FGF21

FGF21 (fibroblast growth factor 21) is an atypical, endocrine member of the FGF family secreted primarily by the liver. Unlike classical FGFs, it lacks a high-affinity heparin-binding domain and acts as a circulating hormone rather than a local paracrine signal. FGF21 is induced by diverse nutritional and metabolic stresses — fasting, protein restriction, mitochondrial dysfunction, cold — and drives coordinated metabolic adaptations including enhanced adipose glucose uptake, thermogenesis, improved insulin sensitivity, and lipid mobilization. Transgenic overexpression in mice extends median lifespan ~36%, placing FGF21 among the most potent single-gene longevity interventions identified in mammals ¹.

Identity

• UniProt: Q9NSA1 (FGF21_HUMAN) — Swiss-Prot reviewed entry
• NCBI Gene: 26291
• HGNC: 3678
• Ensembl: ENSG00000105550
• Mouse ortholog: Fgf21 (high sequence conservation)
• Length: 209 amino acids (canonical isoform, including 28-aa signal peptide; mature secreted form: residues 29–209)
• Molecular weight: ~22.3 kDa (calculated); glycosylation adds ~1–2 kDa in vivo

Domain structure

FGF21 shares the beta-trefoil fold of canonical FGFs but has two key divergences that confer its endocrine character ²:

• Reduced heparin-binding affinity — due to substitutions in the heparin-binding loop; prevents sequestration in extracellular matrix, enabling systemic distribution. Kharitonenkov 2005 confirmed that none of the FGF21-induced in vitro responses were heparin-regulated.
• Disordered C-terminal tail (residues ~143–209) — intrinsically disordered and structurally flexible. Subsequent structural work established this region as critical for binding the obligate co-receptor beta-Klotho (klotho); this was not determined in Kharitonenkov 2005 (the co-receptor mechanism remained unresolved at that time). needs-canonical-id
• Signal peptide (residues 1–28) — cleaved cotranslationally; mature protein is secreted via the classical ER-Golgi pathway.

Receptor complex and signaling

FGF21 cannot activate FGFRs alone at physiological concentrations ². Kharitonenkov 2005 showed that FGF21 induces tyrosine phosphorylation of FGFR1 and FGFR2 in differentiated 3T3-L1 adipocytes in a heparin-independent manner, but the underlying co-receptor mechanism was not resolved in that paper. Subsequent work established that productive FGF21 signaling requires a ternary complex with:

• Obligate co-receptor: Beta-Klotho (klotho) — a single-pass transmembrane protein expressed in metabolically active tissues (liver, adipose, pancreas, brain)
• Signaling receptor: FGFR1c (primary in adipose and hypothalamus), FGFR2c, or FGFR3c

The FGF21 / beta-Klotho / FGFR complex activates canonical MAPK (ERK1/2) downstream signaling ². PI3K-AKT activation is reported in subsequent pharmacological studies but was not established in Kharitonenkov 2005. needs-canonical-id

Tissue distribution of beta-Klotho directly determines FGF21 responsiveness — tissues lacking beta-Klotho expression (e.g., skeletal muscle) are effectively FGF21-insensitive despite expressing FGFRs. Kharitonenkov 2005 confirmed that FGF21 did not stimulate glucose uptake in muscle L6-GLUT4myc myoblasts/myotubes, consistent with tissue-specific co-receptor requirement.

Expression: primary sources

FGF21 is expressed in multiple tissues but hepatic secretion dominates circulating levels:

Tissue	Context	Regulatory input
Liver (primary)	Fasting, PPARα activation, ketogenic diet	PPARα → FGF21 promoter
Adipose (WAT/BAT)	Cold, PPARγ activation	PPARγ, β3-adrenergic
Skeletal muscle	Exercise, mitochondrial stress	ATF4, ISR
Pancreas	Acute amino acid depletion	GCN2-eIF2α-ATF4
Brain	Ketogenic diet, neurological stress	ATF4

Induction stimuli

1. Fasting and PPARα

The canonical induction pathway: prolonged fasting → hepatic fatty acid oxidation → PPARα activation → FGF21 transcription ³. FGF21 mRNA levels were increased ~28-fold after a 12-hour fast in wild-type mice; PPARα-null mice showed only ~5-fold induction under the same conditions, demonstrating that PPARα is required for the full fasting-induced upregulation of FGF21 but that PPARα-independent pathways also contribute. PPARα directly binds to two PPRE elements in the FGF21 promoter (ChIP confirmed, −1119 to −1044 region). FGF21 then acts in an endocrine loop to sustain the fasting metabolic state (ketogenesis, lipolysis in WAT, and torpor).

2. Dietary protein / amino acid restriction

FGF21 is the primary endocrine signal mediating the systemic effects of dietary protein restriction ⁴. The pathway runs through the GCN2 → eIF2α → ATF4/ATF5 arm of the integrated-stress-response:

Low dietary amino acids → GCN2 kinase activation (gcn2) → eIF2α phosphorylation → ATF4/ATF5 bind amino acid response elements (AAREs) in the FGF21 promoter → FGF21 induction

Laeger 2014 specifically demonstrated this using dietary protein restriction (low-protein / LP diet, ~5% casein in mice), not methionine restriction alone. FGF21 was induced in both mice and rats on LP diet and in humans after 28 days on a low-protein diet. Both GCN2 and PPARα are required: Gcn2-KO and Ppara-KO mice each showed markedly blunted LP-induced FGF21. PPARα does not appear to be transcriptionally activated during protein restriction per se, but plays a permissive/constitutive role. FGF21 induction by methionine-restriction is consistent with this axis (methionine depletion activates GCN2) but Laeger 2014 does not specifically test methionine restriction. needs-replication for methionine-specific link.

3. Mitochondrial dysfunction (“mitokine” role)

Mitochondrial stress (UPRmt, respiratory chain inhibition) induces FGF21 via ATF4 in multiple tissues ⁵. This positions FGF21 as a bona fide mitokine — a mitochondria-to-systemic-circulation signal coordinating whole-organism responses to organelle stress.

4. Other induction contexts

• Ketogenic diet — hepatic PPARα activation (overlapping with fasting pathway)
• Cold exposure — adipose PPARγ and sympathetic activation → brown adipose FGF21 → thermogenesis amplification
• Excess simple carbohydrates — liver ChREBP-mediated induction (counterintuitive; reflects metabolic stress)

Physiological effects

Effect	Tissue	Magnitude / notes
Glucose uptake ↑	Adipose (via GLUT1)	Insulin-independent; distinct from GLUT4 mechanism
Thermogenesis ↑	BAT	Enhances UCP1, promotes browning of WAT
Insulin sensitivity ↑	Liver, adipose	Improves glucose homeostasis in obesity models
Triglycerides ↓	Plasma	Robust across species; mediated by VLDL reduction
Body weight ↓	Systemic	Observed in pharmacological doses; partly food intake
Growth suppression	Systemic	FGF21 suppresses the GH/IGF-1 axis — a key aging-relevant trade-off (#gap/needs-mechanistic-detail)

FGF21 notably suppresses the GH/IGF-1 axis — transgenic mice with chronic FGF21 overexpression are dwarfed — linking it to the somatotropic signaling reduction seen in many longevity models.

Aging relevance

Lifespan extension by FGF21 overexpression

Transgenic mice with hepatic FGF21 overexpression (FGF21-Tg; apoE promoter; C57Bl/6J background) live longer than wild-type controls in a sex-stratified manner ¹. Median lifespans: males 27.9→36.2 mo (~30% extension); females 28.8→40.1 mo (~39% extension); combined-sex 28.1→38.2 mo (~36% extension); Tg n=77 vs WT n=67; HR=0.22, 95% CI 0.15–0.34, p=2.7×10⁻¹²; log-rank test. The lifespan extension was accompanied by improved insulin sensitivity (hyperinsulinemic-euglycemic clamp confirmed), reduced circulating IGF-1, and suppression of GH/IGF-1 signaling in liver — specifically attenuated hepatic JAK2/STAT5 signaling downstream of the GH receptor. Zhang 2012 explicitly ruled out AMPK activation, mTOR inhibition, and NAD+ increases as mediating mechanisms in this model (phospho-AMPK, phospho-S6, phospho-4E-BP1, and NAD+ were unchanged in Tg vs WT tissues). The proposed mechanism is JAK2/STAT5 attenuation in the GH/IGF-1 axis (NOT mTOR/AMPK/sirtuin pathways). Bone loss was observed in Tg mice (reduced trabecular bone volume, p=0.013). The beta-Klotho co-receptor requirement for this lifespan effect was not tested in Zhang 2012. needs-replication

Dimension	Status	Notes
Pathway conserved in humans?	yes	FGF21/beta-Klotho/FGFR1c axis conserved; PPARα induction conserved
Phenotype conserved in humans?	partial	Metabolic effects in human pharmacology trials; lifespan extension not testable
Replicated in humans?	no	No equivalent transgenic model possible; indirect via FGF21 analog trials

FGF21 levels in human aging — a paradox

Circulating FGF21 is elevated rather than depressed in older adults and in metabolic disease states (obesity, T2D, NAFLD). This is interpreted as FGF21 resistance — a state where target tissues become refractory to FGF21 signaling despite high ligand levels, analogous to insulin resistance ⁶. The likely mechanism: age-related decline in beta-Klotho expression in key target tissues reduces signaling competence. needs-human-replication — the resistance hypothesis is mechanistically plausible but not rigorously demonstrated in humans via interventional evidence.

FGF21 as mediator of protein restriction (and potentially CR and methionine restriction)

FGF21 is required for the normal behavioral and metabolic responses to dietary protein restriction in rodents ⁴. Fgf21-KO mice on a low-protein (LP) diet failed to show the normal increases in food intake, increased energy expenditure (~15% increase in WT on LP), and body weight/composition changes seen in wild-type mice. This positions FGF21 as a necessary downstream effector of protein restriction — not merely a correlate.

Note on methionine restriction and caloric restriction: Laeger 2014 demonstrates that protein restriction (not energy restriction alone) is the key driver of FGF21 induction. FGF21 induction during fasting and ketogenic diets is also attributable to the low-protein content of those conditions rather than energy restriction per se. The role of FGF21 as a required mediator of methionine-restriction longevity specifically is mechanistically plausible (methionine depletion activates GCN2) but not directly demonstrated in Laeger 2014. needs-replication for methionine-restriction and caloric-restriction-specific claims.

AMPK activation

FGF21 is reported to activate ampk in multiple tissues at pharmacological doses, providing a potential mechanistic bridge between nutrient-sensing hormone signaling and the canonical pro-longevity AMPK pathway ⁷. However, Zhang 2012 found that phospho-AMPK levels were NOT increased in liver, muscle, or adipose of FGF21-Tg mice compared to WT, and mTOR targets (phospho-S6, phospho-4E-BP1) and NAD+ were similarly unchanged. This suggests that AMPK/mTOR/sirtuin pathways do not mediate the lifespan extension seen in FGF21 transgenic mice, and that any AMPK activation by FGF21 may be dose- or context-dependent ¹. needs-mechanistic-detail

Therapeutic angles

FGF21 analogs — MASH / metabolic disease

Native FGF21 has poor pharmacokinetics (short half-life, proteolytic instability at the C-terminus). Second-generation long-acting analogs engineered for clinical use include:

Compound	Class	Status (MASH)
Efruxifermin (AMG-876)	Fc fusion	Phase 3 (MASH)
Pegozafermin (BIO89-100)	PEGylated FGF21 analog	Phase 2b/3 (MASH)
Pegbelfermin (BMS-986036)	PEGylated analog	Phase 2 completed

A 2024 meta-analysis of 8 RCTs (n=963) found FGF21 analog treatment significantly improved fibrosis outcomes vs placebo: ≥1 fibrosis stage improvement with no MASH worsening (RR=1.83, 95% CI 1.27–2.62) and ≥2-point NAS score improvement (RR=2.85, 95% CI 2.06–3.95), with acceptable safety profile ⁸.

Aging-specific indication potential

No FGF21 analog has been tested in aging or longevity-focused clinical trials as of 2026-05-06. The metabolic-syndrome indication pipeline (MASH, T2D, obesity) is the active translational pathway; aging-specific repurposing would likely require demonstration that FGF21 resistance can be overcome (e.g., by co-targeting beta-Klotho decline) or that supraphysiological FGF21 can bypass resistance. no-human-aging-trial

Pathway membership

• fgf-signaling — FGF21 is an atypical endocrine member; acts via FGFR1c–4c with beta-Klotho co-receptor (R25+ implicit stub)
• integrated-stress-response — downstream of GCN2-eIF2α-ATF4; major induction arm
• ppar-alpha-pathway — downstream of hepatic PPARα; fasting/ketogenic induction arm
• deregulated-nutrient-sensing — FGF21 links amino acid sensing to systemic metabolic state
• altered-intercellular-communication — FGF21 is a prototypical inter-organ signal (liver→adipose, liver→brain)

Key interactors

• klotho (beta-Klotho / KLB) — obligate co-receptor; tissue expression determines FGF21 responsiveness; age-related KLB decline hypothesized to drive FGF21 resistance
• fgf23 — sibling FGF19-subfamily endocrine FGF; uses alpha-Klotho (klotho) as co-receptor (vs FGF21’s beta-Klotho); shares the structural feature of low heparan-sulfate affinity that permits endocrine circulation; FGF15/19, FGF21, and FGF23 together comprise the three Klotho-coreceptor-requiring endocrine FGFs. Aging-axis-wise the two are complementary: FGF21 = mitokine/metabolic-stress signal; FGF23 = phosphate/bone-kidney/CV-aging signal.
• fgfr1c — primary signaling receptor in adipose and hypothalamus
• atf4 — transcription factor driving FGF21 induction under ISR and mitochondrial stress
• ppar-alpha — transcription factor driving FGF21 induction during fasting and ketogenesis
• gcn2 — upstream sensor in amino acid restriction → FGF21 axis
• ampk — downstream effector activated by FGF21 signaling

Cross-references

• methionine-restriction — FGF21 is a candidate required effector of MR-induced metabolic benefits via the GCN2-eIF2α-ATF4/ATF5 axis; direct demonstration in methionine-restriction-specific experiments not yet established (Laeger 2014 demonstrates this for general protein restriction) needs-replication
• caloric-restriction — Laeger 2014 evidence suggests FGF21 induction during CR may be primarily driven by reduced dietary protein rather than energy restriction per se; FGF21’s role in mediating CR longevity benefits is indirect needs-replication
• ketogenic-diet — strong FGF21 inducer via hepatic PPARα
• integrated-stress-response — GCN2-eIF2α-ATF4-FGF21 arm (R20)
• atf4 — R20 verified; direct transcriptional activator of FGF21 promoter
• gcn2 — R20 verified; upstream sensor in amino acid restriction
• klotho — verified-partial; co-receptor page
• deregulated-nutrient-sensing — hallmark page
• altered-intercellular-communication — hallmark page

Limitations and gaps

• FGF21 resistance mechanism in humans — the beta-Klotho-decline hypothesis is supported by mouse data and correlative human data but lacks rigorous interventional proof. needs-human-replication
• Lifespan extension translatability — the 36% lifespan extension in FGF21 transgenic mice involves chronic supraphysiological FGF21 from birth; acute pharmacological dosing in clinical trials does not approximate this. needs-human-replication
• AMPK/mTOR pathway involvement unclear — Pharmacological FGF21 studies report AMPK activation (Salminen 2016 review), but Zhang 2012 showed AMPK, mTOR targets, and NAD+ are unchanged in FGF21 transgenic mice; the lifespan extension mechanism operates independently of these canonical longevity pathways. needs-mechanistic-detail
• Bone safety signal — FGF21 overexpression causes bone loss in mice (reduced trabecular bone volume, p=0.013 in Zhang 2012); mechanism involves increased marrow adipogenesis and decreased osteoblast differentiation. Relevant for any aging indication given baseline osteopenia risk. long-term-unknown
• GH/IGF-1 suppression trade-off — chronic FGF21 elevation suppresses somatotropic axis; relevance to sarcopenia risk in older adults not characterized. needs-mechanistic-detail
• FGF21 signaling in brain aging — beta-Klotho is expressed in CNS; FGF21 crosses BBB; cognitive effects in aging models are preliminary. needs-human-replication
• Canonical ID verification — UniProt Q9NSA1, HGNC 3678, Ensembl ENSG00000105550 pulled from database lookup on 2026-05-06; recommend confirming HGNC ID against genenames.org on next lint pass. needs-canonical-id

Footnotes

Footnotes

1. doi:10.7554/eLife.00065 · Zhang Y et al. · eLife 2012 · in-vivo (mouse; C57Bl/6J; FGF21-Tg n=77, WT n=67) · randomized · log-rank p=2.7×10⁻¹² · HR=0.22 [0.15–0.34] · model: C57Bl/6J FGF21 transgenic · median lifespan sex-stratified: males 27.9→36.2 mo (~30%); females 28.8→40.1 mo (~39%); combined 28.1→38.2 mo (~36%); improved insulin sensitivity (euglycemic clamp); GH/IGF-1 axis suppression via reduced hepatic JAK2/STAT5; AMPK/mTOR/NAD+/sirtuin pathways not involved; bone loss in Tg mice; ~408 citations ↩ ↩² ↩³

2. doi:10.1172/JCI23606 · Kharitonenkov A et al. · J Clin Invest 2005 · in-vitro (3T3-L1, human primary adipocytes) + in-vivo (ob/ob, db/db mice; ZDF rats; groups ≥5/group) · pharmacological characterization of FGF21 as novel metabolic regulator; insulin-independent GLUT1-mediated glucose uptake in adipocytes (EC₅₀ ~0.5 nM); heparin-independent FGFR1/FGFR2 phosphorylation; MAPK (ERK1/2) activation; non-mitogenic; glucose and triglyceride lowering in diabetic rodents; no hypoglycemia up to 8 mg/kg/d; note: beta-Klotho co-receptor mechanism not identified in this paper; ~2053 citations ↩ ↩² ↩³

3. doi:10.1016/j.cmet.2007.05.003 · Inagaki T et al. · Cell Metabolism 2007 · in-vivo (mouse; 129S4/Sv background PPARα-null + C57Bl/6 WT; n=4–5/group) · established FGF21 as PPARα-regulated fasting hormone; ~28-fold fasting induction in WT vs ~5-fold in PPARα-null mice; PPARα binds FGF21 promoter directly (ChIP); FGF21 promotes ketogenesis, lipolysis, and torpor; ~1474 citations ↩

4. doi:10.1172/jci74915 · Laeger T et al. · J Clin Invest 2014 · in-vivo (mouse + rat; C57Bl/6; Fgf21-KO, Gcn2-KO, Ppara-KO; n=5–10/group) + human RCT subset (n=8–9) · FGF21 established as required endocrine signal of dietary protein (not energy) restriction; GCN2-eIF2α-ATF4/ATF5 + PPARα induction axis; Fgf21-KO mice on LP diet lack normal food intake, EE, and body weight responses; circulating FGF21 increased ~171% in humans on LP diet (P=0.008); ~572 citations ↩ ↩²

5. doi:10.1016/j.cellsig.2017.08.009 · Salminen A et al. · Cellular Signalling 2017 · review · ISR stimulation of FGF21 expression; mitokine role; systemic enhancer of longevity framing; ~99 citations ↩

6. doi:10.1016/j.arr.2017.05.004 · Salminen A et al. · Ageing Research Reviews 2017 · review · FGF21 regulation of longevity; FGF21 resistance in aging; interaction with energy metabolism and stress responses; ~119 citations ↩

7. doi:10.1007/s00109-016-1477-1 · Salminen A et al. · J Mol Med 2016 · review · FGF21 activation of AMPK signaling; metabolic regulation and aging; ~108 citations ↩

8. doi:10.1002/cpt.3278 · Jeong C et al. · Clin Pharmacol Ther 2024 · meta-analysis · n=963 (8 RCTs) · FGF21 analogs (efruxifermin, pegbelfermin, pegozafermin) for MASH; ≥1 fibrosis stage improvement with no MASH worsening RR=1.83 (95% CI 1.27–2.62); ≥2-point NAS improvement RR=2.85 (95% CI 2.06–3.95); acceptable safety profile; note: PDF unavailable (download failed); verified against abstract via Crossref; ~21 citations ↩