for studies of pancreas biology. These include lines wherein Cre recombinase is expressed under control of the elastase (Ela), rat insulin II (RIP), acinar procarboxypeptidase (proCPA1), or pancreas-specific transcription factor 1a (Ptf1a, p48) promoters,^24–27 Each of these lines yields unique and characteristic spatial patterns of Cre expression in vivo and differing degrees of inter-individual variability of expression.²⁸ Some of these lines, including the Pdx1-CreER variant of the Pdx1-Cre line,¹⁵ have been engineered to allow for temporal as well as spatial control of Cre expression. Pdx1-Cre mice are arguably the best characterized of these lines, yielding widespread, early, and comparatively consistent, Cre-expression throughout the pancreatic epithelial compartment. For these reasons, we chose to work with Pdx1-Cre mice in developing this model. For certain studies or to address specific biological questions, the choice of a line with more spatially or temporally restrictive Cre expression may be more appropriate or may provide complementary information.²⁸

We found that Pdx1-Cre/mTmG mice demonstrate intense, generalized, membrane-targeted tdTomato expression with bright, membrane-targeted GFP expression localized to the pancreas and duodenum. In mTmG mice lacking Cre recombinase, expression of membrane-targeted tdTomato (mT) fluorescent protein is particularly bright in the pancreas; this finding does not appear to be due to membrane-targeting of the mTmG transgene-encoded fluorescent protein product, as previous studies have also shown robust pancreatic expression of fluorescent proteins that were not membrane-targeted.²⁹

The finding of mosaic GFP expression in Pdx1-Cre/mTmG pancreata is interesting and is consistent with previous studies demonstrating mosaic pancreatic expression of Cre recombinase under pdx1 promoter control.^15,30 Mosaic expression of pdx1 promoter-driven constructs is not unique to Pdx1-Cre transgenics. Expression constructs engineered using this same 4.3 kilobase pdx1 promoter cassette to drive expression of proteins other than Cre recombinase have also yielded mosaic patterns of pancreatic expression.^31–33 Studies looking at endogenous PDX-1 expression during murine pancreatic morphogenesis have found that PDX-1 expression levels rise and fall at slightly different times in each of the embryonic pancreatic epithelial components: acini, ducts, and islets.³⁴ Owing to the variable timecourse of endogenous PDX-1 protein expression in different epithelial elements of the pancreas, snapshot studies of endogenous PDX-1 expression do not yield an answer as to whether or not the normal murine pancreas harbors isolated parenchymal cells that escaped a PDX-1-expressing heritage (endogenous mosaics). However, pdx1 knockouts form only rudimentary pancreata in the absence of PDX-1 expression.¹⁶ Expression of genomic fragments of the pdx1 locus encompassing this approximately 4.3 kilobase pdx1 promoter cassette have been shown to rescue the lethal, apancreatic phenotype seen in pdx1^−/− knockout mice, suggesting that this promoter construct is sufficient to recapitulate endogenous levels of pancreatic PDX-1 expression in vivo.^32,35 Studies have shown significant variability in PDX-1 expression levels between adjacent cells of developing epithelial cell cords and acinar cell clusters in the embryonic murine pancreas.³⁶ It seems likely that the mosaicism of mT and mG expression seen in adult Pdx1-Cre/mTmG pancreata is the manifestation of a threshold effect of Pdx1-Cre expression on mTmG transgene rearrangement during embryonic pancreatic organogenesis. As such, the conversion from mT to mG expression is a digital readout of an analog system of Pdx1-Cre expression. Genetic variables such as strain background may influence expression of Pdx1-Cre or the process of Cre-mediated mT to mG conversion. However, the mosaicism demonstrated in Figure 2C occurred in Pdx1-Cre/mTmG littermates born to breeders that had been backcrossed onto C57Bl/6 for several generations. This finding argues that the genetic process of Pdx1-Cre-mediated conversion from mT to mG fluorescence, while largely deterministic at the whole organ level, is also influenced by stochastic variables acting on individual cells during pancreatic organogenesis.

We also observed mosaic expression of GFP in the small intestine of Pdx1-Cre/mTmG mice, although this expression pattern is probably more accurately characterized as a dilutional tapering off of GFP-expressing cell clusters as one moves distally from the duodenum, which shares its ontogeny with the pancreas, towards the more caudal regions of the small intestine that arise from embryonic tissues that fall outside the pdx1 expression domain. Whereas, PDX-1 expression is required for development of the pancreas, PDX-1 expression is dispensable for morphogenesis of the duodenum.³⁷ Although the duodena of pdx1 knockout mice are structurally normal at the macroscopic level, pdx1 knockout mice have been reported to show functional deficits that correlate with cytologic abnormalities at the microscopic level, including morphological changes in the epithelium of the duodenal villi, absence of Brunner’s glands, and decreased numbers of duodenal enteroendocrine cells.^16,35 Subsequent studies looking at roles for PDX-1 in adult duodenum have documented mosaic expression of both upstream regulators³⁸ and putative downstream target genes³⁹ of pdx1/PDX-1 in the duodenum. This mosaicism has sometimes been a confounding variable in the interpretation of gene knockout studies of duodenal development and function. Future studies of duodenal biology involving conditional gene targeting may benefit from the multi-spectral lineage tracking enabled by use of mTmG reporter mice in combination with duodenal Cre-expression mice such as the Pdx1-Cre mice used in the present study.

Despite the mosaicism that occurs in the pancreas as a whole, conversion of red (mT) to green (mG) fluorescence appears to be complete within individual pancreatic cells of Pdx1-Cre/mTmG mice. Co-expression of mT and mG fluorescence, which would result in the appearance of a yellow fluorescence signal in superimposed images of red and green fluorescence emission, was not seen in individual cells of the adult pancreas. Muzumdar et al¹⁰ documented an mT to mG transitional state in hepatocytes of tamoxifen-treated, mTmG mice crossed with actin-CreER mice, as well as in a variety of tissues when mTmG reporter mice were crossed with other Cre-expressing transgenic lines; the authors did not report findings specific to the pancreas. The absence of tdTomato perdurance within individual, Cre-recombined, GFP-expressing pancreatic cells in Pdx1-Cre/mTmG mice is testimony to the remote timing of initial Pdx1-Cre expression during ontogeny of the pancreas, allowing time for washout of mT fluorescent protein and replacement by newly-expressed mG fluorescent protein. PDX-1 expression by developing pancreatic acinar and duct cells decreases at mid-gestation, and it is expressed at low levels in adult pancreatic acini and ducts. By contrast, adult pancreatic islet β cells express high levels of PDX-1, and it is required for their maintenance and for insulin expression.^33,40,41 Confocal images of cryosectioned adult Pdx1-Cre^/mTmG pancreata, acinar cells demonstrate bright GFP (mG) fluorescence, whereas mG fluorescence of pancreatic islet cells is comparatively weak. This seemingly contradictory finding may reflect lower ongoing levels of mTmG transgene expression from the Gt(ROSA)26Sor genomic locus in islet cells as compared to acinar cells. Alternatively, the finding may correlate with structural, functional, or dynamic differences between the cell membranes of islet and acinar cells.

This fluorescent mouse model demonstrates many valuable characteristics and overcomes a number of hurdles that can limit the utility of a fluorescent mouse model for experimental study. Genetic expression of mT and mG is stable, heritable, and non-diffusible from cell to cell; qualities important for studies that entail tracking of living cells and their progeny over time. As such, this model would prove ideal for a variety of lineage tracing experiments. Lineage tracing approaches, pioneered in the field of developmental biology, are now being used to address questions in the fields of stem cell biology and cancer biology, amongst others.^42–45 The use of multiple fluorescent protein markers, as in this binary fluorescent Pdx1-Cre/mTmG model, expands the types of questions that can be addressed in lineage tracing experiments. For example, multispectral approaches entailing the use of two, or more, fluorescent protein markers have recently been used to address questions of clonality in mouse models of neural and intestinal cancer biology.^8,46 Although toxicity sometimes limits the utility of fluorescent protein markers for in vivo studies of cell biology, we saw no evidence to suggest toxicity of mT or mG expression in this Pdx1-Cre/mTmG model. Low expression of genetically encoded fluorescent markers is another obstacle sometimes encountered. By contrast, the robust mT and mG fluorescence we observed in Pdx1-Cre/mTmG pancreata should make this mouse model useful for a variety of studies that entail intravital imaging.

In summary, this dual color, double transgenic mouse model is optimal for multi-spectral fluorescence imaging of the pancreas at the macroscopic level and for fluorescence imaging of individual cells and cell processes at the microscopic level of the pancreatic microenvironment. This model should prove to be a valuable tool for a variety of fluorescence imaging studies, ranging from pancreas development to pancreatic cancer biology.

Supplementary Material

Acknowledgments

Contract grant sponsor: NIH, Contract Grant Number: R01-CA132971 (to M.B. and AntiCancer, Inc)

Confocal microscopy was performed using equipment belonging to the UCSD Neuroscience Microscopy Shared Facility core, which is funding by shared facility grant P30 NS047101. Processing of tissues for histological review was performed by the UCSD Moores Cancer Center Histology Shared Facility core laboratory.

Contributor Information

Cynthia S. Snyder, UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California.

Austin R. Harrington, UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California

Sharmeela Kaushal, UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California.

Evangeline Mose, UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California.

Andrew M. Lowy, Division of Surgical Oncology, Department of Surgery and UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California.

Robert M. Hoffman, Department of Surgery and UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California and AntiCancer, Inc., San Diego, California.

Michael Bouvet, Division of Surgical Oncology, Department of Surgery and UCSD Moores Cancer Center, University of California, San Diego, La Jolla, California and VA Healthcare System, San Diego, California.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.