AgRP

What is AgRP?

Agouti-related peptide (AgRP) is a 132-amino-acid endogenous neuropeptide encoded by the AGRP gene, discovered in parallel in 1997 by Marcus Ollmann and Gregory Barsh at Stanford (Science) and Jennifer Shutter and colleagues at Amgen (Genes & Development) as a hypothalamic homolog of the skin agouti signaling protein. The full-length pre-pro-peptide is processed to a C-terminal cysteine-rich active fragment (approximately residues 83–132 in humans) that retains full pharmacological activity at melanocortin receptors. AgRP is produced almost exclusively in a discrete neuronal population in the arcuate nucleus of the hypothalamus — the so-called 'AgRP neurons' — where it is co-expressed with neuropeptide Y (NPY) and GABA in the same cells. Pharmacologically, AgRP is one of the few known endogenous inverse agonists in mammalian biology: it does not merely block agonist binding at MC3R and MC4R but actively suppresses the constitutive activity of these receptors, pushing the melanocortin system toward the 'hunger' state even in the absence of α-MSH. AgRP neurons sit at the center of the leptin–melanocortin appetite circuit. They are activated by fasting, ghrelin, and negative energy balance; inhibited by leptin, insulin, and PYY; and project to the paraventricular hypothalamus and other downstream sites where they drive voracious feeding through a combination of MC4R antagonism, NPY release at Y1/Y5 receptors, and GABAergic inhibition. Their counterpart, the anorexigenic POMC neurons of the same arcuate nucleus, releases α-MSH to oppose them. The circuit has been characterized with unusual experimental precision: optogenetic activation of as few as 800 AgRP neurons evokes voracious feeding within minutes (Aponte, Atasoy, Sternson, Nature Neuroscience 2011), while their ablation in adult mice causes fatal anorexia (Luquet, Palmiter, Science 2005; Gropp, Brüning, Nature Neuroscience 2005). AgRP is not a therapeutic product — no one administers AgRP as a drug, because doing so would drive hunger. Its importance is as the upstream hunger signal in the pathway that setmelanotide (MC4R agonist) was designed to treat from the opposite direction: setmelanotide bypasses and opposes AgRP-driven tone at MC4R to restore satiety in patients whose leptin–POMC–α-MSH input to this circuit is broken.

What AgRP Is Investigated For

AgRP is foundational neuroendocrine biology, not a peptide anyone buys. It is the central nervous system's primary hunger-driving signal and the defining marker of the arcuate-nucleus neuronal population whose activity causally controls feeding behavior in mice — voracious eating when these neurons fire, starvation when they are ablated. The therapeutic relevance of AgRP is inverse: the entire MC4R-agonist class of obesity drugs, most notably setmelanotide (Imcivree, FDA-approved for POMC/PCSK1/LEPR deficiency, Bardet-Biedl syndrome, and acquired hypothalamic obesity), works by pushing MC4R in the opposite direction from the AgRP-driven hunger tone. Understanding AgRP is the biological foundation for understanding why some monogenic obesities exist (loss of upstream POMC signaling leaves AgRP unopposed), why setmelanotide works (it competes against AgRP at the same receptor), and why the leptin–melanocortin pathway is one of the most important target spaces in obesity therapeutics. There is no consumer use case for AgRP itself: exogenous AgRP administration would drive hunger and weight gain, the opposite of what most consumers researching hypothalamic appetite peptides are looking for. The honest framing is that AgRP completes the conceptual picture of the arcuate appetite circuit alongside NPY (co-expressed in the same neurons), α-MSH (the counter-regulatory agonist), POMC (the upstream precursor), leptin and ghrelin (the peripheral modulators), and setmelanotide (the downstream pharmacological restoration).

History & Discovery

The story of AgRP begins with the coat-color genetics of mice. The agouti locus — responsible for the banded yellow-tipped fur of wild-type mice and, in dominant gain-of-function alleles (lethal yellow, Ay), for a syndrome of yellow fur, obesity, and diabetes — had been studied since the early 20th century, and the agouti gene itself was cloned in 1992. The mechanistic question was why a skin signaling protein would also produce obesity when expressed ectopically. By the mid-1990s, melanocortin-receptor pharmacology had advanced far enough to suggest that agouti protein acted as an antagonist of melanocortin receptors on melanocytes (blocking α-MSH to shift pigmentation toward pheomelanin) and that something analogous might exist in the hypothalamus to antagonize MC4R and drive obesity. In 1997, two groups converged on the answer in parallel. Jennifer Shutter and colleagues at Amgen, publishing in Genes & Development (March 1997; PMID 9119224), used a homology-based search strategy to identify a novel gene they called ART (agouti-related transcript), predicted to encode a 132-amino-acid protein with ~25% identity to human agouti. In situ hybridization localized ART expression to the arcuate nucleus of the hypothalamus, the median eminence, and the adrenal medulla — and crucially, hypothalamic ART expression was upregulated roughly 10-fold in ob/ob (leptin-deficient) and db/db (leptin-receptor-deficient) obese mice. Later in the same year, Marcus Ollmann, Brian Wilson, and Gregory Barsh at Stanford, together with Ira Gantz at Michigan, published in Science (October 1997; PMID 9311920). They independently cloned the gene — calling it agouti-related protein (AgRP) — and demonstrated that recombinant human AgRP was a potent, selective antagonist of MC3R and MC4R in vitro, and that transgenic mice ubiquitously expressing human AGRP cDNA developed obesity without pigmentation changes. Between them, these two papers established AgRP as the hypothalamic hunger counter-regulator to α-MSH — a second endogenous melanocortin-receptor antagonist in the mammalian genome, expressed in the exact brain region predicted by the emerging leptin–POMC–MC4R satiety model. The pharmacology deepened over the following decade. In 2001, Wouter Nijenhuis and colleagues (Molecular Endocrinology; PMID 11145747) showed that AgRP(83-132) was not a simple competitive antagonist but an inverse agonist at MC4R, suppressing the receptor's constitutive activity below unliganded baseline. Tolle and Low (Diabetes 2008; PMID 17909095) extended this in vivo using POMC-deficient mice, demonstrating that AgRP modified energy balance even in the absence of α-MSH — functional confirmation of the inverse-agonist model. The neural-circuit era of AgRP research opened in 2005 with two parallel papers on AgRP-neuron ablation. Serge Luquet and Richard Palmiter (Science; PMID 16254186) and, independently, Eva Gropp with Jens Brüning (Nature Neuroscience; PMID 16158063) used genetic ablation strategies to eliminate AgRP-expressing neurons in adult mice, with a consistent and striking result: adult mice lost their AgRP neurons and then acutely stopped eating, with many animals dying of starvation. Neonatal ablation (Luquet) produced compensation and little phenotype, making the adult result all the more pointed — the circuit is actively required for normal feeding in the mature brain. The optogenetic and chemogenetic era arrived in 2011. Yexica Aponte, Deniz Atasoy, and Scott Sternson at Janelia (Nature Neuroscience; PMID 21209617) showed that direct channelrhodopsin-2 activation of roughly 800 AgRP neurons produced voracious feeding within minutes in sated mice, without any prior training — establishing sufficiency. Michael Krashes and Bradford Lowell (JCI; PMID 21364278) used DREADD chemogenetics to drive rapid, reversible AgRP-neuron activation with similar feeding effects, adding reversibility and pharmacological control. Atasoy and Sternson (Nature 2012; PMID 22801496) deconstructed the downstream circuit, identifying the AgRP → paraventricular hypothalamus GABAergic projection as the acute driver of feeding and the AgRP → oxytocin-neuron projection as a clinically resonant node implicated in Prader-Willi syndrome. More recent work has refined the picture. Chen and Knight (Cell 2015; PMID 25703096) used fiber photometry to show that AgRP neuron activity falls within seconds of food sighting or smelling — before any ingestion — reframing these cells as state-representation neurons rather than simple hunger metering. A 2024 paper identified a second AgRP-expressing population in the hindbrain, extending the anatomy beyond the classical arcuate. On the translational side, the 2020 FDA approval of setmelanotide — an MC4R agonist designed to oppose the AgRP-driven hunger tone that goes unopposed in POMC, PCSK1, and LEPR deficiency — closed a 25-year loop between the molecular discovery of AgRP and a clinically approved drug built around its pharmacology.

How It Works

AgRP is one of the brain's strongest 'eat now' signals. A small cluster of neurons in the arcuate nucleus of your hypothalamus makes AgRP (along with another hunger peptide, NPY). These neurons fire when you're fasted or hungry, and the AgRP they release blocks and reverses the 'you're full' signal carried by α-MSH at the MC4 receptor. If you activate these neurons in a mouse, it eats ravenously; if you kill these neurons in an adult mouse, it stops eating and starves. Obesity drugs like setmelanotide work by pushing the MC4 receptor the other way, essentially overriding AgRP's hunger signal.

AgRP is encoded by the AGRP gene on human chromosome 16 and is expressed almost exclusively in a discrete neuronal population (~10,000 neurons in the mouse) of the arcuate nucleus of the hypothalamus, with minor expression in the adrenal medulla and a small hindbrain population identified more recently. The full-length 132-residue pre-pro-peptide is processed to a mature cysteine-rich C-terminal fragment — AgRP(83-132) in humans, with characteristic inhibitor cystine-knot topology — that retains essentially all melanocortin-receptor activity of the parent molecule. At MC3R and MC4R, AgRP is an inverse agonist rather than a simple antagonist (Nijenhuis et al., Molecular Endocrinology 2001; Tolle and Low, Diabetes 2008). These receptors display measurable constitutive, ligand-independent activity through Gs/cAMP signaling, and AgRP binds to stabilize an inactive receptor conformation, suppressing that basal tone below unliganded baseline in addition to competing with α-MSH. Functionally this produces a push–pull system in which α-MSH from POMC neurons activates MC4R (satiety) and AgRP from the paired AgRP/NPY neurons inverse-agonizes it (hunger), with net MC4R output determined by the balance between the two. Arcuate AgRP neurons are GABAergic and co-express NPY. They project to the paraventricular hypothalamus (PVH), the bed nucleus of the stria terminalis, the parabrachial nucleus, and other downstream feeding circuits. Atasoy and Sternson (Nature 2012) mapped the hunger-driving circuit onto the AgRP → PVH projection and showed that GABAergic inhibition of PVH neurons is the most rapid component of AgRP-evoked feeding, while NPY and AgRP contribute more slowly. AgRP neurons are activated by fasting, ghrelin (via GHSR on AgRP neurons), and glucocorticoids, and inhibited by leptin (via LepRb), insulin, PYY(3-36) (via presynaptic Y2 autoreceptors), and sensory cues predicting food availability — Chen and Knight (Cell 2015) used fiber photometry to show that the mere sight and smell of food rapidly deactivates AgRP neurons and activates POMC neurons, before any calories are ingested. The causal role of this neuronal population in feeding has been established with unusual clarity. Luquet and Palmiter (Science 2005) used diphtheria-toxin-mediated ablation of NPY/AgRP neurons and showed that while neonatal ablation produced compensation, ablation in adult mice caused acute severe anorexia and death from starvation. Gropp, Brüning, and colleagues (Nature Neuroscience 2005) reached the same conclusion with a parallel genetic approach. Aponte, Atasoy, and Sternson (Nature Neuroscience 2011) showed that optogenetic channelrhodopsin-2 activation of as few as 800 AgRP neurons evoked voracious feeding within minutes, and Krashes and Lowell (JCI 2011) extended this with DREADD chemogenetic activation showing rapid, reversible feeding induction and corresponding increases in fat mass. Together, these studies establish AgRP neurons as both necessary and sufficient for feeding. In the clinical context, the AgRP-opposed pathway is the target space for the most precision-selected obesity drug approved to date. Setmelanotide (Imcivree) is a synthetic MC4R agonist that restores satiety signaling in patients with POMC, PCSK1, or LEPR loss-of-function (in whom α-MSH production or leptin sensing is broken and AgRP-driven tone at MC4R is unopposed), Bardet-Biedl syndrome, and acquired hypothalamic obesity. Human MC4R loss-of-function mutations are the single most common monogenic cause of obesity because they remove the receptor that AgRP acts upon. Plasma AgRP measurements, while not a standardized clinical test, rise in anorexia nervosa and cachectic states (Moriya et al. 2006) and fall after bariatric surgery and with leptin administration — consistent with peripheral AgRP reflecting central hypothalamic hunger-signal state.

Evidence Snapshot

Research Gaps & Open Questions

What the current literature has not yet settled about AgRP:

• 01Whether human AgRP neurons behave identically to well-characterized rodent AgRP neurons — the circuit has been mapped in mice with exceptional precision, but human post-mortem and imaging evidence for the equivalent circuit is indirect, and whether sensory-cue rapid deactivation (Chen and Knight 2015) operates identically in humans is unestablished.
• 02The functional significance of the recently-described hindbrain AgRP-expressing population (2024) and whether it operates in parallel with or downstream of the classical arcuate AgRP neurons.
• 03Whether selectively modulating AgRP-neuron activity (rather than the downstream MC4R) could produce a better therapeutic profile — in principle, suppressing AgRP firing would spare α-MSH-mediated satiety while reducing hunger drive, but no such drug has reached clinical development.
• 04The long-term consequences of sustained MC4R-agonist therapy (setmelanotide) for AgRP-neuron adaptation — whether AgRP expression or firing increases compensatorily, potentially contributing to the plateau or tolerance that obesity drugs often show over years.
• 05Whether plasma AgRP measurement can be standardized into a clinically useful biomarker of hypothalamic energy-balance state (cachexia, anorexia nervosa, refeeding), or whether peripheral AgRP is too noisy a surrogate for central neuronal activity to support that use.
• 06The role of AgRP signaling in non-feeding behaviors — reward processing, anxiety, exercise, and compulsive behaviors — which recent rodent work has begun to map but which remains poorly characterized in humans.
• 07How AgRP signaling interacts with GLP-1 receptor agonism therapeutically — whether semaglutide-class drugs engage AgRP-neuron activity, whether dual-pathway targeting could add benefit, and whether the anorexia-like tolerability concerns with MC4R agonism reflect unopposed AgRP-neuron suppression.

Forms & Administration

AgRP is not available as an approved therapeutic in any form and is not administered clinically in humans. Synthetic AgRP(83-132) — the active C-terminal fragment — and related constructs are supplied by biochemical vendors as research reagents for laboratory and preclinical animal work, typically delivered by intracerebroventricular injection in rodent studies. There is no consumer route of administration, no validated human dose, no approved indication, and no clinical reason to administer exogenous AgRP given that its biological effect is to drive hunger. Pharmaceutical development in the AgRP target space has moved in the opposite direction — MC4R agonists such as setmelanotide, and investigational compounds that modulate AgRP-neuron activity upstream — rather than toward administering AgRP itself.

Common Questions

Safety Profile

Legal Status

Regulatory status changes over time. Verify current local rules with a qualified professional.

Myths & Misconceptions