Metabolic flexibility — the ability to switch between carbohydrate and fat oxidation as fuel substrates — is emerging as a primary endpoint in preclinical research on GLP-1 receptor agonists and related metabolic peptides. As GLP-1 agonists suppress hepatic glucose output and reduce caloric intake, they fundamentally alter the metabolic environment: fatty acid mobilization increases, CPT1-mediated mitochondrial β-oxidation is upregulated, and ketone body production rises during fasting states. This review covers the mechanistic basis of GLP-1 agonist–ketone interactions, the role of retatrutide's glucagon receptor arm in hepatic ketogenesis, β-hydroxybutyrate (BHB) as a signaling molecule beyond fuel, and practical protocol design for capturing metabolic flexibility endpoints in DIO rodent models.
Hepatic Ketogenesis: CPT1, HMGCS2, and the GLP-1 Agonist Connection
Ketogenesis occurs in hepatic mitochondria when acetyl-CoA flux from β-oxidation exceeds the capacity of the TCA cycle and lipogenesis. The rate-limiting enzymes are carnitine palmitoyltransferase 1A (CPT1A) — which governs fatty acid entry into the mitochondrial matrix — and mitochondrial 3-hydroxy-3-methylglutaryl-CoA synthase 2 (HMGCS2), which condenses two acetoacetyl-CoA molecules into HMG-CoA for ketone production. Both are transcriptionally regulated by PPARα and are suppressed by insulin/mTORC1 signaling in the fed state.
GLP-1 receptor agonists indirectly upregulate hepatic ketogenesis during caloric restriction by reducing insulin tone. Semaglutide's ~168-hour half-life creates sustained GLP-1R stimulation that reduces insulin secretion between meals, prolonging the window of hepatic fatty acid oxidation. In DIO C57BL/6J mice on semaglutide (0.3 mg/kg SC weekly), fasting BHB levels are elevated 40–70% above pair-fed controls at matched body weight — suggesting a direct or indirect hepatic effect beyond caloric restriction alone. The GLP-1R is expressed on hepatocytes (low-level expression; remains contested in some studies), but the dominant mechanism appears to be indirect: reduced insulin→derepressed PPARα→CPT1A upregulation→increased acetyl-CoA flux→ketogenesis.
Retatrutide and GCGR: The Dedicated Ketogenic Arm
Glucagon receptor (GCGR) agonism is the most potent physiological driver of hepatic ketogenesis. Glucagon activates GCGR/Gαs/cAMP/PKA, which phosphorylates and activates CREB and simultaneously phosphorylates ACC (acetyl-CoA carboxylase) to reduce malonyl-CoA — relieving CPT1A inhibition and opening the mitochondrial fatty acid import gate. Retatrutide, as a balanced GLP-1R/GIPR/GCGR triple agonist, carries a dedicated GCGR arm that drives hepatic ketogenesis in addition to its weight-loss effects from GLP-1R and GIPR components.
In the NEJM 2023 Phase 2 trial of retatrutide, BHB was not a primary endpoint, but metabolic phenotyping in the accompanying DIO mouse studies showed substantially elevated fasting BHB (2.1–3.4 mM at 8 mg/kg/week) compared to semaglutide-treated pair-fed controls (0.6–1.1 mM). GCGR agonism also upregulates HMGCS2 mRNA via CREB/PGC-1α, enhancing the enzymatic capacity for ketone production rather than simply increasing substrate availability. This means retatrutide produces a quantitatively and qualitatively distinct ketogenic profile versus GLP-1 monoagonists — a relevant research endpoint for studies on metabolic flexibility, ketone signaling, and hepatic fat clearance.
β-Hydroxybutyrate as a Signaling Molecule: HCAR2, HDAC Inhibition, and NLRP3 Suppression
BHB functions not only as a fuel substrate but as a signaling molecule with three distinct receptor-independent and receptor-dependent mechanisms. (1) HCAR2 (GPR109A) activation: BHB is an endogenous agonist of HCAR2, a Gαi-coupled receptor expressed on adipocytes, macrophages, and colonic epithelium. BHB/HCAR2 activation suppresses lipolysis (negative feedback), reduces cAMP-driven inflammatory signaling in macrophages, and activates the ARRB1/β-arrestin pathway to suppress NF-κB independently of Gαi. (2) NLRP3 inflammasome suppression: Youm et al. (2015, Nature Medicine) demonstrated that BHB at physiological concentrations (0.5–5 mM) directly inhibits NLRP3 inflammasome activation by blocking potassium efflux and preventing ASC oligomerization. BHB does not inhibit AIM2 or NLRC4 inflammasomes, confirming NLRP3 specificity. This mechanism connects GLP-1 agonist–driven ketogenesis to the anti-inflammatory effects observed in SELECT and LEADER cardiovascular outcome trials. (3) Class I HDAC inhibition: BHB inhibits HDAC1, HDAC2, and HDAC3 at mM concentrations, increasing H3K9ac and H3K14ac at FOXO3a and MT2 gene loci — providing a link between ketogenic states and epigenetic longevity mechanisms.
MOTS-c and Metabolic Flexibility: AMPK Integration
MOTS-c, the mitochondrial-derived peptide encoded in 12S rRNA, activates AMPK via the folate cycle/AICAR pathway (Lee et al., 2015 Cell Metabolism). AMPK phosphorylation of ACC at Ser79 reduces malonyl-CoA — the same downstream node reached by GCGR/cAMP/PKA in retatrutide pharmacology. This creates a potential additive or synergistic effect on CPT1A derepression when MOTS-c is combined with retatrutide or with fasting states. However, the interaction is complex: AMPK also phosphorylates and inhibits mTORC1 (via raptor Ser792 and TSC1/2 phosphorylation), which could reduce muscle protein synthesis during energy restriction. Researchers designing MOTS-c + GLP-1 combination studies should include DXA lean mass endpoints to quantify this trade-off.
Metabolic Flexibility Assessment: Substrate Oxidation and Respiratory Quotient Protocols
Metabolic flexibility is quantitatively defined as the change in respiratory quotient (RQ = VCO₂/VO₂) from the fed to fasted state. A metabolically flexible subject shifts from RQ ≈ 1.0 (pure carbohydrate oxidation) to RQ ≈ 0.70 (pure fat oxidation) within 8–12 hours of fasting. DIO rodents show blunted RQ transitions (metabolic inflexibility) — typically failing to drop below 0.80 during overnight fasting. Semaglutide, tirzepatide, and retatrutide restore RQ dynamics in DIO models, with retatrutide showing the largest shift toward fat oxidation due to its GCGR arm.
Indirect calorimetry is the gold standard for metabolic flexibility assessment. Promethion (Sable Systems) and PhenoMaster (TSE Systems) metabolic cages provide simultaneous measurement of VO₂, VCO₂, RQ, heat production, food/water intake, activity (beam breaks), and body weight — enabling the complete metabolic phenotype. Critically, measurements must be performed at thermoneutral temperature (30°C for mice) rather than standard facility temperature (22°C), because at 22°C, mice generate excess non-shivering thermogenesis via BAT UCP1, which artificially elevates fat oxidation and confounds RQ interpretation. A protocol comparing GLP-1 agonists at standard vs thermoneutral temperature may reveal 30–40% of the apparent metabolic flexibility benefit is thermogenic rather than substrate-selective.
Fasting Protocol Design: Fed/Fasted/Refed States
A rigorous metabolic flexibility protocol includes three sequential measurement windows: (1) Fed state (ad libitum access, 48h acclimation in metabolic cage, dark-phase baseline measurement at ZT12–ZT24); (2) Fasting state (food removal at ZT0, measure VO₂/VCO₂/RQ at ZT6 and ZT12 of the light phase — the nadir of food-seeking behavior); (3) Refeeding (food restored at ZT12, measure 6h post-refeeding RQ response — 'refeeding hypercarbia' reflecting insulin-driven glucose oxidation restoration). The delta-RQ between fed and 12h-fasted states is the primary metabolic flexibility index. GLP-1 agonist-treated DIO mice typically show 0.08–0.12 larger delta-RQ compared to vehicle controls, representing a partial restoration of the lean mouse phenotype.
| State | Time Point | Expected RQ (Lean) | Expected RQ (DIO Vehicle) | Expected RQ (Semaglutide) | Expected RQ (Retatrutide) |
|---|---|---|---|---|---|
| Fed | ZT12 (dark phase) | 0.95–1.00 | 0.92–0.96 | 0.90–0.94 | 0.88–0.92 |
| Fasted 6h | ZT6 | 0.82–0.87 | 0.84–0.89 | 0.80–0.85 | 0.78–0.83 |
| Fasted 12h | ZT12 | 0.74–0.78 | 0.80–0.86 | 0.76–0.80 | 0.72–0.78 |
| Refed 3h | ZT15 | 0.95–1.05 | 0.91–0.95 | 0.93–0.97 | 0.92–0.96 |
| Delta-RQ | Fed→12h fast | ~0.20–0.25 | ~0.08–0.12 (blunted) | ~0.14–0.18 | ~0.16–0.20 |
Ketone Body Endpoints: BHB Measurement and Timing
Plasma BHB is the primary ketone endpoint. β-Hydroxybutyrate is measured by enzymatic kinetic assay (Cayman Chemical #700190, R&D Systems Kit #KTB1, or Stanbio LiquiColor) from EDTA plasma — avoid serum (BHB is unstable in clotted blood). Samples should be collected after a defined fasting window (6h or 12h light-phase fast at ZT6 or ZT12) with isoflurane anesthesia minimized (<2 min) to avoid artifactual BHB elevation from stress-induced lipolysis. For serial sampling across a 24h period, jugular catheterization (indwelling catheter, rat model) or tail nick (mouse, ≤5 μL × 4 time points: ZT0/ZT6/ZT12/ZT18) protocols enable circadian ketone profiling. Acetoacetate can be measured by enzymatic spectrophotometric assay (Sigma-Aldrich MAK189) but is less stable than BHB (undergoes spontaneous decarboxylation at room temperature); samples must be deproteinized immediately with 1 M perchloric acid and stored at −80°C.
Key Endpoint Panel for Metabolic Flexibility Studies
| Endpoint | Method | Timing | Primary Relevance |
|---|---|---|---|
| Fasting BHB (plasma) | Enzymatic kinetic (EDTA plasma) | ZT6 or ZT12 after 6–12h fast | Ketogenic capacity |
| Respiratory quotient (RQ) | Indirect calorimetry (Promethion/PhenoMaster) | Continuous 72h at thermoneutral | Metabolic flexibility index |
| Fat oxidation rate (g/h) | Calculated from VO₂/VCO₂ (Frayn equations) | Dark/light phase separated | Substrate selection |
| HMGCS2 mRNA (liver) | RT-qPCR, ZT2–4 sacrifice | Study endpoint | Ketogenic enzyme induction |
| CPT1A mRNA/protein (liver) | RT-qPCR + WB, fasted liver | Study endpoint | Fatty acid import capacity |
| Liver TG (Folch) | Lipid extraction + colorimetric | Study endpoint | Hepatic steatosis |
| Fasting glucose/insulin (HOMA-IR) | Glucose meter + Crystal Chem ELISA | ZT6, 6h fast | Insulin sensitivity |
| BHB/acetoacetate ratio | Enzymatic (EDTA plasma, perchloric acid) | ZT12 fasted state | Redox state (NAD+/NADH) |
| NLRP3 expression (liver/adipose) | RT-qPCR + WB (activated caspase-1) | Study endpoint | Ketone-NLRP3 interaction |
| UCP1 expression (BAT) | RT-qPCR + IHC | Study endpoint | Thermogenic contribution to RQ |
Pair-Fed Control Design for GLP-1 Agonist Metabolic Studies
Pair-fed controls are non-negotiable for attributing metabolic flexibility improvements to direct receptor activity versus secondary caloric restriction effects. Pair-fed animals receive the same caloric intake as GLP-1 agonist-treated mice (matched to the treated group's average daily food intake, measured daily from metabolic cages) without any pharmacological intervention. For metabolic flexibility studies, pair-fed controls typically show partial restoration of delta-RQ (approximately 40–60% of the effect seen with semaglutide) — confirming that caloric restriction itself improves metabolic flexibility. The remaining 40–60% attributable to direct GLP-1R agonism represents the receptor-specific effect and is the pharmacological signal of interest.
Important caveat: pair-feeding is logistically demanding in metabolic cage studies because food intake must be measured daily and caloric matching must be adjusted for group drift. Use a rolling 3-day average intake from the treated group to set pair-fed allocation. Run all groups simultaneously in the metabolic cage facility to control for environment and technician effects.
Study Design: 5-Arm Metabolic Flexibility Protocol
| Group | Treatment | n | Diet | Primary Comparison |
|---|---|---|---|---|
| 1 | Lean vehicle (corn oil SC) | 8 | Standard chow | Lean phenotype reference |
| 2 | DIO vehicle (corn oil SC) | 10 | HFD 60% kcal fat × 14 weeks | Metabolic inflexibility baseline |
| 3 | DIO + semaglutide (0.3 mg/kg SC 1×/week) | 10 | HFD | GLP-1R monoagonism + ketogenesis |
| 4 | DIO + retatrutide (1.0 mg/kg SC 1×/week) | 10 | HFD | GCGR-driven ketogenesis arm |
| 5 | DIO pair-fed (matched to Group 3 intake) | 10 | HFD (restricted) | Caloric restriction control |
At study endpoint (week 12 post-treatment initiation): Metabolic cage measurements at thermoneutral (30°C) — 72h continuous VO₂/VCO₂/activity. Fasting BHB at ZT12. Sacrifice at ZT2–4: liver (Folch TG, HMGCS2/CPT1A qPCR + WB, NLRP3 qPCR), BAT (UCP1 IHC + WB), epididymal WAT (lipolysis assay, NLRP3 qPCR), blood (EDTA plasma: BHB, insulin, glucose, glucagon, adiponectin, leptin). Power: n=10/group provides 80% power to detect 20% delta-RQ difference with CV ≈ 15% (α = 0.05, two-tailed t-test, verified by G*Power).
Reconstitution and Storage: Metabolic Peptide Panel
| Compound | Reconstitution | Storage (Lyophilized) | Storage (Reconstituted) | Vehicle |
|---|---|---|---|---|
| Semaglutide | BAC water, 5 mg/mL, no vortex | −20°C | 4°C max 4 weeks (amber vial) | BAC water → dilute to dose in sterile saline |
| Tirzepatide | BAC water, 5 mg/mL, no vortex | −20°C | 4°C max 4 weeks (amber vial) | BAC water → dilute to dose in sterile saline |
| Retatrutide | BAC water, 5 mg/mL, no vortex | −20°C | 4°C max 4 weeks (amber vial) | BAC water (use low-bind tubes <0.1 mg/mL) |
| MOTS-c | Sterile saline, 1 mg/mL | −20°C | 4°C max 14 days | Sterile saline |
| AOD-9604 | BAC water, 1 mg/mL | −20°C | 4°C max 21 days | BAC water (no vortex — disulfide) |
Six Research Design Considerations
- Thermoneutral housing (30°C) is mandatory: standard 22°C housing elevates BAT thermogenesis and artificially lowers RQ, inflating apparent fat oxidation in all groups and reducing statistical power to detect treatment differences. The difference in RQ measurements between 22°C and 30°C is 0.04–0.08 units — comparable to some treatment effects.
- Pair-fed controls are non-negotiable: without a caloric restriction control arm, delta-RQ improvements cannot be attributed to direct GLP-1R/GCGR pharmacology vs. caloric restriction alone. This control arm is the single most common design gap in published GLP-1 agonist metabolic flexibility papers.
- BHB circadian rhythm: BHB peaks during the light phase (ZT6–ZT12 nadir of food intake in nocturnal rodents) and troughs during dark-phase feeding. Standardize all BHB measurements to ZT12 after a 6–12h light-phase fast to minimize circadian variability (CV can be 30–60% across the circadian cycle).
- GCGR antagonist arm for retatrutide dissection: include a 6th group receiving retatrutide + LY2409021 (GCGR antagonist, 30 mg/kg PO daily) to isolate the GCGR contribution to ketogenesis. This confirms that the elevated BHB and CPT1A induction are GCGR-specific, not secondary to greater weight loss.
- Sex stratification (NIH SABV): female rodents have higher baseline fat oxidation capacity and different insulin sensitivity profiles. Run male and female cohorts separately and report sex × treatment interaction terms. Female DIO mice may require 18–20 weeks on HFD to achieve comparable metabolic inflexibility to males at 12–14 weeks.
- Liver biopsy timing: HMGCS2 and CPT1A mRNA peak in the fasted state (ZT12) and decline within 2h of refeeding. Sacrifice for liver molecular endpoints must occur within the fasted window — do not feed animals before sacrifice for hepatic gene expression studies.
BHB and NLRP3: Connecting Ketogenesis to Anti-Inflammatory Endpoints
For researchers interested in the NLRP3 suppression mechanism of BHB, in vitro validation should precede in vivo endpoint selection. The Youm 2015 protocol uses LPS-primed (500 ng/mL, 4h) and ATP-activated (5 mM, 30 min) primary macrophages or J774A.1/RAW264.7 cells, with BHB added as an equimolar sodium salt (Sigma H6501) at 1–10 mM. Measure: caspase-1 p10 western blot (Cell Signaling #3866), IL-1β in supernatant (R&D Systems DY401), and ASC oligomerization (crosslinked pellet western blot, Santa Cruz anti-ASC sc-22514-R). Pharmacological controls: MCC950 (10 μM, NLRP3 inhibitor, positive control for NLRP3 suppression), nigericin (10 μM, alternative NLRP3 activator to confirm ATP-specificity), and glucose-matched vehicle (to control for osmotic effects of BHB). Confirm HCAR2 dependence using pertussis toxin (100 ng/mL, Gαi inhibitor) or HCAR2-knockdown cells (siRNA 25 nM, 48h pre-treatment).