GDF15

GDF15 (growth differentiation factor 15) is a stress-responsive secreted hormone and the strongest single-protein predictor of all-cause mortality currently known. It is one of seven plasma-protein DNAm surrogates embedded in the grimage-2019 clock, a leading peripheral biomarker of mitochondrial stress, and the critical mediator of metformin’s weight-lowering (but not glucose-lowering) effects. Its receptor — GFRAL, expressed exclusively in the area postrema and nucleus tractus solitarius of the brainstem — was identified simultaneously by four independent groups in 2017, establishing GDF15 as a hormonally active ligand rather than a locally acting cytokine.

Identity

• UniProt: Q99988 (GDF15_HUMAN); Swiss-Prot curated
• NCBI Gene: 9518
• HGNC: 30142; approved symbol GDF15
• Ensembl: ENSG00000130513
• Mouse ortholog: Gdf15 (one-to-one)
• GenAge-models ID: 23886 — Gdf15, Mus musculus, pro-longevity effect; knockout/overexpression models show +43% mean lifespan (mouse); evidence from single study
• Chromosomal location: 19p13.11 (human)

Multiple-discovery aliases

GDF15 was independently discovered and named multiple times before its identity as a single protein was established:

Alias	Discovery context	Year
MIC-1 (macrophage inhibitory cytokine 1)	Macrophage-derived inhibitory cytokine found in myeloid cells	Bootcov 1997 1
NAG-1 (NSAID-activated gene 1)	Induced by NSAIDs and PPARγ ligands	~2000
PLAB (placental TGF-β)	High expression in placenta	~2000
GDF15	TGF-β superfamily annotation	HGNC current canonical name

The unified name GDF15 is now standard. The alias MIC-1 persists in much older literature.

Structure

GDF15 belongs to the TGF-β superfamily but is a distant outlier — only ~25% sequence identity to other family members ¹. This divergence is reflected in its unique receptor (GFRAL) and distinct signaling biology.

• Precursor: 308 amino acids; contains signal peptide (aa 1–29) and pro-peptide (aa 30–194)
• Mature chain: 114 amino acids (aa 195–308), produced by furin-type protease cleavage at Arg194
• Secreted form: disulfide-linked homodimer (the bioactive form in plasma)
• Cystine-knot architecture: six conserved cysteines form the characteristic TGF-β cystine knot; an additional interchain disulfide (Cys273) links the two monomers
• PTMs: N-linked glycosylation at Asn70 (in the pro-peptide)

Receptor and signaling: GFRAL/RET complex

In 2017, four independent groups simultaneously identified GFRAL (GDNF family receptor alpha-like) as the cognate receptor for GDF15 ²³⁴⁵. This represented one of the fastest convergent-discovery events in recent receptor biology.

Critical anatomical constraint: GFRAL mRNA and protein are expressed almost exclusively in the area postrema (AP) and nucleus tractus solitarius (NTS) of the hindbrain — brainstem regions accessible to circulating signals due to the absence of a normal blood-brain barrier ³. GFRAL is essentially absent from peripheral tissues, heart, liver, kidney, adipose, and skeletal muscle. This means GDF15 acts as a classical endocrine hormone — synthesized peripherally (liver, kidney, epithelium) and detected centrally.

Signal transduction:

1. GDF15 homodimer binds GFRAL extracellular domain
2. GFRAL recruits RET (the receptor tyrosine kinase co-receptor used by GDNF-family ligands) ²
3. GFRAL–RET complex signals via RET kinase activity → downstream ERK, AKT, PLC-γ
4. AP/NTS neurons relay the signal via vagal afferents → reduced appetite, nausea, emesis

GFRAL-null and GDF15-null mice are fully insensitive to the body-weight and food-intake effects of exogenous GDF15 ⁵², confirming GFRAL is non-redundant for these responses.

Function

Appetite suppression and aversion

GDF15 is, on a per-molecule basis, among the most potent known appetite suppressants. It drives anorexia and nausea through AP/NTS GFRAL signaling, activating circuits overlapping with malaise and emesis pathways. This mechanistic basis explains:

• Cancer cachexia: tumor-derived GDF15 suppresses appetite and drives wasting needs-replication (mechanistic dissection in humans)
• Pregnancy nausea: elevated fetal GDF15 (from placenta) is the leading candidate mechanism for hyperemesis gravidarum — variants impairing GDF15 secretion protect against HG
• Metformin GI intolerance: intestinal GDF15 induction by metformin may underlie GI side effects at therapeutic doses

Stress response

GDF15 is induced robustly by:

• Mitochondrial dysfunction and the mitochondrial unfolded protein response (UPRmt) — mtDNA-encoded OXPHOS disorders (MELAS, KSS, Pearson syndrome) show dramatically elevated serum GDF15 ⁶ needs-replication
• Integrated stress response (ISR) / eIF2α-ATF4 axis — cytosolic and ER stress converge on GDF15 transcription via ATF4 binding sites in the GDF15 promoter
• Inflammation (NF-κB), hypoxia (HIF-1α), oxidative stress, p53 activation, and NSAID/PPAR-γ ligand treatment

The resulting secreted GDF15 circulates as a sentinel signal of systemic stress, detectable in plasma at pg/mL concentrations.

Mortality biomarker

GDF15 is one of the most robustly validated predictors of all-cause mortality in elderly populations, consistently outperforming conventional clinical risk factors in several large prospective cohort studies.

Wiklund 2010 — in a Swedish population-based cohort, serum MIC-1/GDF15 was an independent predictor of all-cause mortality across multiple age strata; higher circulating GDF15 associated with substantially increased mortality risk, independent of BMI and comorbidities ⁷. needs-replication — this specific cohort paper needs independent replication across diverse populations, though subsequent cohort evidence broadly consistent.

GrimAge component: GDF15 is one of the seven plasma-protein DNAm surrogates incorporated into the grimage-2019 clock ⁸, which is currently the strongest single-composite DNA methylation predictor of all-cause mortality. The fact that a DNAm-based proxy of GDF15 is a top mortality predictor reinforces GDF15’s biological relevance to mortality across the life course.

Epidemiology of age-trajectory: circulating GDF15 increases steeply with chronological age — rising roughly 2–3-fold between ages 40 and 80 in population cross-sections. Whether this rise is mechanistically involved in aging trajectories or is a bystander reflection of accumulating disease burden and mitochondrial dysfunction is the central open question in GDF15 aging biology. no-mechanism

Dimension	Status
GDF15 elevation correlates with mortality in humans?	yes (multiple large cohorts)
GDF15 elevation is causally upstream of aging?	unknown; observational only
Mendelian randomization evidence?	limited; not well-powered MR instruments yet

Metformin connection

Coll et al. 2020 (Nature) demonstrated that metformin — at therapeutic doses and over relevant timescales — induces GDF15 expression predominantly in enterocytes of the distal small intestine and colon, and in periglomerular renal tubular cells ⁹. Expression was assessed by Gdf15 mRNA in a tissue panel and confirmed by in situ hybridisation; the liver, while capable of responding to biguanides in cell-based assays, was not the dominant in vivo source under oral dosing. This finding was established in:

• Gdf15-null mice: insensitive to metformin-induced weight loss on high-fat diet
• Gfral-null mice: same phenotype
• GFRAL antagonist antibody co-treatment: reverses metformin weight loss in obese wild-type mice
• Human CAMERA trial sub-study: plasma GDF15 rises with metformin dose

Dissociation of mechanisms: Metformin’s glucose-lowering and insulin-sensitizing effects are fully preserved in Gdf15-null mice — GDF15 is therefore necessary for weight-lowering but GDF15-independent glucose lowering is confirmed ⁹. This mechanistic dissociation is clinically relevant: therapeutic weight loss from metformin in T2D is partly GDF15-mediated, but the antidiabetic benefit does not require GDF15.

This finding also suggests that at least part of metformin’s GI intolerance (nausea, diarrhea) may be GDF15/GFRAL-mediated, a hypothesis still under investigation.

Mitochondrial stress biomarker

GDF15 is currently the strongest peripheral biomarker of mitochondrial dysfunction available in clinical settings. Serum GDF15 is markedly elevated in:

• Primary mitochondrial disorders caused by mtDNA mutations (MELAS, Pearson, Kearns-Sayre) ⁶
• Patients with mitochondria-toxic drug exposures (nucleoside reverse transcriptase inhibitors)
• States of high mitochondrial stress (critical illness, advanced heart failure, sepsis)

The mechanism is via UPRmt → ATF4 → GDF15 transcription in mitochondrially-stressed cells. GDF15 therefore functions as a systemic reporter of intracellular mitochondrial stress — a peripheral signal visible in a blood test. In the context of aging, rising GDF15 with age is consistent with the known accumulation of mitochondrial dysfunction as a mitochondrial-dysfunction hallmark.

no-mechanism — whether GDF15 induction in response to mitochondrial stress is adaptive (a protective hormetic signal) or maladaptive (driving cachexia and anorexia as collateral damage) is unresolved.

Aging trajectory

GDF15 rises with age more consistently than almost any other circulating protein. It is among the most reliably age-upregulated analytes across multiple proteomics aging atlases ⁷. unsourced — a confirmed-DOI proteomics atlas citation showing GDF15 age trajectory should be added on verification pass (Lehallier 2019 proteomics clock or similar is likely). The aging-related rise appears to reflect:

1. Accumulating mitochondrial dysfunction in tissues → UPRmt → GDF15 secretion
2. Increasing burden of low-grade chronic disease
3. Possible direct contribution of the chronic-inflammation hallmark (NF-κB drives GDF15)

Whether elevated GDF15 in elderly individuals is mechanistically harmful (driving muscle wasting, anorexia, and cachexia via GFRAL signaling) or simply a bystander biomarker is a central unsettled question. The GenAge entry (mouse, +43% lifespan from Gdf15 manipulation) suggests modulating GDF15 levels can influence lifespan, but the directionality (knockout vs overexpression) and biological mechanism in that mouse model require independent confirmation. needs-replication

Therapeutic landscape

Anti-GDF15 antibodies (cancer cachexia)

Blocking GDF15 is a rational strategy for cancer cachexia and anorexia of aging — restoring appetite by interrupting GFRAL signaling. Two antibodies are in advanced development:

• Ponsegromab (Pfizer): anti-GDF15 monoclonal antibody; Phase 2 results published in NEJM 2025 (doi:10.1056/nejmc2500502) showing weight gain in cancer cachexia patients with elevated GDF15; Phase 3 optimization underway (2026 modeling papers). long-term-unknown
• NGM120 (NGM Bio): anti-GFRAL monoclonal antibody (targets receptor rather than ligand); Phase 1/2 data reported; less advanced than ponsegromab

Clinical target: patients with cancer cachexia, elevated baseline GDF15, and weight loss — the GDF15-high subgroup may be the most responsive.

GDF15 agonism (obesity, metabolic disease)

Since GDF15 → GFRAL is among the most potent appetite-suppression circuits identified, GDF15 agonists are being developed as anti-obesity agents. This represents the opposite therapeutic direction from the cachexia application: recombinant GDF15 analogs or GDF15 mimetics that activate GFRAL to reduce body weight.

• Phase 1/2 trials are ongoing with GDF15 analogs for obesity/metabolic disease (multiple sponsors)
• The challenge is GDF15’s nausea-inducing effects at high doses; titration strategies and tolerance are under investigation
• Receptor selectivity (GFRAL vs any off-target) appears favorable given GFRAL’s narrow brainstem expression

Aging and sarcopenia

Whether blocking GDF15 in elderly individuals with elevated baseline GDF15 and sarcopenia/anorexia would improve outcomes (muscle mass, physical function, survival) is an active research question. The dual signal — GDF15 as mortality predictor AND GDF15 as driver of anorexia/wasting — makes GDF15 a pharmacologically prioritized target in aging biology. No aging-specific GDF15 trial results have been published as of 2026-05-05. long-term-unknown needs-human-replication

Limitations and gaps

1. Causal vs. biomarker ambiguity — GDF15 is robustly elevated with aging and disease, but whether it drives outcomes or is a passive readout is unresolved. MR instruments for GDF15 are not yet well-established. no-mechanism

2. GenAge mouse data context — The +43% lifespan claim (GenAge 23886) requires primary source verification (study design, genetic manipulation type, direction of effect). needs-replication needs-canonical-id — exact primary paper citation for the GenAge mouse entry not yet confirmed.

3. GFRAL antagonism and aging — Whether restoring appetite in GDF15-high elderly individuals improves clinical outcomes is unknown; trials are in cancer cachexia, not aging per se. needs-human-replication

4. Sex differences — GDF15 dynamics during pregnancy (extremely high, due to placental production) complicate cross-sex comparisons of aging trajectories. Whether female baseline GDF15 aging trajectories differ from male requires careful cohort analysis. dose-response-unclear

5. Tissue-of-origin heterogeneity — Liver, kidney, intestine, placenta, macrophages, and tumors all secrete GDF15; the contribution of each compartment to circulating levels in aging is unclear. no-mechanism

6. Druggability tier note — Tier 1 assigned based on ponsegromab clinical-stage program (anti-GDF15 mAb in Phase 2/3); this reflects the ligand as target. GFRAL is the receptor target for NGM120. Open Targets API was unavailable during seeding; druggability tier should be verified against current Open Targets platform data. needs-canonical-id

Cross-references

• grimage-2019 — GDF15 is one of 7 plasma-protein surrogates in GrimAge (verified page); the DNAm-GDF15 surrogate is a top GrimAge component
• metformin — GDF15 mediates metformin’s weight-lowering effect (Coll 2020); verified-partial on metformin.md
• mitochondrial-dysfunction — GDF15 as peripheral UPRmt reporter hallmark
• chronic-inflammation — NF-κB induces GDF15; GDF15 rise contributes to chronic inflammation signaling
• deregulated-nutrient-sensing — GFRAL/RET → appetite suppression; mechanistic link to systemic nutrient sensing
• tgf-beta — parent superfamily pathway (implicit stub)
• gfral — cognate receptor; brainstem-restricted (implicit stub)
• ret — co-receptor required for GFRAL signaling (implicit stub)
• integrated-stress-response — ATF4-driven GDF15 transcription (implicit stub)
• mus-musculus — GenAge pro-longevity data in mouse

Footnotes

Footnotes

1. doi:10.1073/pnas.94.21.11514 · Bootcov MR et al. · PNAS 94(21):11514–11519, 1997 · in-vitro + secretion assays · model: U937 myelomonocytic and primary macrophage cell lines · first characterization of MIC-1/GDF15 as TGF-β superfamily member; 308-aa precursor, 15–29% identity to other TGF-β superfamily members (distant outlier); citation count ~1,128 (100th percentile in archive); local PDF available (PMC23523) ↩ ↩²

2. doi:10.1038/nm.4392 · Mullican SE et al. · Nature Medicine 23:1150–1157, 2017 · in-vivo + cell-based · model: mice (Gfral-null, WT) + nonhuman primates · established GFRAL as GDF15 receptor; GDF15 promotes weight loss via GFRAL; archive status: closed-access (not_oa) no-fulltext-access ↩ ↩² ↩³

3. doi:10.1038/nature24042 · Hsu JY et al. · Nature 550:255–259, 2017 · in-vivo · model: mice + brainstem-restricted receptor expression characterization · GFRAL expression restricted to area postrema + NTS; archive status: closed-access (not_oa) no-fulltext-access ↩ ↩²

4. doi:10.1038/nm.4393 · Emmerson PJ et al. (Eli Lilly) · Nature Medicine 23:1215–1219, 2017 · in-vivo · model: mice + cell-based binding assays · confirmed GFRAL mediates GDF15 metabolic effects; archive status: closed-access (not_oa) no-fulltext-access ↩

5. doi:10.1038/nm.4394 · Yang L et al. (Novo Nordisk) · Nature Medicine 23:1158–1166, 2017 · in-vivo · model: Gfral-null mice · GFRAL required for GDF15 anti-obesity effects; archive status: closed-access (not_oa) no-fulltext-access ↩ ↩²

6. doi:10.1002/ana.24506 · Yatsuga S et al. · Ann Neurol 78(5):814–823, 2015 · observational · model: 48 patients with mitochondrial disease (MELAS, KSS, Pearson, and other primary mtDNA disorders) vs 146 controls · GDF15 was the most discriminating biomarker for mitochondrial disease; markedly elevated serum GDF15 in MELAS, Kearns-Sayre, Pearson syndrome; archive status: not downloaded — note: the seeder attributed this to a “Suomalainen 2011” paper that does not exist as described; the Suomalainen lab’s primary contribution in this area was establishing FGF21 as a biomarker (Lehtonen et al. 2016, Neurology, doi:10.1212/WNL.0000000000003374); the landmark GDF15 paper is Yatsuga/Koga 2015 ↩ ↩²

7. doi:10.1111/j.1474-9726.2010.00629.x · Wiklund FE et al. · Aging Cell 9(6):1057–1064, 2010 · observational (prospective cohort) · model: Swedish population cohorts · MIC-1/GDF15 as independent predictor of all-cause mortality; archive status: download failed (bronze OA URL returned 403; full-text not accessible) no-fulltext-access — DOI confirmed correct via Crossref; quantitative claims (effect size, HR, cohort n) not verified against full text ↩ ↩²

8. doi:10.18632/aging.101684 · Lu AT et al. · Aging 2019 · observational (cohort, training + 5 validation cohorts, n=6,935 total) · GrimAge clock; GDF15 is one of 7 plasma-protein DNAm surrogates; composite achieves HR=1.10/yr AgeAccelGrim, meta-P=2.0E-75 · local PDF available (PMC6366976) ↩

9. coll-2019-gdf15-metformin · doi:10.1038/s41586-019-1911-y · in-vivo + human RCT sub-study · model: Gdf15-null, Gfral-null mice; human CAMERA trial (metformin dose-escalation sub-study) · Nature 578:444–448 (2020); metformin induces intestinal GDF15; GDF15/GFRAL necessary for weight-lowering, not glucose-lowering; local PDF available ↩ ↩²