Why the Same Gene Makes Glucagon in One Cell and GLP-1 in Another

The economy of one gene

Glucagon raises blood glucose. GLP-1 lowers it and suppresses appetite. GLP-2 grows intestinal mucosa. Oxyntomodulin does several of these things weakly at once. They are not related hormones that evolved in parallel - they are pieces of a single precursor protein, cut apart differently depending on which cell is doing the cutting. That is unusual even by the standards of prohormone biology, because the products are not variations on a theme but hormones with opposing metabolic effects, and the same gene is transcribed in the pancreatic alpha cell, the intestinal L cell and parts of the brain.

The standard account

The version in most textbooks and reviews is simple. Prohormone convertases are the enzymes that cut precursors at basic residue pairs. The intestinal L cell expresses PC1 - also written PC1/3 - and cutting proglucagon with PC1 liberates the intestinal set: glicentin, oxyntomodulin, GLP-1 and GLP-2. The pancreatic alpha cell expresses PC2, and cutting with PC2 liberates glucagon along with the major proglucagon fragment. Which peptides a cell makes therefore follows from which convertase it expresses. It is a satisfying explanation, and it is the one worth checking.

What the reconstitution experiments actually found

Dhanvantari, Seidah and Brubaker tested it directly in 1996 by expressing proglucagon in cells and supplying convertases one at a time. PC1 behaved as advertised: in endocrine GH3 and AtT-20 cells it produced glicentin, oxyntomodulin, GLP-1(1-37), GLP-1(7-37) and GLP-2. PC2 did not. In one system it failed to process proglucagon at all; elsewhere it produced glicentin and stopped there. The authors state the negative result plainly - no enzyme or combination they tested produced glucagon. Adding furin, PACE4, PC5a or PC5b did not help. Nor did co-expressing PC2 with chromogranin A, chromogranin B, secretogranin II, or 7B2, the chaperone PC2 requires to mature. Supplying the enzyme credited with making glucagon, in the presence of its own chaperone, did not make glucagon.

And the same result in a real cell

That could have been an artefact of reconstituting the system in cells that do not normally do this. Damholt and colleagues tested it in primary canine intestinal L cells, which do express proglucagon endogenously. They confirmed by immunohistochemistry and Western blot that PC1/3, PC2 and mature 7B2 were all present in those cells. The peptides produced were the classical intestinal profile - mainly glicentin, oxyntomodulin, GLP-1(7-37) and GLP-2 - and they detected, in their phrase, absolute minimal amounts of glucagon. A cell containing proglucagon, PC2 and 7B2 together still did not make glucagon.

What that leaves unexplained, and why it matters

The authors' conclusion is that glucagon formation must require an additional element, or that the components are held in separate intracellular compartments so that the enzyme and the substrate never meet. Either way, the determinant of tissue-specific processing is not simply which convertase is present, because the presence of PC2 is demonstrably not sufficient. Something about the alpha cell beyond its enzyme complement - additional factors, or the physical organisation of its secretory pathway - decides the outcome. The clean version of the story survives as a useful summary and fails as a mechanism, and the papers that showed this are thirty and twenty-seven years old.

Why a peptide reference library cares

Because this is where the modern compound landscape comes from. Dual and triple agonists are built by combining activities that proglucagon already contains: GLP-1 with glucagon, GLP-1 with GIP and glucagon, oxyntomodulin as a naturally occurring dual agonist. Reviews of proglucagon-derived peptides as therapeutics describe exactly this - designing analogues around a set of related activities that a single precursor encodes. Understanding that these are fragments of one protein explains why their receptor selectivities overlap and why a molecule can be tuned along a spectrum between them rather than being either one thing or the other.

Quick reference