In the present study, we describe the design and use of a new “indirect" fluorescent reporter system that drives mutually exclusive expression of either EGFP for INS+ cells or mCherry for INS− cells. This new approach relies on a human insulin promoter-driven Cre-mediated shift in reporter color from red to green in single transgene construct. This results in several advantages over other approaches that use single- or dual-color reporters to distinguish cellular phenotypes.
A single-color reporter system with the insulin promoter driving expression of a fluorescent protein 
serves to identify a subset of cells fulfilling two conditions: (1) successful transduction/transfection and (2) activation of the insulin promoter. However, this system does not allow for straightforward calculation of the percentage of transfected/transduced cells. Among those cells that are successfully transfected/transduced, it is not clear what fractions have activated the insulin promoter.
The “direct" dual-color reporter systems that have been described use the ubiquitous CMV or a cell-/tissue-specific promoter to drive one color and another cell-/tissue-specific reporter for the second color 
. Thus, the cells in which the cell-/tissue-specific promoter is activated should express two colors, while all other cells should express only one color. In principle, this system has the advantage compared to the single-color system of identifying all cells that have been successfully transfected or transduced, so the efficiency of transfection/transduction can be easily calculated. However, a problem with this system became evident when we employed this approach (). For cells in which the tissue-specific promoter had been activated, the relative levels of fluorescence for the two reporter colors was highly variable, due to variability in the relative strength of the two promoters driving fluorescent protein expression, differences in the relative fluorescence intensities, and/or relative degradation rate of the proteins. In a small subset of cells, the only fluorescence seen was that expected to be due to the activation of the cell-/tissue-specific promoter. It is also possible that in these cells, the CMV promoter was inactivated 
The “indirect" dual-color reporter system that we have introduced reports all cells that have been transduced/transfected (so efficiency of transduction/transfection is easily calculated), and it also leads to mutually exclusive marking of cells with one or the other fluorescent protein. Regardless of which fluorescent protein gets expressed, expression is under control of the same strong and ubiquitous CMV promoter. We thus observe only a single color for each phenotype that reports successful INS gene expression by a discrete change in color from red to green ( and ).
Another advantage of our indirect dual-color reporter relates to possible off-target effects of the introduced plasmid or viral vector. Introduction of plasmid or viral vectors into cells may alter cellular phenotype 
, due to heterologous expression of non-native viral and inserted proteins. Controlling for such changes using a single-color reporter is not possible, since different cultures with control vectors are used to assess phenotypes 
. In contrast, our approach results in separate colors arising from the same transfected cells using a single transgene construct. The color resulting from the Cre-mediated recombination event (i.e., removal of the mCherry cassette) thus allows GFP+
cells to serve as an appropriate control for non-insulin-positive cell phenotypes. A very similar dual-color system to ours was reported 
, but the investigators used two separate lentiviral constructs that could not eliminate the complexity of cell-to-cell transduction variability (
, ). Nevertheless, our reporter may need to discern gradients of insulin promoter activity (red-to-green cells over time) in transiently transfected cells using real-time imaging.
Using our new reporter, we found that presumably clonal HIT-T15 cells 
were heterogeneous with respect to INS gene expression (). Approximately 70% of transfected cells were INS+
beta-like-cells, while roughly 30% were INS−
, non-beta-like-cells (possibly of mixed cell phenotypes). This result might be due to unanticipated differentiation of the cells into different phenotypic classes during culture or could possibly indicate a non-clonal nature of the reference HIT-T15 cell line we obtained from ATCC. It was recently reported that pancreatic beta-cell identity is maintained by DNA methylation-mediated repression of Arx 
. Deletion of DNA methyl-transferase gene in INS+
beta-cells converts them into GCG+
alpha-cells by derepression of Arx transcription repressor in beta-cells 
. The red population of cells showed a low level of INS and Pdx1 transcript expression and a high level of GCG and Arx transcript expression (), suggesting conversion of beta-cells into alpha-cells, likely by over-expression of Arx through lack of DNA methylation-mediated repression. Regardless, our results argue that care must be taken in interpretation of previous publications on gene expression and functional profiling of HIT-T15 cells 
with respect to identification of genes and properties as being exclusive correlates of authentic beta-like phenotypes. Some of the previously reported findings may instead be confounded by the significant population of non-beta-cells.
Our fluorescent reporter enables rapid identification and FACS purification of a small percentage of unambiguously identified INS+
beta-like-cells in a mixed population with a significant proportion of non-beta-like-cells (). We found that GFP+
cells expressed insulin transcript, while mCherry+
cells expressed glucagon (). Furthermore, in comparison to mCherry+
cells produced significantly larger inward voltage-gated calcium currents, glucose-stimulated elevation of [Ca2+
, and larger voltage-stimulated secretion ( and ). These characteristics are all indicative of authentic beta-cells 
. In contrast, mCherry+
cells exhibited greater variability in the amplitude of calcium currents and exocytotic responses, perhaps indicating a mixture of different cellular phenotypes in this population. In the future, we will use our reporter to isolate purely homogeneous beta- or beta-like-cells, which will later be used for more accurate genetic, epigenetic, and functional profiling. Our approach should also allow us to compare gene expression and functional phenotypes of beta- or beta-like-cells with other types of pancreatic cells that may be present in the mCherry+
class of cells (α, δ, and PP cells, or perhaps even exocrine acinar cells).
Human islet transplantation using the Edmonton protocol is an effective treatment for type I diabetes 
. Its widespread applicability, however, is limited because of a scarcity of donor tissue. Properly and correctly differentiated beta-cells from human pluripotent stem cells could potentially overcome this limitation. Most current human pluripotent stem cell differentiation protocols 
have limited reproducibility, low yield of beta-like-cells, and most importantly, the absence of glucose responsiveness 
. We expect that our reporter will facilitate easy and quick evaluation of new differentiation protocols designed to produce clinically useful beta-cells—it could be stably introduced into human pluripotent stem cells 
or transduced into different stage cells using an adenoviral vector 
. In addition, the reporter system may be used as a high-throughput screening for reprogramming non-insulin-producing cells to insulin-producing cells. Further additions and modifications to our approach could be easily incorporated, depending on the particular phenotypic property desired. For example, other promoters could be used to mark different cell- or tissue-specific lineages and to further test their homo/heterogeneity.