The hunger neurons GLP-1 drugs were supposed to silence turn out to be doing the work

Yale researchers report in the Proceedings of the National Academy of Sciences that semaglutide, the active ingredient in Ozempic, activates the brain's AgRP hunger neurons rather than suppressing them, and that mice engineered to lack those neurons no longer sustain weight loss on the drug ^[1][2][3]. If the result carries into humans, the mechanistic account the field has been working from, that GLP-1 drugs act by turning hunger circuitry down, has the sign wrong on one of its central terms ^[4].

The starting puzzle was durability, not appetite. Earlier weight loss medications reduce appetite nearly as effectively as semaglutide but do not produce the same sustained loss, which led the Yale group to suspect the drug is doing something beyond making animals eat less ^[5][6]. GLP-1 therapies can hold 10 to 15 percent or more of body weight off, against the modest reductions typical of older drugs ^[7][8]. The gap is the interesting number: same appetite effect, different outcome.

The team ran a mouse model tracking body weight, food intake, metabolism and energy expenditure during semaglutide treatment, then used genetic techniques to eliminate or silence AgRP neurons and test whether the drug's lasting effects survived without them ^[9]. They did not. Electron microscopy, molecular biology and electrophysiology then showed the neurons being activated rather than suppressed ^[9][1]. The researchers' reading is that when GLP-1 treatment opens a calorie deficit, the brain raises AgRP activity, and those same neurons help coordinate the loss of fat ^[10].

Two things about the prior model are worth being precise about. AgRP neurons were generally understood to work mainly against weight loss ^[11]. And the widely discussed explanation, that GLP-1 drugs suppress hunger-promoting neurons, had not been directly tested in vivo during chronic GLP-1 treatment before this work ^[4]. That is an unusually load-bearing assumption to have gone unmeasured for as long as it has.

The consequence for discovery programs is the part worth sitting with, and it is inference rather than anything the Yale team tested. A candidate screened on appetite suppression, or on quieting AgRP firing, would look like a success on the old model and could be selecting against exactly the mechanism this paper associates with durable fat loss ^[4][1][2]. First author Mateus d'Avila, a doctoral candidate in Tamas Horvath's lab in the Department of Comparative Medicine at Yale School of Medicine, put the claim broadly: "This completely changes how we think about the mechanism involved in these medications and provides new insight into the biology underlying their long-term effects, opening an avenue for the development of more efficient drugs" ^[12].

The obvious discount applies. These are mice, and the authors say further work is needed to establish whether the same mechanism operates in humans ^[13].

What to watch: whether the necessity result replicates with silencing rather than ablation alone, since a circuit that is required when deleted from development is not the same claim as one required moment to moment ^[9]. Watch also whether human tolerability data, the nausea and appetite reports that companies already collect, can be read against a model in which hunger signaling rises during treatment ^[10]. And watch which preclinical obesity targets get requalified, because the ones justified purely by appetite suppression now rest on a premise the source describes as untested until now ^[4][6].