Hypothesis

Berberine lowers hepatic glucose output not only by activating AMPK via a lysosomal AXIN1‑dependent route but also by suppressing the epigenetic regulator UHRF1, leading to DNA hypomethylation of gluconeogenic gene promoters. This epigenetic shift reduces basal gluconeogenesis independently of AMPK and synergizes with metformin’s PEN2‑dependent lysosomal AMPK activation, producing greater overall inhibition of hepatic glucose production than either drug alone.

Mechanistic Rationale

• Berberine decreases UHRF1 expression in hepatocytes [5], a scaffold that recruits DNMT1 to hemimethylated DNA during replication.
• Reduced UHRF1 diminishes DNMT1 activity, causing passive DNA demethylation at promoters of key gluconeogenic enzymes (e.g., G6PC, PCK1).
• Hypomethylation paradoxically suppresses transcription of these genes because their promoters contain CpG islands that, when unmethylated, bind repressive transcription factors such as KLF4, a mechanism documented in other metabolic contexts.
• This epigenetic repression lowers basal hepatic glucose production without requiring AMPK phosphorylation, consistent with observations that berberine stimulates glucose uptake even when AMPK is silenced [5].
• Metformin activates lysosomal AMPK through PEN2 [1], increasing phospho‑AMPK and downstream inhibition of gluconeogenic transcription via CREB-regulated coactivator 2 (CRTC2) phosphorylation.
• When both agents are present, metformin‑driven AMPK activation further phosphorylates and stabilizes PGC‑1α, enhancing mitochondrial oxidation and reducing acetyl‑CoA availability for DNMT1, thereby amplifying the berberine‑initiated demethylation loop.
• The combined effect yields a dual hit: (1) AMPK‑mediated acute inhibition of gluconeogenic flux and (2) a sustained epigenetic dampening of gluconeogenic gene expression.

Testable Predictions

1. Hepatocyte‑specific UHRF1 knockout mice will exhibit reduced basal hepatic glucose production comparable to wild‑type mice treated with berberine, and adding berberine will not further lower glucose output in the knockout.
2. Chromatin immunoprecipitation sequencing (ChIP‑seq) for 5‑methylcytosine in liver tissue will show significant hypomethylation at the G6PC and PCK1 promoters after berberine treatment, an effect absent in UHRF1‑deficient hepatocytes.
3. Co‑treatment of primary hepatocytes with berberine and metformin will produce a greater decrease in phospho‑CREB and CRTC2 nuclear localization than either agent alone, correlating with increased AMPK phosphorylation.
4. Rescue experiments overexpressing UHRF1 in berberine‑treated hepatocytes will restore gluconeogenic gene expression and blunt the glucose‑lowering effect, despite intact AMPK activation.
5. Microbiome‑depleted mice (via broad‑spectrum antibiotics) will retain the berberine‑induced UHRF1 downregulation and glucose‑lowering effect, indicating that the epigenetic mechanism operates independently of gut‑flora modulation.

Experimental Design

• In vitro: Use HepG2 and primary mouse hepatocytes transfected with siRNA against UHRF1 or overexpressing UHRF1. Treat with berberine (10 µM), metformin (1 mM), or both. Measure AMPK phosphorylation (Western blot), UHRF1/DNMT1 levels, 5‑methylcytosine at gluconeogenic promoters (dot‑blot or bisulfite PCR), and glucose output in the media.
• In vivo: Generate Alb‑Cre;UHRF1^fl/fl mice and littermate controls. Administer berberine (200 mg/kg/day orally), metformin (250 mg/kg/day via drinking water), or both for 2 weeks. Perform hyperinsulinemic‑euglycemic clamps to assess hepatic glucose production, collect liver for bisulfite sequencing, and quantify AMPK activity.
• Controls: Include vehicle‑treated groups, AMPK‑deficient (Liver‑specific AMPKα1/α2 knockout) mice to verify AMPK‑independent effects, and antibiotic‑treated cohorts to exclude microbiome contributions.

If the data confirm that berberine’s suppression of UHRF1 drives epigenetic repression of gluconeogenic genes and that this action adds to metformin’s AMPK‑mediated inhibition, the hypothesis will be supported. Failure to observe UHRF1‑dependent promoter demethylation or lack of additive glucose lowering in UHRF1‑deficient models would falsify the proposed mechanism.