Pyrroloquinoline quinone prevents developmental programming of microbial dysbiosis and macrophage polarization to attenuate liver fibrosis in offspring of obese mice

Abstract

Increasingly, evidence suggests that exposure to maternal obesity creates an inflammatory environment in utero, exerting long-lasting postnatal signatures on the juvenile innate immune system and microbiome that may predispose offspring to development of fatty liver disease. We found that exposure to a maternal Western-style diet (WD) accelerated fibrogenesis in the liver of offspring and was associated with early recruitment of proinflammatory macrophages at 8-12 weeks and microbial dysbiosis as early as 3 weeks of age. We further demonstrated that bone marrow-derived macrophages (BMDMs) were polarized toward an inflammatory state at 8 weeks of age and that a potent antioxidant, pyrroloquinoline quinone (PQQ), reversed BMDM metabolic reprogramming from glycolytic toward oxidative metabolism by restoring trichloroacetic acid cycle function at isocitrate dehydrogenase. This resulted in reduced inflammation and inhibited collagen fibril formation in the liver at 20 weeks of age, even when PQQ was withdrawn at 3 weeks of age. Beginning at 3 weeks of age, WD-fed mice developed a decreased abundance of Parabacteroides and Lactobacillus, together with increased Ruminococcus and decreased tight junction gene expression by 20 weeks, whereas microbiota of mice exposed to PQQ retained compositional stability with age, which was associated with improved liver health. Conclusion: Exposure to a maternal WD induces early gut dysbiosis and disrupts intestinal tight junctions, resulting in BMDM polarization and induction of proinflammatory and profibrotic programs in the offspring that persist into adulthood. Disrupted macrophage and microbiota function can be attenuated by short-term maternal treatment with PQQ prior to weaning, suggesting that reshaping the early gut microbiota in combination with reprogramming macrophages during early weaning may alleviate the sustained proinflammatory environment, preventing the rapid progression of nonalcoholic fatty liver disease to nonalcoholic steatohepatitis in offspring of obese mothers. (Hepatology Communications 2018;2:313-328).

Figure 1

PQQ supplementation reduces steatosis and retinoid loss in livers of WD‐fed mice at 20 weeks of age. (A) H&E staining of liver sections from 20‐week‐old offspring. Locations of central veins and portal triads are indicated (magnification × 200). (B) CARS microscopy was used to visualize lipid droplets in 12‐μm‐thick cryosections from mouse livers at 20 weeks. Representative images from eight mice per group are shown for offspring of dams treated with or without PQQ and for offspring removed from PQQ at weaning. Image magnification × 60, resulting in a field of view of ∼1.4 × 1.4 µm²; scale bars are 50 µm. Background subtraction was performed with Fluoview software. (C) Distribution of droplet areas was calculated as described in Materials and Methods. Images were acquired from three separate liver cryosections from six mice per group. Points represent data, the red line denotes the mean, pink boxes denote the 75% confidence interval and colored boxes denote the 25% confidence interval for each diet group. (D) Spontaneous Raman spectra examined over a spectral range from 1,400 to 1,700 cm⁻¹ were averaged from 12 lipid droplets from livers of WD (blue trace), WDPQQ (green trace), and WDPQQ/WD (red trace)‐fed offspring at 20 weeks of age. (E) Ratio of abundance of 1,593 cm⁻¹/1,440 cm⁻¹ for n = 5 mice/group. The peak at 1,440 cm⁻¹ is a background peak used for normalization. Data are mean ± SEM, * P < 0.05 as compared to WD. (F) Correlation between large droplet size and retinoid signal in hepatic lipid droplets. CH, WD, and WDPQQ in A‐E represent both maternal and offspring diets. Abbreviations: AU, arbitrary unit; CV, central veins; H&E, hematoxylin and eosin; PT, portal triad.

Figure 2

PQQ supplementation diminishes collagen deposition in livers from WD‐fed offspring. (A) Fixed liver sections were stained with picrosirius red and viewed under cross‐polarized light at 100 × to visualize collagen fibrils. a) CH, b) WD, c) WDPQQ, d) WDPQQ/WD. (B) Quantification of positive pixels present in picrosirius red‐stained sections. (C) SHG imaging was performed on fixed liver sections at a magnification of 20×. a) CH, b) WD, c) WDPQQ, d) WDPQQ/WD. (D) Quantification of mean intensity of the SHG signal. (E) Group‐averaged modified NAS, including Brunt criteria, SHG, and CARS indices. Data are mean ± SEM, n = 5‐6/group. Representative images are shown for each group. One‐way analysis of variance with Tukey correction was used to compare between groups. *P < 0.05, ****P < 0.0001 compared to WD. CH, WD, and WDPQQ in A‐E represent both maternal and offspring diets. Abbreviations: AU, arbitrary unit; CV, central veins; NAS, NAFLD activity score; PT, portal triad; SHG, second‐harmonic generation.

Figure 3

PQQ reverses WD‐induced up‐regulation of profibrotic genes in mouse liver. mRNA expression of genes important for (A) modification of the extracellular matrix, (B) profibrotic signaling, (C) induction of proinflammatory and oxidative stress‐related signaling pathways, and (D) infiltration of activated proinflammatory macrophages was measured in livers of adult male offspring. Endogenous normalization was to 18S rRNA (A‐C) or Hprt1 (D). Data are mean ± SEM; n = 5‐6/group. Mann‐Whitney U test was used to compare groups pairwise. *P < 0.05 compared with WD, ^# P < 0.05 compared with WDPQQ. CH, WD, and WDPQQ represent both maternal and offspring diets. Abbreviations: ECM, extracellular matrix; mRNA, messenger RNA; rRNA, ribosomal RNA.

Figure 4

PQQ attenuates recruitment of inflammatory macrophages and LPS‐induced activation of BMDMs. (A) Percent abundance of FACS‐sorted hepatic macrophage populations obtained from CH, WD, and WDPQQ‐fed offspring at 12 weeks. (B) mRNA expression levels of proinflammatory and profibrotic genes in liver. Endogenous normalization was to 18S rRNA and Gapdh. (C) mRNA expression levels of proinflammatory and profibrotic genes in BMDMs stimulated with or without LPS for 24 hours. Endogenous normalization was to 18S rRNA and Gapdh. Data are mean ± SEM, n = 4/group. Mann‐Whitney U test was used to compare groups pairwise. ***P < 0.001 compared with WD. Abbreviations: FACS, fluorescence‐activated cell sorting; mRNA, messenger RNA; n.s., not significant; rRNA, ribosomal RNA.

Figure 5

PQQ shifts cellular metabolism toward OXPHOS in activated BMDMs. (A) FLIM maps are shown in the top panel, phasor plots and quantitation of percentage free versus bound NADH are displayed below for BMDMs differentiated from 8‐12‐week‐old offspring fed CH, WD, or WDPQQ. Green signal in the FLIM map corresponds to pixels occupied by free NADH (more glycolysis), and red corresponds to pixels with more bound NADH (more OXPHOS). In unstimulated conditions, cells had low levels of free NADH, suggesting oxidative phosphorylation as the primary metabolic program. In contrast, on stimulation with LPS and IFNγ (+LPS/IFNγ), macrophages exhibited significantly increased free NADH, suggesting increased glycolysis. PQQ treatment shifted the NADH signal back to more bound NADH (more OXPHOS) for all groups. (B) Signal intensities of metabolites identified by ultrahigh‐performance liquid chromatography–tandem mass spectrometry. Representative data are shown. Abbreviations: AU, arbitrary units; US, unstimulated.

Figure 6

PQQ‐induced microbial changes may persist into adulthood. 16S rRNA sequencing was performed on cecal contents from n = 5‐7 mice/group. Relative abundance of the top 99.5% most abundant (A) phyla and (B) genera for adults and weanlings were plotted as a percent of OTU abundance. Permutational analysis of variance P values for pairwise comparisons are noted (*P < 0.05, **P < 0.01, ***P < 0.001). (C) mRNA levels of zona occludens 1 (Tjp1) in RNA isolated from cecum, normalized to 18S. (D) Count of OTUs with P < 0.05 when compared between weanlings and adults plotted for each diet group. (E) Mean percentage abundance of OTUs with significant changes between adults and weanlings only in WD‐fed mice. Data are mean ± SEM. Abbreviations: mRNA, messenger RNA; OTU, operational taxonomic unit; rRNA, ribosomal RNA.