We previously reported that serum levels of the gut-derived tryptophan metabolite indole-3-propionic acid (IPA) are lower in patients with liver fibrosis. In this study, we investigated the transcriptome and DNA methylome in obese livers in relation to serum IPA levels, as well as the role of IPA in inducing phenotypic inactivation of hepatic stellate cells (HSCs) in vitro.
       The study included 116 obese patients without type 2 diabetes mellitus (T2DM) (age 46.8 ± 9.3 years; BMI: 42.7 ± 5.0 kg/m²) who underwent bariatric surgery at the Kuopio Bariatric Surgery Center (KOBS). Circulating IPA levels were measured by liquid chromatography-mass spectrometry (LC-MS), liver transcriptome analysis was performed by total RNA sequencing, and DNA methylation analysis was performed using the Infinium HumanMethylation450 BeadChip. Human hepatic stellate cells (LX-2) were used for in vitro experiments.
       Serum IPA levels correlated with the expression of genes involved in apoptotic, mitophagic, and longevity pathways in the liver. The AKT serine/threonine kinase 1 (AKT1) gene was the most abundant and dominant interacting gene in the liver transcript and DNA methylation profiles. IPA treatment induced apoptosis, decreased mitochondrial respiration, and altered cell morphology and mitochondrial dynamics by modulating the expression of genes known to regulate fibrosis, apoptosis, and survival of LX-2 cells.
       Taken together, these data support that IPA has potential therapeutic effects and can induce apoptosis and shift the HSC phenotype toward an inactive state, thereby expanding the possibility of inhibiting liver fibrosis by interfering with HSC activation and mitochondrial metabolism.
       The prevalence of obesity and metabolic syndrome has been associated with an increasing incidence of metabolically associated fatty liver disease (MASLD); the disease affects 25% to 30% of the general population [1]. The main consequence of MASLD etiology is liver fibrosis, a dynamic process characterized by continuous accumulation of fibrous extracellular matrix (ECM) [2]. The main cells involved in liver fibrosis are hepatic stellate cells (HSCs), which exhibit four known phenotypes: quiescent, activated, inactivated, and senescent [3, 4]. HSCs can be activated and transdifferentiate from a quiescent form into proliferative fibroblast-like cells with high energy demands, with increased expression of α-smooth muscle actin (α-SMA) and type I collagen (Col-I) [5, 6]. During liver fibrosis reversal, activated HSCs are eliminated via apoptosis or inactivation. These processes include downregulation of fibrogenic genes and modulation of prosurvival genes (such as NF-κB and PI3K/Akt signaling pathways) [7, 8], as well as changes in mitochondrial dynamics and function [9].
       Serum levels of the tryptophan metabolite indole-3-propionic acid (IPA), produced in the intestine, have been found to be decreased in human metabolic diseases including MASLD [10–13]. IPA is associated with dietary fiber intake, is known for its antioxidant and anti-inflammatory effects, and attenuates the diet-induced non-alcoholic steatohepatitis (NASH) phenotype in vivo and in vitro [11–14]. Some evidence comes from our previous study, which demonstrated that serum IPA levels were lower in patients with liver fibrosis than in obese patients without liver fibrosis in the Kuopio Bariatric Surgery Study (KOBS). Furthermore, we showed that IPA treatment could reduce the expression of genes that are classical markers of cell adhesion, cell migration and hematopoietic stem cell activation in a human hepatic stellate cell (LX-2) model and is a potential hepatoprotective metabolite [15]. However, it remains unclear how IPA induces liver fibrosis regression by activating HSC apoptosis and mitochondrial bioenergetics.
       Here, we demonstrate that serum IPA is associated with the expression of genes enriched in apoptosis, mitophagy, and longevity pathways in the liver of obese but non-type 2 diabetes (KOBS) individuals. Furthermore, we found that IPA can induce the clearance and degradation of activated hematopoietic stem cells (HSCs) via the inactivation pathway. These results reveal a novel role for IPA, making it a potential therapeutic target to promote liver fibrosis regression.
       A previous study in the KOBS cohort showed that patients with liver fibrosis had lower circulating IPA levels compared with patients without liver fibrosis [15]. To exclude the potential confounding effect of type 2 diabetes, we recruited 116 obese patients without type 2 diabetes (mean age ± SD: 46.8 ± 9.3 years; BMI: 42.7 ± 5.0 kg/m2) (Table 1) from the ongoing KOBS study as the study population [16]. All participants gave written informed consent and the study protocol was approved by the Ethics Committee of North Savo County Hospital in accordance with the Declaration of Helsinki (54/2005, 104/2008 and 27/2010).
       Liver biopsy specimens were obtained during bariatric surgery and histologically assessed by experienced pathologists according to previously described criteria [17, 18]. The assessment criteria are summarized in Supplementary Table S1 and have been described previously [19].
       Fasting serum samples were analyzed by untargeted liquid chromatography-mass spectrometry (LC-MS) for metabolomics analysis (n = 116). Samples were analyzed using a UHPLC-qTOF-MS system (1290 LC, 6540 qTOF-MS, Agilent Technologies, Waldbronn, Karlsruhe, Germany) as described previously19. Identification of isopropyl alcohol (IPA) was based on retention time and comparison of the MS/MS spectrum with pure standards. The IPA signal intensity (peak area) was considered in all further analyses [20].
       Whole liver RNA sequencing was performed using Illumina HiSeq 2500 and data were preprocessed as described previously [19, 21, 22]. We performed targeted differential expression analysis of transcripts affecting mitochondrial function/biogenesis using 1957 genes selected from the MitoMiner 4.0 database [ 23 ]. Liver DNA methylation analysis was performed using the Infinium HumanMethylation450 BeadChip (Illumina, San Diego, CA, USA) using the same methodology as described previously [24, 25].
       Human hepatic stellate cells (LX-2) were kindly provided by Prof. Stefano Romeo and were cultured and maintained in DMEM/F12 medium (Biowest, L0093-500, 1% Pen/Strep; Lonza, DE17-602E, 2% FBS; Gibco, 10270-106). To select the working dose of IPA, LX-2 cells were treated with different concentrations of IPA (10 μM, 100 μM and 1 mM; Sigma, 220027) in DMEM/F12 medium for 24 h. Furthermore, to investigate the ability of IPA to inactivate HSCs, LX-2 cells were co-treated with 5 ng/ml TGF-β1 (R&D systems, 240-B-002/CF) and 1 mM IPA in serum-free medium for 24 h. For the corresponding vehicle controls, 4 nM HCL containing 0.1% BSA was used for TGF-β1 treatment and 0.05% DMSO was used for IPA treatment, and both were used together for the combination treatment.
       Apoptosis was assessed using the FITC Annexin V Apoptosis Detection Kit with 7-AAD (Biolegend, San Diego, CA, USA, Cat# 640922) according to the manufacturer’s instructions. Briefly, LX-2 (1 × 105 cells/well) were cultured overnight in 12-well plates and then treated with multiple doses of IPA or IPA and TGF-β1. The following day, floating and adherent cells were collected, trypsinized, washed with PBS, resuspended in Annexin V binding buffer, and incubated with FITC-Annexin V and 7-AAD for 15 min.
       Mitochondria in living cells were stained for oxidative activity using Mitotracker™ Red CMXRos (MTR) (Thermo Fisher Scientific, Carlsbad, CA). For MTR assays, LX-2 cells were incubated at equal densities with IPA and TGF-β1. After 24 h, living cells were trypsinized, washed with PBS, and then incubated with 100 μM MTR in serum-free medium at 37 °C for 20 min as described previously [ 26 ]. For live cell morphology analysis, cell size and cytoplasmic complexity were analyzed using forward scatter (FSC) and side scatter (SSC) parameters, respectively.
       All data (30,000 events) were acquired using NovoCyte Quanteon (Agilent) and analyzed using NovoExpress® 1.4.1 or FlowJo V.10 software.
       Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured in real time using a Seahorse Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA) equipped with a Seahorse XF Cell Mito Stress according to the manufacturer’s instructions. Briefly, 2 × 104 LX-2 cells/well were seeded onto XF96 cell culture plates. After overnight incubation, cells were treated with isopropanol (IPA) and TGF-β1 (Supplementary Methods 1). Data analysis was performed using Seahorse XF Wave software, which includes the Seahorse XF Cell Energy Phenotype Test Report Generator. From this, a Bioenergetic Health Index (BHI) was calculated [27].
       Total RNA was transcribed into cDNA. For specific methods, see reference [15]. Human 60S ribosomal acidic protein P0 (RPLP0) and cyclophilin A1 (PPIA) mRNA levels were used as constitutive gene controls. The QuantStudio 6 pro Real-Time PCR System (Thermo Fisher, Landsmeer, The Netherlands) was used with the TaqMan™ Fast Advanced Master Mix Kit (Applied Biosystems) or the Sensifast SYBR Lo-ROX Kit (Bioline, BIO 94050), and relative gene expression fold was calculated using comparative Ct value cycling parameters (ΔΔCt) and the ∆∆Ct method. Details of the primers are provided in Supplementary Tables S2 and S3.
       Nuclear DNA (ncDNA) and mitochondrial DNA (mtDNA) were extracted using the DNeasy blood and tissue kit (Qiagen) as described previously [28]. The relative amount of mtDNA was calculated by calculating the ratio of each target mtDNA region to the geometric mean of the three nuclear DNA regions (mtDNA/ncDNA), as detailed in Supplementary Methods 2. Details of the primers for mtDNA and ncDNA are provided in Supplementary Table S4.
       Live cells were stained with Mitotracker™ Red CMXRos (MTR) (Thermo Fisher Scientific, Carlsbad, CA) to visualize intercellular and intracellular mitochondrial networks. LX-2 cells (1 × 104 cells/well) were cultured on glass slides in corresponding glass-bottomed culture plates (Ibidi GmbH, Martinsried, Germany). After 24 h, live LX-2 cells were incubated with 100 μM MTR for 20 min at 37 °C and cell nuclei were stained with DAPI (1 μg/ml, Sigma-Aldrich) as described previously [29]. Mitochondrial networks were visualized using a Zeiss Axio Observer inverted microscope (Carl Zeiss Microimaging GmbH, Jena, Germany) equipped with a Zeiss LSM 800 confocal module at 37 °C in a humidified atmosphere with 5% CO2 using a 63×NA 1.3 objective. We acquired ten Z-series images for each sample type. Each Z-series contains 30 sections, each with a thickness of 9.86 μm. For each sample, images of ten different fields of view were acquired using ZEN 2009 software (Carl Zeiss Microimaging GmbH, Jena, Germany), and mitochondrial morphology analysis was performed using ImageJ software (v1.54d) [30, 31] according to the parameters detailed in Supplementary Methods 3.
       The cells were fixed with 2% glutaraldehyde in 0.1 M phosphate buffer, followed by fixation with 1% osmium tetroxide solution (Sigma Aldrich, MO, USA), gradually dehydrated with acetone (Merck, Darmstadt, Germany), and finally embedded in epoxy resin. Ultrathin sections were prepared and stained with 1% uranyl acetate (Merck, Darmstadt, Germany) and 1% lead citrate (Merck, Darmstadt, Germany). Ultrastructural images were obtained using a JEM 2100F EXII transmission electron microscope (JEOL Ltd, Tokyo, Japan) at an accelerating voltage of 80 kV.
       The morphology of LX-2 cells treated with IPA for 24 h was analyzed by phase-contrast microscopy at 50x magnification using a Zeiss inverted light microscope (Zeiss Axio Vert.A1 and AxioCam MRm, Jena, Germany).
       Clinical data were expressed as mean ± standard deviation or median (interquartile range: IQR). One-way analysis of variance (continuous variables) or χ² test (categorical variables) were used to compare differences between the three study groups. The false positive rate (FDR) was used to correct for multiple testing, and genes with FDR < 0.05 were considered statistically significant. Spearman correlation analysis was used to correlate CpG DNA methylation with IPA signal intensity, with nominal p values ​​(p < 0.05) reported.
       Pathway analysis was performed using a web-based gene set analysis tool (WebGestalt) for 268 transcripts (nominal p < 0.01), 119 mitochondria-associated transcripts (nominal p < 0.05), and 4350 CpGs out of 3093 liver transcripts that were associated with circulating serum IPA levels. The freely available Venny DB (version 2.1.0) tool was used to find overlapping genes, and StringDB (version 11.5) was used to visualize protein-protein interactions.
       For the LX-2 experiment, samples were tested for normality using the D’Agostino-Pearson test. Data were obtained from at least three biological replicates and subjected to one-way ANOVA with Bonferroni post hoc test. A p-value of less than 0.05 was considered statistically significant. Data are presented as mean ± SD, and the number of experiments is indicated in each figure. All analyses and graphs were performed using GraphPad Prism 8 statistical software for Windows (GraphPad Software Inc., version 8.4.3, San Diego, USA).
       First, we investigated the association of serum IPA levels with liver, whole-body, and mitochondrial transcripts. In the total transcript profile, the strongest gene associated with serum IPA levels was MAPKAPK3 (FDR = 0.0077; mitogen-activated protein kinase-activated protein kinase 3); in the mitochondria-related transcript profile, the strongest associated gene was AKT1 (FDR = 0.7621; AKT serine/threonine kinase 1) (Additional file 1 and Additional file 2).
       We then analyzed global transcripts (n = 268; p < 0.01) and mitochondria-associated transcripts (n = 119; p < 0.05), ultimately identifying apoptosis as the most significant canonical pathway (p = 0.0089). For mitochondrial transcripts associated with serum IPA levels, we focused on apoptosis (FDR = 0.00001), mitophagy (FDR = 0.00029), and TNF signaling pathways (FDR = 0.000006) (Figure 1A, Table 2, and Supplementary Figures 1A-B).
       Overlapping analysis of global, mitochondria-associated transcripts, and DNA methylation in human liver in association with serum IPA levels. A represents 268 global transcripts, 119 mitochondria-associated transcripts, and DNA methylated transcripts that are mapped to 3092 CpG sites associated with serum IPA levels (p values ​​< 0.01 for global transcripts and DNA methylated, and p values ​​< 0.05 for mitochondrial transcripts). The major overlapping transcripts are shown in the middle (AKT1 and YKT6). B The interaction map of the 13 genes with the highest interaction score (0.900) with other genes was constructed from the 56 overlapping genes (black line region) that were significantly associated with serum IPA levels using the online tool StringDB. Green: Genes mapped to the Gene Ontology (GO) cellular component: mitochondria (GO:0005739). AKT1 is the protein with the highest score (0.900) for interactions with other proteins based on the data (based on text mining, experiments, databases, and co-expression). Network nodes represent proteins, and edges represent connections between proteins.
       Since gut microbiota metabolites can regulate epigenetic composition through DNA methylation [32], we investigated whether serum IPA levels were associated with liver DNA methylation. We found that the two major methylation sites associated with serum IPA levels were near proline-serine-rich region 3 (C19orf55) and heat shock protein family B (small) member 6 (HSPB6) (Additional file 3). DNA methylation of 4350 CpG (p < 0.01) was correlated with serum IPA levels and was enriched in longevity regulatory pathways (p = 0.006) (Figure 1A, Table 2, and Supplementary Figure 1C).
       To understand the biological mechanisms underlying the associations between serum IPA levels, global transcripts, mitochondria-associated transcripts, and DNA methylation in human liver, we performed an overlap analysis of the genes identified in the previous pathway analysis (Figure 1A). The results of pathway enrichment analysis of the 56 overlapping genes (inside the black line in Figure 1A) showed that the apoptosis pathway (p = 0.00029) highlighted two genes common to the three analyses: AKT1 and YKT6 (YKT6 v-SNARE homolog), as shown in the Venn diagram (Supplementary Figure 2 and Figure 1A). Interestingly, we found that AKT1 (cg19831386) and YKT6 (cg24161647) were positively correlated with serum IPA levels (Additional file 3). To identify potential protein interactions between gene products, we selected 13 genes with the highest common region score (0.900) among 56 overlapping genes as input and constructed an interaction map. According to the confidence level (marginal confidence), the AKT1 gene with the highest score (0.900) was at the top (Figure 1B).
       Based on the pathway analysis, we found that apoptosis was the major pathway, so we investigated whether IPA treatment would affect the apoptosis of HSCs in vitro. We previously demonstrated that different doses of IPA (10 μM, 100 μM, and 1 mM) were nontoxic to LX-2 cells [15]. This study showed that IPA treatment at 10 μM and 100 μM increased the number of viable and necrotic cells. However, compared with the control group, cell viability decreased at 1 mM IPA concentration, while the cell necrosis rate remained unchanged (Figure 2A, B). Next, to find the optimal concentration to induce apoptosis in LX-2 cells, we tested 10 μM, 100 μM, and 1 mM IPA for 24 h (Figure 2A-E and Supplementary Figure 3A-B). Interestingly, IPA 10 μM and 100 μM decreased the apoptosis rate (%), however, IPA 1 mM increased late apoptosis and apoptosis rate (%) compared to control and was therefore chosen for further experiments (Figures 2A–D).
       IPA induces apoptosis of LX-2 cells. Annexin V and 7-AAD double staining method was used to quantify the apoptotic rate and cell morphology by flow cytometry. BA cells were incubated with 10 μM, 100 μM and 1 mM IPA for 24 h or with F–H TGF-β1 (5 ng/ml) and 1 mM IPA in serum-free medium for 24 h. A: living cells (Annexin V -/ 7AAD-); B: necrotic cells (Annexin V -/ 7AAD+); C, F: early (Annexin V +/ 7AAD-); D, G: late (Annexin V+/7AAD.+); E, H: percentage of total early and late apoptotic cells in apoptotic rate (%). Data are expressed as mean ± SD, n = 3 independent experiments. Statistical comparisons were performed using one-way ANOVA with Bonferroni post hoc test. *p < 0.05; ****p < 0.0001
       As we have shown previously, 5 ng/ml TGF-β1 can induce HSC activation by increasing the expression of classical marker genes [15]. LX-2 cells were treated with 5 ng/ml TGF-β1 and 1 mM IPA in combination (Fig. 2E–H). TGF-β1 treatment did not change the apoptosis rate, however, IPA co-treatment increased late apoptosis and apoptosis rate (%) compared with TGF-β1 treatment (Fig. 2E–H). These results indicate that 1 mM IPA can promote apoptosis in LX-2 cells independently of TGF-β1 induction.
       We further investigated the effect of IPA on mitochondrial respiration in LX-2 cells. The results showed that 1 mM IPA decreased the oxygen consumption rate (OCR) parameters: non-mitochondrial respiration, basal and maximal respiration, proton leak and ATP production compared to the control group (Figure 3A, B), while the bioenergetic health index (BHI) did not change.
       IPA reduces mitochondrial respiration in LX-2 cells. The mitochondrial respiration curve (OCR) is presented as mitochondrial respiration parameters (non-mitochondrial respiration, basal respiration, maximal respiration, proton leak, ATP generation, SRC and BHI). Cells A and B were incubated with 10 μM, 100 μM and 1 mM IPA for 24 h, respectively. Cells C and D were incubated with TGF-β1 (5 ng/ml) and 1 mM IPA in serum-free medium for 24 h, respectively. All measurements were normalized to DNA content using the CyQuant kit. BHI: bioenergetic health index; SRC: respiratory reserve capacity; OCR: oxygen consumption rate. Data are presented as mean ± standard deviation (SD), n = 5 independent experiments. Statistical comparisons were performed using one-way ANOVA and Bonferroni post hoc test. *p < 0.05; **p < 0.01; and ***p < 0.001
       To gain a more comprehensive understanding of the effect of IPA on the bioenergetic profile of TGF-β1-activated LX-2 cells, we analyzed mitochondrial oxidative phosphorylation by OCR (Fig. 3C,D). The results showed that TGF-β1 treatment could reduce the maximum respiration, respiratory reserve capacity (SRC) and BHI compared with the control group (Fig. 3C,D). In addition, the combination treatment decreased basal respiration, proton leak and ATP production, but SRC and BHI were significantly higher than those treated with TGF-β1 (Fig. 3C,D).
       We also performed the “Cellular Energy Phenotype Test” provided by Seahorse software (Supplementary Fig. 4A–D). As shown in Supplementary Fig. 3B, both OCR and ECAR metabolic potentials were decreased after TGF-β1 treatment, however, no difference was observed in the combination and IPA treatment groups compared to the control group. Furthermore, both basal and stress levels of OCR were decreased after combination and IPA treatment compared to the control group (Supplementary Fig. 4C). Interestingly, a similar pattern was observed with combination therapy, where no change in basal and stress levels of ECAR was observed compared to TGF-β1 treatment (Supplementary Fig. 4C). In HSCs, the reduction in mitochondrial oxidative phosphorylation and the ability of combination treatment to restore SCR and BHI after exposure to TGF-β1 treatment did not alter the metabolic potential (OCR and ECAR). Taken together, these results indicate that IPA may reduce bioenergetics in HSCs, suggesting that IPA may induce a lower energetic profile that shifts the HSC phenotype toward inactivation (Supplementary Figure 4D).
       The effect of IPA on mitochondrial dynamics was investigated using three-dimensional quantification of mitochondrial morphology and network connections as well as MTR staining (Figure 4 and Supplementary Figure 5). Figure 4 shows that, compared with the control group, TGF-β1 treatment decreased the mean surface area, branch number, total branch length, and branch junction number (Figure 4A and B) and changed the proportion of mitochondria from spherical to intermediate morphology (Figure 4C). Only IPA treatment decreased the mean mitochondrial volume and changed the proportion of mitochondria from spherical to intermediate morphology compared with the control group (Figure 4A). In contrast, sphericity, mean branch length, and mitochondrial activity assessed by mitochondrial membrane potential-dependent MTR (Figure 4A and E) remained unchanged and these parameters did not differ between groups. Taken together, these results suggest that TGF-β1 and IPA treatment appear to modulate mitochondrial shape and size as well as network complexity in living LX-2 cells.
       IPA alters mitochondrial dynamics and mitochondrial DNA abundance in LX-2 cells. A. Representative confocal images of live LX-2 cells incubated with TGF-β1 (5 ng/ml) and 1 mM IPA for 24 h in serum-free medium showing mitochondrial networks stained with Mitotracker™ Red CMXRos and nuclei stained blue with DAPI. All data contained at least 15 images per group. We acquired 10 Z-stack images for each sample type. Each Z-axis sequence contained 30 slices, each with a thickness of 9.86 μm. Scale bar: 10 μm. B. Representative objects (mitochondria only) identified by applying adaptive thresholding to the image. Quantitative analysis and comparison of mitochondrial morphological network connections were performed for all cells in each group. C. Frequency of mitochondrial shape ratios. Values ​​close to 0 indicate spherical shapes, and values ​​close to 1 indicate filamentous shapes. D Mitochondrial DNA (mtDNA) content was determined as described in Materials and Methods. E Mitotracker™ Red CMXRos analysis was performed by flow cytometry (30,000 events) as described in Materials and Methods. Data are presented as mean ± SD, n = 3 independent experiments. Statistical comparisons were performed using one-way ANOVA and Bonferroni post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001
       We then analyzed the mtDNA content in LX-2 cells as an indicator of mitochondrial number. Compared with the control group, the mtDNA content was increased in the TGF-β1-treated group (Figure 4D). Compared with the TGF-β1-treated group, the mtDNA content was decreased in the combination treatment group (Figure 4D), suggesting that IPA may reduce the mtDNA content and possibly the mitochondrial number as well as mitochondrial respiration (Figure 3C). Moreover, IPA seemed to reduce the mtDNA content in the combination treatment but did not affect MTR-mediated mitochondrial activity (Figures 4A–C).
       We investigated the association of IPA with the mRNA levels of genes associated with fibrosis, apoptosis, survival, and mitochondrial dynamics in LX-2 cells (Figure 5A–D). Compared with the control group, the TGF-β1-treated group showed increased expression of genes such as collagen type I α2 chain (COL1A2), α-smooth muscle actin (αSMA), matrix metalloproteinase 2 (MMP2), tissue inhibitor of metalloproteinase 1 (TIMP1), and dynamin 1-like gene (DRP1), indicating increased fibrosis and activation. Furthermore, compared with the control group, TGF-β1 treatment reduced the mRNA levels of nuclear pregnane X receptor (PXR), caspase 8 (CASP8), MAPKAPK3, inhibitor of B-cell α, enhancer of nuclear factor κ gene light peptide (NFκB1A), and inhibitor of nuclear factor κB kinase subunit β (IKBKB) (Figure 5A–D). Compared with TGF-β1 treatment, combination treatment with TGF-β1 and IPA reduced the expression of COL1A2 and MMP2, but increased the mRNA levels of PXR, TIMP1, B-cell lymphoma-2 (BCL-2), CASP8, NFκB1A, NFκB1-β, and IKBKB. IPA treatment significantly decreased the expression of MMP2, Bcl-2-associated protein X (BAX), AKT1, optic atrophy protein 1 (OPA1), and mitochondrial fusion protein 2 (MFN2), whereas the expression of CASP8, NFκB1A, NFκB1B, and IKBKB was increased compared with the control group. However, no difference was found in the expression of caspase-3 (CASP3), apoptotic peptidase activating factor 1 (APAF1), mitochondrial fusion protein 1 (MFN1), and fission protein 1 (FIS1). Collectively, these results suggest that IPA treatment modulates the expression of genes associated with fibrosis, apoptosis, survival, and mitochondrial dynamics. Our data suggest that IPA treatment reduces fibrosis in LX-2 cells; at the same time, it stimulates survival by shifting the phenotype towards inactivation.
       IPA modulates the expression of fibroblast, apoptotic, viability, and mitochondrial dynamics genes in LX-2 cells. Histograms display mRNA expression relative to endogenous control (RPLP0 or PPIA) after LX-2 cells were induced with TGF-β1 and IPA in serum-free medium for 24 h. A indicates fibroblasts, B indicates apoptotic cells, C indicates surviving cells, and D indicates mitochondrial dynamics gene expression. Data are presented as mean ± standard deviation (SD), n = 3 independent experiments. Statistical comparisons were performed using one-way ANOVA and Bonferroni post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001
       Then, changes in cell size (FSC-H) and cytoplasmic complexity (SSC-H) were assessed by flow cytometry (Figure 6A,B), and changes in cell morphology after IPA treatment were assessed by transmission electron microscopy (TEM) and phase contrast microscopy (Supplementary Figure 6A-B). As expected, cells in the TGF-β1-treated group increased in size compared to the control group (Figure 6A,B), showing the classic expansion of rough endoplasmic reticulum (ER*) and phagolysosomes (P), indicating hematopoietic stem cell (HSC) activation (Supplementary Figure 6A). However, compared with the TGF-β1-treated group, the cell size, cytoplasmic complexity (Fig. 6A,B), and ER* content were decreased in the TGF-β1 and IPA combination treatment group (Supplementary Fig. 6A). Furthermore, IPA treatment decreased the cell size, cytoplasmic complexity (Figs. 6A,B), P and ER* content (Supplementary Fig. 6A) compared with the control group. In addition, the content of apoptotic cells increased after 24 h of IPA treatment compared with the control group (white arrows, Supplementary Fig. 6B). Collectively, these results suggest that 1 mM IPA can stimulate HSC apoptosis and reverse the changes in cell morphological parameters induced by TGF-β1, thereby regulating the cell size and complexity, which may be associated with HSC inactivation.
       IPA alters cell size and cytoplasmic complexity in LX-2 cells. Representative images of flow cytometry analysis. The analysis used a gating strategy specific for LX-2 cells: SSC-A/FSC-A to define the cell population, FSC-H/FSC-A to identify doublets, and SSC-H/FSC-H for cell size and complexity analysis. Cells were incubated with TGF-β1 (5 ng/ml) and 1 mM IPA in serum-free medium for 24 h. LX-2 cells were distributed into lower left quadrant (SSC-H-/FSC-H-), upper left quadrant (SSC-H+/FSC-H-), lower right quadrant (SSC-H-/FSC-H+), and upper right quadrant (SSC-H+/FSC-H+) for cell size and cytoplasmic complexity analysis. B. Cell morphology was analyzed by flow cytometry using FSC-H (forward scatter, cell size) and SSC-H (side scatter, cytoplasmic complexity) (30,000 events). Data are presented as mean ± SD, n = 3 independent experiments. Statistical comparisons were performed using one-way ANOVA and Bonferroni post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001 and ****p < 0.0001
       Gut metabolites such as IPA have become a hot topic of research, suggesting that new targets may be discovered in the gut microbiota. It is therefore interesting that IPA, a metabolite that we have linked to liver fibrosis in humans [15], has been shown to be a potential anti-fibrotic compound in animal models [13, 14]. Here, we demonstrate for the first time an association between serum IPA and global liver transcriptomics and DNA methylation in obese individuals without type 2 diabetes (T2D), highlighting apoptosis, mitophagy and longevity, as well as a possible candidate gene AKT1 regulating liver homeostasis. Another novelty of our study is that we demonstrated the interaction of IPA treatment with apoptosis, cell morphology, mitochondrial bioenergetics and dynamics in LX-2 cells, indicating a lower energy spectrum that shifts the HSC phenotype towards inactivation, making IPA a potential candidate for improving liver fibrosis.
       We found that apoptosis, mitophagy and longevity were the most important canonical pathways enriched in liver genes associated with circulating serum IPA. Disruption of the mitochondrial quality control (MQC) system can lead to mitochondrial dysfunction, mitophagy and apoptosis, thereby promoting the occurrence of MASLD[33, 34]. Therefore, we can speculate that IPA may be involved in maintaining cell dynamics and mitochondrial integrity through apoptosis, mitophagy and longevity in the liver. Our data showed that two genes were common across the three assays: YKT6 and AKT1. It is worth noting that YKT6 is a SNARE protein involved in the process of cell membrane fusion. It plays a role in autophagy and mitophagy by forming an initiation complex with STX17 and SNAP29 on the autophagosome, thereby promoting the fusion of autophagosomes and lysosomes[35]. Furthermore, loss of YKT6 function results in impaired mitophagy[36], while upregulation of YKT6 is associated with the progression of hepatocellular carcinoma (HCC), showing increased cell survival[37]. On the other hand, AKT1 is the most important interacting gene and plays an important role in liver diseases, including PI3K/AKT signaling pathway, cell cycle, cell migration, proliferation, focal adhesion, mitochondrial function, and collagen secretion[38–40]. Activated PI3K/AKT signaling pathway can activate hematopoietic stem cells (HSCs), which are the cells responsible for the production of extracellular matrix (ECM), and its dysregulation may contribute to the occurrence and progression of liver fibrosis[40]. In addition, AKT is one of the key cell survival factors that inhibits p53-dependent cell apoptosis, and AKT activation may be associated with the inhibition of liver cell apoptosis[41, 42]. The obtained results suggest that IPA may be involved in liver mitochondria-associated apoptosis by affecting the decision of hepatocytes between entering apoptosis or survival. These effects may be regulated by AKT and/or YKT6 candidate genes, which are critical for liver homeostasis.
       Our results showed that 1 mM IPA induced apoptosis and decreased mitochondrial respiration in LX-2 cells independent of TGF-β1 treatment. It is noteworthy that apoptosis is a major pathway for fibrosis resolution and hematopoietic stem cell (HSC) activation, and is also a key event in the reversible physiological response of liver fibrosis [4, 43]. Moreover, the restoration of BHI in LX-2 cells after combination treatment provided new insights into the potential role of IPA in the regulation of mitochondrial bioenergetics. Under resting and inactive conditions, hematopoietic cells normally utilize mitochondrial oxidative phosphorylation to produce ATP and have low metabolic activity. On the other hand, HSC activation enhances mitochondrial respiration and biosynthesis to compensate for the energy demands of entering the glycolytic state [44]. The fact that IPA did not affect metabolic potential and ECAR suggests that the glycolytic pathway is less prioritized. Similarly, another study showed that 1 mM IPA was able to modulate mitochondrial respiratory chain activity in cardiomyocytes, human hepatocyte cell line (Huh7) and human umbilical vein endothelial cells (HUVEC); However, no effect of IPA was found on glycolysis in cardiomyocytes, suggesting that IPA may affect the bioenergetics of other cell types [45]. Therefore, we speculate that 1 mM IPA may act as a mild chemical uncoupler, since it can significantly reduce fibrogenic gene expression, cell morphology and mitochondrial bioenergetics without changing the amount of mtDNA [46]. Mitochondrial uncouplers can inhibit culture-induced fibrosis and HSC activation [47] and reduce mitochondrial ATP production regulated or induced by certain proteins such as uncoupling proteins (UCP) or adenine nucleotide translocase (ANT). Depending on the cell type, this phenomenon can protect cells from apoptosis and/or promote apoptosis [46]. However, further studies are needed to elucidate the role of IPA as a mitochondrial uncoupler in hematopoietic stem cell inactivation.
       We then investigated whether the changes in mitochondrial respiration are reflected in mitochondrial morphology in living LX-2 cells. Interestingly, TGF-β1 treatment alters the mitochondrial proportion from spherical to intermediate, with decreased mitochondrial branching and increased expression of DRP1, a key factor in mitochondrial fission [48]. Furthermore, mitochondrial fragmentation is associated with overall network complexity, and the transition from fusion to fission is critical for hematopoietic stem cell (HSC) activation, whereas inhibition of mitochondrial fission leads to HSC apoptosis [49]. Thus, our results indicate that TGF-β1 treatment may induce a decrease in mitochondrial network complexity with decreased branching, which is more common in mitochondrial fission associated with activated hematopoietic stem cells (HSCs). Furthermore, our data showed that IPA could change the proportion of mitochondria from spherical to intermediate shape, thereby reducing the expression of OPA1 and MFN2. Studies have shown that downregulation of OPA1 could cause a decrease in mitochondrial membrane potential and trigger cell apoptosis[50]. MFN2 is known to mediate mitochondrial fusion and apoptosis[51]. The obtained results suggest that induction of LX-2 cells by TGF-β1 and/or IPA appears to modulate mitochondrial shape and size, as well as activation state and network complexity.
       Our results indicate that the combination treatment of TGFβ-1 and IPA could reduce mtDNA and cell morphological parameters by regulating the mRNA expression of fibrosis, apoptosis and survival-related genes in apoptosis-evading cells. Indeed, IPA decreased the mRNA expression level of AKT1 and important fibrosis genes such as COL1A2 and MMP2, but increased the expression level of CASP8, which is associated with apoptosis. Our results showed that after IPA treatment, BAX expression decreased and mRNA expression of TIMP1 family subunits, BCL-2 and NF-κB increased, suggesting that IPA could stimulate survival signals in hematopoietic stem cells (HSCs) that evade apoptosis. These molecules may act as pro-survival signals in activated hematopoietic stem cells, which may be associated with increased expression of anti-apoptotic proteins (such as Bcl-2), decreased expression of pro-apoptotic BAX, and a complex interplay between TIMP and NF-κB [5, 7]. IPA exerts its effects through PXR, and we found that combination treatment with TGF-β1 and IPA increased PXR mRNA expression levels, indicating suppression of HSC activation. Activated PXR signaling is known to inhibit HSC activation both in vivo and in vitro [52, 53]. Our results indicate that IPA may participate in the clearance of activated HSCs by promoting apoptosis, reducing fibrosis and mitochondrial metabolism, and enhancing survival signals, which are typical processes that convert the activated HSC phenotype to an inactivated one. Another possible explanation for the potential mechanism and role of IPA in apoptosis is that it scavenges dysfunctional mitochondria primarily through mitophagy (intrinsic pathway) and the extrinsic TNF signaling pathway (Table 1), which is directly linked to the NF-κB survival signaling pathway (Supplementary Figure 7). Interestingly, IPA-related enriched genes are able to induce pro-apoptotic and pro-survival signals in the apoptotic pathway [54], suggesting that IPA may induce the apoptotic pathway or survival by interacting with these genes. However, how IPA induces apoptosis or survival during HSC activation and its mechanistic pathways remain unclear.
       IPA is a microbial metabolite formed from dietary tryptophan via the gut microbiota. Studies have shown that it has anti-inflammatory, antioxidant, and epigenetic regulatory properties in the intestinal environment.[55] Studies have shown that IPA can modulate intestinal barrier function and reduce oxidative stress, which may contribute to its local physiological effects.[56] In fact, IPA is transported to target organs via the circulation, and since IPA shares a similar major metabolite structure with tryptophan, serotonin, and indole derivatives, IPA exerts metabolic actions resulting in competitive metabolic fates.[52] IPA may compete with tryptophan-derived metabolites for binding sites on enzymes or receptors, potentially disrupting normal metabolic pathways. This highlights the need for further studies on its pharmacokinetics and pharmacodynamics to better understand its therapeutic window.[57] It remains to be seen whether this can also occur in hematopoietic stem cells (HSCs).
       We acknowledge that our study has some limitations. To specifically examine associations related to IPA, we excluded patients with type 2 diabetes mellitus (T2DM). We acknowledge that this limits the broad applicability of our findings to patients with type 2 diabetes mellitus and advanced liver disease. Although the physiological concentration of IPA in human serum is 1–10 μM [11, 20], a concentration of 1 mM IPA was chosen based on the highest non-toxic concentration [15] and the highest rate of apoptosis, with no difference in the percentage of the necrotic cell population. Although supraphysiological levels of IPA were used in this study, there is currently no consensus regarding the effective dose of IPA [52]. Although our results are significant, the broader metabolic fate of IPA remains an active area of ​​research. Moreover, our findings on the association between serum IPA levels and DNA methylation of liver transcripts were obtained not only from hematopoietic stem cells (HSCs) but also from liver tissues. We chose to use human LX-2 cells based on our previous findings from transcriptome analysis that IPA is associated with hematopoietic stem cell (HSC) activation [15], and HSCs are the major cells involved in the progression of liver fibrosis. The liver is composed of multiple cell types, so other cell models such as hepatocyte-HSC-immune cell co-culture system combined with caspase activation and DNA fragmentation as well as the mechanism of action including protein level should be considered to study the role of IPA and its interaction with other liver cell types.