Abstract

   Renal cell carcinoma is characterized by a poor prognosis. Recently,
   renal cell carcinoma has been recognized as a metabolic disease
   associated with fatty acid metabolic reprogramming, although in-depth
   studies on this topic are still lacking. We found that fatty acid
   metabolism reprogramming in renal cell carcinoma is primarily
   characterized by high expression of FABP1. FABP1 + tumors significantly
   impact survival and display distinct differentiation trajectories
   compared to other tumor subclusters. They show elevated expression of
   angiogenesis and cell migration signals, with PLG-PLAT-mediated
   interactions with endothelial cells notably enhanced. Spatial
   transcriptomics show a prominent co-localization of FABP1 + tumors with
   endothelial cells, and their spatial distribution closely aligns with
   that of PLAT + endothelial cells. FABP1 + tumors exhibit a unique
   pattern in spatial transcriptomics, enriched in Extracellular Matrix
   and angiogenesis-related pathways. Through receptor-ligand interaction
   analysis, a novel PLG-PLAT functional axis was found between tumor
   epithelial cells and endothelial cells. Based on results of
   experiments, we infer that FABP1 + tumors can promote plasmin-related
   tumor angiogenesis by triggering the PLG-PLAT signaling axis. Finally,
   utilizing preclinical models, we suggest that targeting the
   FABP1-PLG-PLAT axis may serve as promising strategy enhancing the
   sensitivity of Tyrosine Kinase Inhibitor therapy.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12943-025-02377-9.

   Keywords: Renal cell carcinoma, Fatty acid metabolism, Multi-Omics,
   FABP1, Tumor angiogenesis

Introductions

   Renal cell carcinoma (RCC), a prevalent malignancy originating from the
   kidney epithelium, was associated with 431,288 new diagnoses and
   179,368 fatalities globally in 2020 [[38]1]. The clear-cell variant of
   renal cell carcinoma (ccRCC) is the most common type, accounting for
   roughly 75% of all RCC cases [[39]2]. Patients with ccRCC typically
   experience a worse prognosis compared to those with other histological
   types of RCC[[40]3].

   RCC has been considered a metabolic disease. Oncogenic
   drivers—including mutations, epigenetic silencing, and constitutive
   activation of signaling nodes—reprogram tumor metabolism by hijacking
   bioenergetic and biosynthetic pathways. Hallmarks include exaggerated
   glycolytic flux ("Warburg effect"), disrupted TCA cycle intermediates,
   glutamine anaplerosis, and impaired oxidative phosphorylation,
   culminating in ATP depletion [[41]4, [42]5]. Critically, these
   adaptations are coupled with hypoxia-inducible factor (HIF)-driven
   pseudohypoxia and glutathione-mediated redox buffering, enabling RCC
   survival under nutrient scarcity and oxidative duress [[43]6]. The
   association between patients with RCC and obesity suggests that fatty
   acid metabolic reprogramming plays a critical role in renal cancer
   [[44]7]. Previous studies have identified certain molecules and
   pathways associated with fatty acid metabolic reprogramming in RCC;
   however, a comprehensive understanding of tumor cell metabolic
   reprogramming remains insufficient [[45]8, [46]9].

   Recently, scRNA-sequencing have enabled the investigation of
   tumorigenesis at the cellular level, while the integration of spatial
   transcriptomics allows for more systematic spatial characterization of
   tumor microenvironment interactions. In this study, we explored the
   fatty acid metabolism reprogramming pattern in RCC through scRNA-seq
   combined with multi-omics analysis, identifying its biological features
   as well as interactions with other cells and spatial patterns.

   Additionlly, based on the results of biological experiments, a PLG-PLAT
   functional axis was identified between FABP1 + tumor cells and vascular
   endothelial cell. Plasmin, a key component of the fibrinolytic system,
   is a serine protease derived from its precursor plasminogen [[47]10].
   The two most important plasminogen activators (PAs) in vivo are
   tissue-type plasminogen activator (t-PA, also known as PLAT) and
   urokinase-type plasminogen activator (u-PA). t-PA is primarily
   synthesized and secreted by vascular endothelial cells and is
   continuously released into the bloodstream, widely distributed in
   various tissues throughout the body [[48]11]. Previous study have
   reported that plasmin activates TGF-β, thereby promoting tumor growth
   and metastasis [[49]12]. As a critical signaling molecule in vascular
   biological behavior, investigating the dysregulation of the PLG-PLAT
   axis during tumor progression may provide important insights for cancer
   treatment.

   Through this integrated analysis combining single-cell multi-omics with
   functional biological validation, we will provide valuable insights to
   aid in the treatment of RCC.

Methods

Data availability

   The single-cell sequencing data and spatial transcriptomics sequencing
   (ST-seq) data used in this study were obtained from the Gene Expression
   Omnibus (GEO) database, while the bulk-RNA data came from the TCGA
   database. The bulk RNA data were obtained from the TCGA-KIRC dataset,
   comprising 541 tumor tissues and 72 normal tissues. The single-cell
   data were obtained from [50]GSE210038 [[51]13], [52]GSE237429 and
   [53]GSE242299 [[54]14]. We utilized ST-seq from Series [55]GSE175540
   for the analysis of spatial positioning [[56]15] (Supplement Table 1).

Patient sample

   This research was approval from the Ethics Committees of First
   Affiliated Hospital of Fujian Medical University and adhered to
   established ethical standards throughout (Table S2). The expression
   levels of FABP1 in RCC samples were assessed via qRT-PCR. Survival and
   correlation analyses were performed based on the expression levels.
   Patients were diagnosed independently by three pathologists. Written
   informed consents were obtained from all patients before collection of
   tissues, which were used for and IHC.

Cell culture and reagents

   The human RCC cell strains 786-O and OSRC-2 were sourced from the
   American Type Culture Collection (Manassas, VA, USA) and propagated
   following suggested guidelines. All cells were confirmed to be free
   from mycoplasma contamination and were validated using short tandem
   repeat (STR) profiling before experimentation. The cells were
   maintained in MEM (Gibco) supplemented with 10% FBS (Gibco) and
   incubated at 37 ℃ in a 5% CO₂ humidified environment. For the hypoxia
   treatment, cells were cultured under 1% O[2] at hypoxia station (Don
   Whitley Scientific:H35 hypoxystation, UK).

Lentivirus constructs and transfection

   To knock down PLAT, the PLAT-shRNA sequence was cloned into the
   pLVX-shRNA2 interference vector, while an empty vector served as the
   negative control. The complete coding sequence of PLG was inserted into
   the pcDNA3.1 expression vector through digestion with KpnI and BamHI
   restriction endonucleases. For cellular delivery, the recombinant
   pcDNA3.1-PLG construct was introduced into target cells employing Lipo
   Plus transfection reagent under optimized conditions.

   For lentiviral particle generation, HEK293T cells were co-transfected
   with the essential packaging plasmids RRE, REV, and VSVG. To establish
   stable gene expression, target cells were exposed to the
   custom-produced lentiviral particles for a 24-hour transduction period.
   Following viral infection, polyclonal cell populations with stable
   integration were isolated via puromycin-based antibiotic selection.
   Cell lines expressing luciferase-green fluorescent protein (GFP) were
   generated by transducing cells with a lentivirus based on the pLEX
   vector and encoded the luciferase-GFP fusion gene.

CRISPR/Cas9 knockout (KO)

   To generate FABP1 knockout (KO) cell lines, we first validated the
   functional importance of intronic complementary sequences and their
   adjacent downstream regions. The gene editing procedure was performed
   following conventional CRISPR-Cas9 protocols. Specifically,
   custom-designed single-guide RNAs (sgRNAs) were cloned into the
   LentiCRISPRv2 backbone (item #52961, puromycin resistance). Lentiviral
   particles were then produced by co-transfecting HEK293T cells with the
   LentiCRISPRv2 construct along with packaging plasmids pVSVg (item
   #8454) and psPAX2 (item #12260). The resulting viral supernatant was
   used to transduce renal cell carcinoma (RCC) cells, achieving stable
   FABP1 ablation.

scRNA-seq data processing

   Single-cell data analysis was performed using Seurat v5.0.1. Initial
   quality control involved filtering cells based on the following
   criteria: 1) more than 3 expressed cells, 2) fewer than 200 total
   genes, 3) mitochondrial/ribosomal genes comprising over 20% of total
   gene expression, 4) fewer than 200 or more than 10,000 detected RNA
   features, and 5) total molecule counts below 1,000. The "NormalizeData"
   function was employed to perform library size normalization and obtain
   normalized counts. Specifically, the "LogNormalize" method adjusted the
   gene expression measurements for each cell based on total expression,
   multiplied by a default scaling factor of 10,000, and log-transformed
   the results. The top 2,000 highly variable genes were identified in the
   dataset using the "FindVariableFeatures" function. To remove batch
   effects in the scRNA-Seq data, we utilized "harmony".The 30 top
   principal components were calculated based on the gene expression
   profiles of the highly variable genes. We selected an appropriate
   resolution for the "FindNeighbors" and "FindClusters" functions to
   perform cell clustering. The "RunUMAP" function was used to visualize
   the initial clustering. We annotated the cells using the following
   canonical marker genes: endothelial cells (PECAM1, VWF), neutrophils
   (NAMPT, FCGR3B), dendritic cells (CD1A, HLA-DRA), T cells (CD3D, CD8A,
   CD4), NK cells (NKG7, GNLY, CD160), epithelial cells (EPCAM, KRT8,
   KRT18), fibroblasts (ACTA2, LUM, DCN), mesangial cells (ACTA2, PDGFRB),
   B cells (CD79A), Mast (TPSAB1, MS4A2, KIT) monocytes/macrophages (CD68,
   CD14). Using the "FindAllMarkers" function in Seurat, we identified
   marker genes for each cluster, focusing only on positive markers
   relative to all other cells.

The copy number variation (CNV) analysis

   The CNV evaluation of the epithelial cell subcluster in the tumor group
   was conducted using the "infercnv" R package (version 1.18.1) on the
   chromosome [[57]16]. We selected T cells and epithelial cells from
   normal adjacent tissue as references, setting the parameters denoise =