GPT2-mediated Glutamate to α-Ketoglutaric Acid Conversion in Cardiac Fibroblast Promotes Fibrosis

Visual Abstract

Highlights

• •Identifies GPT2 as a novel metabolic checkpoint that drives cardiac fibrosis by reprogramming glutamate to α-ketoglutaric acid conversion.
• •Discovers that MSC-EVs exert antifibrotic effects via delivery of miR-30c-5p, which specifically targets and inhibits GPT2 expression.
• •Elucidates a precise mechanism where GPT2 catalysis fuels ATP overproduction, directly linking metabolic shift to collagen overproduction.

Summary

Cardiac fibroblasts play a key role in heart fibrosis, but how their metabolism changes during this process is unclear. This study shows that mesenchymal stem cell–derived extracellular vesicles reduce fibrosis in mice after heart pressure overload. Under stress, fibroblasts increase mitochondrial adenosine triphosphate by boosting glutamate metabolism, especially through the enzyme GPT2, which converts glutamate to α-ketoglutarate. This leads to fibroblast activation and excess collagen. Inhibiting GPT2 via microRNA-30c-5p delivered by mesenchymal stem cell–derived extracellular vesicles -reduces fibrosis in both mice and human cells. GPT2 inhibition also works in other organs, suggesting broad therapeutic potential.

Fibrosis is a common pathological process of tissue repair and the important cause of adverse cardiac remodeling after heart failure,¹ which is characterized by excessive extracellular matrix synthesis and deposition. Cardiac fibrosis is mainly mediated by the resident cardiac fibroblasts, which are activated into myofibroblasts under stress condition.² Though myofibroblasts play an important role in maintaining the structural integrity of the heart during pathological conditions,³ sustained myofibroblast activation can lead to progressive interstitial fibrosis, which contributes to left ventricular dilation and dysfunction.⁴ Therefore, it is necessary to identify the molecular mechanisms that maintain myofibroblast activation and explore new therapies to interrupt the progression of cardiac fibrosis.

So far, many studies have confirmed the therapeutic effect of bone marrow mesenchymal stem cells on acute myocardial infarction, mainly through the paracrine mechanism. This suggests that mesenchymal stem cell–derived extracellular vesicles (MSC-EVs) transplantation may provide a new strategy for the treatment of cardiac fibrosis. Current studies have shown that EVs are the main components of paracrine function of stem cells. Through EVs as carriers, various cytokines and RNA and other active components can be transported between different cells to promote myocardial repair. Meanwhile several studies have reported that myofibroblast activation is accompanied by metabolism alteration, such as increased glycometabolism.5, 6, 7 Previous metabolism research on cardiac fibrosis has revealed the roles of mitochondrial reactive oxidation species production and mitochondrial Ca²⁺ uptake in myofibroblast activation.^8,9 However, the detailed changes in cellular metabolism and specific metabolic targets in cardiac fibrosis remain poorly understood. MSC-EVs are lipid bilayer vesicles that have been reported to have therapeutic effects on cardiac remodeling.^10,11 Besides, MSC-EVs are ideal capsules that can be easily remodeled, making them an effective tool for delivering molecules to intervene in specific targets. Therefore, we took advantage of these characteristics of MSC-EVs to investigate the metabolic alterations and key molecular targets in fibroblasts under disease conditions.

In the present study, we demonstrated that GPT2-mediated glutamate to α-ketoglutarate (α-KG) conversion plays a crucial role in cardiac fibroblast activation under stress condition, this biological process was confirmed through metabolomics that proposed that GPT2 catalyzes the conversion from glutamate to α-KG, leading to increased adenosine triphosphate (ATP) level and subsequent deactivation of adenosine monophosphate–activated protein kinase (AMPK), ultimately promotes cardiac fibrosis after pressure overload condition. Moreover, microRNA-30c-5p (miR-30c-5p) in EVs was determined to inhibit GPT2 expression. The profibrosis utilization of GPT2 and the metabolic alterations were also verified in multiple organs and human fibroblasts. To summarize, we unraveled glutamate to α-KG conversion and ATP level alteration in mitochondria as pivotal targets for treating cardiac fibrosis under pressure-overload conditions. Among these biological processes, GPT2 is the key coordinator in accommodating fibrosis procedures in failing hearts.

Methods

Extended methods are provided in the Supplemental Appendix.

Human heart tissue sample and ethics statement

Human cardiac fibroblasts (hCFs) were obtained from the donated hearts of nondiseased patients. The baseline characteristics of patients are detailed in Supplemental Table 1. All subjects were duly informed and written consent by the patient or their relatives. All studies were approved by the Ethics Review Committee from the Second Affiliated Hospital of Zhejiang University.

Experimental material was taken from left ventricle free wall. Sample preparation was done at 4 °C. Sample procurement and preparation were performed according to a human research subject protocol. To isolate cardiac fibroblasts, hearts were excised and rinsed in cold Hank balanced salt solution. Then tissues were minced and digested with type II collagenase (100 U/mL) (Worthington) and pancreatin (0.6 mg/mL) (Sigma) at 37 °C for 15 minutes. Supernatant was collected and resuspended in Dulbecco modified Eagle medium (DMEM; Gibco, Thermo Fisher Scientific, Inc) with 10% fetal bovine serum (Thermo Fisher Scientific, Inc) and 1% antibiotic solution. Cell pellets were subjected to digestion until the digestion fluid became clear (5-6 times). All the collected and resuspended supernatant was centrifuged for 10 minutes at 600g and resuspended in DMEM with 10% fetal bovine serum and 1% antibiotic solution. Cells were plated in 12-well plates or 100-mm dishes (Corning) and allowed to attach for 48 hours before the first media change.

Mouse model of TAC

Transverse aortic constriction (TAC) surgery was employed to induce pressure overload as a well-established model of cardiac fibrosis.¹² Mice were anesthetized by an intraperitoneal injection of pentobarbital sodium (50 mg/kg, Sigma) and intubated prior to opening the thoracic cavity. A left parasternal incision was performed until the third rib. The lobes of the thymus were separated to expose the transverse aorta. A blunt 27-gauge needle was tied to the transverse aorta between the brachiocephalic and left carotid arteries using a 6-0 silk suture. The needle was removed and the thoracic wall was sutured with 4-0 silk suture, then the thoracic cavity was closed and mice were allowed to recover.¹³ Sham-operated mice underwent surgery but without suture placement around the aorta. At no point were the group sizes increased after start of the experiment. Mice were randomly assigned to sham or TAC surgery. If sham and TAC mice were housed together, we performed an ear notch during the surgery to distinguish the treatments at the time of harvest.

Ultrasound-guided intramyocardial injection

At 7 days post-TAC, mice were subjected to 2.5%-3.5% isoflurane anesthesia with 100% medical O₂ (1 L/min). Mice were placed supine on an adjustable stage (Vevo2100, Visual Sonics) with the left rib cage orientated to face the syringe holder. Then the intramyocardial injection was performed via 30-gauge needle by using micromanipulator. A single injection of 1 mg/kg EVs diluted in phosphate-buffered saline (PBS) to the total volume of 20 μL or 20-μL PBS vehicle control solution was slowly injected into the anterior wall of the left ventricle in a controlled manner by a blinded operator. Successful injection was visualized by a change in brightness in the injected area on the echocardiogram.

Metabolomic profiling

We analyzed the lysates of neonatal rat cardiac fibroblasts (NRCFs) in a transforming growth factor-β (TGF-β) group and a TGF-β+EVs group by targeted metabolomics. To measure intracellular metabolite pools, cells were washed with 2 mL of ice-cold PBS 3 times and metabolites were extracted with 200 μL of 80:20 (volume/volume) methanol: water solution at −20 °C. The cells were quickly scraped on dry ice. All metabolite extractions were stored at −80 °C for at least 1 hour, followed by centrifugation, twice at 20,000g for 20 minutes to obtain protein-free metabolite extraction. Metabolomic analysis of metabolite pool was conducted by Dalian Dashuo Information Technology Co, LTD.

Seahorse XF analysis and metabolites detection

The Mito Stress Test Kit (103015–100, Agilent Technologies) was used to measure the oxygen consumption rate (OCR) in fibroblasts treated differently. The Agilent Seahorse XF Real-Time ATP Rate Assay Kit (103592–100, Agilent Technologies) was used to detect the ATP production rates of mitochondrial oxidative phosphorylation. A Seahorse Bioscience XF96 extracellular flux analyzer was employed to measure OCR and ATP production rate. Before metabolism measurement, the probe plate was hydrated with high-performance liquid chromatography grade water in a CO₂-free incubator and 20,000 NRCFs per well were plated into XF96 cell culture microplates overnight. To evaluate OCR and ATP production rate, 10 mmol/L glucose, 2 mmol/L glutamine (Gln), and 1 mmol/L sodium pyruvate were added to XF basal media (103575-100, Agilent Technologies). For OCR values detection, mitochondrial parameters were measured by sequential addition of 1.5 μmol/L oligomycin, 2.0 μmol/L carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone, and 0.5 μmol/L rotenone (Agilent Technologies). For ATP production rate detection, only 1.5 μmol/L oligomycin and 0.5 μmol/L rotenone were needed. Finally, OCR and ATP production rates of mitochondrial oxidative phosphorylation and glycolysis were determined and analyzed on the Agilent Seahorse Bioscience XF96 Extracellular Flux Analyzer according to the manufacturer’s instructions and protocols (Seahorse Bioscience). Gln (MAK438), glutamate (MAK004) and α-KG (MAK054; all from Sigma-Aldrich) were evaluated by a metabolite detection assay kit according to the manufacturer’s instructions and protocols. The materials used in the study are listed in the Supplemental Tables 2 to 4.

Quantification and statistical analysis

All statistical analyses were performed using GraphPad Prism (version 8.0, GraphPad Software). For in vitro study, all biological replicates using primary cultured cells correspond to independent experiments from distinct expansions and passage numbers, with technical replicates. Because each experimental data set is an average of a huge number of cultured cells, the data were assumed normally distributed based on the central limit theorem. Shapiro-Wilk test (P < 0.05) was used to test normality of all data obtained from in vivo study. Data with normal distribution are represented as the mean ± SEM. Comparisons among multiple groups were performed using 1-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons. For comparisons between 2 groups, analyses were performed using Student’s t-test. All experiments were performed independently, only within-test corrections were made. P < 0.05 was considered statistically significant.

Results

Dampening ATP provision restricts activation of cardiac fibroblast and inhibits fibrosis under stress condition

Energy metabolism has been reported to play an essential role in myofibroblast activation and organ fibrosis.⁶ To assess ATP concentration changes of CFs during cardiac fibrosis and EV treatment, we extracted EVs from MSCs that have not been genetically edited and identified them through 3 approaches (Supplemental Figures 1A to 1C). The efficient uptake of EVs by CFs was also verified (Supplemental Figure 1D). Then we constructed pressure-overload induced cardiac remodeling models in mice using TAC surgery and administered EVs by intramyocardial injection at 1, 3, and 5 weeks post TAC (Supplemental Figures 2A and 2B). The retention of intramyocardially injected DiD (Beyotime, C1995S)-labeled MSC-EVs in mice was detected by in vivo imaging at 4, 7, 10, and 14 days post injection (Supplemental Figures 2C and 2D). The stability of the model was ensured by measuring the pressure at the site of constriction (Supplemental Figure 2E). As shown by echocardiography, pathological staining, and immunoblotting, EV treatment significantly improved TAC-induced cardiac dysfunction and cardiac fibrosis compared with the PBS treatment (Figures 1A to 1I, Supplemental Figures 2F to 2I). Then we isolated CFs from the heart of each group and measured the ATP contents. Interestingly, ATP levels were increased in CFs isolated from TAC-induced pressure-overload hearts. In contrast, EV treatment is associated with decreased ATP levels after TAC surgery (Figure 1J), suggesting that the provision of ATP was associated with fibroblast-mediated cardiac fibrosis.

Figure 1.

Dampening ATP Provision Restricts Activation of CFs and Inhibits Fibrosis Under Stress Condition

(A) Representative images of M-mode echocardiography in each group at different time points. (B and C) Quantitative analysis of echocardiography; n = 6 in each group. (D) Representative images of picrosirius red staining from heart sections on mice from each experimental group. Bar = 1.5 mm. (E) Summary data of fibrosis area of picrosirius red staining of heart sections; n = 6 in each group. (F) Representative images of periostin immunostaining images from heart sections on mice from each experimental group. Bar = 50 μm. (G) Summary data of the proportion of periostin positive area in each high-power field (HPF); n = 6 in each group. (H) Fibrotic protein expressions including fibronectin, periostin, and α-smooth muscle actin (α-SMA) were evaluated in sham-operated mice and aortic constricted mice that either received phosphate-buffered saline (PBS) or extracellular vesicle (EV) transplantation 6 weeks after transverse aortic constriction (TAC). (I) Quantitative analysis of fibrotic proteins is shown; n = 6 in each group. (J) Adenosine triphosphate (ATP) content in cardiac fibroblasts (CFs) isolated from sham and TAC mice that either received PBS or EV injection; n = 6 in each group. (K) Fibrotic proteins were detected in Dulbecco modified Eagle medium (DMEM)-treated neonatal rat cardiac fibroblasts (NRCFs) and transforming growth factor-β (TGF-β) stimulated NRCFs that either received no therapy or EVs. (L) Quantitative results of fibrotic proteins were plotted; n = 6 in each group. (M) ATP concentration in different groups of NRCFs was measured by the Firefly method; n = 6 in each group. (N) Calculated ATP production rate was detected by Seahorse XFe Extracellular Flux Analyzer. (O) Quantitative analysis of calculated ATP production rate generated by glycolysis (glycoATP) and mitochondrial oxidative phosphorylation (mitoATP) is shown; n = 7 in DMEM group, n = 6 in TGF-β group, n = 8 in TGF-β+EVs group. (P) The continuous oxygen consumption rates (OCRs) of fibroblasts were recorded and analyzed using Seahorse XFe Extracellular Flux Analyzer. (Q) Quantitative analysis of basal and maximal respiration was plotted; n = 6 in DMEM group, n = 8 in TGF-β group, n = 8 in TGF-β+EVs group. (R) Immunoblots for fibrotic proteins were conducted in DMEM-treated NRCFs and TGF-β–stimulated NRCFs that received negative control (NC) treatment, EVs, 2-DG, or CPI-613. (S) Quantitative bar graphs of fibrotic proteins; n = 6 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in B, C, E, G, I, J, L, M, O, Q, and S. All data are presented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. cTnI = cardiac troponin I; DAPI = 4ʹ,6-diamidino-2,-phenylindole; EF = ejection fraction; FCCP = carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone; FS = fractional shortening; OLIGO = oligonucleotide; rot/AA = rotenone/Antimycin A; 2-DG = 2-Deoxy-D-glucose; CPI-613 = Devimistat.

To further investigate the alterations of ATP concentration in fibroblasts under stress condition, we subjected NRCFs to TGF-β stimulation with either PBS or EV intervention and then used Seahorse XFe Extracellular Flux Analyzer to measure the ATP production rate of each group. As shown in Figures 1K and 1L, TGF-β stimulation significantly up-regulated fibrotic protein expression in fibroblasts, which was significantly attenuated by EV treatment. Similar to the results of in vivo experiments, the activation of fibroblasts induced by TGF-β in vitro also led to an increase in intracellular ATP concentration, whereas MSC-EVs could counteract this effect (Figure 1M). Meanwhile, the calculated ATP production rate, which consists of both glycolysis and mitochondrial ATP production rate, was drastically elevated in TGF-β–stimulated fibroblasts, whereas EV treatment significantly reduced this elevation, especially the mitochondrial ATP production rate (Figures 1N and 1O). Additionally, we measured OCR, which reflects the mitochondrial respiratory activity of the fibroblasts. As depicted in Figures 1P and 1Q, TGF-β stimulation led to a rapid and sustained increase in the OCR of NRCFs, which was effectively reversed by EV treatment. To further confirm the necessity of ATP provision in the myofibroblast activation process, we added 2-DG (2-Deoxy-D-glucose), the competitive inhibitor of glucose metabolism, or CPI-613 (Devimistat), the unique inhibitor of mitochondrial metabolism, respectively, to the fibroblasts exposed to the TGF-β stimulation. Both inhibitors significantly reduced the expression levels of collagen and myofibroblast marker proteins (Figures 1R and 1S), indicating that ATP provision, especially from mitochondria, may contribute to myofibroblast activation and collagen deposition.

Taken together, the data displayed herein clearly demonstrate that adequate energy supply, which is mainly reflected by ATP concentration, is necessary for myofibroblast activation, and the process of ATP provision is reprogrammed to a higher level during the pathogenesis of cardiac fibrosis. Depriving ATP provision in fibroblasts through reducing mitochondrial oxygen consumption capacity can effectively inhibit fibrosis.

Glutamate to α-KG conversion mediated regulation of ATP plays a pivotal role in myofibroblast activation

To investigate the underlying mechanisms regulating ATP concentration in fibroblasts under stress conditions, we conducted targeted metabolomics analysis on TGF-β–stimulated fibroblasts treated with PBS or EVs. Among the metabolites with marked differences between the 2 groups, L-glutamic acid, which is also named glutamate, was ranked at the top based on fold of change and significance (Figures 2A and 2B). Kyoto Encyclopedia of Genes and Genomes pathway analysis suggested that alanine, aspartate, and glutamate metabolism were strongly involved in the regulation process from EV-treated fibroblasts with TGF-β stimulation (Figure 2C). Hence, combining the top altered metabolite and metabolic pathway, we determined glutamic to α-KG conversion as the key pathway changed under TGF-β stimulation and reversed by EV treatment. First, glutamate content assay confirmed that the intracellular glutamate content was significantly decreased in TGF-β–treated fibroblasts, whereas EV treatment prevented the glutamate decrease despite no difference in Gln content (Figures 2D and 2E). Meanwhile, the content of α-KG, downstream derivate of glutamate within mitochondria, which also play a significant role in oxidative phosphorylation, was significantly elevated in TGF-β–exposed fibroblasts, while being refrained after EV treatment (Figure 2F). Furthermore, fibroblasts isolated from TAC- and EV-treated mice via flow cytometry further corroborated consistent changes in glutamate and α-KG levels aligned with in vitro findings (Supplemental Figures 3A and 3B), indicating the blockade of metabolic transformation from glutamate to α-KG plays the main role during myofibroblast activation.

Figure 2.

Glutamate to α-KG Conversion Regulates ATP Concentration, Playing a Pivotal Role in Myofibroblast Activation

(A) Heatmap plot of the TGF-β–stimulated fibroblasts received control or EV treatment for 24 hours. (B) Volcano map visualized the statistical results of differential metabolites. (C) Metabolomic pathway analysis assisted with Kyoto Encyclopedia of Genes and Genomes analysis was conducted to identify metabolic pathways correlated to EVs treatment. (D to F) The content of glutamine (Gln) (D), glutamate (E), and α-ketoglutarate (α-KG) (F) were measured in DMEM-treated NRCFs and TGF-β–stimulated NRCFs that either received no therapy or EVs by metabolite assay kit; n = 3 in each group. (G) Immunoblots for extracellular matrix proteins were conducted in DMEM-treated NRCFs and TGF-β–stimulated NRCFs that received different treatments, respectively. (H) Quantitative bar graphs of fibrotic proteins; n = 6 in each group. (I) Fibrotic proteins were detected in DMEM-treated NRCFs and TGF-β–stimulated NRCFs that received no therapy or EVs, in the absence and presence of α-KG supplementation. (J) Quantitative bar graphs of fibrotic proteins; n = 6 in each group. (K) Phosphorylation (p) level of SMAD2/3 and adenosine monophosphate-activated protein kinase (AMPK) of fibroblasts treated differently were analyzed by Western blot. (L) The summary data of blot (K) were shown; n = 6 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in D, E, F, H, J, and L. All data are presented as mean ± SEM., ∗p < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. P > 0.05 is not significant (NS). ADP = adenosine diphosphate; FC = fold-change; t = total; UDP = uridine phosphorylase; other abbreviations as in Figure 1.

To further validate whether glutamate to α-KG conversion is necessary for TGF-β–induced myofibroblast activation, we first supplemented Gln, the upstream substrate of glutamate, or depleted Gln by either directly removing Gln in the cell medium or adding a selective glutaminase inhibitor BPTES. Notably, glutamate reduction by removing Gln from the medium or adding BPTES successfully inhibited TGF-β–stimulated profibrotic protein expression, whereas reinforcement of Gln augmented profibrotic features (Figures 2G and 2H). Second, we added exogenous cell-permeable α-KG in the culture medium of CFs and noted that α-KG itself cannot directly lead to myofibroblast activation under normal conditions, which may be due to the low energy demand of CFs at quiescent status. However, exogenous α-KG supplement strongly inhibited the antifibrosis effects of EVs on fibroblasts under TGF-β stimulation (Figures 2I and 2J), accompanied by escalated basal and maximal mitochondrial respiration rate, resulting in increased calculated ATP production rate (Supplemental Figures 3C to 3E). Thus, we reasoned that glutamate to α-KG conversion, which in turn leads to augmented mitochondrial ATP provision, is crucial during TGF-β–induced myofibroblast activation, and that curbing this metabolic process is sufficient to inhibit the activation of CFs.

AMPK is a crucial energy sensor that detects the intracellular ATP/AMP ratio.^9,14 AMPK activation inhibits TGF-β signaling in myofibroblast differentiation.^15,16 To further investigate the underlying mechanism by which α-KG and mitochondrial ATP mediate myofibroblast activation, we tested the phosphorylation of AMPK and downstream SMAD2/3. We discovered that AMPK phosphorylation was elevated in EV-treated NRCFs compared with the TGF-β group, whereas SMAD2/3 was dephosphorylated in the EV-treated group compared with the TGF-β group. On the other hand, restoration of energy supply by α-KG supplement counteracted the effect of EVs on AMPK, leading to dephosphorylation of AMPK and subsequent phosphorylation of SMAD2/3 (Figures 2K and 2L). To summarize, the conversion of glutamate to α-KG up-regulates mitochondrial ATP in CF, resulting in dephosphorylation of AMPK and SMAD2/3 phosphorylation, which leads to myofibroblast activation.

GPT2 orchestrates glutamate to α-KG conversion during myofibroblast activation

To deeply investigate the potential molecular mechanism that mediates glutamate to α-KG conversion alteration in CFs under stress, we screened gene and protein expression of related regulatory enzymes along glutamic conversion associate pathways, covering deamination, transamination, and decarboxylation. We demonstrated that GPT2 was markedly increased by TGF-β stimulation in fibroblasts and down-regulated by EV treatment at both the protein and mRNA levels, whereas the expression of GPT1, glutamic oxalacetic transaminase, GLS, glutamic acid decarboxylase, and glutamate dehydrogenase remained unaffected (Figures 3A and 3B, Supplemental Figures 4A to 4C). Western blot analysis confirmed up-regulated GPT2 expression after TGF-β stimulation, which was decreased under EV treatment (Supplemental Figures 4D and 4E). Therefore, these data indicate that GPT2 expression, which was inhibited by EV treatment, regulates the pivotal downstream conversion of glutamate to α-KG in CFs.

Figure 3.

GPT2 Orchestrates Glutamate to α-KG Conversion Under Fibrotic Condition in CFs

(A) Glutamic metabolism associated enzymes were screened by Western blot. (B) Quantitative summary of changes in glutamic metabolism associated enzymes; n = 3 in each group. The content of glutamate (C) and α-KG (D) were measured in TGF-β–treated NRCFs with and without GPT2 small, interfering RNA (siRNA), DMEM-treated NRCFs and TGF-β–stimulated NRCFs that either received scramble or EVs were served as control treatment; n = 3 in each group. (E) Calculated ATP production rate was detected by Seahorse XFe Extracellular Flux Analyzer. (F) Quantitative analysis of calculated ATP production rate; n = 4 in each group. (G) The continuous OCRs of fibroblast were recorded and analyzed using Seahorse XFe Extracellular Flux Analyzer. (H) Quantitative analysis of basal and maximal respiration; n = 8 in DMEM group, n = 8 in TGF-β group, n = 6 in TGF-β+EVs group, n = 7 in TGF-β+NC group, n = 7 in TGF-β+siGPT2 group. (I) Fibrotic proteins and GPT2 levels in TGF-β–treated NRCFs with and without GPT2 siRNA, DMEM-treated NRCFs, and TGF-β–stimulated NRCFs that either received scramble or EVs served as control treatment. (J) Quantitative summary of changes in fibrosis and GPT2; n = 6 in each group. The content of glutamate (K) and α-KG (L) were measured in NRCFs of different groups; n = 3 in each group. (M) The calculated ATP production rate in NRCFs was detected by Seahorse XFe Extracellular Flux Analyzer. (N) Quantitative analysis of calculated ATP production rate; n = 4 in DMEM group, n = 5 in TGF-β group, n = 5 in TGF-β+EVs group, n = 4 in TGF-β+EVs+NC group, n = 5 in TGF-β+EVs+OE-GPT2 group. (O) The continuous OCRs of fibroblast were recorded and analyzed using Seahorse XFe Extracellular Flux Analyzer. (P) Quantitative analysis of basal and maximal respiration was plotted; n = 4 in DMEM group, n = 4 in TGF-β group, n = 4 in TGF-β+EVs group, n = 4 in TGF-β+EVs+NC group, n = 5 in TGF-β+EVs+OE-GPT2 group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in B, C, D, F, H, J, K, L, N, and P. All data are presented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ^NSP > 0.05. GDH = glucose dehydrogenase; GOT = glutamic oxalacetic transaminase; OE = overexpression; other abbreviations as in Figures 1 and 2.

GPT2 is located in mitochondria and functionally catalyzes glutamate to α-KG.¹⁷ To identify whether GPT2 is a key factor in regulating fibroblast energy production and the myofibroblast activation process, the glutamate and α-KG contents of CFs after small interfering (si)-GPT2 or negative control (NC) intervention were evaluated under TGF-β stimulation. As expected, glutamate content was significantly elevated when GPT2 was knocked down. Contrarily, α-KG was significantly down-regulated in the context of si-GPT2 compared with NC (Figures 3C and 3D), indicating that GPT2 absence inhibited the conversion of glutamate to α-KG, thus inducing the accumulation of glutamate. In addition, real-time ATP production rate measurement exhibited a decreased calculated ATP production rate in the absence of GPT2 (Figures 3E and 3F). Further OCR quantification showed that GPT2 knockdown inhibited the overall mitochondrial oxidation level of fibroblast, which was manifested by the downward movement of the overall OCR curve in the si-GPT2 group (Figure 3G). Moreover, the calculated mitochondrial basal respiration and maximum respiration capacity were also significantly decreased in the si-GPT2 group compared with the NC group (Figure 3H). Finally, knocking down GPT2 markedly decreased TGF-β–induced profibrotic protein expression compared with NC under TGF-β stimulation (Figures 3I and 3J). These results indicate that GPT2 inhibition can effectively cut off mitochondrial ATP supply and reduce fibrosis.

To further understand the effect of EVs and GPT2 on the activity of glutamate to α-KG conversion, ¹³C tracer analysis was applied to demonstrate the process (Supplemental Figure 5A). As shown in Supplemental Figure 5B, M+5-labeled metabolites, including glutamate and α-KG from [U-¹³C] Gln, were significantly reduced in MSC-EV–treated NRCFs as well as in si-GPT2 group when compared with corresponding control groups. Similar results were observed in the tricarboxylic acid (TCA) cycle of M+ 4-labeled metabolites (including succinate, fumarate, and malate) from [U-¹³C] Gln. These findings suggest that suppressing the expression of GPT2 inhibits TGF-β–induced significant increase in glutaminolysis in NRCFs, which directly decreases the content of intracellular α-KG and hampers the TCA cycle.

To further confirm the pivotal role of GPT2 in fibroblast metabolism, the glutamate and α-KG contents of CFs after overexpression of GPT2 were evaluated under TGF-β stimulation and EV treatment. We found that overexpression of GPT2 in fibroblasts abolished the inhibition effect of EVs on glutamate to α-KG conversion under TGF-β stimulation, resulting in overconsumed glutamate and increased α-KG content compared with the NC group (Figures 3K and 3L). Furthermore, as expected, GPT2 overexpression counteracted the inhibitory effect of EVs on fibroblast mitochondrial ATP production rate compared with NC (Figures 3M and 3N). The inhibited oxygen-consuming capacity by EVs reflected by OCR was also reversed by GPT2 compensation (Figures 3O and 3P). Meanwhile, GPT2 overexpression abolished the antifibrosis effect of EVs in fibroblast and exhibited increased expression of fibronectin, collagen I, and periostin, as well as fibroblast proliferation (Supplemental Figures 6A to 6C). Altogether, these data clearly illustrate that GPT2 overexpression is sufficient to promote the conversion of glutamate to α-KG in the context of EV utilization, thus reversing the protective effect conferred in fibroblast under TGF-β stimulation and deteriorating fibrotic process.

GPT2 in CFs exacerbates cardiac remodeling under pressure-overload conditions

We next investigated whether inhibiting GPT2 expression in CFs could reduce cardiac remodeling in mice under pressure-overload conditions. To specifically regulate GPT2 expression in CFs, we generated periostin promoter-specific GPT2 knockdown lentivirus and a control lentivirus, delivered to the heart via intramyocardial injection (Supplemental Figure 7A). The early expression of periostin and the corresponding decrease in GPT2 expression confirmed the efficacy of lentiviral injection (Supplemental Figures 7B and 7C). Meanwhile, we isolated CFs, cardiomyocytes, endothelial cells, and immune cells, respectively, from lentivirus-injected mice 6 weeks after TAC (Supplemental Figure 7D). We detected significant down-regulation of GPT2 in PDGFRA-positive fibroblasts, but nearly no changes in cardiomyocytes and endothelial cells, as well as in immune cells (Supplemental Figures 7E to 7G). As shown on echocardiography, mice in the GPT2 knockdown group performed better cardiac function and reduced cardiac remodeling 6 weeks after TAC surgery (Figures 4A to 4C, Supplemental Figures 7H and 7I). Picrosirius red staining showed that mice in the CF-GPT2 knockdown group exhibited less cardiac fibrosis area under pressure-overload condition compared to the NC group (Figures 4D and 4E, Supplemental Figures 7K and 7L), which was further confirmed by interstitial periostin staining (Figures 4F and 4G). Western blot indicated that GPT2 knockdown down-regulated fibrotic protein expression induced by pressure-overload (Figures 4H and 4I).These data demonstrated that GPT2 knockdown in CFs alleviates cardiac remodeling and improves cardiac function under pressure-overload conditions.

Figure 4.

Targeting GPT2 in CFs Regulates Cardiac Fibrosis Under Pressure-overload Conditions

(A) Representative images of M-mode echocardiography (baseline and 6 weeks after TAC). (B and C) Quantitative analysis of echocardiography in each group at different time points (baseline and 6 weeks after TAC); n = 6-9 in each group. (D) Representative images of picrosirius red staining from heart sections on mice from each experimental group. Bar = 1.5 mm. (E) Summary data of fibrosis area; n = 6 in each group. (F) Representative images of periostin immunostaining from heart sections on mice from each experimental group. Bar = 50 μm. (G) Summary data of the proportion of periostin positive area to the total number of cells in each HPF; n = 6-7 in each group. (H) Fibrotic protein expressions were evaluated in each group 6 weeks after TAC. (I) Quantitative analysis of fibrotic proteins is shown; n = 6 in each group. (J) Representative images of M-mode echocardiography (baseline and 6 weeks after TAC). (K and L) Quantitative analysis of echocardiography in each group at different time points. (M) Representative images of picrosirius red staining from heart sections on mice from each experimental group. Bar = 1.5 mm. (N) Summary data of fibrosis area; n = 6 in each group. (O) Representative images of periostin immunostaining from heart sections on mice from each experimental group. Bar = 50 μm. (P) Summary data of the proportion of periostin-positive area to the total number of cells in each HPF; n = 6-7 in each group. (Q) Fibrotic protein expressions were evaluated in each group 6 weeks after TAC. (R) Quantitative analysis of fibrotic proteins is shown; n = 6 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in B, C, E, G, I, K, L, N, P, and R. All data are presented as mean ± SEM. ∗∗P < 0.01, ∗∗∗P < 0.001. sh = short hairpin; other abbreviations as in Figures 1 and 3.

To further confirm whether GPT2 expression in CFs has a profibrotic effect under pressure-overload conditions, we generated periostin promoter-specific GPT2 overexpression lentivirus and its control lentivirus. Lentiviruses were delivered to the heart of mice by intramyocardial injection, and then mice were subjected to TAC or sham surgery (Supplemental Figure 8A). Western blot confirmed the specific overexpression of GPT2 in CFs following lentiviral injection (Supplemental Figures 8B to 8D). Six weeks after TAC, echocardiography showed that mice in the GPT2 overexpression group exhibited significantly reduced cardiac function compared with mice in the NC group (Figures 4J to 4L, Supplemental Figures 8E to 8G). Picrosirius red staining revealed expanded cardiac fibrosis area in mice with CF-GPT2 overexpression under pressure-overload condition (Figures 4M and 4N, Supplemental Figures 8H and 8I), and interstitial periostin staining further confirmed this result (Figures 4O and 4P). Concurrently, Ki67 staining further confirmed that GPT2 overexpression promotes cell proliferation during fibroblast activation (Supplemental Figures 8J and 8K). Western blot also indicated that GPT2 overexpression significantly up-regulated fibrotic protein expression induced by pressure-overload (Figures 4Q and 4R). Taken together, these in vivo observations suggest that GPT2 in CFs plays a pivotal role in exacerbating cardiac remodeling under pressure-overload conditions.

miR-30c-5p inhibits the expression of GPT2 in CFs

Because the expression of the GPT2 can be inhibited by EVs, we reasoned that miRNAs are likely responsible for the negative regulation of GPT2 and the interception of glutamate to α-KG conversion in fibroblasts. We conducted sequencing of EV-encapsulated miRNAs and performed cross-referenced screening against multiple biological databases (Targetscan, miRDB, and miRwalk) to identify putative miRNAs targeting the GPT2 3' untranslated region (UTR) (Figure 5A). The integrated analysis revealed consistently high expression of miR-30c-5p, which aligned with both the bioinformatic predictions and experimental RNA validation (Figure 5B). To further confirm that miR-30c-5p can negatively regulate the expression of GPT2, we generated luciferase reporter plasmids incorporated the indicated region of GPT2 miRNA, NC, or a mutation form, followed by conducting dual luciferase reporter assay. The binding of miR-30c-5p with GPT2 miRNA 3' UTR was confirmed as reflected by inhibited luciferase activity in wild-type mice, whereas no significant difference in luciferase activity was detected in 3ʹ UTR mutation group (Figures 5C and 5D). To determine whether miR-30c-5p can be delivered to fibroblasts via EVs for future translational utilization, a Cy5 tag was added to the mimic of miR-30c-5p. Confocal microscopy revealed effective uptake of miR-30c-5p by fibroblasts (Supplemental Figure 9A).

Figure 5.

miR-30c-5p Inhibits the Expression of GPT2 in CFs

(A) Upset plot generated from different databases and micro-RNA sequence (miRNA-seq) indicated high expression of miR-30c-5p. (B) Polymerase chain reaction quantification of miR-30c-5p in the fibroblasts treated with EVs. (C and D) Dual luciferase reporter assay confirmed the binding of miR-30c-5p with GPT2 3' untranslated region (3utr). Sequence information was shown in (C). (E) Western blot identification for GPT2 and fibrotic proteins in DMEM-treated NRCFs and TGF-β–stimulated NRCFs that incubated with EVs. EVs were obtained from mesenchymal stem cells that were pretreated with anti-miR-30c-5p (EVs^anti-miR) or scramble (EVs^NC). (F) Quantitative analyses of (E) were plotted; n = 6 in each group. (G) The continuous OCRs of fibroblast treated as in (E) were recorded and analyzed using Seahorse XFe Extracellular Flux Analyzer. (H) Quantitative analysis of basal and maximal respiration were plotted; n = 4 in each group. (I) The continuous OCRs of NRCFs in TGF-β–treated NRCFs with and without mimic of miR-30c-5p were recorded and analyzed using Seahorse XFe Extracellular Flux Analyzer, and DMEM-treated NRCFs and TGF-β stimulated NRCFs that either received scramble or EVs served as control treatment. (J) Quantitative analysis of basal and maximal respiration were plotted; n = 5 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in G, H, and J. Independent 2-sample Student’s t-test was conducted in B and D. All data are presented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ^NSP > 0.05. m- = mouse-; mmu- = Mus musculus-; MUT = mutant; pmirGLO = pmirGLO Dual-Luciferase miRNA Target Expression Vector; WT = wild-type; other abbreviations as in Figure 1.

To validate whether miR-30c-5p delivered by EVs can inhibit GPT2 expression and subsequent glutamate to α-KG conversion in CFs, we generated EVs containing anti–miR-30c-5p oligonucleotide (EVs^anti-miR), mimic of miR-30c-5p (EVs^over-miR), or NC (EVs^NC), respectively. As shown in Western blotting, EVs^anti-miR treatment could not effectively inhibit TGF-β–induced GPT2 and fibrotic protein expression in NRCF compared with the EVs^NC treatment group (Figures 5E and 5F). Meanwhile, EVs^anti-miR was unable to inhibit TGF-β–induced glutamate to α-KG conversion, whereas EVs^over-miR could further inhibit glutamate to α-KG conversion compared to scramble control (EVs^NC) (Supplemental Figures 9B to 9E). OCR was measured simultaneously, whereas EVs^NC significantly inhibited TGF-β–induced mitochondrial oxygen consumption elevation including both basal and maximal levels, neutralizing miR-30c-5p in EVs, which caused them to lose these effects (Figures 5G and 5H). Additionally, miR-30c-5p also significantly inhibited the proliferation of fibroblasts stimulated by TGF-β (Supplemental Figure 9F). Correspondingly, direct administration of miR-30c-5p mimics also significantly inhibited TGF-β–induced mitochondrial oxygen consumption elevation (Figures 5I and 5J). These consequences indicate that miR-30c-5p interferes with the expression of GPT2, leading to a decrease in glutamate to α-KG conversion in CFs.

Modification of miR-30c-5p counterbalances GPT2 expression and restricts fibrosis in vivo

To evaluate whether the suppression of GPT2 by miR-30c-5p plays a pivotal role under pressure-overload condition in vivo, and to explore whether miR-30c-5p can be delivered to CFs via extracellular vesicles in vivo for clinical translational practice. Mice were injected intramyocardially with PBS, EVs, EVs^NC, or EVs^over-miR, respectively, after surgically induced TAC. Echocardiography revealed that mice from the EVs^over-miR group had better cardiac performance as displayed by enhanced ejection fraction, fractional shortening, and smaller left ventricle chambers compared with mice from the EVs^NC group at 6 weeks after TAC (Figures 6A and 6B, Supplemental Figures 10A to 10D). Meanwhile, picrosirius red staining revealed less cardiac fibrosis area in the EVs^over-miR group compared with the EVs^NC group (Figures 6C and 6D, Supplemental Figures 10E and 10F), and immunofluorescence staining showed a significant reduction in interstitial periostin-positive area (Supplemental Figures 10G and 10H). As shown in Western blotting, the EVs^over-miR group showed a lower expression of GPT2, which in turn modulated the expression level of fibrotic proteins (Figures 6E and 6F).

Figure 6.

Modification of miR-30c-5p Counterbalances GPT2 Expression and Restrict Fibrosis In Vivo

(A and B) Quantitative analysis of echocardiography in sham-operated mice and aortic-constricted mice that either received no therapy or different EVs transplantation at different time points (baseline and 6 weeks after TAC). EVs were obtained from mesenchymal stem cells that were pretreated with EVs^over-miR)or EVs^NC; n = 6 in each group. (C) Representative images of picrosirius red staining from heart sections on mice from each experimental group, scale bar = 1.5 mm. (D) Summary data of fibrosis area; n = 6 in each group. (E) Fibrotic protein expressions were evaluated in mice treated as in (A) 6 weeks after TAC. (F) Quantitative analysis of fibrotic proteins is shown; n = 6 in each group. (G and,H) Quantitative analysis of echocardiography in sham-operated mice and aortic constricted mice that either received no therapy or different EVs transplantation at different time points (baseline and 6 weeks after TAC). EVs were obtained from mesenchymal stem cells that were pretreated with EVs^anti-miR or EVs^NC; n = 6 in each group. (I) Representative images of picrosirius red staining from heart sections on mice from each experimental group. Bar = 1.5 mm. (J) Summary data of fibrosis area; n = 6 in each group. (K) Fibrotic protein expressions were evaluated in mice treated as in (G) 6 weeks after TAC. (L) Quantitative analysis of fibrotic proteins is shown; n = 6 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in A, B, D, F, G, H, J, and L. All data are presented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ^NSP > 0.05. Abbreviations as in Figures 1 and 5.

To further demonstrate that miR-30c-5p in EVs negatively regulates GPT2 expression and thus exerts therapeutic effects under pressure-overload conditions in vivo, EVs^anti-miR were injected intramyocardially to mice after TAC surgery. Echocardiography showed that neutralization of miR-30c-5p in EVs by oligonucleotide deteriorated cardiac function in mice compared with the EVs^NC group (Figures 6G to 6H, Supplemental Fig. 11A to 11C). Meanwhile, picrosirius red staining detected expanded fibrosis area in EVs^anti-miR–treated mice than in EVs^NC under pressure-overload conditions (Figures 6I and 6J, Supplemental Figures 11D and 11E). Interstitial periostin staining further validated this result (Supplemental Figures 11F and 11G). Western blot also revealed a notable elevation in the expression of GPT2 and fibrotic protein in the hearts of EVs^anti-miR–treated mice compared to the EVs^NC group (Figures 6K and 6L). Collectively, these observations from both gain and loss of function perspectives indicate that miR-30c-5p and its downstream target GPT2 potently regulate cardiac fibrosis after TAC.

GPT2 and miR-30c-5p have conserved function on glutamate to α-KG conversion in hCFs and fibrotic diseases in other organs

To validate the clinical relevance of GPT2 and miR-30c-5p in fibrosis regulation, we isolated CFs from the left ventricle-free wall of patients without heart disease after obtaining informed consent (Supplemental Table 1). Consistent with the previous observations, GPT2 and fibrosis genes expression were significantly up-regulated under TGF-β stimulation in hCFs. Though EVs treatment can significantly inhibit these up-regulations, α-KG supplementation could effectively abolish the inhibited effect (Figures 7A and 7B). Meanwhile, the increase in ATP concentration induced by TGF-β stimulation can also be inhibited by EVs in hCFs (Figure 7C). To verify the effect of direct intervention in GPT2 expression in hCFs, we constructed siRNAs and overexpressed lentiviruses of human GPT2. Western blotting shows that hCFs with GPT2 silencing expressed lower levels of fibrotic proteins under TGF-β stimulation (Figures 7D and 7E), whereas overexpression of GPT2 significantly reversed the antifibrosis effect of EVs (Figures 7F and 7G). Consistent with these molecular changes, a mimic of miR-30c-5p pretreated EVs inhibited GPT2 and fibrotic proteins expression in TGF-β–stimulated hCFs, more effectively than EVs only. Neutralization of miR-30c-5p in EVs significantly abolishes the ability of EVs to suppress GPT2 and fibrotic proteins expression in TGF-β–stimulated hCFs (Figures 7H to 7K).

Figure 7.

GPT2 and miR-30c-5p Have Conserved Function on Glutamate to α-KG Conversion in hCFs

(A) Fibrotic proteins were detected in DMEM and TGF-β–stimulated human cardiac fibroblasts (hCFs) that received no therapy or EVs, in the absence and presence of α-KG supplementation. (B) Quantitative bar graphs of fibrotic proteins and GPT2; n = 3 in each group. (C) ATP content was assessed in different treated hCFs; n = 3 in each group. (D) Fibrotic proteins and GPT2 levels in TGF-β–treated hCFs with and without GPT2 siRNA treatment. (E) Quantitative summary of changes in fibrotic proteins and GPT2; n = 3 in each group. (F) Fibrotic proteins and GPT2 levels in TGF-β–treated hCFs with and without GPT2 overexpression. (G) Quantitative summary of changes in fibrotic proteins and GPT2; n = 3 in each group. (H) Fibrotic proteins and GPT2 were detected in DMEM hCFs and TGF-β–stimulated hCFs that incubated with EVs. EVs were obtained from mesenchymal stem cells that were pretreated with EVs^over-miR or EVs^NC. (I) Quantitative summary of changes in fibrotic proteins and GPT2; n = 3 in each group. (J) Fibrotic proteins and GPT2 were detected in DMEM hCFs and TGF-β–stimulated hCFs that incubated with EVs. EVs were obtained from mesenchymal stem cells that were pretreated with inhibitor of miR-30c-5p (EVs^anti-miR) or EVs^NC. (K) Quantitative summary of changes in fibrotic proteins and GPT2; n = 3 in each group. One-way analysis of variance followed by Tukey post hoc test for multiple pairwise comparisons was conducted in B, C, E, G, I, K, and M. All data are presented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Abbreviations as in Figure 1, Figure 2, Figure 3 and 5.

To investigate whether regulating GPT2 expression has a similar antifibrosis effect and glutamate to α-KG conversion regulation mechanisms as in other organs, we isolated lung and kidney primary fibroblasts from mice. Consistent with the previous results, GPT2 and fibrosis genes expression were significantly up-regulated under TGF-β stimulation in lung fibroblasts, and exogenous α-KG supplement significantly reversed the inhibitory effect of EVs. Direct knockdown of GPT2 in lung fibroblasts inhibited the expression of TGF-β–induced fibrotic proteins expression (Supplemental Figures 12A and 12B). Besides, a mimic of miR-30c-5p inhibited the expression of GPT2 and fibrotic proteins, whereas EVs^anti-miR reduced the beneficial effect of EVs^NC (Supplemental Figures 12C and 12D). Meanwhile, these experiments recapitulated the same results in fibroblasts isolated from mice kidneys (Supplemental Figures 12E to 12H), indicating the general significance of GPT2 expression and glutamate to α-KG conversion during the organ fibrosis process, and miR-30c-5p have the potential therapeutic effects for organ fibrosis on the basis of the inhibition of GPT2.

Discussion

Cardiac fibrosis, characterized by the excessive deposition of extracellular matrix proteins, is a pivotal pathological process in adverse cardiac remodeling and the progression to heart failure.¹⁸ Central to this maladaptive scarring is the phenotypic transformation of resident CFs into hyperactive and collagen-secreting myofibroblasts. A growing evidence underscores that this activation is not merely a response to soluble signals but is underpinned by fundamental metabolic reprogramming, wherein fibroblasts shift their energy production and biosynthetic pathways to meet the demands of proliferation, contraction, and matrix synthesis.¹ Whereas the involvement of general metabolic pathways like glycolysis¹⁹ has been documented, the specific metabolic checkpoints and enzymatic regulators that orchestrate this reprogramming in CFs remain incompletely mapped, leaving a significant gap in our ability to design targeted metabolic therapies. In this study, we identify GPT2 as a critical and previously overlooked metabolic node essential for cardiac fibrosis. We demonstrate that under the duress of pressure-overload in vivo or profibrotic stimulation in vitro, GPT2 expression is specifically up-regulated in CFs. This enzyme catalyzes the transamination of glutamate to yield α-KG, a reaction that serves as a crucial anaplerotic entry point into the TCA cycle. The consequent enhancement of mitochondrial oxidative phosphorylation fuels a state of ATP overproduction, which we posit is a necessary bioenergetic prerequisite for sustaining the energetically costly myofibroblast phenotype and its resultant excessive collagen secretion. More innovatively, we delineate a precise therapeutic countermeasure: MSC-EVs serve as natural delivery vehicles for miR-30c-5p, a microRNA that specifically targets and silences GPT2 expression. This axis potently disrupts the profibrotic metabolic flow, yielding significant antifibrotic effects both in cellular models and in the pressure-overloaded murine heart, thereby revealing a novel mechanism for the recognized therapeutic benefits of MSC-EVs.

Studies have shown that the metabolic state of CFs changes under cardiac pathological conditions.^5,6,20,21 Following tissue injury, fibroblasts undergo significant metabolic changes to properly perform their cellular functions, such as proliferation and activation. It has been reported that myofibroblast activation is tightly correlated to glutaminolysis in the lungs.²² Augmented conversion from Gln to glutamate and its downstream metabolites confers substrates to collagen production and increased demand of energy. In vivo studies have also demonstrated that inhibition of glutamate to α-KG conversion ameliorates bleomycin-induced pulmonary fibrosis.²³ Our findings position GPT2 within a broader and evolving understanding of nitrogen metabolism in fibrosis. Recent seminal work has elegantly illuminated the glutaminolysis pathway, wherein GLS converts Gln to glutamate, which is then further metabolized to α-KG, driving fibroblast activation and cardiac fibrosis.^8,24 Our discovery of GPT2 reveals a complementary and reinforcing metabolic circuit. Whereas glutaminolysis utilizes Gln as the primary nitrogen donor, GPT2 provides a direct conduit for converting free glutamate pools into α-KG. This indicates a sophisticated layer of metabolic flexibility and redundancy, whereby activated fibroblasts, under pathological stress, can harness multiple substrate streams including extracellular Gln via GLS and potentially intracellular glutamate via GPT2 to ensure a robust and sustained elevation of intracellular α-KG. This convergence on α-KG as a central metabolite underscores its nonredundant, pleiotropic role in fibrosis. Beyond its canonical function as a TCA cycle intermediate for ATP generation, α-KG is an indispensable cofactor for a family of α-KG–dependent dioxygenases. These enzymes, including JmjC-domain containing histone demethylases and the ten-eleven translocation family of DNA demethylases, are key regulators of the epigenetic landscape.²⁵ Therefore, GPT2-driven α-KG production likely exerts a dual, synergistic profibrotic influence, which directly fuels the bioenergetic engine required for cellular work and simultaneously potentiating the epigenetic machinery necessary for establishing and maintaining the profibrotic transcriptional program, such as the expression of α-smooth muscle actin and collagen gene,⁸ the latter of which warrants further investigation. This dual role elevates GPT2 from a simple metabolic enzyme to a key regulatory checkpoint that integrates cellular energy status with gene expression control, making it a strategically attractive therapeutic target.

The identification of miR-30c-5p as the upstream endogenous inhibitor of GPT2, packaged within therapeutically active MSC-EVs, adds a crucial layer of mechanistic depth and therapeutic specificity to our study. This defined axis, from vesicular cargo to metabolic enzyme, moves beyond the often-observed but poorly explained effects of stem cell therapies. It offers a rational, coordinated therapeutic strategy against the fibrotic niche. By down-regulating GPT2, miR-30c-5p simultaneously attenuates the energetic supply that sustain the myofibroblast state. The utilization of MSC-EVs as the delivery platform is particularly salient. These natural nanovesicles possess inherent biocompatibility, low immunogenicity, and a demonstrated tropism for injured tissues, including the stressed heart.^26,27 They act as protective capsules for their miRNA cargo, ensuring its delivery to target cells while avoiding rapid degradation. Our observation of EV uptake specifically by activated fibroblasts in fibrotic areas further supports the potential for targeted intervention. The use of engineered or naturally selected EVs as vehicles for specific regulatory RNAs exemplifies a promising frontier in precision medicine for fibrotic diseases.

Furthermore, our data suggest that the pathogenic role of this GPT2-mediated pathway exhibits a degree of conservation that extends beyond the rodent heart. The inhibition of GPT2 effectively attenuated profibrotic responses not only in hCFs but also in mouse fibroblasts derived from lung and kidney tissues. This indicates that the reliance on glutamate conversion via GPT2 may represent a common metabolic vulnerability shared by activated fibroblasts across different species and organ systems. In idiopathic pulmonary fibrosis, pulmonary fibroblasts exhibit a significant increase in glycolytic flux,^28,29 which serves not only as a pathway for rapid ATP generation, but also provides essential carbon skeletons for the synthesis of proteins and lipids. Similarly, in pressure-overload–induced cardiac fibrosis models, CFs also demonstrate a reliance on glycolytic activity.²⁹ Moreover, an increased dependence on Gln metabolism constitutes another major commonality. Research based on metabolic flux analysis confirmed that modulating Gln metabolic reprogramming can alleviate pulmonary fibrosis.³⁰ Similarly, in CFs, inhibiting Gln metabolism can block their activation and collagen production.²⁰ Such conservation implies that the fundamental metabolic demands of the fibrotic phenotype that characterized by the need for increased energy and biosynthetic precursors are met through similar pathways regardless of tissue origin, which significantly amplifies the therapeutic effects. Importantly, our in vivo models indicate that interventions via fibroblast-specific lentivirus achieve significant antifibrotic efficacy without observable detrimental effects on cardiomyocyte function under stress. This cell-type specificity is a critical therapeutic advantage, because it suggests the possibility of disrupting the fibrotic process without adversely affecting the already compromised energy metabolism of cardiomyocytes in the failing heart.

Study limitations

We acknowledge several limitations that contextualize our findings and guide future research: First, our detailed mechanistic metabolic analyses were performed primarily in neonatal rodent fibroblasts, a standard model that robustly recapitulates key activation pathways.³¹ Whereas we have validated the antifibrotic effect of miR-30c-5p in adult mouse CFs, potential differences in metabolic baselines between neonatal and adult cells warrant further investigation. Second, all experiments were performed using male mice, which might limit the universality of our findings. Finally, the human data are derived from a limited sample size. Future studies involving larger cohorts with detailed metabolic profiling of fibroblasts from patients with diverse etiologies of heart failure are essential for clinical translation.

Conclusions

Our results demonstrate that miR-30c-5p inhibits the expression of GPT2 in CFs, hinders the glutamate to α-KG conversion, leading to a decrease in intracellular ATP levels, and thus activates the AMPK phosphorylation and eventually inhibits cardiac fibrosis under pressure-overload circumstance. Importantly, the metabolic pathway mediated by GPT2 is partially conserved across different organs, like in lung and kidney under fibrotic pathological condition, rendering a silver lining in the treatments in a wide range of fibrotic diseases in different organs. Collectively, this study reveals the relationship between glutamate to α-KG conversion and myofibroblast activation, which may lead to the innovative development of new therapeutic targets and strategies for the treatment of tissue fibrosis.

Perspectives.

COMPETENCY IN MEDICAL KNOWLEDGE: Fibrosis is not merely a passive scarring process but can be actively driven by specific metabolic reprogramming in fibroblasts, centered on the GPT2-mediated glutamate conversion, informing future strategies to prevent or reverse adverse myocardial remodeling.

TRANSLATIONAL OUTLOOK: GPT2 inhibitors or miR-30c-5p mimics may serve as therapeutic agents for heart failure or fibrosis in other organs and metabolites associated with glutamate conversion could represent potential biomarkers for fibrotic diseases.

Data availability

The data that support the findings of this study are available from the corresponding author on reasonable request.

Funding Support and Author Disclosures

This work was supported by the National Key R&D Program of China (2023YFA1800700 to X.Y.H.), grants from National Natural Science Foundation of China (82225004 and 82430013 to X.Y.H.), Noncommunicable Chronic Diseases - National Science and Technology Major Project of China (2024ZD0521800 to X.Y.H.), the National Key R&D Program of China (2019YFA0110400 to J.W.), grants from National Natural Science Foundation of China (U22A20267 and 82030014 to Dr Wang., 82070253 Dr Hu., 82370240 to Dr. Ni), The Key R&D projects of Zhejiang Province (2021C03097 to Dr Wang., 2023C03086 to Dr Hu). Financial support was provided by Binjiang Institute of Zhejiang University (ZY202205SMKY001 to Dr Wang., ZY202205SMKY001 to Dr Hu). Fundamental Research Funds for the Central Universities (226-2023-00156 to Dr Ni). The authors have reported that they have no relationships relevant to the contents of this paper to disclose.