Tissue-Specific Expression of GLP1R in Mice: Is the Problem of Antibody Nonspecificity Solved?

Accurate localization of receptors is of importance in the assignment of role for ligand-receptor signaling in the tissue of interest and conceivably derives functional meaning to receptor signaling in that tissue. For example, expression of glucagon-like peptide 1 receptor (GLP1R) on the β-cells of the pancreas indicates that GLP-1 via binding to GLP1R on these cells helps with secretion of insulin from the pancreas. Similarly, GLP1R expression in the liver takes on a different meaning due to the change in organ and organ function. Hence, it is imperative that tissue- and cell-specific expression profile is known for accuracy of conclusions drawn about role of GLP-1/GLP1R signaling in modulation of the function of an organ. So far, this has been elusive due to the paucity of “good” antibodies against GLP1R that produce accurate, reliable, and reproducible data (1,2). The commercially available antibodies as of today are thought to be either nonspecific (i.e., bind to other proteins in addition to GLP1R) or inaccurate (i.e., do not bind to GLP1R at all or are not sensitive enough to detect GLP1R in quantifiable amounts). The reasons for this are many. First, G-protein–coupled receptors (GPCRs) are in general expressed in low numbers in vivo, making it difficult to generate sufficient material for immunization and can take many months to years to produce sufficient amounts. Second, the epitopes are made up of small discontinuous segments, making them inaccessible to standard peptide or recombinant protein technology as these small segments in isolation cannot fold or assemble together. Third, the epitopes are largely dependent on the membrane for structure and purification, and thus removal of the membrane typically results in denaturation of the protein and destruction of the epitopes. Alternatively, immunizing with whole cells and thus maintaining the structure means immunizing with a very impure mix of protein, and the immune response is greatly diminished if not nonexistent. Fourth, GPCRs are heavily glycosylated, yielding different sizes of bands on a gel that may not be recognized as “real” bands. Fifth, GPCRs are highly conserved in their transmembrane domains, making cross-reactivity a major problem.

In this issue, Richards et al. (1) have overcome the known problems with GLP1R antibodies by generating transgenic mice expressing “humanized” Cre recombinase (iCre) under control of GLP1R promoter, in the process bypassing the whole gamut of issues inherent to GLP1R antibodies (1–5) (Fig. 1). Their goal was to determine the action of GLP-1 and its metabolites in postprandial effects—pancreas versus other organs (heart, kidney, skeletal muscle, neurons) and if there is constitutive expression/secretion. As a result, some novel data were generated in addition to existing data being confirmed or refuted. GLP1R was predominantly coexpressed with α-smooth muscle actin, suggesting and confirming a vasodilatory role in the vasculature (1,6). GLP1R was expressed in atrial cardiomyocytes, but not in ventricular cardiomyocytes, consistent with atrial natriuretic peptide in the atrium and an indirect role for GLP1R in lowering blood pressure (7). Neurons express GLP1R and primary cultures of GLP1R-fluorescent enteric neurons respond to GLP-1 stimulation ex vivo, demonstrating that elegant ligand-receptor interaction studies can be designed (1,8).

Richards et al. generated GLP1R transgenic mice using a bacterial artificial chromosome (BAC)-based construct that confers the advantages of possessing (∼71.4 kb upstream and ∼93.2 kb downstream of GLP1R coding sequence) long stretches of DNA that contain all the enhancer and promoter elements in their native state (9–12). This is opposed to using shorter cDNA sequences of genes to generate transgenic mice that generally lack copy number and expression specificity (could integrate in the vicinity of another gene and take on the tissue specificity of that gene or take on the level of expression of the local gene due to control by the local promoter). In addition, while longer genes integrate in less copy numbers (one to three), shorter sequences can form long concatemers of 8–12 and overexpress at very high levels (10). In general, using BACs for generating transgenic constructs gives one much better control over where the gene is expressed, at what levels, and how the gene influences other genes in the vicinity. In addition, generation of mutations in the GLP1R enhancer/promoter sequence that alter the tissue specificity and expression level offer a powerful tool to study gene and protein regulation in vivo (11,12). Most importantly, localization of receptor expression is important to determine 1) what the site of action of GLP-1 is, 2) if abundance of GLP1R expression can determine how much effect occurs, 3) which cell types in a tissue express GLP1R (colocalization studies can determine, but may not be definitive and limited by antibody resolution and second antibody’s reputation), and 4) what the function of GLP1R is in that tissue (e.g., increased firing of enteric neurons in gastric antrum vs. intestinal epithelium).

Despite the significant advancement in GLP1R expression/function by Richards et al., use of transgenes with iCre driven by GLP1R–enhancer/promoter elements will not replace the need for antibodies in the study of other receptor functions (such as binding to signaling partners, receptor cycling, and determination of motifs important for this process), antagonizing GLP-1/GLP1R signaling and generation of novel drugs (13), and of GLP1R expression in ex vivo samples (unless GLP1R-iCre mouse is used to do all the studies). Other drawbacks of transgene expression include lack of control on BAC site of insertion; lack of tissue specificity, cell specificity, and copy number; and that influence of other neighborhood genes, although rare, may still exist. In addition, iCre expression results in permanent expression of red fluorescent protein (RFP) in progeny of parental cells that may no longer express GLP1R (embryonal vs. neonatal vs. adult). Furthermore, very low iCre expression may not be sufficient to cleave floxed RFP, suggesting that false negatives may exist (Richards et al. mention intestinal epithelia). mRNA transcript–based detection is probably the gold standard for identification of tissue-specific expression, but some caveats still need to be recognized. First, tissue-specific alternate transcripts/splice variants may not be detected by using a single set of primers. Second, smaller or bigger PCR products than anticipated may be detected and would need sequencing in order to estimate relevance (14). Of importance, the above issues are exemplified by two studies, one using chicken GLP1R and the other rat GLP1R, which show that tissue-specific variants of GLP1R do exist and these may explain the extrapancreatic actions of GLP-1/-2 (15). Third, sometimes primers used to detect a target sequence do not work across species due to sequence variations (16). Fourth, expression may be extremely low (quantitative PCR needs to be done or efficiency of primers may be too low). Other concerns exist. Extrapolation of cross-species data should be done with caution; all organisms are not alike, so what is applicable in the mouse may not be applicable in the human (17). At least two different isoforms of GLP have been identified (GLP-1 and GLP-2), cross-reactivity to other receptors (glucose-dependent insulinotropic polypeptide) exists, and, last, a second GLP1R (http://www.glucagon.com) may exist.

In summary, Richards et al. (1) have demonstrated that studies of reliable tissue-specific expression of GLP1R and other GPCRs are feasible. The concerns pertinent to this methodology are accessibility to most researchers, cost-effectiveness of making multiple transgenic models for multiple transmembrane receptors, and issues inherent to the use of a genetically altered transgenic mouse model.