Endothelial-to-mesenchymal transitionFibrosisFMO3Gut MicrobiomeMyofibroblast differentiation

Gut microbe-derived metabolite trimethylamine N-oxide activates PERK to drive fibrogenic mesenchymal differentiation

  • 2022
  • Seok-Jo Kim, Swarna Bale, Priyanka Verma, et al.
Why this matters for your health

Systemic sclerosis (SSc, scleroderma) is a serious chronic disease with high mortality and no effective therapy. Its hallmarks are blood-vessel damage and fibrosis (tissue scarring) in several organs. Gut dysbiosis is common in SSc, but how it might drive skin and vascular damage was unknown. This study proposes a pathway from gut bacteria (making trimethylamine, TMA) through host FMO3 enzyme (making TMAO) and the PERK kinase to the conversion of local cells into scar-forming myofibroblasts. If confirmed, it could point to new drug targets. The evidence is from cell culture and patient samples only, so it remains a hypothesis.

Summary

The team used cultured human skin fibroblasts, microvascular endothelial cells (HMVECs) and fat-derived stem cells, plus skin biopsies and fibroblasts from SSc patients and healthy controls. TMAO (usually 50–200 µM) pushed all three cell types toward a myofibroblast state, with more collagen and fibronectin and the loss of endothelial or fat-cell features. Blocking PERK with a drug or siRNA prevented these effects. The TMAO-making enzyme FMO3, normally found in the liver, was detected in skin fibroblasts and endothelial cells and was raised by TGF-β1. FMO3 was higher in SSc skin biopsies (several datasets) and in SSc fibroblasts, which also showed more active PERK. The authors propose an FMO3-TMAO-PERK axis in SSc and call for animal studies.

Key findings

  • In normal skin fibroblasts, TMAO increased COL1A1, COL1A2, ACTA2 and FN-EDA gene expression, raised procollagen I, fibronectin and TGF-β1 secretion, and increased F-actin stress fibers and focal adhesions, signs of myofibroblasts.
  • TMAO did not activate Smad2 (unlike TGF-β1), so its effects are unlikely to run through the classic TGF-β pathway. TMA, the precursor, caused weaker responses and PERK activation only at high doses.
  • In microvascular endothelial cells, TMAO lowered VE-cadherin (CDH5) and CD31 (PECAM1) while raising COL1A1, ACTA2, SNAI2 and TAGLN. RNA-seq showed more mesenchymal and fewer endothelial genes, and TMAO sped up cell migration. Histone analysis showed higher H3K4me3 and other H3 methylation marks, suggesting chromatin remodeling.
  • In differentiated fat cells, TMAO reduced lipid-filled (Oil Red O-positive) cells and cut adipocyte genes (adiponectin, PPARγ, FABP4), while raising fibrotic genes (Fn-EDA, αSMA). TMAO did not change adipocyte markers during ordinary stem-cell differentiation.
  • TMAO rapidly and dose-dependently raised PERK phosphorylation in fibroblasts and endothelial cells (with higher Foxo1). TGF-β1 did not activate PERK.
  • The PERK inhibitor GSK2606414 (10 µM) and PERK siRNA blocked TMAO’s fibrotic effects and kept CDH5 and CD31 from falling. Thapsigargin, a chemical ER-stress inducer, did not reproduce them, so the effect appears independent of ER stress.
  • TGF-β1 raised FMO3 (not FMO1) in skin explants, fibroblasts and HMVECs.
  • FMO3 was elevated in SSc skin across five transcriptome datasets (fold-change 1.69–4.29), for example 1.78-fold in the PRESS cohort and 2.71-fold in the GENISOS cohort. Freshly sorted early-diffuse SSc fibroblasts showed more than 4-fold higher FMO3.
  • Single-cell RNA-seq (6 SSc, 2 controls) placed FMO3 mainly in fibroblasts and pericytes, with the highest share in CLDN1+ fibroblasts.
  • Immunostaining showed more FMO3-positive dermal cells in SSc biopsies (7 SSc vs 8 controls, p < 0.0001). Cultured SSc fibroblasts (n = 4) had more FMO3, collagen I and phospho-PERK than healthy ones (n = 3).
  • Limitations: all TMAO effects were shown in cultured cells, not animals. The study did not show that fibroblast FMO3 actually converts TMA to TMAO. Downstream mechanisms of PERK were not worked out. Several human comparisons used small groups, and TMAO levels in the patients were not measured.
  • Small inconsistencies in the PDF: the Figure 3 legend says endothelial cells were treated for 72 h, while the text and RNA-seq methods say 48 h. The PERK inhibitor is named GSK2606414 in the results but “AMG PERK44” in the methods. The histone assay is described in the methods for fibroblasts but in the results for endothelial cells. The data-availability statement lists three GEO datasets, while Table 1 lists five cohorts, two of them unpublished or without a GEO ID. Check the original before quoting exact details.
  • Funding came from NIH grants. Dr. Hazen reports patents, royalties and consulting or research funding from companies. The other authors declare no competing interests.

Read the full paper
VIEW PDF
Citation

Kim SJ, Bale S, Verma P, Wan Q, Ma F, Gudjonsson JE, Hazen SL, Harms PW, Tsou PS, Khanna D, Tsoi LC, Gupta N, Ho KJ, Varga J. Gut microbe-derived metabolite trimethylamine N-oxide activates PERK to drive fibrogenic mesenchymal differentiation. iScience. 2022;25(8):104669. doi:10.1016/j.isci.2022.104669

Related Research Papers
n