Shared Tissue-Level Molecular Signatures and Immunometabolic Associations in Nonalcoholic Steatohepatitis and Chronic Kidney Disease: Potential Endothelial Relevance

Abstract

Purpose

NASH and CKD are closely associated epidemiologically and share metabolic, inflammatory, and fibrotic features. This study integrated bulk-tissue transcriptomic datasets from both diseases to investigate shared tissue-level molecular signatures and their potential endothelial and immunometabolic relevance.

Methods

Transcriptomic datasets from the Gene Expression Omnibus (GEO) for NASH and CKD were integrated to screen for common differentially expressed genes (co-DEGs). A protein-protein interaction network was constructed to identify ten hub genes. Single-sample Gene Set Enrichment Analysis (ssGSEA) and single-gene GSEA were performed to elucidate functional pathways. In vitro, human umbilical vein endothelial cells (HUVECs) were stimulated with palmitic acid (PA) and indoxyl sulfate (IS) to model NASH and CKD, respectively. Quantitative real-time PCR (qPCR), Western blotting, tube formation assays, and migration assays were employed to validate the expression of hub genes and functional phenotypic changes in endothelial cells.

Results

A total of 156 commonly differentially expressed genes were identified. Protein–protein interaction network analysis revealed 10 hub genes, all of which showed a downregulated expression pattern. Immune infiltration analysis indicated that CCL2 and ICAM1 were positively correlated with the abundance of MDSCs and Th1 cells, respectively. Single-gene GSEA further showed that downregulation of these hub genes was closely associated with suppressed mitochondrial energy metabolism, mitophagy, and antioxidant defense pathways. qPCR and Western blotting confirmed downregulation in cell models, and treatment with PA/IS significantly impaired HUVEC tube formation and migration.

Conclusion

This study identified a shared tissue-level molecular signature with potential endothelial relevance across NASH and CKD. The findings suggest that endothelial alterations may participate in an interconnected metabolic–redox–immune–fibrotic network. Because bulk-tissue data cannot determine cellular origin, the observed downregulation may also reflect changes in cellular states or tissue composition. However, the cellular origin and underlying causal mechanisms remain to be established.

Introduction

Nonalcoholic fatty liver disease (NAFLD) is the most prevalent chronic liver disease worldwide, with an estimated prevalence of 25.2%.¹ As a progressive disorder, NAFLD begins with hepatic steatosis resulting from excessive lipid accumulation in hepatocytes and may advance to typical nonalcoholic steatohepatitis (NASH). The pathological hallmarks of NASH include hepatic steatosis, hepatocyte injury (ballooning degeneration), inflammation, and varying degrees of fibrosis.^2–4 The disease can further progress to cirrhosis, hepatocellular carcinoma, and end-stage liver disease.⁵ Meanwhile, chronic kidney disease (CKD), often referred to as a silent epidemic, has a global prevalence of approximately 10% and continues to rise with population aging,^6,7 posing an increasingly severe global health threat. Studies have shown that NAFLD is closely associated with the development and progression of CKD.^8,9 Both conditions share multiple metabolic risk factors, including obesity,¹⁰ diabetes,¹¹ and hypertension,¹² although the common molecular mechanisms remain unclear.

From a systems-level perspective, NASH and CKD may be viewed as interconnected chronic disorders with partially overlapping pathological processes, including metabolic stress, mitochondrial dysfunction, oxidative stress and redox imbalance, chronic inflammation, and extracellular matrix deposition.^13–15 Lipid overload in NASH and the accumulation of uremic toxins in CKD represent distinct but partially overlapping forms of metabolic stress. These disturbances may impair mitochondrial energy metabolism and autophagy, increase reactive oxygen species production, and weaken antioxidant defenses, thereby interacting with persistent inflammation and fibrosis.^15–18 Rather than acting independently or in a strictly sequential manner, these processes may form an interconnected regulatory network through reciprocal feedback. Oxidative stress and redox imbalance may therefore serve as shared regulatory hubs linking metabolic disturbances with endothelial alterations and immune activation.

Endothelial cells lie at the interface of metabolic, vascular, immune, and fibrotic processes and may represent an important component of this regulatory network.^19,20 In NASH, liver sinusoidal endothelial cells may lose their fenestrae and undergo basement membrane deposition and capillarization, resulting in impaired permeability and material exchange. These alterations may aggravate hepatocellular injury, promote inflammatory-cell recruitment, and contribute to hepatic fibrogenesis.²¹ Renal endothelial cells comprise multiple specialized subpopulations adapted to distinct local microenvironments.²² In CKD, endothelial abnormalities may include increased permeability, vascular rarefaction, inflammatory activation, and metabolic remodeling.^23,24 These alterations may contribute to glomerulosclerosis and tubulointerstitial fibrosis and thereby promote CKD progression. Endothelial dysfunction may therefore represent a shared pathological feature of NASH and CKD and participate in the proposed liver–kidney axis. However, it should not be regarded as an isolated or sufficient explanation for the association between the two diseases. Parenchymal injury, immune-cell infiltration, stromal activation, endothelial adaptation or senescence, shifts in endothelial subpopulations, and broader changes in tissue composition may also contribute. It remains unclear to what extent the shared tissue-level molecular features of NASH and CKD are related to endothelial biology and how these alterations participate in the broader immunometabolic network.

This study identified a shared tissue-level molecular signature with potential endothelial relevance across NASH and CKD and suggested its possible involvement in immune, mitochondrial metabolic, and redox-regulatory networks. The HUVEC experiments provided complementary evidence that metabolic stress can induce molecular and functional endothelial alterations. However, because bulk-tissue data cannot determine cellular origin, future single-cell or spatial transcriptomic studies will be needed to definitively establish the endothelial origin of this signature and its functional relevance.

Materials and Methods

Data Sources for NAFLD and CKD

Disease-related datasets for NAFLD and CKD were identified by searching relevant keywords in the GEO database (https://www.ncbi.nlm.nih.gov/geo/), and gene expression matrices were subsequently downloaded. Using R software (version 4.5.2), the GSE89632 and GSE66494 datasets were selected for differential expression gene (DEG) analysis. The GSE89632 dataset included samples from 20 patients with simple steatosis (SS), 19 patients with NASH, and 24 healthy controls (HC), and was generated using the GPL14951 platform (Illumina HumanHT-12 WG-DASL V4.0 R2 expression beadchip). The GSE66494 dataset comprised samples from 54 patients with CKD and 8 healthy controls, and was generated using the GPL17077 platform [Agilent-014850 Whole Human Genome Microarray 4x44K G4112F (Probe Name Version)].

Functional Enrichment Analysis of Common Differentially Expressed Genes

To further explore the biological functions and signaling pathways associated with the common DEGs between the NAFLD and CKD groups, Gene Ontology (GO) analysis was employed to evaluate the biological functions of these genes, focusing on three classification domains: cellular component (CC), molecular function (MF), and biological process (BP). Additionally, the Kyoto Encyclopedia of Genes and Genomes (KEGG) database was utilized to identify signaling pathways related to the common DEGs. GO and KEGG pathway enrichment analyses were performed using the clusterProfiler package (version 4.18.4) in R, with a significance threshold set at P.adj < 0.05. The final results were visualized using the ggplot2 package (version 4.0.2) in R.

Protein-Protein Interaction (PPI) Network Construction and Key Gene Screening

Protein-protein interaction (PPI) network analysis was performed using the STRING online tool (https://cn.string-db.org), with a confidence threshold set at 0.4. The results were visualized using Cytoscape software (version 3.9.1). The CytoHubba plugin within Cytoscape was used to identify ten key genes among the common DEGs.

Selection of Candidate Genes for in vitro Validation

To guide experimental validation, we selected a functionally representative subset from the ten PPI-identified hub genes. A densely interconnected submodule comprising TLR2, ICAM1, CCL2, and CXCL1 was prioritized based on its higher connectivity and closer functional coordination (deeper color intensity in Figure 1B), covering innate immune sensing (TLR2), endothelial adhesion (ICAM1), chemokine-mediated recruitment (CCL2), and inflammatory endothelial activation (CXCL1). To further extend the functional breadth, EDN1 was additionally selected as a representative of vascular tone and endothelial homeostasis. This strategy ensured coverage of key biological pathways while maintaining experimental feasibility.

Figure 1.

PPI Network and Hub Gene Identification. (A) The 156 co-DEGs were imported into the STRING database, and a PPI network was constructed to identify interacting genes. (B) Using the cytoHubba plugin in Cytoscape software, ten hub genes were identified: TLR2, ICAM1, CXCL1, CCL2, THBS1, EDN1, RRS1, BIRC3, RELB, and JAK3.

ssGSEA Immune Infiltration Analysis

Single-sample Gene Set Enrichment Analysis (ssGSEA) immune infiltration analysis was performed using the GSVA package (version 2.4.8) in R (version 4.5.2) on both the NASH and CKD datasets to evaluate the relative abundance of immune cell subpopulations in each sample. Boxplots were generated using the ggplot2 package (version 4.0.2) to visually compare the abundance differences of various immune cell subpopulations between healthy controls and disease groups. Correlations among different immune cell subpopulations were displayed using heatmaps generated with the corrplot package (version 0.9.5). Furthermore, the ten key genes identified previously were incorporated into the analysis and integrated with the immune cell abundance matrix. Spearman correlation coefficients were calculated using the psych package (version 2.3.3), and the correlations between key genes and immune cells were displayed via heatmaps. For key gene-immune cell pairs exhibiting significant correlations, scatter plots were generated using the ggplot2 package, with regression lines, correlation coefficients, and P-values included.

RT-qPCR

Total RNA was extracted from cells using a total RNA extraction kit. The concentration and purity of the RNA were determined using a Nanodrop spectrophotometer. One microgram of RNA was used to synthesize cDNA according to the manufacturer’s instructions for the reverse transcription kit. Quantitative PCR (qPCR) was performed in a 10 μL reaction system containing: 2 μL of cDNA, 5 μL of SYBR Green Master Mix, 0.5 μL each of forward and reverse primers, and ddH₂O added to reach the final volume. The reaction conditions were as follows: 95°C for 3 minutes, followed by 40 cycles of 95°C for 10 seconds and 60°C for 30 seconds. GAPDH was used as an internal reference gene, and relative gene expression levels were calculated using the 2⁻ΔΔCt method. The corresponding primer sequences are listed in Table 1 below.

Table 1.

Primer Sequence for RT-qPCR

Gene (ID)	Forward (5’-3’)	Reverse (5’-3’)
EDN1	TACTTCTGCCACCTGGACAT	CTTCCTCTCACTAACTGCTGAT
TLR2	TCCAGCACACGAATACACAGT	AGGAACAGAGCACAGCACAT
CCL2	TTCTGTGCCTGCTGCTCAT	ACTTGCTGCTGGTGATTCTTC
CXCL1	CTGCACCCAAACCGAAGTC	AGCTTCAGGGTCAAGGCAAG

Cell Culture

The HUVECs used in the present study were purchased from American Type Culture Collection (cat. no. CRL-1730) through the Institute of Infectious Diseases, Ditan Hospital, Beijing, China (20). These cells were obtained at passage 2 and used between passages 2 and 5 in all experiments (21). Cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; cat. no. A4192101; Gibco; Thermo Fisher Scientific, Inc), supplemented with 1% penicillin/streptomycin (cat. no. 15140–122; Gibco; Thermo Fisher Scientific, Inc). and 10% fetal bovine serum (FBS; cat. no. A5670701; Gibco; Thermo Fisher Scientific, Inc.) and incubated in a humidified atmosphere containing 5% CO2 at 37°C. HUVECs were passaged when reaching 80% confluence.

HUVECs were selected because they provide a reproducible and widely used model for assessing general endothelial stress responses and permit matched molecular and functional assays. Palmitic acid and indoxyl sulfate were used to model selected lipid- and uremic-toxin-related stressors, respectively.

Cell Migration Assay

Prior to cell seeding, the bottom of each well in a 6-well plate was marked with a line. A total of 2.5×10⁵ cells/mL cell suspension (2 mL per well) was seeded. When cell confluence reached 95%, the cells were serum-starved for 16 hours. A vertical scratch was made using a 200 μL pipette tip, and the cells were washed twice with PBS. Fresh culture medium was then added. Images were captured at 0 and 24 hours post-scratching. The wound area was measured using ImageJ, and the migration rate was calculated as follows: Migration rate (%) = (1 − wound area at endpoint / wound area at initial) × 100.

Angiogenesis Assay

Matrigel was thawed at 4°C overnight, and all equipment and culture plates were pre-cooled at −20°C. On ice, DMEM and Matrigel were mixed at a 1:1 ratio. A total of 100 μL of the Matrigel mixture was added to each well of a 24-well plate and incubated at 37°C for 45 minutes to allow polymerization. Cells were digested and seeded at a density of 1.5×10⁵ cells per well in 200 μL of suspension. After 6 hours, tubular structures were observed under a microscope, and five random fields were captured per well. The total tube length, number of nodes, and number of branches were analyzed using the ImageJ Angiogenesis Analyzer.

Statistical Analysis

All statistical analyses were performed using R software. DEGs were identified with the thresholds of |log2(Fold Change)| > 1 and adjusted P-value (P.adj) < 0.05. For GO and KEGG pathway enrichment analyses, a P.adj < 0.05 was considered statistically significant. Two statistical methods were employed for the validation of co-DEGs: For data meeting the assumptions of normality and homogeneity of variance, assessed by the Shapiro–Wilk test and Levene’s test, respectively, an independent t-test was used. When Levene’s test indicated a violation of the homogeneity of variance assumption, Welch’s corrected t-test was applied. All experiments were performed with at least three independent biological replicates (n = 3 per group). Data are presented as mean ± standard deviation (SD) unless otherwise specified. For all statistical tests, a P-value < 0.05 was considered to indicate statistical significance.

Results

Dataset Characteristics

In the GEO database, “nonalcoholic fatty liver disease” and “chronic kidney disease” were used as keywords for data retrieval. All included datasets were derived from human tissue samples, and two datasets were ultimately selected for analysis. The basic characteristics of each dataset are presented in Table 2.

Table 2.

Dataset Characteristics

Datasets	Organism	Source	N	Platforms
GSE89632	Liver	Human	63	GPL14951
GSE66494	Kidney	Human	62	GPL17077

Common Differentially Expressed Genes in NASH and CKD

Two datasets were included in this study: the NASH dataset GSE89632 and the CKD dataset GSE66494, both of which had undergone data normalization to ensure the reliability of subsequent results. Between NASH patients and healthy controls, 1892 DEGs were identified, of which 1029 were upregulated and 863 were downregulated and are displayed using heatmaps and volcano plots (Figure 2A, Figure 3A and C). Between CKD patients and their respective healthy controls, 4289 DEGs were detected, including 782 upregulated and 3507 downregulated genes which are displayed using heatmaps and volcano plots (Figure 2A, Figure 3B and D). Venn diagram analysis revealed 156 co-DEGs shared between NASH and CKD, comprising 39 upregulated and 117 downregulated genes (Figure 2A).

Figure 2.

Screening and Enrichment Analysis of Shared Differentially Expressed Genes. (A) Venn diagram illustrating the expression of common DEGs between the two datasets. (B) GO analysis of the 156 co-expressed DEGs, including biological process (BP) and molecular function (MF) categories. (C) KEGG pathway enrichment analysis revealed that the 156 co-DEGs were significantly enriched in multiple signaling pathways (adjusted p< 0.05).

Figure 3.

Visualization of Differentially Expressed Genes in theGSE89632 and GSE66494 Databases. (A, B) Heatmaps displaying the top 50most significantly differentially expressed genes in the NASH and CKDdatabases, respectively. Red and blue colors indicate the most significantlyupregulated and downregulated differentially expressed genes (DEGs) in thesamples. (C, D) Volcano plots representing the DEGs in the NASH and CKDdatasets, respectively.

GO and KEGG Enrichment Analysis of Co-DEGs

To explore the biological functions of the co-DEGs, GO and KEGG pathway enrichment analyses were performed. As illustrated in the figure, GO analysis indicated enrichment in BP such as positive regulation of leukocyte migration, regulation of granulocyte chemotaxis, and cellular response to tumor necrosis factor. In terms of MF, enrichment was observed in cytokine activity, chemoattractant activity, G protein-coupled receptor binding, and chemokine receptor binding (Figure 2B).

KEGG pathway enrichment analysis demonstrated that the co-DEGs were significantly enriched in several inflammation- and immune-related pathways, including the TNF signaling pathway, NF-κB signaling pathway, NOD-like receptor signaling pathway, IL-17 signaling pathway, and chemokine signaling pathway. Additionally, enrichment was noted in the AGE-RAGE signaling pathway, which is associated with metabolic and vascular dysfunction (Figure 2C).

PPI Network Analysis

As shown in the figure, the 156 co-DEGs were imported into the STRING database. PPI network analysis constructed using STRING identified interacting genes (Figure 1A). Subsequently, using the cytoHubba plugin in Cytoscape software, ten hub genes were screened out: TLR2, ICAM1, CXCL1, CCL2, THBS1, EDN1, RRS1, BIRC3, RELB, and JAK3 (Figure 1B).

ssGSEA Immune Infiltration Analysis in NASH and CKD Datasets

Immune infiltration analysis was performed separately for the NASH and CKD datasets using the ssGSEA algorithm in R. As shown in Figure 4, based on the ssGSEA scores of each sample, box plots were used to compare the relative abundance differences of various immune cell subpopulations between HC and NASH patients (Figure 4A). Furthermore, correlations among different immune cell subpopulations were visualized using a heatmap (Figure 4B). The ten key genes identified were further incorporated into the analysis, and their correlations with each immune cell type were displayed in a heatmap (Figure 4C). Notably, a significant positive correlation was observed between the key gene CCL2 and MDSCs (R = 0.59, p = 0.009), and between ICAM1 and Type 1 T helper cells (R = 0.77, p < 0.001) (Figure 4D). As depicted in the figure, the same analytical approach was applied to the CKD dataset (Figure 5A–C). In this dataset, a significant positive correlation was also found between CCL2 and MDSCs (R = 0.54, p < 0.001), and between ICAM1 and Type 1 T helper cells (R = 0.75, p < 0.001) (Figure 5D).

Figure 4.

GSEA Results for the NASH Dataset. (A) Box plots showing differences inthe relative abundance of various immune cell subpopulations between HC andNASH patients. (B) Heatmap displaying correlations among different immune cellsubpopulations in the NASH dataset. (C) Heatmap illustrating thecorrelations between the ten key genes and various immune cell types in theNASH dataset. (D) Individual demonstration of the significant positivecorrelations between the key gene CCL2 and myeloid-derived suppressor cells(MDSCs), and between ICAM1 and Th1 cells. P < 0.05, P <0.01 < 0.001, P ≥ 0.05.

Figure 5.

GSEA Results for the CKD Dataset. (A) Box plots showing differences in the relative abundance of various immune cell subpopulations between HC and CKD patients. (B) Heatmap displaying correlations among different immune cell subpopulations in the CKD dataset. (C) Heatmap illustrating the correlations between the ten key genes and various immune cell types in the CKD dataset. (D) Individual demonstration of the significant positive correlations between the key gene CCL2 and MDSCs, and between ICAM1 and Th1 cells.* P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant (P ≥ 0.05).

Single-Gene GSEA Enrichment Analysis

In the NASH dataset, single-gene GSEA enrichment analysis was performed on the key genes CCL2 and EDN1. The results indicated that CCL2 was primarily enriched in mitochondrial energy metabolism, NAD coenzyme synthesis, and the PINK1-mediated mitophagy pathway. Conversely, EDN1 was significantly enriched in mitochondrial respiratory chain complex I, the PINK1-mediated mitophagy pathway, and complex I dysfunction induced by abnormal α-synuclein (Figure 6A).

Figure 6.

Single-Gene GSEA. (A) Single-gene GSEA of the key genes in the NASH dataset. (B) Single-gene GSEA of the key genes in the CKD dataset.

In the CKD dataset, single-gene GSEA enrichment analysis was conducted on the key genes CCL2 and ICAM1. The results showed that CCL2 was mainly enriched in the KEAP1-NRF2 antioxidant defense pathway, inflammation/fibrosis signaling pathways, and energy metabolism pathways. ICAM1 was significantly enriched in the KEAP1-NRF2 antioxidant defense pathway, mitochondrial energy metabolism pathways, and the PINK1-mediated mitophagy pathway (Figure 6B).

HUVEC Endothelial Dysfunction Induced by PA and IS Stimulation

Human umbilical vein endothelial cells (HUVECs) were treated with palmitic acid (PA) and indoxyl sulfate (IS) to simulate the in vitro environments of NASH and CKD, respectively. The mRNA expression levels of EDN1, CCL2, CXCL1, and TLR2 were quantified by qPCR, and the protein expression levels of ICAM1 and TLR2 were evaluated by Western blotting in both experimental and control groups. qPCR results demonstrated that, compared with the control group, mRNA expression of EDN1, CCL2, and CXCL1 was significantly downregulated in both PA and IS treatment groups, while TLR2 expression showed no significant change (Figure 7D). Western blotting results further indicated that protein expression levels of ICAM1 and TLR2 were downregulated in both PA and IS treatment groups (Figure 7E).

Figure 7.

In vitro Validation of Key Gene Expression and Endothelial Cell Phenotype Changes (A and B) Wound healing assay evaluating the migration ability of HUVECs in each group. (C) Tube formation assay assessing the angiogenic capacity of HUVECs in each group. (D) Quantitative real-time PCR (qPCR) detecting the mRNA expression levels of key genes in HUVECs from each group. (E) Western blotting detecting the protein expression levels of key genes in HUVECs from each group. * P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant (P ≥ 0.05). Data are shown as mean ± SD (n = 3 independent experiments).

Wound healing assay results revealed that, compared with the control group, cell migration ability was significantly inhibited in both the PA and IS treatment groups (Figure 7A and B). In the tube formation assay, the PA group exhibited significant tube formation defects, with key parameters such as total tube length and the number of branching nodes being significantly lower than those in the control group (P < 0.05). Notably, although the IS treatment group did not show a statistically significant difference in tube formation ability, it exhibited a decreasing trend consistent with that of the PA group (Figure 7C).

Discussion

Nonalcoholic fatty liver disease (NAFLD) and chronic kidney disease (CKD) are two major global public health concerns. Meta-analyses have shown that NAFLD is independently associated with an increased risk of incident CKD and that the risk of adverse renal outcomes increases in a graded manner with the severity of hepatic fibrosis.^9,25 Conversely, the presence of CKD is associated with increased all-cause mortality among patients with NAFLD compared with the general population.²⁶ Collectively, these findings support a close epidemiological association between NAFLD and CKD. However, despite the well-established relationship between these conditions, their shared molecular basis remains incompletely understood.

By integrating transcriptomic datasets from NASH and CKD, we identified 156 common differentially expressed genes and ten hub genes associated with inflammatory, immune, and endothelial-related biological processes using a protein–protein interaction network. Subsequent immune-infiltration analysis and single-gene GSEA revealed associations between the hub genes and immune-cell signatures, as well as enrichment of relevant functional pathways. In vitro experiments further evaluated changes in ICAM1, TLR2, CCL2, and CXCL1 expression alongside alterations in endothelial functional phenotypes.

The findings should be interpreted by distinguishing direct observations from mechanistic inferences. Because the transcriptomic datasets were derived from whole-liver and whole-kidney tissues, the observed gene downregulation may reflect transcriptional regulation within endothelial cells, but it may also result from changes in endothelial abundance or subpopulations, endothelial adaptation or senescence, immune-cell infiltration, or broader alterations in tissue composition. Although genes such as ICAM1, EDN1, and THBS1 have established endothelial functions, they are not endothelial-specific markers. Previous single-cell studies have demonstrated liver sinusoidal endothelial-cell heterogeneity and alterations in the C-Kit⁺ endothelial subpopulation in NASH.²⁷ Similarly, a human kidney single-cell, multi-omic, and spatial atlas showed that the fibrotic microenvironment integrates vascular, immune, stromal, and epithelial signals.²⁸ Although these studies support the biological plausibility of endothelial involvement, they do not determine the cellular origin of the bulk-tissue signals identified in the present study.

A systems-level interpretation may better explain these coordinated findings than a single linear pathway. Lipid overload in NASH and the accumulation of uremic toxins in CKD represent distinct but partially overlapping forms of metabolic stress that may disrupt mitochondrial energy metabolism and quality control, thereby increasing reactive oxygen species production and impairing antioxidant defenses.^17,29,30 Redox imbalance may subsequently interact with inflammatory signaling, immune-cell recruitment, and extracellular matrix remodeling, whereas persistent inflammation may further aggravate oxidative stress and tissue injury.^31,32 Mitochondrial dysfunction, oxidative stress, immune activation, metabolic reprogramming, and fibrosis may therefore form a mutually reinforcing regulatory network^29,32 (Figure 8). Within this framework, oxidative stress and inflammation may function as systems-level hubs that sustain chronic pathological states, while endothelial alterations may represent both a consequence of and a contributor to network feedback. This interpretation is consistent with emerging systems-biology concepts of chronic disease.³³

Figure 8.

Proposed systems-level model of the NASH–CKD axis. Lipid overload and uremic toxin accumulation may be associated with an interconnected network involving mitochondrial dysfunction, oxidative stress/redox imbalance, immune activation, endothelial alterations, and fibrosis. Dashed lines indicate proposed associations without implying directionality or causality. Created with BioGDP.com.³⁴

The ssGSEA results showed that CCL2 expression was positively correlated with MDSC scores and ICAM1 expression with Th1-cell scores in both the NASH and CKD datasets, suggesting coordinated variation between the selected genes and specific inferred immune states. However, ssGSEA estimates relative immune-cell abundance from transcriptomic data rather than directly measuring cell numbers.³⁵ These correlations therefore do not demonstrate that the selected genes directly mediate immune-cell recruitment or establish causal relationships. Single-gene GSEA further showed that the expression of selected hub genes was associated with pathways related to mitochondrial energy metabolism, PINK1-mediated mitophagy, and KEAP1–NRF2 antioxidant defense. Nevertheless, pathway enrichment associations should not be interpreted as evidence that these pathways are directly activated or suppressed. Overall, these findings support the possible presence of an interconnected mitochondrial–redox–immune regulatory network, although its directionality and underlying mechanisms require further investigation.

The HUVEC experiments provided complementary rather than organ-specific evidence. PA and IS both reduced cell migration and altered selected transcript or protein levels. PA significantly impaired tube formation, whereas the IS-associated reduction did not reach statistical significance. These findings demonstrate that PA and IS can induce molecular and functional alterations in HUVECs. However, HUVECs reflect general endothelial stress responses and cannot reproduce the intrinsic phenotypes or native tissue microenvironments of liver sinusoidal or kidney-specific endothelial cells. Accordingly, these findings should be interpreted as complementary evidence rather than organ-specific validation.

This study has several limitations. Whole-tissue transcriptomic data cannot resolve the cellular origin of the observed expression changes. The datasets were derived from independent NASH and CKD cohorts rather than patients with confirmed comorbidity. Furthermore, HUVECs do not reproduce the specialized structures or transcriptional programs of liver sinusoidal, glomerular, or peritubular endothelial cells. Future studies should incorporate matched clinical cohorts, comparable cell-type deconvolution across both organs, single-cell or spatial analyses, and primary organ-specific endothelial models to investigate cellular origin and causality.

Conclusion

In conclusion, this study identified shared tissue-level molecular signatures with potential endothelial relevance across independent NASH and CKD datasets and demonstrated their associations with immune regulation, mitochondrial metabolism, redox homeostasis, and fibrosis-related pathways. Together with the HUVEC findings, these results reveal a potential relationship between metabolic stress and endothelial molecular and functional alterations, providing a new perspective and a testable conceptual framework for understanding the NASH–CKD axis from an immunometabolic network perspective. Given the limitations of bulk-tissue data and the HUVEC model, the cellular origin and underlying mechanisms require further investigation.

Funding Statement

The present study was supported by the Special Fund of Capital Health Research and Development (no. 2024-1-1203), the Dengfeng Talent Support Program of Beijing Municipal Administration of Hospitals (no. DFL20221601) and the High-level Public Health Technical Personnel Construction Project (no. Subject leaders-03-21).

Data Sharing Statement

All public data used in this study were obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo/), with dataset accession numbers provided in the study. All analysis codes and original experimental data generated in this study are available from the corresponding author upon reasonable request.

Ethics Approval and Informed Consent

This study analyzed patient tissue transcriptomic data obtained from a public database. The research did not involve sensitive personal information or commercial interests, nor did it include any intervention or behavior that disrupts public conduct. In accordance with Article 32 of the *Measures for Ethical Review of Life Sciences and Medical Research Involving Humans* (2023 Edition) (Document No.: Guowei Kejiao Fa [2023] No. 4) of China, studies that use lawfully obtained publicly available data, or data acquired through observation without interfering with public behavior, causing harm to subjects, or involving sensitive personal information or commercial interests, are exempt from ethical review. This exemption is intended to reduce unnecessary burdens on researchers and to promote the conduct of life sciences and medical research involving human subjects. This study fully complies with the aforementioned criteria and therefore qualifies for exemption from ethical review.

Author Contributions

Xin Li: Conceptualization, Funding acquisition, Project administration, Resources, Writing – original draft, Supervision, Writing – review and editing

Rui Qiao: Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing – original draft, Supervision, Writing – review and editing

Jian Zhang: Investigation, Methodology, Visualization, Writing – original draft

Yinghui Gao: Software, Writing – original draft. All authors gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this study.