HM17321

What is HM17321?

HM17321 is an investigational once-weekly peptide from Hanmi Pharmaceutical, engineered as a selective agonist at corticotropin-releasing factor receptor 2 (CRFR2) and derived from urocortin-2, the endogenous CRFR2-selective member of the corticotropin-releasing factor peptide family. It is being developed for obesity and associated conditions including type 2 diabetes and cardiovascular disease. The reason it attracted a major pharmaceutical partner is what it is trying not to do. GLP-1-based obesity drugs work, but a substantial fraction of the weight they remove is lean tissue rather than fat — commonly cited at roughly a quarter of total weight lost, and measured at 25% to 39% across incretin drugs in a 2026 meta-analysis of 20 randomized trials — which matters increasingly as these drugs move into older patients, longer durations, and populations where sarcopenia is already a concern. HM17321 targets a completely different receptor with a completely different biology: CRFR2 is expressed in skeletal muscle, heart, and vasculature rather than being primarily an appetite receptor, and CRFR2 activation has been shown since the early 2000s to increase muscle mass in healthy animals and to reduce muscle loss during atrophy. Hanmi's stated design goal is simultaneous weight loss and lean mass preservation, and its pipeline materials go further, describing increased lean mass rather than merely preserved lean mass. On 24 August 2026 Hanmi licensed HM17321 to Genentech, part of the Roche Group, for $190 million upfront within a deal whose total potential value the company put at approximately $2.3 billion inclusive of that upfront payment, plus tiered royalties on future sales. Hanmi retained South Korean rights and will complete the Phase 1 trial, after which Genentech assumes development from Phase 2 onward. The honest position is that this is an early asset with a compelling thesis and no human results. The Phase 1 trial (NCT07219589) began dosing on 6 November 2025 and is still recruiting, with primary completion estimated for March 2027. Every efficacy claim in the public domain comes from mouse, rat, and non-human-primate studies presented as conference abstracts by the sponsor. Hanmi has publicly targeted a 2031 launch.

What HM17321 Is Investigated For

HM17321 is a Phase 1 asset, and the appropriate way to read it is as a well-funded bet on a specific hypothesis rather than as a drug with demonstrated effects. The hypothesis is that CRFR2 agonism can drive fat loss while sparing or building skeletal muscle, addressing the most-discussed shortcoming of incretin-based obesity therapy. The supporting biology is real and predates the current obesity boom by two decades: Hinkle and colleagues showed in 2003 that CRFR2 activation increases skeletal muscle mass in healthy rodents and reduces muscle mass and function loss during atrophy, and CRFR2-selective sauvagine analogues were being developed specifically for muscle atrophy as early as 2005. A 2025 review in Skeletal Muscle by Vidal and colleagues frames the CRFR2/urocortin axis explicitly in terms of preserving muscle during incretin-driven weight loss, which is the clinical thesis in one sentence. Hanmi's own preclinical data has been presented across roughly nine conference abstracts at ADA 2025 and ADA 2026 and an EASD 2025 poster, covering body composition in diet-induced-obese mice, glycemic control, sarcopenic mice, combinations with myostatin/activin inhibition and with amylin analogs, cardiorenal effects in spontaneously hypertensive rats, and cardioprotection in rodent and non-human-primate heart failure models. The EASD poster is the most informative public document: in diet-induced-obese mice, HM17321 reduced body weight more than its own pair-fed control group — implying an effect beyond appetite suppression alone — while increasing lean mass and reducing fat mass, where semaglutide reduced both. Two details in that same poster deserve to travel with the headline. Food intake was significantly reduced, so this is not a purely appetite-neutral mechanism. And the apparent increase in energy expenditure held when adjusted for body weight but not when adjusted for lean mass, with the poster itself concluding the effect was largely driven by the increased lean mass rather than by direct thermogenesis. There is also a genuine scientific contradiction that any honest treatment has to include. Chen and colleagues reported in PNAS in 2006 that urocortin-2-knockout mice have increased insulin sensitivity, are protected from fat-induced insulin resistance, and — strikingly — show decreased fat with increased lean mass on a high-fat diet, with a CRFR2-selective antagonist reproducing the knockout glucose phenotype in wild-type animals. Their conclusion was that suppressing the urocortin-2/CRFR2 pathway might benefit insulin-resistant states, which is the opposite of the direction HM17321 and its competitors are betting on. Chao and colleagues supported that mechanistically via mTOR in 2015. This is unresolved, and it is the single most important reason to treat the glycemic-control claims as unproven rather than merely early. One further caveat on positioning: Hanmi's first-in-class framing is contestable, though not because anyone is ahead of it. Gubra's GUB-UCN2 (NCT07702890) is a rival urocortin-2/CRHR2 analog, but it began dosing on 27 July 2026 — nearly nine months after Hanmi — and its primary completion is estimated for December 2028 against Hanmi's March 2027. What Gubra has is a broader trial rather than an earlier one: a Phase 1/2a design in 188 participants spanning lean, obese and type 2 diabetic groups, up to 16 weeks of dosing, standalone and on top of standard of care, with registered muscle endpoints. Since Hanmi's Phase 1 registers only pharmacokinetic secondary outcomes, Gubra will probably produce efficacy-grade body-composition data first even though it started later. A third entrant has since appeared: Novo Nordisk published acylated urocortin-2 data in rats and minipigs in October 2026.

History & Discovery

HM17321 sits at the intersection of two histories: a thirty-year-old peptide family that never produced an approved drug, and a two-year-old commercial problem created by the success of GLP-1 agonists. The peptide side begins in 1995, when Joan Vaughan, Wylie Vale, and colleagues at the Salk Institute identified urocortin-1 as a mammalian relative of fish urotensin I and corticotropin-releasing factor. In 2001 two laboratories working in parallel described two further family members: Reyes, Lewis, Vaughan, Sawchenko, Vale and colleagues reported urocortin-2 and then urocortin-3 in PNAS, while Hsu and Hsueh at Stanford identified the same peptides from an expressed-sequence-tag search and named them stresscopin-related peptide and stresscopin in Nature Medicine. The finding that mattered commercially was selectivity: urocortin-2 and urocortin-3 bind CRFR2 with essentially no CRFR1 activity, which meant the peripheral CRFR2 biology could be engaged pharmacologically without triggering the cortisol-and-anxiety arm of the stress axis. Skeletal muscle became the most interesting CRFR2 story almost immediately. In 2003 Hinkle and colleagues published two papers showing that CRFR2 activation increases muscle mass and function in healthy animals and reduces the loss of both during atrophy from nerve damage, corticosteroids, and disuse. Isfort and colleagues at Procter & Gamble developed CRFR2-selective sauvagine analogues explicitly for muscle atrophy, publishing in J Med Chem in 2005, and Hall and colleagues tested a CRFR2 agonist with exercise in mdx muscle in 2007. None of it reached approval. In parallel, the Christchurch Heart Institute in New Zealand pursued the cardiovascular arm — Davis, Richards and colleagues infused urocortin-2 into patients with heart failure in 2007, and the UNICORN trial in 2013 remains the only randomized therapeutic trial of urocortin-2 in humans. That programme also stopped, and the UC San Diego group under Hammond pivoted to AAV-delivered urocortin-2 gene transfer rather than peptide infusion. By the late 2010s the CRFR2 axis was a well-characterised target with no drug. What revived it was the incretin era. As semaglutide and tirzepatide moved from diabetes into mass-market obesity treatment, the composition of the weight being lost became a live clinical question — a substantial share of it is lean tissue, commonly put at around a quarter, which matters more as these drugs reach older patients, longer treatment durations, and people already at risk of sarcopenia. A 2022 review by Christoffersen and colleagues in Obesity laid out the strategic case for targeting energy expenditure, fat oxidation, and lean mass preservation rather than appetite alone. CRFR2, with two decades of muscle-anabolism literature behind it, was an obvious candidate, and a 2025 review by Vidal and colleagues in Skeletal Muscle framed the axis explicitly around preserving muscle during incretin-driven weight loss. Hanmi Pharmaceutical, a Korean company with a long-running metabolic pipeline, built HM17321 as a CRFR2-selective urocortin-2 analog engineered for once-weekly subcutaneous dosing, describing the discovery process as using AI and structural modelling. Preclinical data appeared as conference abstracts at ADA 2025 — covering body composition and glycemic control across obesity models — and at EASD 2025, where a poster compared HM17321 head-to-head against semaglutide in diet-induced-obese mice with pair-fed controls and reported increased lean mass with reduced fat mass where semaglutide reduced both. A further tranche of ADA 2026 abstracts extended into sarcopenic mice, combination with myostatin/activin inhibition, combination with amylin analogs, cardiorenal effects in hypertensive rats, and cardioprotection in rodent and non-human-primate heart failure models. Hanmi submitted an IND to the FDA in late September 2025 and reported clearance in November 2025; the Phase 1 trial NCT07219589 began dosing on 6 November 2025 at a single site in Cincinnati. The commercial validation came on 24 August 2026, when Hanmi licensed HM17321 to Genentech for $190 million upfront within a deal it valued at approximately $2.3 billion inclusive of that payment, plus tiered royalties, retaining South Korea. Hanmi completes Phase 1; Genentech develops from Phase 2. Roche's business development lead described the intent as selectively reducing fat mass while improving muscle mass and function. Hanmi has publicly targeted a 2031 launch. The programme is not unopposed, though Hanmi retains the clinical lead. Gubra's GUB-UCN2 (NCT07702890) began dosing on 27 July 2026, nearly nine months after HM17321, in a broader Phase 1/2a trial of 188 lean, obese and type 2 diabetic participants running to December 2028. And in October 2026 Novo Nordisk published the first peer-reviewed, non-Hanmi preclinical validation of the mechanism — an acylated urocortin-2 analogue tested in diet-induced obese rats and Göttingen minipigs — making this a three-company field rather than a two-company one. And the field is proceeding against a specific piece of published evidence pointing the other way: Chen and colleagues' 2006 PNAS finding that urocortin-2 knockout improves insulin sensitivity and produces less fat with more lean mass on a high-fat diet, with a CRFR2 antagonist reproducing the glucose phenotype. Thirty years after urocortin-1 was cloned, whether CRFR2 should be switched on or off for metabolic benefit is still an open question — and two companies have now bet substantial capital on the answer being 'on.'

How It Works

Most obesity drugs on the market copy gut hormones like GLP-1 that tell your brain you are full. They work, but a meaningful share of the weight people lose on them is muscle rather than fat, which matters more the older or frailer someone is. HM17321 goes after a different target entirely. It is a modified version of urocortin-2, a natural body signal that acts on a receptor called CRFR2. That receptor sits on muscle, heart, and blood vessels rather than mainly on appetite circuits, and switching it on has been shown in animals to build muscle and to protect muscle from wasting. The bet is that losing weight through this pathway strips fat while keeping — or even adding — muscle. In mice it appears to do both, and it also reduces how much the animals eat. Whether any of that holds in people is unknown; the first human trial is still running.

The corticotropin-releasing factor family comprises four mammalian ligands acting on two class B G-protein-coupled receptors. CRF itself and urocortin-1 bind both CRFR1 and CRFR2. Urocortin-2 (also called stresscopin-related peptide) and urocortin-3 (stresscopin) are CRFR2-selective, with essentially no activity at CRFR1 — a selectivity established by Reyes and colleagues in PNAS in 2001 and independently by Hsu and Hsueh in Nature Medicine the same year. HM17321 is a synthetic analog built on the urocortin-2 scaffold and engineered for CRFR2 selectivity plus a half-life supporting once-weekly subcutaneous dosing; Hanmi has described the molecule as discovered using AI and structural modelling but has not disclosed the sequence, the modifications, or the half-life-extension chemistry. The receptor split is the entire therapeutic premise. CRFR1 mediates hypothalamic-pituitary-adrenal activation, ACTH and cortisol release, and anxiogenesis — the classical stress response, and an unacceptable liability in a chronically dosed metabolic drug. CRFR2 has a largely peripheral distribution: skeletal muscle, cardiac myocytes, vascular smooth muscle, and gastrointestinal tissue, alongside discrete central sites. Both receptors couple principally to Gs, raising cAMP and activating PKA, but the anatomical separation means a CRFR2-selective agonist can in principle be given long-term without driving the stress axis. CRHR2-null mice are anxious and stress-hypersensitive (Bale et al., Nature Genetics 2000), indicating that tonic CRFR2 signalling is physiologically meaningful rather than vestigial. The skeletal muscle biology is the differentiating claim and it predates the obesity application by two decades. Hinkle and colleagues demonstrated in 2003, in two companion papers, that CRFR2 activation increases muscle mass and function in non-atrophying skeletal muscle and reduces the loss of muscle mass and function during atrophy induced by nerve damage, corticosteroids, and disuse. Isfort and colleagues developed CRFR2-selective sauvagine analogues explicitly for skeletal muscle atrophy in 2005, and Hall and colleagues tested a CRFR2 agonist alongside exercise in mdx and wild-type muscle in 2007 — so the target has prior industry history that did not reach approval, which is worth knowing before treating the current programs as novel. A 2025 review by Vidal and colleagues in Skeletal Muscle synthesises the contemporary position: urocortin treatment produces decreased food intake, muscle hypertrophy, and improved skeletal muscle glucose uptake, but the molecular mechanisms remain to be elucidated. Hanmi's preclinical package, presented as conference abstracts, reports dose-dependent weight loss with improved body composition across obesity models, effects in sarcopenic mice, additive effects in combination with myostatin/activin inhibition and with amylin analogs, cardiorenal effects in spontaneously hypertensive rats, and cardioprotection in rodent and non-human-primate heart failure models. The EASD 2025 poster compared HM17321 at 100 nmol/kg every other day against semaglutide at 20 nmol/kg every other day in diet-induced-obese mice, each with pair-fed controls, and reported that HM17321 increased lean mass while reducing fat mass where semaglutide reduced both, raised absolute gastrocnemius weight, and reduced intramuscular triglyceride more than semaglutide. In vitro work showed lipolysis in 3T3-L1 and human visceral white adipocytes and increased MyoD with differentiation and hypertrophy in C2C12 and human skeletal muscle cells. Two nuances from that same poster should not be dropped: food intake was significantly reduced, and the energy-expenditure advantage over semaglutide held when adjusted for body weight but not when adjusted for lean mass, with the authors attributing it largely to the increased lean mass rather than to direct thermogenesis. The counter-hypothesis is substantial and unresolved. Chen and colleagues reported in PNAS in 2006 that urocortin-2-knockout mice show increased insulin sensitivity, protection from fat-induced insulin resistance, and decreased fat with increased lean mass on a high-fat diet; that a CRFR2-selective antagonist reproduced the knockout glucose phenotype in wild-type mice; and that urocortin-2 inhibited insulin-induced Akt and ERK1/2 phosphorylation in myotubes. Their conclusion was that suppression of the urocortin-2/CRFR2 pathway may provide benefit in insulin-resistant states — the opposite direction from HM17321's design. Chao and colleagues extended the mechanism through mTOR in 2015. HM17321, Gubra's GUB-UCN2 and now Novo Nordisk's acylated urocortin-2 are all betting against this literature, and the clinical programs will effectively adjudicate it. The balance of evidence shifted somewhat in October 2026: Novo's rats showed improved glucose tolerance and its minipigs improved insulin sensitivity under CRF2 agonism, which is the opposite direction from the knockout work and the first substantial pharmacological counterweight to it. The contradiction is now genuinely two-sided rather than one-sided, but it is not resolved — knockout and pharmacological agonism are not mirror experiments, and no human glycemic data exist for any CRF2 agonist.

Evidence Snapshot

Research Gaps & Open Questions

What the current literature has not yet settled about HM17321:

• 01Any human data at all — no efficacy, safety, tolerability, or pharmacokinetic results have been published, and the Phase 1 trial's estimated primary completion is March 2027.
• 02Whether the rodent body-composition advantage translates to humans. Mouse gastrocnemius weight and DEXA-measured human lean mass with functional testing are not the same endpoint, and the history of muscle-anabolic agents failing to convert lean-mass gains into functional benefit is long.
• 03The direction of effect on insulin sensitivity — directly disputed. Chen et al. (PNAS 2006) found urocortin-2 knockout improves insulin sensitivity and produces less fat with more lean mass on a high-fat diet, with a CRFR2 antagonist reproducing the glucose phenotype, and Chao et al. (2015) supported this via mTOR. HM17321 bets the opposite way. This is the most consequential unresolved question in the programme.
• 04Whether chronic weekly CRFR2 agonism produces clinically limiting hypotension. The pooled infusion data show a meaningful blood-pressure fall, and the apparent absence of tachyphylaxis means tolerance may not develop — an unusual and underappreciated liability profile for a chronically dosed drug.
• 05CRFR2 versus CRFR1 selectivity — no EC50, Ki, or selectivity ratio has been published, so the central safety argument for avoiding cortisol elevation is unquantified.
• 06How much of the effect is appetite-mediated versus directly anabolic. Hanmi's own poster shows significantly reduced food intake alongside a weight reduction exceeding the pair-fed control, and the energy-expenditure advantage disappeared on lean-mass adjustment — so the mechanistic decomposition is unsettled even in mice. Novo Nordisk's 2026 rat work reached the same place independently, reporting flatly that energy expenditure was not affected.
• 07Peptide sequence, modifications, and half-life-extension chemistry are all undisclosed, which limits independent assessment of the molecule.
• 08Competitive position in what is now a three-company field — Hanmi, Gubra (GUB-UCN2, NCT07702890) and Novo Nordisk, whose acylated urocortin-2 published preclinically in October 2026. Hanmi started first and will finish Phase 1 first, but Gubra's broader Phase 1/2a carries the registered muscle and body-composition endpoints that Hanmi's PK-focused Phase 1 does not, so the first efficacy-grade human read on the lean-mass thesis may not come from HM17321 at all.
• 09Whether the cardiovascular and cardiorenal preclinical signals represent a genuine additional benefit or simply the vasodilatory effect viewed favourably. The Christchurch and UNICORN human infusion studies showed real hemodynamic improvement in heart failure but were never carried to outcomes.

Forms & Administration

HM17321 is designed as a once-weekly subcutaneous injection and is administered in the Phase 1 trial as single and multiple ascending subcutaneous doses under sequential dose-escalation with safety review between cohorts. No dose levels, pharmacokinetic parameters, or formulation details have been publicly disclosed. There is no approved form, no commercially available product, and no compounding pathway. The only legitimate access is enrollment in Hanmi's registered Phase 1 trial (NCT07219589) at the Medpace Clinical Pharmacology Unit in Cincinnati, Ohio. Any material sold as HM17321 outside that trial is unverified research chemical of unknown identity and potency, and because no human dose or pharmacokinetic data exist, no defensible self-administration protocol could be constructed even by someone willing to accept the risk.

Common Questions

Who HM17321 Is NOT For

Drug & Supplement Interactions

No human drug-interaction data exist, and none will until the Phase 1 programme reports. Everything below is inferred from urocortin-2 pharmacology and should be read as anticipated rather than documented. The interaction that matters most is hemodynamic. Urocortin-2 is a potent vasodilator, and pooled human infusion data show mean arterial pressure falling roughly 9 mmHg with heart rate rising roughly 6 bpm. Combining a CRFR2 agonist with antihypertensives, nitrates, hydralazine, alpha-blockers, PDE5 inhibitors, or other vasodilators would predictably be additive. Because obesity populations have high background rates of treated hypertension, this is likely to be a central question in Phase 2 rather than an edge case. Beta-blockers might attenuate the reflex tachycardia without blunting the vasodilation. Diuretics and renin-angiotensin-aldosterone-system agents target the same neurohormonal axis that urocortin infusion suppresses, and the UNICORN trial observed a transient rise in plasma renin activity coinciding with blood-pressure falls plus reduced urine volume and creatinine clearance during infusion. Whether that matters for weekly subcutaneous dosing is unknown. If HM17321 is combined with incretin or amylin agents — which Hanmi has explored preclinically and which is the most likely clinical positioning — the interaction surface widens to include additive appetite suppression and gastrointestinal effects, and additive hypotension given that GLP-1 agonists also modestly lower blood pressure. Insulin and insulin secretagogues are worth flagging because the direction of CRFR2's effect on insulin sensitivity is genuinely disputed in the preclinical literature, so hypoglycemia risk in combination cannot be predicted with confidence in either direction. Glucocorticoid interactions would become relevant if CRFR1 selectivity proves imperfect, since CRFR1 engagement raises ACTH and cortisol. No selectivity ratio has been published, so this remains an open question rather than a quantified risk.

Safety Profile

Legal Status

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

Myths & Misconceptions