Interrogating the central melanocortin system: choosing the “right” model

The British statistician George Box famously quipped, “Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful” (2). This statement speaks to an inconvenient truth of the biomedical sciences—no biological system can be perfectly replicated using an artificial, testable model. In particular, the use of genetic animal models comes with a host of caveats that must be carefully considered when designing experiments or interpreting data. However, as the second part of Box’s infamous comment implies, the overwhelmingly complex nature of living biological systems necessitates the use of such models to interrogate their inner workings and to understand the basic mechanisms underlying observed physiology. It is therefore imperative to judiciously and deliberately choose the model that will best address a specific hypothesis and to take into account the limitations associated with that model when drawing conclusions.

Investigation of the central melanocortin system is an excellent example of the need to carefully choose an appropriate model with acceptable caveats to answer a particular biological question. The central melanocortin system comprises neurons expressing the pro-opiomelanocortin (POMC) prohormone in the arcuate nucleus of the hypothalamus (ARC) and in the nucleus of the solitary tract (NTS), as well as the downstream neurons that mediate the effects of POMC prohormone products (e.g., α-MSH) via the central melanocortin receptors MC3R and MC4R. The central melanocortin system has been shown to play important roles in food intake, body weight regulation, and cardiovascular function (3, 4, 10). POMC neurons in the NTS have been shown to respond to peripheral metabolic signals (1) and coordinate appropriate behavioral responses (8). However, the detection of Pomc mRNA and protein within the NTS has presented a challenge to researchers interested in brain stem POMC function, with some laboratories reporting the inability to measure Pomc expression using several methodologies. Furthermore, POMC neurons are notoriously small and difficult to visualize, further complicating analysis of NTS POMC neuron signaling, gene expression, and electrophysiological properties.

Because of the inherent difficulties in studying POMC neurons, several reporter mouse lines have been generated in which the POMC promoter drives the expression of a fluorophore (Pomc-Gfp) or Cre recombinase (Pomc-Cre^GSB and Pomc-Cre^Lowl). Although the use of these models has yielded multiple insights into the physiology of ARC POMC neurons, studies evaluating NTS POMC neurons have been more difficult to interpret and in some cases have been contradictory. For example, Pomc-Gfp neurons in the NTS have been shown to express leptin receptors (6) and respond to peripherally injected leptin (5). In contrast, Pomc-Cre-identified neurons in the NTS do not exhibit STAT3 phosphorylation or early gene expression in response to leptin treatment (7). These differences may be explained by the recent work of Hentges and colleagues at Colorado State University, who evaluated the expression patterns of POMC in the NTS using multiple POMC mouse models (9). This was accomplished by crossing Pomc-Gfp mice with Pomc-Cre^GSB mice and then enabling Cre-dependent expression of ChR2 tagged with a fluorophore (mCherry) by injecting an adeno-associated viral vector (AAV) directly into the NTS. Remarkably, Pomc-Gfp-positive cells did not colocalize with ChR2^mCherry-positive cells, suggesting that these two methods of POMC neuronal identification targeted separate populations of Pomc-expressing neurons. To further evaluate the anatomic distribution of these two POMC populations in the NTS, Hentges and colleagues utilized the Cre recombinase reporter mouse line GtROSA26Sor-tdTOM (ROSA-tom), in which Cre expression drives expression of the td-tomato fluorophore. This reporter mouse was crossed with mice expressing the Pomc-Gfp transgene and one of two commonly used Pomc-Cre transgenes (Pomc-Cre^GSB or Pomc-Cre^Lowl), which yielded mice expressing Pomc-Gfp, Pomc-Cre (GSB or Lowl), and ROSA-tom. As with their first experiment, Pomc-Gfp did not colocalize with Pomc-Cre-expressing cells. Indeed, POMC cells expressing Pomc-Cre were positioned caudal and lateral to cells identified by Pomc-Gfp. In addition, differences in amino acid (i.e., Gad2 vs. vGlut2 expression) phenotype were observed between Pomc-Gfp and Pomc-Cre cells as well, further underscoring the discrepancies between the cells identified by these two transgenes.

The reason for the differential expression of the Pomc-Gfp and Pomc-Cre transgenes in the NTS is unclear. One possibility is that Pomc can be transiently transcribed during development; however, Hentges and colleagues controlled for this possibility by the use of yet another Pomc-Cre model with inducible expression by tamoxifen (Pomc-Cre:ERT2), and the results were the same. Interestingly, this disparate expression pattern appeared to be confined to the NTS, as colocalization between Pomc-Gfp and Pomc-Cre was observed the arcuate nucleus. Given that these findings were restricted to the NTS, a more probable explanation lies in the promoter regions used for generating the transgenes, since the Pomc-Cre lines likely include more extensive regulatory elements. Regardless, this study highlights two important points. First of all, it could serve as an explanation for the contradictory findings regarding the biology of NTS POMC neurons. Second, and most importantly, this study serves as a reminder of the importance of carefully choosing the appropriate model for a particular experiment and understanding the limitations inherent in that model system.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

G.L.C.Y. drafted manuscript; G.L.C.Y. edited and revised manuscript; G.L.C.Y. approved final version of manuscript.