Abstract
Deciphering the reprogramming of glucose metabolism in
cancer-associated fibroblasts (CAFs) within the ovarian cancer (OC)
microenvironment is essential for understanding tumor progression.
While CAFs are known to influence tumor metabolism, the specific
mechanisms underlying their role in metabolic adaptation remain
unclear. Here, we show that GLUT1 is highly expressed in CAFs and
promotes glucose uptake, glycolysis, and lactate production, which in
turn drives OC cell proliferation and migration via the
TGF-β1/p38/MMP2/MMP9 pathway. Single-cell RNA sequencing and
bioinformatics analyses identify GLUT1 as a key metabolic regulator in
CAFs, and 3D bioprinting models further confirm its role in shaping the
tumor microenvironment. These findings highlight GLUT1 as a potential
therapeutic target for OC and provide new insights into tumor
metabolism and metastasis.
Subject terms: Cancer, Cell biology
__________________________________________________________________
Single-cell and multi-omics analysis reveals that CAF-derived GLUT1
promotes metabolic reprogramming and ovarian cancer progression through
the TGF-β1/p38/MMP2/MMP9 signaling axis.
Introduction
Ovarian cancer (OC) is a prevalent and lethal gynecologic
malignancy^[36]1,[37]2. The incidence of OC is steadily increasing,
posing significant challenges to clinical management^[38]3. Unlike
cancers in other organs, OC often presents no conspicuous symptoms in
its early stages, leading to late-stage diagnoses in a majority of
patients, which in turn complicates treatment^[39]4–[40]6. The tumor
microenvironment is a crucial factor in the development and progression
of OC. Investigating the impact of other cells within the tumor
microenvironment on tumor growth, invasion, and migration is essential
for revealing potential therapeutic targets and improving patient
outcomes^[41]7,[42]8.
In the tumor microenvironment, a vital component influencing tumor
growth, invasion, and metastasis comprises various elements such as
cells, stroma, extracellular matrix, and intercellular signaling
molecules, all playing crucial roles in regulating cancer
progression^[43]9,[44]10. Cancer-associated fibroblasts (CAFs), a
significant group within the microenvironment, exhibit a complex origin
derived from mesenchymal cells, fibroblasts, and other cell types,
engaging in close interactions with tumor cells^[45]11–[46]13. Studies
have demonstrated that CAFs provide nutrients, maintain
microenvironmental homeostasis, promote tumor growth and invasion, and
contribute to the development of drug resistance, highlighting the
importance of in-depth research to unravel the mechanisms driving tumor
development^[47]14.
Aberrant glycometabolism in tumor cells is widely recognized as a
common feature believed to be enhanced to varying degrees in several
cancers, including OC^[48]15,[49]16. Dysregulated glycometabolism can
fuel the rapid growth and proliferation of tumor cells, sustaining
their survival^[50]17. As key cells within the microenvironment, CAFs
play a crucial role in regulating tumor cell growth and
metabolism^[51]18,[52]19. Therefore, exploring the characteristics of
glucose metabolism reprogramming in CAFs is essential for understanding
the regulatory mechanisms of energy metabolism within the tumor
microenvironment to elucidate the patterns of tumor development. GLUT1
acts as the primary pathway for glucose uptake into cells. Within CAFs,
GLUT1-mediated metabolic reprogramming not only sustains their own
metabolic demands but also plays a crucial role in shaping the tumor
microenvironment to support cancer progression. CAFs with high GLUT1
activity contribute to the metabolic heterogeneity and adaptability of
tumors^[53]20,[54]21. Therefore, targeting GLUT1 in CAFs and its role
in metabolic reprogramming offers a potential therapeutic strategy to
disrupt the support CAFs provide to cancer cells, potentially leading
to new directions in cancer treatment.
To begin the study, OC tissue samples were collected for scRNA-seq to
analyze differences in cell composition and communication.
Subsequently, through integrating transcriptomics and bioinformatics
analyses, targets significantly associated with CAF glucose metabolism
reprogramming were identified and validated through relevant cellular
experiments, including assessments of cell viability, proliferation,
migration, and protein expression, among other aspects. Additionally,
the role of GLUT1 in establishing the tumor microenvironment and tumor
progression was validated using 3D bioprinting models and animal
models.
This study aims to comprehensively reveal the characteristics of CAF
glucose metabolism reprogramming in the OC tumor microenvironment and
its relationship with tumor progression by combining single-cell
multi-omics technology and 3D bioprinting models. Furthermore, it seeks
to explore the role and potential mechanisms of the GLUT1 gene in OC
development. This research not only provides potential molecular
targets for the diagnosis and treatment of OC but also elucidates the
glucose metabolism reprogramming characteristics of CAFs and the
molecular mechanisms driving OC metastasis. By integrating single-cell
multi-omics technology and 3D bioprinting models, this study can lay a
solid foundation for a deeper understanding of the pathophysiological
mechanisms of OC and pave the way for the development of more effective
treatment strategies and personalized therapeutic approaches in the
future.
Results
scRNA-seq reveals cellular heterogeneity in OC and normal tissue cells
classification
To elucidate the cellular heterogeneity and molecular features
underlying ovarian cancer (OC), we applied a multi-omics approach
integrating single-cell RNA sequencing (scRNA-seq) and bulk RNA
sequencing (RNA-seq). An overview of the experimental design, including
sample collection, sequencing strategies, and downstream analyses, is
illustrated in Fig. [55]1A. This comprehensive workflow allowed us to
identify differentially expressed genes (DEGs), integrate data from
both platforms, and pinpoint key genes involved in glucose metabolism
reprogramming within the tumor microenvironment.
Fig. 1. Cell clustering and communication analysis of scRNA-seq data.
[56]Fig. 1
[57]Open in a new tab
A Technical flowchart of ovarian tissue sequencing (Created by
BioRender). B Visualization of cell annotation results based on UMAP
clustering, with each color representing a different cell
subpopulation. C Visualization of the grouped cell annotation results
based on UMAP clustering. D Proportions of different cell
subpopulations in each sample, with different colors representing
different cell types. E T-test analysis of the differences in cell
proportions between normal and OC samples (normal, n = 3; OC, n = 3).
Statistically significant differences are highlighted with dashed black
boxes. Quantitative data are expressed as Mean ± SD, with three samples
per group. For comparisons between two groups, *P < 0.05, **P < 0.01,
ns indicates no statistical significance. F Cell communication chord
diagram for normal samples and interaction chord diagrams between each
of the 9 cell types and the other 8 cell types. The thickness of the
lines represents the number of pathways and the strength of
interactions. G Cell communication chord diagram for OC samples and
interaction chord diagrams between each of the 9 cell types and the
other cell types. The thickness of the lines represents the number of
pathways and the strength of interactions.
In our clinical study, tissues from patients undergoing OC surgery,
including three ovarian tumor tissues (OC, n = 3) and three adjacent
normal tissues (normal, n = 3), were collected and subjected to
scRNA-seq analysis. In the normal adjacent tissues we collected
(n = 3), the proportion of adipocytes was approximately 20%, while
stromal cells accounted for about 40%. Microscopic examination revealed
no presence of cancer cells, indicating that the selected adjacent
tissues remained unaffected by tumor invasion, thus demonstrating their
validity as control tissues. We utilized the Seurat package to
integrate the data and initially examined various parameters of all
cells, including the number of genes expressed (nFeature_RNA), total
mRNA molecules (nCount_RNA), and percentage of mitochondrial genes
(percent.mt). The results indicated that most cells had
nFeature_RNA < 5000, nCount_RNA < 20000, and percent.mt <20%
(Figure [58]S1A). Subsequently, by applying quality control criteria of
200 < nFeature_RNA < 5000 percent.mt <20%, low-quality cells were
filtered out, resulting in an expression matrix of 19435 genes and
24,789 cells. Correlation analysis of sequencing depth revealed r
values of 0.04 for nCount_RNA with percent.mt and 0.85 for nCount_RNA
with nFeature_RNA (Figure [59]S1B), indicating satisfactory data
quality for further analysis.
Further analysis of the filtered cells involved calculating variance by
selecting highly variable genes and identifying the top 2000 variably
expressed genes for downstream analysis (Figure [60]S1C). Cell cycle
scoring was conducted using the CellCycleScoring function
(Figure [61]S1D), and preliminary data normalization was performed.
Subsequently, based on the selected highly variable genes, PCA was
employed for linear dimensionality reduction. The main gene expression
heatmap for PC_1 – PC_6 was presented (Figure [62]S1E), along with the
cell distribution in PC_1 and PC_2 (Figure [63]S1F), revealing some
batch effects among the samples.
Data batch correction was carried out using the harmony package to
reduce batch effects and improve the accuracy of cell clustering
(Figure [64]S1G). Furthermore, an ElbowPlot was used to rank PCs by
standard deviation, showing that PC_1-PC_20 adequately captured
information from the selected highly variable genes with significant
analytical implications (Figure [65]S1H). Following batch correction,
the results demonstrated the effective elimination of batch effects
across samples (Figure [66]S1I, J).
Subsequently, a non-linear dimensionality reduction using the UMAP
algorithm on the top 20 PCs was executed, and clustering patterns at
various resolutions were displayed using the cluster package
(Figure [67]S2). Through UMAP clustering analysis, all cells were
classified into 24 cell clusters (Figure [68]S1K, L). By leveraging
known cell lineage-specific marker genes obtained from literature and
CellMarker online database, 9 cell types were identified as ovarian
stromal cells (OSCs), T cells, natural killer cells (NK cells),
macrophages, CAFs, endothelial cells, pericyte, epithelial cells or
cancer cells, and B cells (Fig. [69]1B, C). Furthermore, the UMAP
expression charts for these 9 cell marker genes were displayed, where
DCN represented a marker gene for OSCs, CD3D for T cells, KLRD1 for NK
cells, CD68 for macrophages, THY1 for CAFs, PECAM1 for endothelial
cells, RGS5 for pericyte, EPCAM for epithelial cells (cancer cells),
and CD79A for B cells (Figure [70]S3A, B, Table [71]S1). Apart from
THY1, other CAF marker genes such as FAP, COL1A1, and PDGFR exhibited
similar expression patterns to THY1. We provided a detailed description
of the cell composition and distribution of nine cell types across six
samples. A t-test was conducted to compare the differences in cell
numbers between normal and OC samples. The analysis showed that,
compared to normal samples, the numbers of epithelial cells (cancer
cells) and CAFs were significantly increased in OC samples, while the
number of macrophages was significantly reduced (Fig. [72]1D, E).
Furthermore, to understand the functional differences underlying these
numerical variations, we investigated cell-cell communication mediated
by ligand-receptor interactions. Using the “CellChat” package in R, we
analyzed the communication and interactions between different cell
phenotypes. The results indicated that, compared to normal samples, the
communication links between CAFs and epithelial cells (cancer cells),
macrophages, and T cells were enhanced in OC samples (Fig. [73]1F, G).
These results highlight the importance of CAFs in OC progression.
Additionally, compared to normal samples, the signaling pathways
involved in the interactions between CAFs and other cells in OC samples
mainly include the VEGF signaling pathway, TNF signaling pathway, and
HLA signaling pathway (Figure [74]S3C, D), suggesting that CAFs may
influence OC progression either through direct interactions with cancer
cells or via immune cells. These findings demonstrate that CAFs are key
mediators of cell-cell communication in OC.
The scRNA-seq analysis results indicate that OC samples and their
adjacent normal tissues can be divided into 24 clusters, successfully
identifying 9 cell subpopulations. Among these, the numbers of
epithelial cells (cancer cells) and CAFs were significantly increased,
while the number of macrophages was significantly reduced. CAFs were
found to be key cells mediating cell-cell communication in OC.
The key roles of CAFs and glucose metabolism reprogramming in OC: GLUT1 as a
potential therapeutic target
To further investigate the role of CAF glucose metabolism reprogramming
in OC progression, differential gene expression analysis was conducted
between CAFs in adjacent normal samples and OC samples. A total of 564
differentially expressed genes were identified, with 425 genes
significantly upregulated in OC samples and 139 genes significantly
downregulated (Fig. [75]2A). Additionally, high-throughput RNA-seq was
performed on normal samples (n = 3) and OC samples (n = 3), resulting
in the identification of 1238 differentially expressed genes, including
630 upregulated genes and 608 downregulated genes (Fig. [76]2B).
Fig. 2. Screening of key OC genes based on single-cell sequencing and
transcriptome sequencing analysis.
[77]Fig. 2
[78]Open in a new tab
A Volcano plot showing differentially expressed genes between
fibroblasts in normal and OC samples. The red dots on the left side of
the dashed line represent genes highly expressed in OC samples, while
the dots on the right represent genes with lower expression in OC
samples. B Volcano plot of differentially expressed genes in RNA-seq
between three normal samples and three OC samples. The red upward
triangles represent upregulated genes, the green downward triangles
represent downregulated genes, and the black dots represent
non-differential genes. C Venn diagram showing the overlap of
differentially expressed genes between RNA-seq and CAFs in scRNA-seq. D
Bubble plot (left) and chord diagram (right) of GO enrichment analysis
for the 48 overlapping genes. In the bubble plot, circle color
represents the significance of enrichment, with colors ranging from
blue to red indicating increasing significance. Circle size represents
the number of enriched genes. E Bubble plot (left) and chord diagram
(right) of KEGG enrichment analysis for the 48 overlapping genes. In
the bubble plot, circle color represents the significance of
enrichment, with colors ranging from blue to red indicating increasing
significance. Circle size represents the number of enriched genes. F
Volcano plot showing differentially expressed genes between epithelial
cells in normal and OC samples. The red dots on the left side of the
dashed line represent genes highly expressed in OC samples, while the
dots on the right represent genes with lower expression in OC samples.
G Venn diagram showing the overlap of differentially expressed genes
between RNA-seq and epithelial cells in scRNA-seq. H Venn diagram
showing the overlap of differentially expressed genes from RNA-seq,
CAFs in scRNA-seq, epithelial cells in scRNA-seq, and 221 genes related
to glycometabolism reprogramming, resulting in the identification of
four key genes: SLC2A1, KRT18, KRT8, and HK2. I Relevance score of the
four key genes in the GeneCards database. J RNA-seq analysis results of
the four key genes in normal and OC samples, with three samples per
group. * indicates a comparison between two groups, *P < 0.05,
***P < 0.001.
By analyzing the intersection of scRNA-seq and RNA-seq data, 48
overlapping genes were obtained (Fig. [79]2C), showing consistent
expression patterns between the two datasets. Their expression levels
in RNA-seq are shown in Table [80]S2. These 48 overlapping genes were
subjected to GO functional enrichment analysis, revealing involvement
in biological processes such as female pregnancy, multi-organism
reproductive process, and multicellular organism development. In terms
of cellular components, enrichment was observed in locations like
secretory granule lumen, cytoplasmic vesicle lumen, and platelet alpha
granule lumen. Meanwhile, the molecular functions of these genes were
predominantly associated with activities like endopeptidase inhibitor
activity, peptidase inhibitor activity, and endopeptidase regulator
activity (Fig. [81]2D). Furthermore, KEGG pathway enrichment analysis
on the 48 overlapping genes showed significant enrichment in pathways
including MicroRNAs in cancer, Complement and coagulation cascades,
Diabetic cardiomyopathy, Cysteine and methionine metabolism, and
Insulin resistance, with a notable presence in the
Glycolysis/Gluconeogenesis pathway (Fig. [82]2E). Additionally, we
found that hypoxia-related pathways, such as the HIF-1 signaling
pathway, were enriched (Fig. [83]2E). Furthermore, we performed
differential gene expression analysis between epithelial cells from
adjacent normal samples and OC samples, identifying 325 differentially
expressed genes, of which 155 were significantly upregulated and 170
were significantly downregulated in OC samples (Fig. [84]2F). By
intersecting these 325 differentially expressed genes with the
differentially expressed genes obtained from RNA-seq, we identified 37
overlapping genes (Fig. [85]2G). The expression of these 37 overlapping
genes is shown in Table [86]S3. To further validate the relationship
between CAFs glycolytic reprogramming and OC progression, we obtained
221 genes related to ‘glycometabolism reprogramming’ from the GeneCards
database. A Venn analysis was performed between this gene set and the
564 differentially expressed genes in CAFs, the 325 differentially
expressed genes in epithelial cells, and the 1238 RNA-seq
differentially expressed genes, yielding 4 overlapping genes
significantly associated with glycolytic reprogramming, OC, and CAFs:
SLC2A1, KRT18, KRT8, and HK2 (Fig. [87]2H). The relevance of these four
genes to glucose metabolism reprogramming is presented in Fig. [88]2I,
highlighting SLC2A1 and HK2 as genes with high relevance. All four
intersecting genes showed significantly elevated expression in OC
samples compared to normal samples (Fig. [89]2J).
We identified four key genes linked to CAFs and glucose metabolism
reprogramming in OC development. To further investigate the role of
these key genes in OC, we analyzed their protein expression patterns in
normal and tumor samples using The HPA database. The results revealed a
significant increase in the expression of these four key genes in OC
tumor tissues compared to normal ovarian tissues, with SLC2A1, KRT18,
and KRT8 showing strong expression, while HK2 exhibited moderate
expression (Figure [90]S4A).
To delve into the impact of these four key genes on the prognosis of
OC, we extracted RNAseq data in TPM format from The Cancer Genome Atlas
(TCGA) and Genotype-Tissue Expression (GTEx) databases and conducted
ROC curve analysis. The analysis indicated that the AUC values of the
ROC curves for all four genes were greater than 0.5, with SLC2A1,
KRT18, and KRT8 having AUC values exceeding 0.9, suggesting excellent
diagnostic potential in predicting the outcome of OC. In contrast, HK2
had an AUC value of 0.664, indicating relatively weaker diagnostic
performance in predicting the outcome of OC (Figure [91]S4B).
Furthermore, based on the expression levels of the four key genes
associated with prognosis, we stratified TCGA-OV patients into high and
low-expression groups, performed proportional hazards assumption tests,
conducted survival regression fitting, and plotted Kaplan-Meier
survival curves. The results demonstrated that patients with high
expression of SLC2A1 and HK2 had poorer survival outcomes, while those
with low expression had better outcomes (HR > 1, p < 0.05). Conversely,
there was no significant difference in survival outcomes based on the
expression of KRT18 and KRT8 genes (Figure [92]S4C). In addition, we
extracted the expression of the 4 overlapping genes in TCGA OC samples
(Figure [93]S4D). The changes in expression levels of these 4 genes
were consistent with the sequencing results. Therefore, considering the
correlation scores between the four key genes and glucose metabolism
reprogramming, gene expression in the HPA database, ROC curves, and
Kaplan-Meier survival curves, we ultimately identified SLC2A1.
To further validate the above analysis results, we obtained the numbers
of various cell subtypes from different tissues in the Ovary Cancer sc
Database. The results showed that, compared to normal ovarian tissues,
the numbers of epithelial cells (cancer cells) and fibroblasts were
both increased in OC tissues (Figure [94]S5A–C). Additionally, the
SLC2A1 gene was significantly expressed in both epithelial cells and
fibroblasts (Figure [95]S5D, E). Furthermore, we obtained the
expression levels of the SLC2A1 gene from the [96]GSE139555,
[97]GSE147082, [98]GSE158722, and [99]GSE168652 datasets in the TISCH
database, and the results showed that SLC2A1 was also expressed in
fibroblasts across all four datasets (Figure [100]S5F–I). This data
further validated our sequencing data analysis results.
Reprogramming of GLUT1 by CAFs in OC promotes tumor proliferation and
invasion
Through multi-omics sequencing and bioinformatics analysis, we have
preliminarily identified potential therapeutic targets significantly
associated with both CAFs and glucose metabolism reprogramming in
OC—GLUT1. To further elucidate the regulatory relationship between
GLUT1 in CAFs and OC, we initially demonstrated the expression of
SLC2A1 using scRNA-seq in normal and OC samples. The results revealed a
significant upregulation of SLC2A1 in OC samples compared to normal
samples, particularly in epithelial cells (cancer cells) and CAFs
(Figure [101]S6A). Additionally, we detected the expression of the CAF
marker proteins α-SMA and GLUT1 in normal and OC samples through
immunohistochemical staining. The results showed that, compared to
normal samples, the expression of the CAF marker protein α-SMA was
higher in OC tissues, indicating a significant enrichment of CAFs in OC
tumors. Moreover, the expression of GLUT1, the tumor marker KRT8, and
the CAF markers FAP and THY1 were also significantly upregulated in OC
tissues (Figure [102]S6B). The localization of GLUT1 expression was
close to that of THY1 and FAP. Furthermore, RT-qPCR and Western blot
analysis further confirmed the upregulation of α-SMA and GLUT1 proteins
in OC tissues (Figure [103]S6C, D).
Next, to investigate whether GLUT1 regulates CAF-mediated tumor
microenvironment, we isolated various cell types from tumor tissues of
OC patients post-resection to obtain CAFs (Figure [104]S6E). Quality
assessment of the obtained CAFs was conducted by immunofluorescence
staining for the expression of endothelial cell marker CD31, epithelial
cell marker Cytokeratin, and fibroblast marker Vimentin. The results
indicated minimal expression of CD31 and Cytokeratin in the obtained
fibroblasts, while Vimentin was prominently expressed
(Figure [105]S6F), suggesting minimal contamination from epithelial and
endothelial cells, ensuring good quality for subsequent analyses.
Additionally, we detected the expression of GLUT1 in CAFs
(Figure [106]S6F). To further explore the impact of CAFs on the
proliferation and migration of OC cells, we co-cultured CAFs with OC
cells at a 2:1 ratio (Figure [107]S7A) and assessed OC cell viability
and proliferation ability through CCK8 and EDU experiments. The results
(Figure [108]S7B, C) showed a significant enhancement in the viability
and proliferation capacity of OC cells when co-cultured with CAFs
(SKOV3+CAFs and A2780+CAFs groups) compared to OC cells cultured alone
(SKOV3 and A2780 groups). Colony formation assays indicated that CAFs
enhanced the clustering ability of OC cells (Figure [109]S7D).
Furthermore, following treatment with 10 μM cisplatin, flow cytometry
analysis of cell apoptosis revealed a significant reduction in
apoptosis levels in OC cells when co-cultured with CAFs
(Figure [110]S7E). Subsequent Transwell and wound healing assay results
demonstrated a significant enhancement in the migration and invasion
abilities of OC cells when co-cultured with CAFs (Figure [111]S7F–I).
GLUT1 in CAFs promotes metabolic reprogramming and enhances the invasive
ability of OC cells
To further investigate the impact of GLUT1 expression in CAFs on the
proliferation and migration of OC cells, as well as its specific
regulatory mechanisms, we constructed CAF cells with GLUT1
overexpression and silencing through lentiviral transfection
(Fig. [112]3A). The efficiency of GLUT1 silencing or overexpression was
validated by RT-qPCR and Western blot analysis, and the cell line with
the best efficiency was selected for subsequent experiments
(Figure [113]S8A–D).
Fig. 3. Impact of GLUT1 in CAFs on OC cell proliferation and migration via
glucose metabolism reprogramming.
[114]Fig. 3
[115]Open in a new tab
A Schematic illustration of lentiviral transfection process to
establish GLUT1-silenced or overexpressed CAFs (Created by BioRender).
B Effect of GLUT1 silencing or overexpression on ECAR in CAFs. C Impact
of GLUT1 silencing or overexpression on OCR in CAFs. D Quantitative
analysis of metabolites in energy metabolism pathways using LC-MS,
including six glucose intermediates in glycolysis, OXPHOS, and the
pentose phosphate pathway. E Influence of GLUT1 silencing or
overexpression on glucose uptake rate in CAFs. F Effect of GLUT1
silencing or overexpression on LDH enzyme activity in CAFs. G Impact of
GLUT1 silencing or overexpression on lactic acid production in CAFs. H
Influence of GLUT1 silencing or overexpression on ATP generation in
CAFs. I The MTT to detect the impact of GLUT1 silencing or
overexpression on the activity of CAFs. J Colony formation assay to
assess the impact of GLUT1 silencing or overexpression in CAFs on the
colony-forming ability of SKOV3 cells. K EDU experiment to evaluate the
influence of GLUT1 silencing or overexpression in CAFs on the
proliferation capacity of SKOV3 cells, where EDU-positive cells are
depicted in red indicating cells in the proliferative phase;
EDU-negative cells are shown in blue (Scale bar=25 μm). L Transwell
assay to determine the effects of GLUT1 silencing or overexpression in
CAFs on the migration and invasion abilities of SKOV3 cells (Scale
bar=50 μm). M Wound healing assay to evaluate the impact of GLUT1
silencing or overexpression in CAFs on the migration capability of
SKOV3 cells (Scale bar=100 μm). N RT-qPCR analysis to examine the
effects of GLUT1 silencing or overexpression in CAFs on the expression
of TGFB1, MAPK14, MMP2, and MMP9 mRNA in SKOV3 cells. O Western blot
analysis to assess the effects of GLUT1 silencing or overexpression in
CAFs on the protein expression of TGF-β1, p38, p-p38, MMP2, and MMP9 in
SKOV3 cells. P Effects of GLUT1 silencing or overexpression on TGFB1,
MAPK14, MMP2, and MMP9 mRNA expression in CAFs, detected by RT-qPCR. Q
Effects of GLUT1 silencing or overexpression on TGF-β1, p38, p-p38,
MMP2, and MMP9 protein expression in CAFs, detected by Western blot.
The quantitative data in the figures are presented as Mean ± SD, with
each cell experiment group repeated three times. A connection between
the two groups indicates a significant difference.
To delve deeper into the influence of GLUT1 on the metabolic
reprogramming of CAFs, we employed Seahorse equipment to measure the
eECAR (ECAR for aerobic glycolysis) and OCR (OCR for OXPHOS) of CAFs in
different intervention groups. The results indicated that silencing
GLUT1 significantly decreased the glycolysis and OXPHOS of CAFs, while
overexpressing GLUT1 markedly increased both processes (Fig. [116]3B,
C). To further elucidate the energy metabolism of CAFs, we
quantitatively analyzed key metabolites in the glycolysis pathway,
OXPHOS, and pentose phosphate pathway using liquid chromatography-mass
spectrometry (LC-MS). The results revealed that cells in the sh-GLUT1
group exhibited significantly reduced metabolic activity in the
mentioned pathways compared to the sh-NC group, with specific
metabolites such as D-glucose 1-phosphate, D-glucose 6-phosphate, and
α-D-ribose 5-phosphate showing a notable decrease, whereas the oe-GLUT1
group displayed enhanced metabolic activity (Fig. [117]3D).
Furthermore, we assessed the impact of silencing and overexpressing
GLUT1 in CAFs on glucose uptake. The results showed that compared to
the sh-NC group, silencing GLUT1 (sh-GLUT1 group) inhibited glucose
uptake in CAFs, while overexpressing GLUT1 (oe-GLUT1 group)
significantly enhanced glucose uptake (Fig. [118]3E). Additionally,
assessments of LDH enzyme activity and lactic acid production revealed
that overexpressing GLUT1 significantly increased LDH enzyme activity
and lactic acid production in CAFs, whereas silencing GLUT1 inhibited
LDH enzyme activity and reduced lactic acid production (Fig. [119]3F,
G). Moreover, overexpressing GLUT1 enhanced ATP generation in CAFs
while silencing GLUT1 suppressed ATP production (Fig. [120]3H). These
findings further underscore the promotive role of GLUT1 in the
metabolic reprogramming of CAFs. Furthermore, we assessed the impact of
GLUT1 on the activity of CAFs through MTT analysis. The results
revealed a decrease in cellular activity in the sh-GLUT1 group compared
to the sh-NC group. Conversely, the oe-GLUT1 group displayed a
significant increase in CAFs cellular activity compared to the oe-NC
group (Fig. [121]3I).
To investigate the impact of CAF metabolic reprogramming on OC cell
proliferation and invasion, we co-cultured different intervention
groups of CAF cells with SKOV3 cells. Cell colony formation and EdU
assays were conducted to assess SKOV3 cell clustering and proliferation
capabilities. The results showed a significant decrease in SKOV3 cell
clustering and proliferation ability in the sh-GLUT1 CAFs group
compared to the sh-NC CAFs group. Conversely, in the oe-GLUT1 CAFs
group compared to the oe-NC CAFs group, a significant enhancement in
SKOV3 cell clustering and proliferation ability was observed
(Fig. [122]3J, K). Transwell experiments demonstrated that
overexpression of GLUT1 in CAFs significantly increased the migration
and invasion abilities of SKOV3 cells while silencing GLUT1 markedly
inhibited these cellular functions (Fig. [123]3L, M).
RT-qPCR and Western blot results showed a significant decrease in the
expression of TGF-β1 protein in SKOV3 cells, as well as its encoding
gene TGFB1, phosphorylated p38 protein and its encoding gene MAPK14,
and MMP2 and MMP9 proteins and their encoding genes in the sh-GLUT1
group compared to the sh-NC group. Conversely, an increase in the
expression of the relevant proteins and their encoding genes was
observed in the oe-GLUT1 group compared to the oe-NC group
(Fig. [124]3N, O). Silencing or overexpressing GLUT1 in CAFs was
measured for various targets such as TGFB1, and the results showed no
significant changes in the expression levels of TGF-β1 protein and its
encoding gene TGFB1, phosphorylated p38 protein and its encoding gene
MAPK14, or MMP2 and MMP9 proteins and their encoding genes in the
different groups of CAFs (Fig. [125]3O, P). The precursor forms of MMP2
and MMP9 (pro-MMP2 and pro-MMP9) are inactive enzymes, typically
present as larger molecular weight proteins (such as 72 kDa for
pro-MMP2 and 92 kDa for pro-MMP9). Upon removal of the precursor
portion or activation by certain enzymes, MMPs are converted into their
active forms (such as 53 kDa for MMP2 and 82 kDa for MMP9), enabling
them to degrade collagen and other matrix components. The WB results in
Fig. [126]3N show that the molecular weights of MMP2 and MMP9 are
53 kDa and 82 kDa, respectively, indicating their active forms, which
can degrade the extracellular matrix (ECM), creating space for cell
migration and enhancing the invasive abilities of both cancer cells and
CAFs.
In CAFs, high levels of active MMP2 and MMP9 contribute to the
remodeling of the tumor microenvironment, thereby promoting tumor
progression and metastasis. To further validate whether CAFs promote OC
cell proliferation and invasion capabilities through lactic acid, we
inhibited lactic acid expression in CAFs using an exogenous LDH
inhibitor - Galloflavin (Gal) and co-cultured them with SKOV3 cells.
Initially, after Gal treatment, both LDH enzyme activity and lactic
acid production in CAFs significantly decreased (Figure [127]S9A, B).
Subsequently, CCK-8 and EdU assays revealed a marked reduction in CAFs’
activity and proliferation capabilities towards SKOV3 cells post-Gal
treatment (Figure [128]S9C, D), accompanied by a significant decrease
in cell clustering ability (Figure [129]S9E). Additionally, a decrease
in SKOV3 cell migration and proliferation abilities was detected
post-Gal treatment (Figure [130]S9F, G). RT-qPCR and Western blot
analysis showed a significant decrease in the expression of TGF-β1,
p-p38, MMP2, and MMP9 proteins, as well as their encoding genes in
SKOV3 cells after Gal treatment (Figure [131]S9H, I), further
confirming that CAFs activate the TGF-β1/p38/MMP2/MMP9 signaling
cascade through lactic acid, thereby enhancing the proliferation and
invasion abilities of OC cells.
Interaction between CAFs and SKOV3 Cells in 3D in Vitro OC Model and Effect
of GLUT1 on OC Biological Characteristics
Cancer is a complex pathological phenomenon where interactions between
cells drive cancer development. In the most severe cases, these
interactions can even lead to cancer metastasis. Traditional
two-dimensional cell culture models are overly simplistic and fail to
accurately reflect the complexity of tissues. While animal experiments
offer advantages, they are hindered by long experimental periods, poor
replicability, and high costs. In contrast, three-dimensional in vitro
cancer models not only mimic the tumor microenvironment but also
enhance the predictability of cancer toxicity and drug sensitivity.
Utilizing 3D bioprinting technology, we constructed a tumor model of
CAFs and SKOV3 cells in OC to further investigate the impact of GLUT1
in CAFs on OC biological characteristics (Fig. [132]4A).
Fig. 4. Validation of the impact of GLUT1 in CAFs on OC cell biological
characteristics through 3D bioprinting model.
[133]Fig. 4
[134]Open in a new tab
A Schematic representation of the construction of an OC tumor model
with co-cultured CAFs and SKOV3 cells (Created by BioRender). B
Bioprinting experiment conducted during the hydrogel optimization
phase. C H&E staining image of the 3D bioprinting structure. D
Live/dead cell staining of the 3D printed structure, where green
fluorescence represents live cells and red fluorescence indicates dead
cells (scale bar=25 μm). E Cell viability assessment in the 3D printed
tissues using the CCK-8 assay. F Measurement of lactic acid content in
the 3D printed structures of each group. G Immunohistochemical staining
to detect the expression changes of TGF-β1, p-p38, MMP2, and MMP9
proteins in the 3D printed tissues of each group (scale bar=30 μm). H
Live/Dead cell staining of the 3D printed structure, with green
fluorescence indicating live cells and red fluorescence indicating dead
cells (scale bar = 25 μm). I Cell viability in the 3D printed tissue
measured by CCK8 assay. J Lactate content measurement in each group of
3D printed structures. K Immunohistochemical staining to detect the
expression changes of TGF-β1, p-p38, MMP2, and MMP9 proteins in each
group of 3D printed tissues (scale bar = 30 μm). Data are expressed as
Mean ± SD, with each experiment repeated 3 times. The quantitative data
in the figures are presented as Mean ± SD, with each experiment group
repeated three times. A connection between the two groups indicates a
significant difference.
Initially, following the formulation reported in the
literature^[135]22, we precisely reproduced the theoretical square
shape at 37 °C to determine the final concentration ratios used. As
shown in Fig. [136]4B, hydrogels formed by 1% SA + 15% Gel and 2%
SA + 13% Gel exhibited excessive flowability, compromising shape
integrity, while 2% SA + 15% Gel effectively reproduced the desired
shape. Therefore, we selected 2% SA + 15% Gel as the composition of the
biochemical ink for 3D printing to construct the CAFs and SKOV3 cell OC
tumor model. Additionally, the survival rate of CAFs reached 80%, and
the cell proliferation rate reached 150%, indicating that our model
meets the requirements.
Subsequently, we observed the growth status of the cells in the
3D-printed structures using H&E staining. The results demonstrated that
in the SKOV3 group, the growth rate and cell clusters significantly
increased in structures co-cultured with CAFs compared to structures
constructed solely with SKOV3 cells. Additionally, downregulating GLUT1
expression in SKOV3 cells within the SKOV3+CAFs-sh-GLUT1 group
inhibited the proliferative effect of CAFs on SKOV3 cells, leading to a
reduction in cell clusters (Fig. [137]4C). Furthermore, cell viability
assays on the 3D bioprinted structures revealed that CAFs increased the
survival rate of SKOV3 cells and decreased the rate of cell death.
Conversely, silencing GLUT1 expression in CAFs significantly
counteracted the pro-survival role of CAFs on SKOV3 cells
(Fig. [138]4D). Results from the CCK8 assay further confirmed this,
where CAFs promoted SKOV3 cell proliferation, a process inhibited by
downregulating GLUT1 in CAFs (Fig. [139]4E).
Next, we measured the lactic acid production within the 3D bioprinted
structures. The results showed that, consistent with in vitro cell
experiments, CAFs significantly elevated the lactic acid content within
the tumor microenvironment, whereas silencing GLUT1 led to a
significant reduction in lactic acid levels within the tumor
microenvironment (Fig. [140]4F).
Lastly, immunohistochemistry staining was conducted to assess the
changes in the expression of TGF-β1, p-p38, MMP2, and MMP9 proteins in
the 3D-printed tissues of each group. These results were consistent
with the 2D cell culture findings, where co-culturing SKOV3 cells with
CAFs resulted in a notable increase in the expression of TGF-β1, p-p38,
MMP2, and MMP9 proteins, enhancing tumor cell proliferation through
activating the TGF-β1/p38/MMP2/MMP9 signaling pathway. Conversely, in
the scenario of silencing GLUT1 in CAFs, there was a clear inhibition
of the CAFs’ activation of the TGF-β1/p38/MMP2/MMP9 signaling pathway
in SKOV3 cells (Fig. [141]4G).
Additionally, we supplemented the analysis by investigating the effects
of Oxygen/Glucose Deprivation (OGD) on OC cells. The experimental
results showed that OGD increased the survival rate and decreased the
mortality of SKOV3 cells (Fig. [142]4H). In contrast, silencing GLUT1
in CAFs significantly inhibited the effect of OGD in promoting SKOV3
cell survival. The CCK8 assay further confirmed this, showing that OGD
promoted SKOV3 cell proliferation, whereas downregulating GLUT1
expression in CAFs suppressed this effect (Fig. [143]4I). Subsequently,
we measured lactate production and found that OGD significantly
increased lactate content in the tumor microenvironment, while
silencing GLUT1 significantly reduced lactate levels (Fig. [144]4J).
Finally, immunohistochemical staining was used to detect the changes in
TGF-β1, p-p38, MMP2, and MMP9 protein expression. Under OGD conditions,
the expression of TGF-β1, p-p38, MMP2, and MMP9 proteins in SKOV3 cells
significantly increased, thereby enhancing tumor cell proliferation via
activation of the TGF-β1/p38/MMP2/MMP9 signaling pathway. However, when
GLUT1 was silenced in CAFs, the OGD-induced activation of the
TGF-β1/p38/MMP2/MMP9 signaling pathway in SKOV3 cells was significantly
inhibited (Fig. [145]4K).
Study on the regulation of GLUT1 expression in CAFs in the tumor
microenvironment on the proliferation and metastasis of OC cells
In this study, we investigated the impact of GLUT1 expression in CAFs
on the proliferation and metastasis of OC cells in a subcutaneous OC
model transplanted in mice. We established xenograft models by using
SKOV3, SKOV3+CAFs, and SKOV3+CAFs-shGLUT1 groups (Fig. [146]5A).
Starting from the 8th day after model establishment, we measured tumor
volume in vitro every 4 days, and upon 36 days post-modeling, we
dissected and photographed the tumors. The results showed that compared
to the SKOV3 group, the tumors in the SKOV3+CAFs group grew larger,
while the tumors in the SKOV3+CAFs-sh-GLUT1 group were smaller
(Fig. [147]5B). The tumor volume data obtained using calipers confirmed
this trend, especially in the later stages of the experiment
(Fig. [148]5C). Furthermore, the weight of the xenografts in the
SKOV3+CAFs group significantly increased compared to the SKOV3 group,
whereas the weight decreased significantly in the SKOV3+CAFs-sh-GLUT1
group (Fig. [149]5D).
Fig. 5. Impact of GLUT1 knockdown on tumor growth in tumor-bearing mice.
[150]Fig. 5
[151]Open in a new tab
A Schematic diagram of subcutaneous xenograft nude mouse model
experiment (n = 6) (Created by BioRender). B Dissection images of
subcutaneous xenograft mice in each group on day 36 (n = 6). C Line
graph showing tumor volume growth in subcutaneous xenograft mice from
day 8 to 36 (n = 6). D Statistical analysis of tumor weight in
subcutaneous xenograft mice on day 36 (n = 6). E Schematic diagram of
peritoneal metastasis xenograft mouse model experiment (n = 3). F IVIS
Lumina II imaging system measuring tumor cell peritoneal metastasis in
various groups of peritoneal metastasis models in mice (n = 3). G
Representative abdominal dissection image of peritoneal metastasis nude
mice, white arrows indicate tumors (n = 3). H Dissection image of
intraperitoneal xenograft tumors in peritoneal metastasis nude mice
(n = 3); I TUNEL assay to detect apoptosis of tumor cells in each group
of mice (Scale bar=50 μm, n = 6). J Immunohistochemical staining to
examine the expression changes of TGF-β1, p-p38, MMP2, and MMP9
proteins in tumor tissues of each group of mice (Scale bar=50 μm,
n = 6). Quantitative data in the figures are presented as Mean ± SD,
and a connection between the two groups indicates a significant
difference.
Additionally, we constructed a peritoneal metastasis nude mouse model
by intraperitoneal injection of tumor cells to study the metastasis of
different cell groups of OC cells (Fig. [152]5E). Using the IVIS Lumina
II imaging system for live fluorescence imaging detection showed that
compared to the SKOV3 group, the fluorescence signals of SKOV3 cells in
the abdominal cavity of mice in the SKOV3 + CAFs group were
significantly enhanced, while they were notably reduced in the
SKOV3+CAFs-sh-GLUT1 group (Fig. [153]5F). Further confirmation of these
results came from the anatomical images of the mouse abdominal cavity
and the dissected abdominal tumors, showing that compared to the SKOV3
group, the number of peritoneal metastatic nodules in the SKOV3+CAFs
group increased by approximately 300%, whereas it decreased by about
100% in the SKOV3+CAFs-sh-GLUT1 group (Fig. [154]5G, H).
Moreover, TUNEL staining results indicated that compared to the sole
construction of subcutaneous tumors with SKOV3 cells, CAFs suppressed
the apoptosis of SKOV3 cells in vitro, while further silencing GLUT1 in
CAFs inhibited CAFs’ regulation of apoptosis in SKOV3 cells
(Fig. [155]5I). Immunohistochemical staining results revealed that CAFs
upregulated the expression of TGF-β1, p-p38, MMP2, and MMP9 proteins in
the xenografts while downregulating GLUT1 in CAFs inhibited CAFs’
regulation of these proteins (Fig. [156]5J).
Discussion
This study delves into the characteristics of glucose metabolism
reprogramming in CAFs within the OC tumor microenvironment by employing
single-cell multi-omics technology in conjunction with 3D bioprinting
models. In contrast to previous studies, this research emphasizes the
comprehensive utilization of various experimental methods, confirming
the significant role of GLUT1 in OC from cellular to animal models,
thus addressing existing gaps in research^[157]23–[158]25.
Our scRNA-seq revealed discrepancies in the cellular composition and
intercellular communication within OC tissue, compared to adjacent
normal tissue. We identified nine cell subtypes in the ovarian tissue,
with the fibroblast subtype showing the greatest disparity between
normal and OC tissues. This highlights the enriched abundance of CAFs
in OC tissues, which corroborates prior research^[159]26. In OC
tissues, CAFs exhibited notably stronger interactions with epithelial
cells, macrophages, and T cells, enhancing tumor progression and immune
evasion. Hypoxia-related pathways, including the HIF-1 signaling
pathway, were also identified in our RNA-seq analysis, linking hypoxia
and glycolysis to tumor growth, metastasis, and immune
suppression^[160]27,[161]28. Hypoxia, a common feature of the tumor
microenvironment, promotes glycolysis, resulting in the Warburg effect,
where tumor cells prioritize glycolysis for energy
production^[162]27,[163]28.
Additionally, lactate accumulation from glycolysis creates an acidic
environment that facilitates immune evasion by suppressing immune cell
activity, further supporting tumor survival and metastasis^[164]29.
HIF-1, a key regulator under hypoxic conditions, mediates the
expression of downstream genes such as GLUT1 and HK2, which promote
glucose metabolism and tumor cell survival^[165]30,[166]31. CAF-derived
lactic acid activates pathways like TGF-β2 in other cancers, and
similar mechanisms were observed in OC, where lactic acid from CAFs
activates the TGF-β1/p38/MMP2/MMP9 signaling pathway, promoting OC cell
proliferation and migration^[167]32,[168]33. This pathway’s activation
is confirmed through our co-culture experiments with SKOV3 cells,
highlighting the role of GLUT1 in regulating CAFs’ metabolic activity
and promoting OC progression^[169]34.
Research in other cancers, such as breast cancer, has demonstrated that
GLUT1 enhances glycolysis in CAFs, promoting lactate production, which
facilitates tumor progression^[170]21. Our results build upon these
findings, establishing that GLUT1 overexpression in CAFs not only
increases glucose uptake but also accelerates glycolysis, resulting in
lactic acid production and the activation of the TGF-β1/p38/MMP2/MMP9
signaling pathway, enhancing OC cell proliferation and migration. This
metabolic shift within CAFs alters the tumor microenvironment, favoring
tumor growth and immune suppression by influencing the activity of
immune cells such as T cells and regulatory T cells^[171]35.
CAF-mediated metabolic reprogramming is not exclusive to OC. CAFs play
a critical role in the tumor microenvironment of various cancers,
including pancreatic, breast, and lung cancers^[172]36,[173]37. They
are transformed under the influence of growth factors, creating
barriers that inhibit immune cell infiltration and therapeutic drug
penetration, thereby contributing to tumor progression and metastasis.
Our study offers a novel perspective by focusing on the pivotal role of
GLUT1 in the OC microenvironment, demonstrating how it modulates
glucose metabolism in CAFs to facilitate tumor growth and
metastasis^[174]38,[175]39.
One of the significant innovations of this study is the use of 3D
bioprinting technology to model the OC tumor microenvironment.
Traditional 2D cultures and animal models often fail to replicate the
complexity of human tissues. In contrast, 3D bioprinted models more
accurately simulate the spatial arrangement of cells and their
interactions with the extracellular matrix. By mimicking the tumor
microenvironment, these models provide a more reliable platform for
investigating tumor growth, metastasis, and drug
response^[176]40,[177]41. In this study, 3D bioprinting was utilized to
create models of OC tumors that reflect the in vivo conditions more
closely than standard 2D models, demonstrating that the physical and
biochemical properties of the tumor microenvironment significantly
impact tumor behavior^[178]42.
Despite employing advanced technologies and conducting multi-faceted
experimental validations, this study has several limitations. First,
the relatively small sample size may affect the generalizability and
reliability of our findings. To mitigate the impact of inter-individual
heterogeneity, we selected multiple samples from both OC patients and
adjacent normal tissues for sequencing. However, future studies should
expand the sample size and include cases from different clinical stages
and subtypes to improve the robustness and applicability of the
results. Second, this study primarily focuses on the role of GLUT1 in
OC, but its function in different OC subtypes, particularly high-grade
serous ovarian cancer (HGSOC), remains to be further investigated.
Evaluating GLUT1 in HGSOC cell lines could enhance the clinical
relevance of our findings. Moreover, the metabolic regulatory
mechanisms of GLUT1, such as its influence on lactate levels and their
impact on OC cell proliferation and invasion, warrant further
exploration.
Additionally, due to technical limitations, multiplex
immunohistochemistry and spatial transcriptomics were not employed in
this study, which may have restricted our ability to comprehensively
map GLUT1 expression within the tumor microenvironment. Future studies
should integrate these technologies to gain a deeper understanding of
GLUT1’s spatial distribution and its interactions with other cellular
components. Furthermore, while we validated GLUT1’s role in OC using in
vitro and in vivo models, whether these animal models fully
recapitulate the human OC tumor microenvironment remains to be further
examined.
In conclusion, this study highlights the crucial role of GLUT1 in
CAF-mediated glucose metabolism reprogramming and provides new insights
into the metabolic dynamics of the OC tumor microenvironment
(Fig. [179]6). However, the interactions between GLUT1 and other
signaling pathways remain incompletely understood. Future studies
should incorporate single-cell metabolomics and real-time metabolic
flux analysis to elucidate the metabolic interplay between CAFs and OC
cells under different microenvironmental conditions. Additionally,
investigating the functional differences of GLUT1 in CAFs derived from
different sources will further refine our understanding of its role in
OC progression. Future studies should also expand the sample size and
include cases from different clinical stages and subtypes to validate
our findings. Moreover, integrating other omics technologies such as
proteomics and metabolomics will enable a more comprehensive
identification of key molecules involved in glucose metabolism pathways
and provide deeper insights into the pathophysiology of OC. Further
exploration of GLUT1’s role in different tumor types may extend its
application in early tumor diagnosis and treatment. Ultimately,
translating these research findings into clinical practice will offer
more personalized and precise therapeutic strategies for OC patients,
leading to improved clinical outcomes.
Fig. 6.
[180]Fig. 6
[181]Open in a new tab
Schematic representation of the molecular mechanism of fibroblast
glucose metabolism reprogramming in the OC tumor microenvironment and
tumor progression (Created by BioRender).
Materials and Methods
Ethics statement
This study strictly adhered to relevant ethical guidelines and
regulations concerning animal experimentation. All experimental
procedures were approved by the Ethics Committee of Shengjing Hospital
of China Medical University (protocol number: No. 2024PS1387K). All
animals were housed and cared for in conditions consistent with humane
principles and subjected to experiments with efforts to minimize pain.
We have complied with all relevant ethical regulations for animal use.
At the conclusion on the experiments, all nude mice were euthanized
humanely under ether anesthesia.
Clinical specimen collection
Our study selected cancer tissues from three OC patients who underwent
surgery at our hospital between January 2023 and December 2023. Tissue
samples located 2 cm away from the cancerous lesions were also
collected as controls. The ages of these three patients ranged from 30
to 50 years (42.1 ± 6.1), with tumors spreading to areas such as the
uterus, fallopian tubes, and pelvis, all classified as stage III or
higher OC. These patients had not undergone any anti-tumor therapies
such as radiation or chemotherapy before surgery and had no history of
other diseases. Basic information of the patients can be found in
Table [182]S4. After removal, tissue samples were divided into two
parts: one immediately stored in liquid nitrogen and the other fixed in
10% formaldehyde, embedded in paraffin, and stored as slides at −80 °C.
Prior to surgery, all patients read and signed informed consent forms.
This study was approved by the ethics committee and strictly followed
the Helsinki Declaration^[183]43 (protocol number: EC-2024-KS-058). All
ethical regulations relevant to human research participants were
followed.
Sample preparation for clinical sequencing
The ovarian tissues obtained were washed with cold PBS to eliminate
residual tissues outside the ovaries and disrupt the tissues.
Subsequently, the tissues were digested in a solution of DMEM
(11965092, Thermo Fisher, USA) containing 1 mg/mL collagenase (C2674,
Sigma-Aldrich, USA), 1 unit/mL DNase I, and 10% FBS at 37 °C for
30 min. The digested tissues, along with the remaining tissues, were
filtered through a 200-mesh sieve, and the filtrate was centrifuged at
4 °C for 5 min at 50 × g. The supernatant was discarded, and the cell
pellet was resuspended in a complete DMEM medium, followed by two
washes. Red blood cells were removed using red blood cell lysis buffer
(C3702-120ml, Beyotime, Shanghai, China), and dead cells were
eliminated with the Dead Cell Removal Kit (17899, StemCell
Technologies, Canada) via immunomagnetic cell separation to exclude
apoptotic and dead cells. The resulting cells were then resuspended in
PBS to obtain the sequencing sample. The prepared sequencing sample
underwent assessment for cell viability, integrity, and counting using
a microscope and flow cytometer^[184]43.
scRNA-seq and data analysis
Qualifying samples were processed using the C1 Single-Cell Auto Prep
System (Fluidigm, Inc., located in South San Francisco, California,
USA) to capture individual cells. Post-capture, the cells were lysed
within the chip to release mRNA, followed by reverse transcription to
generate cDNA. Subsequently, the lysed and reverse-transcribed cDNA
underwent pre-amplification within a microfluidic chip for downstream
sequencing. The amplified cDNA was then used to construct libraries and
subjected to single-cell sequencing on the HiSeq 4000 Illumina platform
(parameters: paired-end reads, read length of 2 × 75 bp, approximately
20,000 reads per cell).
Data analysis was performed using the “Seurat” package in R software.
Genes are highly variable in expression, meeting the criteria of 200 <
nFeature_RNA < 5000 percent.mt < 20, were selected as the top 2000
genes. To reduce the dimensionality of the scRNA-Seq dataset, Principal
Component Analysis (PCA) was conducted based on these highly variable
top 2000 genes. The top 20 principal components (PCs) were chosen for
further analysis utilizing the Elbowplot function in the “Seurat”
package. Subsequently, the FindClusters function in Seurat was employed
to identify main cell subclusters, with a resolution set to the default
value (res=1). UMAP algorithm was then used for non-linear
dimensionality reduction of the scRNA-seq sequencing data. Marked genes
for various cell subclusters were filtered using the “Seurat” package,
followed by cell annotation through the “SingleR” package. Finally,
cell communication analysis was conducted using the “CellChat” package
in R. Differential Expression Genes (DEGs) within the scRNA-Seq dataset
were screened using the “Limma” package in R. DEGs between different
samples were selected based on |log2FC | > 1 and
P.value < 0.05^[185]6,[186]44.
The single-cell data in Hierarchical Data Format version 5 (HDF5)
format and annotation results for [187]GSE139555, [188]GSE168652,
[189]GSE147082, and [190]GSE158722 were downloaded from the TISCH
database. The R software, specifically the MAESTRO and Seurat packages,
was used to process and analyze the single-cell data, and the t-SNE
method was employed to re-cluster and subgroup the cells^[191]45. The
number of cell subtypes and the expression levels of the SLC2A1 gene in
OC samples were downloaded from the Ovary Cancer sc Database^[192]46.
High-throughput transcriptome sequencing
RNA Purity and Integrity Verification: The concentration of RNA samples
was determined using the Nanodrop ND-1000 spectrophotometer (Thermo
Fisher) to ensure the absence of protein and organic contaminants,
indicated by OD260/280 ratios. RNA samples with an RIN ≥ 7.0 and a
28S:18S ratio ≥ 1.5 were used.
The sequencing libraries were prepared and sequenced by CapitalBio
Technology (Beijing, China), with each sample utilizing a total of 5 μg
of RNA. Initially, the Ribo-Zero Magnetic Kit (MRZG12324, Epicentre,
USA) was employed to remove ribosomal RNA (rRNA) from the total RNA.
Subsequently, the NEB Next Ultra RNA Library Prep Kit (E7760S, NEB,
USA) from Illumina was used for library construction. The RNA fragments
were sheared into approximately 300 base pair (bp) fragments, followed
by cDNA synthesis with first-strand cDNA using reverse transcriptase
and random primers, and second-strand cDNA synthesis in the presence of
dUTP Mix (10x) buffer. End repair of cDNA fragments included the
addition of polyA tails and sequencing adapters. After the addition of
Illumina sequencing adapters, the second strand of cDNA was digested
with USER Enzyme (M5508, NEB, USA) to create a strand-specific library.
Library DNA was amplified, purified, and enriched through PCR. Finally,
the library was analyzed using the Agilent 2100 system and quantified
with the KAPA Library Quantification Kit (kk3605, Merck, USA).
Paired-end sequencing was conducted on the Illumina NextSeq CN500
sequencer (Figure [193]S1)^[194]47,[195]48.
Transcriptome sequencing data analysis
The raw sequencing data quality of paired-end reads was assessed using
FastQC software v0.11.8. Pre-processing of the raw data was performed
using Cutadapt software version 1.18 to remove Illumina sequencing
adapters and poly(A) tail sequences. Reads with N content exceeding 5%
were filtered out using a Perl script. Subsequently, the FASTX Toolkit
software version 0.0.13 was employed to extract reads with a base
quality of at least 20 and comprising 70% of the bases. The BBMap
software was utilized to correct any anomalies in the paired-end
sequences. Finally, the filtered high-quality read fragments were
aligned to the human reference genome using Hisat2 software (version
0.7.12) for further comparative analysis.
Differential expression analysis of mRNA read counts was conducted
using the “Limma” package in R language, employing |log2FC | >1 and
P.value < 0.05 as selection criteria. The “VennDiagram” package in R
language was used for Venn analysis to identify intersecting genes.
Subsequently, the “ClusterProfiler” package in R language was applied
to perform Gene Ontology (GO) functional enrichment analysis on the
intersecting genes, focusing on biological processes (BP), molecular
functions (MF), and cellular components (CC). Visualization of the GO
enrichment results was completed through bubble and circle plots. Using
|log2FC | >1 as a filtering criterion, candidate targets were subjected
to Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis
using the “ClusterProfiler” package in R language, and the results were
depicted in bubble and circle plots^[196]36,[197]48.
Through the GeneCards database ([198]https://www.genecards.org/), 221
genes related to glucose metabolism reprogramming were retrieved using
the search term “Glycometabolism reprogramming”. The gene set obtained
was subjected to Venn analysis with differentially expressed genes in
CAFs and high-throughput transcriptome sequencing differentially
expressed genes using the “VennDiagram” package in R language,
revealing a significant overlap of genes related to cellular glucose
metabolism reprogramming, OC, and CAFs^[199]49.
Analysis of protein expression in clinical specimens
The Human Protein Atlas (HPA) database
([200]https://www.proteinatlas.org/) contains slices from 46 normal
human tissues and over 20 types of human cancers. These slices are
labeled with antibodies targeting over 11,000 human proteins. Staining
intensity is categorized into negative, weak, moderate, and strong
levels based on laser power, detector gain parameters, and visual
appearance of the images^[201]50.
TCGA data analysis
ROC Curve Analysis: RNAseq data in TPM format from TCGA and GTEx,
processed through the Toil pipeline in UCSC XENA, were extracted. The
TCGA data corresponding to ovarian serous cystadenocarcinoma and normal
tissue data from GTEx were obtained. ROC analysis was performed using
the “pROC” package in R, followed by visualization of the results using
the “ggplot2” package^[202]51.
Survival Curve Analysis: RNAseq data processed through the STAR
pipeline and clinical data from the TCGA-OV (ovarian serous
cystadenocarcinoma) project were downloaded and curated from the TCGA
database ([203]https://portal.gdc.cancer.gov). After excluding normal
and samples with missing clinical information, proportional hazard
assumption testing and fitting survival regression analysis were
conducted using the “survival” package in R. Finally, the results were
visualized using the “survminer” and “ggplot2” packages^[204]52.
Immunohistochemistry staining
The paraffin was cooled on ice or in a refrigerator at 4 °C. After
cooling, embedding slicing was performed. The paraffin sections were
left to dry overnight; then, the slides were placed in an oven at 60 °C
for 20 min. Subsequently, the slices were immersed in xylene for
10 min, followed by another 10 min after changing the xylene. Next,
hydration was carried out in absolute alcohol for 5 min, followed by
another 5 min in fresh absolute alcohol, then sequentially hydrated in
95% and 70% alcohol for 10 min each, and finally rinsed in distilled
water for 5 min. The sections were immersed in citrate buffer (pH 6.0),
microwave on high heat for 8 min, and then cooled to room temperature.
PBS (pH 7.2 ~ 7.6) was used to wash the sections three times, each for
3 min.
To deactivate endogenous enzymes, 3% H[2]O[2] was added at room
temperature for 10 min. Afterward, the sections were washed three times
with PBS for 3 min each. Subsequently, they were blocked with normal
goat serum blocking solution (E510009, Sinobiological Co., Ltd.,
Shanghai, China) at room temperature for 20 min.
After incubation, primary antibodies were applied as follows:α-Smooth
Muscle Actin (α-SMA) (ab32575, Dilution: 1:100), Glucose Transporter 1
(GLUT1) (ab115730, Dilution: 1:250), Transforming Growth Factor Beta 1
(TGF-β1) (ab215715, Dilution: 1:500), Phosphorylated p38
Mitogen-ActivatedProteinKinase(p-p38)(ab30838,Dilution:1:20),Anti-Cytok
eratin 8 antibody (KRT8/2174 R) (ab234348, Dilution: 1:200), Fibroblast
Activation Protein (FAP) (ab314456, Dilution: 1:1000), Thy1 (CD90)
(ab97779, Dilution: 1:500), Matrix Metalloproteinase 2 (MMP2) (ab92574,
Dilution: 1:500), Matrix Metalloproteinase 9 (MMP9) (ab76003, Dilution:
1:1000).
All antibodies were sourced from Abcam, UK. Antigen Retrieval: For
antibodies where pH9 antigen retrieval is recommended (such as FAP,
MMP2, MMP9, and TGF-β1), tissue sections were incubated in antigen
retrieval buffer (pH 9) at 95–100 °C for 20 min prior to blocking and
antibody incubation. The primary antibodies were left overnight at
4 °C, rinsed three times in PBS and then incubated with secondary
antibodies goat anti-mouse IgG (ab205719, dilution 1:5000, Abcam,
Cambridge, UK) or goat anti-rabbit IgG (ab6721, dilution 1:5000, Abcam,
Cambridge, UK) for 30 min.
Streptavidin-Biotin Complex (SABC, P0603, Beyotime, Shanghai, China)
was added and kept warm at 37 °C for 30 min in a constant temperature
chamber. A drop of DAB chromogenic reagent (P0203, Beyotime, Shanghai,
China) was added to the specimen, incubated for 6 min, stained with
hematoxylin for 30 s, and then dehydrated in 70%, 80%, 90%, and 95%
ethanol, and absolute ethanol for 2 min each. The sections were
immersed in xylene for 5 min twice, mounted with neutral resin, and
observed and analyzed using a brightfield microscope (BX63, Olympus,
Japan) with five random high-magnification fields per section. Positive
cell rates were calculated as the number of positive cells divided by
the total cell count per field using Image-Pro Plus 6.0 software. Each
experiment was repeated three times.
RT-qPCR
Total RNA was extracted from tissues and cells using the Trizol reagent
kit (A33254, Thermo Fisher, USA). Subsequently, cDNA was prepared using
the reverse transcription kit (RR047A, Takara, Japan). The reaction mix
was prepared using the SYBR® Premix Ex TaqTM II kit (DRR081, Takara,
Japan), and real-time RT-qPCR reactions were performed on an ABI7500
real-time PCR system (Thermo Fisher, USA). The PCR program was set as
follows: initial denaturation at 95 °C for 30 s, followed by 40 cycles
of denaturation at 95 °C for 5 s, annealing at 60 °C for 30 s,
extension at 95 °C for 15 s, and final extension at 60 °C for 60 s.
Amplification curves were then generated. GAPDH was used as the
reference gene, and all RT-qPCR reactions were performed in triplicate
and repeated three times. The fold change in gene expression between
the experimental group and the control group was calculated using the
2^−ΔΔCt method, where ΔΔCT = ΔCt [experimental group] - ΔCt [control
group], and ΔCt = Ct [target gene] - Ct [reference gene]. Ct represents
the cycle threshold at which the fluorescent signal reaches a set
threshold during amplification, indicating exponential growth^[205]53.
Primer details can be found in Table [206]S5.
Western blot
Tissue and cell total protein extraction was performed using RIPA lysis
buffer (P0013B, Beyotime, Shanghai, China) containing 1%
phenylmethanesulfonyl fluoride (PMSF), following the manufacturer’s
instructions. The supernatant was used to determine the total protein
concentration of each sample with a BCA assay kit (P0011, Beyotime,
Shanghai, China). The protein concentration was adjusted to 1 μg/μL,
and each sample volume was set at 100 μL. The samples were boiled at
100 °C for 10 min to denature the proteins and then stored at −80 °C
for later use. Based on the size of the target protein bands, 8%–12%
SDS-PAGE gels were prepared, and 50 μg of protein samples were loaded
into each lane using a micropipette. Electrophoresis was conducted
under constant pressure at 80 V to 120 V for 2 h. After
electrophoresis, wet transfer was performed at a constant current of
250 mA for 90 min to transfer the proteins from the Gel to a PVDF
membrane (1620177, Bio-Rad, USA).
The membrane was blocked at room temperature with 1 × TBST containing
5% skim milk for 1 h, followed by washing with 1 × TBST for 10 min. The
primary antibody (antibody information in Table [207]S6) was incubated
overnight at 4 °C, followed by three washes with 1 × TBST for 10 min
each. Subsequently, three additional 5-min washes with 1 × TBST at room
temperature were performed. The membrane was then incubated with
HRP-conjugated goat anti-rabbit IgG (ab6721, dilution 1:5000, Abcam,
Cambridge, UK) or goat anti-mouse IgG (ab205719, dilution 1:5000,
Abcam, Cambridge, UK) secondary antibodies at room temperature for 1 h.
After three washes with 1 × TBST at room temperature for 5 min each,
the membrane was immersed in ECL reaction solution (1705062, Bio-Rad,
USA) and incubated at room temperature for 1 min. The liquid was
removed, the membrane was covered with plastic wrap, and band exposure
was performed using the Image Quant LAS 4000 C gel imaging system (GE
Healthcare, USA). The gray value ratio of the target band to the
reference band α-tubulin was calculated as the relative protein
expression level to evaluate the protein expression levels^[208]53.
Each experiment was repeated three times. All the original WB images
can be found at Figures [209]S10–[210]S34.
Isolation and identification of CAFs
Tumor tissues from three OC patients were collected and washed twice
with pre-cooled PBS solution (70011044, Thermo Fisher, USA) containing
2% dual antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin,
15140163, Thermo Fisher, USA) to remove blood clots and necrotic tissue
from the surface. The tissues were then minced and added to a 0.1% type
IV collagenase (17104019, Thermo Fisher, USA) solution containing 10%
FBS (10100147 C, Thermo Fisher, USA), transferred into a centrifuge
tube, and digested in a shaking incubator at 37 °C for 30–40 min. The
digestion solution, along with the remaining tissue, was filtered
through a 200-mesh sieve and centrifuged at 4 °C for 5 min (50 × g),
and the supernatant was discarded. The pellet was resuspended in a
complete DMEM medium (11965092, Thermo Fisher, USA) and washed twice.
To remove red blood cells, red blood cell lysis buffer (C3702-120 ml,
Beyotime, Shanghai, China) was used, and the cell density was adjusted
to 10^6 cells/mL. Based on the differences in growth rates and adhesion
abilities between CAFs and other cells, the cell suspension was added
to the first well of a 6-well plate and left to stand for 20 min. The
adherent cells, primarily fibroblasts, were transferred to the second
well, and after another 20 min, the supernatant was discarded. The
adherent cells in both wells were trypsinized, mixed, and cultured in
DMEM with 10% FBS in a humidified incubator at 37 °C and 5% CO[2], with
medium changes every 3 days. Finally, the obtained fibroblasts were
identified by immunofluorescence staining for fibroblast marker
proteins, and CAFs were immortalized by treatment with human telomerase
reverse transcriptase (TERT, ENZ-1016, Nanjing SBL Biotech Co., Ltd.,
Nanjing, China)^[211]53.
Immunofluorescence staining
Cells or tissues were rinsed three times with PBS for 2 min each,
followed by fixation in ice-cold methanol at −20 °C for 30 min. After
removing excess methanol, samples were washed again with PBS three
times for 5 min each. Subsequently, they were incubated in 0.1% Triton
X-100 at room temperature for 15 min and then washed with PBS three
times for 5 min each. Blocking was performed with BSA for 30 min,
followed by the addition of primary antibodies against CD31 (MA3100,
Thermo Fisher, USA, 1:200), Cytokeratin (MA1-06312, Thermo Fisher, USA,
1:200), Vimentin (MA5-11883, Thermo Fisher, USA, 1:250), and GLUT1
(ab115730, Abcam, UK, 1:1000) and incubation at 37 °C for 60 min.
Subsequent washing steps with PBS were carried out for 5 min three
times. FITC-conjugated secondary antibodies, either goat anti-mouse IgG
(A10551, Thermo Fisher, USA, 1:200) or goat anti-rabbit IgG (A-11008,
Thermo Fisher, USA, 1:500), were then applied and incubated in the dark
at 37 °C for 60 min, followed by a 3-min wash with PBS repeated three
times. DAPI staining solution was added for 10 min, followed by three
washes with PBS to remove excess DAPI (C1002, Beyotime, Shanghai,
China). Finally, 20 μL of mounting medium was applied for slide
sealing. After the mounting medium dried, samples were observed and
photographed using a fluorescence microscope. Quantification involved
determining the fluorescent coverage area in fixed fields at 40 ×
magnification, with image analysis performed using ImageJ Pro Plus 6.0
software, calculating the average from six fields of view^[212]53.
Co-localization immunofluorescence staining
Cells or tissues were washed three times with PBS for 2 min each, and
then fixed in ice-cold methanol at −20 °C for 30 min. After removing
excess methanol, they were washed again with PBS three times for 5 min
each. The samples were incubated in 0.1% Triton X-100 at room
temperature for 15 min and then washed three times with PBS for 5 min
each. Blocking was performed with BSA for 30 min, followed by the
addition of Alexa Fluor® 555 Anti-Cytokeratin (ab214391), Alexa Fluor®
647 Anti-CD31 (ab305210), Alexa Fluor® 488 Anti-Vimentin (ab185030),
and Alexa Fluor® 647 Anti-GLUT1 (ab195020). The samples were incubated
at 37 °C in the dark for 60 min and then washed three times with PBS
for 3 min each. DAPI staining solution was added for 10 min, followed
by three washes with PBS to remove excess DAPI (C1002, Beyotime,
Shanghai, China). Finally, 20 μL of mounting medium was applied for
sealing. After the mounting medium dried, images were captured using a
fluorescence microscope. Quantification was conducted by measuring the
fluorescence coverage area in fixed fields at 40 × magnification, with
image analysis performed using ImageJ Pro Plus 6.0 software, and the
average from six fields was calculated^[213]54.
Cell culture and treatment
The cells used in the experiments were purchased from Beijing
Biosciences Biotechnology Co., Ltd., including HEK-293T human embryonic
kidney cells, SKOV3 and A2780 human OC cells, with corresponding
product codes Bio-72947, Bio-73156, and Bio-105918. HEK-293T cells were
cultured in a high glucose DMEM medium (11965084, Thermo Fisher
Scientific, USA) containing 10% FBS and 1% penicillin-streptomycin.
SKOV3 cells were cultured in McCoy’s 5 A medium (16600082, Thermo
Fisher Scientific, USA) with 10% FBS and 1% penicillin-streptomycin.
A2780 cells were cultured in RPMI-1640 medium (11875119, Thermo Fisher,
USA) with 10% FBS and 1% penicillin-streptomycin.
All cells were cultured in a humidified incubator at 37 °C with 5%
CO[2](Heracell™ Vios 160i CR CO[2] incubator, 51033770, Thermo
Scientific™, Germany). When the cells reached 80%–90% confluence, they
were passaged. In the experiments, CAFs and OC cells were co-cultured
at a 2:1 ratio. As described above, different OC cells were cultured
using different media. When CAFs were co-cultured with HEK-293T cells,
they were cultured in a high-glucose DMEM medium (11965084, Thermo
Fisher Scientific, USA) containing 10% FBS and 1%
penicillin-streptomycin. For co-culture with SKOV3 cells, McCoy’s 5 A
medium (16600082, Thermo Fisher Scientific, USA) containing 10% FBS and
1% penicillin-streptomycin was used. For co-culture with A2780 cells,
RPMI-1640 medium (11875119, Thermo Fisher, USA) containing 10% FBS and
1% penicillin-streptomycin was used. Cells were treated with 10 mM
lactate dehydrogenase (LDH) inhibitor Galloflavin (HY-W040118,
MedChemExpress, USA), 20 mM lactic acid (HY-B2227, MedChemExpress,
USA), or 10 μM cisplatin (HY-17394, MedChemExpress, USA) for
24 h^[214]21,[215]53. The medium was replaced with glucose-free DMEM,
and the cells were incubated at 37 °C under 5% carbon dioxide and 95%
nitrogen for 60 min to construct the hypoxia model^[216]55.
Lentivirus and plasmid transfection
In this study, CAF cells were subjected to lentivirus-mediated
overexpression or silencing. The lentivirus packaging service was
provided by GeneEngine (Shanghai, China). The pHAGE-puro series
plasmids and helper plasmids pSPAX2 and pMD2.G were obtained from
Addgene (USA) with catalog numbers #118692, #12260, and #12259,
respectively. Similarly, the pSuper-retro-puro series plasmids and
helper plasmids gag/pol and VSVG were acquired from Addgene (USA) with
catalog numbers #113535, #14887, and #8454, respectively. The
constructed plasmids were co-transfected into HEK293T cells (Bio-72947,
Beijing Baio Bowei Biotechnology Co., Ltd.) using Lipofectamine 2000
reagent (catalog #11668030, Thermo Fisher, USA). The supernatant was
collected after 48 h of cell culture, filtered through a 0.45 µm
filter, and the virus was concentrated by centrifugation. Subsequently,
the concentrated virus was harvested after 72 h, and the titers were
determined from a mixture of the collected viruses.
During the logarithmic growth phase, cells were dissociated using
trypsin. Cells were seeded at 1 × 10^5 cells per well in a 6-well
plate, cultured for 24 h, and infected with lentivirus (MOI = 10,
working titer approximately 5 × 10^6 TU/mL) along with 5 μg/mL
polybrene (catalog #TR-1003, Merck, USA) in the medium when the cell
confluency reached around 75%. After a 4 h infection, the medium was
replaced with fresh medium to dilute the polybrene, and 24 h
post-infection, fresh medium was again provided.
For the establishment of stable cell lines, cells were cultured in a
medium containing 2 μg/mL puromycin (catalog #E607054, GeneEngine,
Shanghai, China) after infection. During passaging, puromycin
concentration was gradually increased in increments of 2, 4, 6, 8, and
10 μg/mL for resistance selection to obtain stable cell lines. When
cells no longer died in the puromycin-containing medium, cells were
collected, and the knockout efficiency was confirmed by Western blot
and RT-qPCR^[217]56. The silent lentivirus sequences are listed in
Table [218]S7, with the optimal silencing sequences selected for
further experimentation.
Cell grouping: sh-NC CAFs cells (GLUT1-silenced lentivirus control
cells); sh-GLUT1 CAFs cells (GLUT1-silenced cells); oe-NC CAFs cells
(cells transfected with empty lentivirus); oe-GLUT1 CAFs cells
(GLUT1-overexpressing cells).
CCK-8 assay
The OC cells of each group were digested and resuspended, adjusting the
cell concentration to 1 × 10^5 cells/mL, and then seeded at a volume of
100 µL per well into a 96-well plate for overnight incubation.
Following the instructions provided in the CCK-8 assay kit (C0041,
Beyotime, Shanghai, China), the cells were treated, and their viability
was assessed using the CCK-8 method after 12, 24, 36, and 48 h of
incubation. During each assessment, 10 µL of the CCK-8 detection
solution was added, followed by incubation at 37 °C in a 5% CO[2]
humidified chamber for 1 h. Subsequently, the absorbance at 450 nm was
measured using an ELISA reader to calculate cell viability^[219]39.
EDU staining
CAF and OC cells were co-cultured in a 2:1 ratio in a 24-well plate
with a seeding density of 1 × 10^5 cells per well, with each cell group
having 3 replicate wells. The cells were treated with EDU
(5-Ethynyl-2’-deoxyuridine) solution (ST067, Beyotime, Shanghai, China)
at a concentration of 10 µmol/L in the culture medium and then
incubated in a CO[2] chamber for 2 h. After removing the culture
medium, the cells were fixed at room temperature for 15 min using a PBS
solution containing 4% paraformaldehyde, washed twice with PBS
containing 3% BSA, incubated at room temperature with PBS containing
0.5% Triton-100 for 20 min, and washed twice again with PBS containing
3% BSA. Subsequently, 100 µL of staining solution was added per well,
followed by a light-protected 30-min incubation at room temperature.
DAPI staining was then applied for 5 min to label the cell nuclei, and
after mounting the coverslip, 6-10 random fields of view were observed
under a fluorescence microscope (FM-600, Shanghai Pudan Optical
Instrument Co., Ltd.) to record the number of positive cells in each
field. The Edu labeling rate (%) was calculated as the percentage of
positive cells divided by the sum of positive and negative cells
multiplied by 100%^[220]57. Each experiment was repeated 3 times.
Formation of cell colonies experiment
CAF cells and the tested OC cells (n = 600) were seeded in a 2:1
concentration ratio in 6-well plates and maintained in a culture medium
for 2 weeks. The culture medium was changed every 3 days. Upon
completion of the incubation, cell colonies were fixed with methanol
and stained with 0.1% crystal violet (C0121, Beyotime, Shanghai, China)
for 15 min. After washing, photographs were taken, and visible cell
colonies were analyzed and quantified using Image-Pro Plus 6.0
software^[221]58.
Transwell experiment
CAF cells were seeded in the lower chamber of Transwell culture plates
(10^5 cells/well), while OC cells were seeded in the upper chamber of
Transwell plates (5 × 10^4 cells/well). A 50 μL layer of Matrigel
(354234, BD Biosciences, USA) was coated in the Transwell chambers and
allowed to solidify at 37 °C for 30 min to conduct an invasion
experiment. After rinsing the Matrigel, cells were diluted in a culture
medium without FBS to the appropriate concentration and seeded in the
plates. After 24 h, the chambers were removed, upper chamber cells were
discarded, and the cells were fixed with 4% PFA at room temperature for
30 min. Subsequently, cells were stained with 0.1% crystal violet for
30 min, five random areas were selected, images were captured under an
inverted microscope (IXplore Pro, Olympus, Japan), and cell counts were
calculated^[222]59. The experiment was repeated three times. For cell
migration experiments, Matrigel was not necessary.
Wound healing assay
On the bottom surface of a 6-well plate, lines were marked at intervals
of 0.5–1 cm using a ruler and a marker, ensuring that each well was
intersected by at least five lines. The CAFs and OC cells were seeded
in a 2:1 ratio into the 6-well plate and allowed to grow to confluency,
followed by scratch assays using a 200 μL pipette tip held
perpendicular to the horizontal lines on the back. Subsequently, the
cells were cultured in a serum-free medium, and the distance between
wounds was measured and recorded under an optical microscope (model:
DM500, Leica) at 0 and 24 h. Images of cell migration for each group
were captured under an inverted microscope to assess their migratory
capabilities. The Image-Pro Plus 6.0 software was employed for the
analysis of wound distances, and the wound healing rate was calculated
using the formula provided in the refs. ^[223]39,[224]60.
[MATH: Woundhealingrate=dis<
mi>tance0h−distance24hdistance0h :MATH]
Here, “distance[0 h]” and “distance[24 h]” respectively indicate the
distance between cell scratches at 0 h and 24 h after scratching.
Flow cytometry
For each well, 1 × 10^6 cells were plated. Following cell collection,
195 µL of Annexin V-FITC binding buffer was added to resuspend the
cells, followed by the addition of 5 µL of Annexin V/FITC solution and
10 µL of the PI solution, and then the cells were incubated at room
temperature in the dark for 15 min. Flow cytometry analysis was
performed within 20 min using the BD FACSCalibur to determine the
apoptotic rate, which is calculated as the sum of the apoptotic cell
proportions in the Q1-UR (upper right) and Q1-LR (lower right)
quadrants^[225]61. The gating strategy for flow cytometry analysis of
apoptosis is detailed in Figure [226]S35.
Metabolic measurements
The Seahorse XFe96 extracellular flux analyzer (Agilent Technologies)
was utilized for metabolic analysis. The extracellular acidification
rate (ECAR) and the oxygen consumption rate (OCR) per well were
calculated. The cells were subjected to XF glycolytic stress or XF Cell
Mito testing by treating them with the following concentrations of
compounds injected into the wells: 10 mM glucose (50-99-7,
Sigma-Aldrich, USA), 2 μM oligomycin (1404-19-9, Sigma-Aldrich, USA),
50 mM 2-deoxy-D-glucose (2-DG) (154-17-6, Sigma-Aldrich, USA), 1 μM
carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (370-86-5,
Sigma-Aldrich, USA), and 0.5 μM rotenone (83-79-4, Sigma-Aldrich, USA).
The XF glycolytic stress or XF Cell Mito Test Kits were procured from
Agilent Technologies (USA)^[227]57,[228]58.
Glucose uptake measurement
The determination of glucose uptake was carried out following the
instructions provided in the glucose uptake assay kit (ab136955, Abcam,
UK). Briefly, cells were starved for 24 h, then 2-deoxyglucose (2-DG)
was added and incubated at 37 °C for 20 min. The cells were washed with
PBS to remove residual 2-DG. Subsequently, the cells were lysed to
extract 2-deoxyglucose-6-phosphate (2-DG6P). Standards and samples were
added to the reaction wells, followed by incubation with the reaction
mixture. Finally, the reaction was terminated by heating, neutralized
after cooling, and absorbance was measured at 412 nm using a microplate
reader to evaluate glucose uptake based on the standard curve^[229]61.
LDH enzyme activity assay
The LDH enzyme activity assay was performed according to the
instructions provided in the LDH assay kit (ab102526, Abcam, UK).
During the assay, LDH reduces NAD+ to NADH, which can be detected using
a specific probe to measure LDH enzyme activity. In brief, cells were
collected and centrifuged at 180 × g for 5 min at 4 °C. The supernatant
was mixed with the reaction reagent, and absorbance was continuously
read at 450 nm wavelength for 25 min^[230]62.
Lactic acid release measurement
The quantification of lactic acid release was conducted following the
instructions in the lactic acid assay kit (MAK064, Sigma-Aldrich, UK).
Simply put, cells were collected, washed with PBS, homogenized in
lactic acid assay buffer, and deproteinized using a 10 kDa filter. The
samples were then mixed with lactic acid assay buffer, lactic acid
enzyme mix, and lactic acid probe. After incubating at room temperature
in the dark for 30 min, absorbance was measured at 570 nm using a
spectrophotometer to determine lactic acid content based on the
standard curve^[231]61.
Measurement of ATP content
To determine the ATP levels, an ATP assay kit (S0026, Beyotime,
Shanghai, China) was utilized following the manufacturer’s protocol.
Initially, cells were collected and lysed using a lysis buffer,
followed by centrifugation at 12,000 g for 6 min to obtain the
supernatant. Subsequently, both the samples and standard samples were
added to the wells, along with the ATP assay reagent, and then
incubated for 3 min. The ATP content was measured using a
spectrophotometer (Promega, Madison, WI, USA) and evaluated based on
the standard curve^[232]63.
Preparation of biochemical ink and 3D bioprinting
Preparation of biochemical ink involved weighing the required amounts
of gelatin (Gelatine, Gel, EFL-GEL-001, Suzhou Institute of Intelligent
Manufacturing) and sodium alginate (Sodium alginate, SA, EFL-Alg-300K,
Suzhou Institute of Intelligent Manufacturing) powders, which were
subjected to one hour of UV decontamination. After separate dissolution
and dispersion, the components were mixed in proportion and added to a
solvent in a sterile glass bottle. The solution was kept overnight at
37 °C for complete dissolution. Following the digestion of cells with
trypsin, they were suspended in the respective culture medium and
incorporated into the dissolved hydrogel in specified ratios (solvent
and cell ratios per group are detailed in Table [233]S8). Gentle
stirring ensured even cell distribution and minimized air bubble
incorporation to obtain the biochemical ink.
For 3D bioprinting, the prepared biochemical ink was poured into the
cartridge and left at room temperature for 15–20 min before being
loaded into the bio 3D bioprinter (Axolotl BIOSYSTEMS, AXO A3) using a
dispensing needle. A 24-well plate was employed as the bioprinting and
incubation container for tumor model construction unless stated
otherwise; all subsequent experiments were conducted in the 24-well
plate. Bioprinting parameters were set as follows: mid-needle diameter
of 23 G (0.60 mm), injector and needle temperature at 37 °C, bed
temperature at 8 °C, printing head speed at 4-5 mm/s, and extrusion
pressure at 10–20 kPa. Subsequent to bioprinting, the obtained 3D
structure was cross-linked with 500 μL of 100 mM calcium chloride
(CaCl[2]) for 7 min, followed by a wash with fresh culture medium. Each
structure was supplemented with 1 mL of complete culture medium and
incubated at 37 °C with 5% CO[2] for downstream experimentation. The
described experiments were repeated three times, consistently resulting
in stable cell-laden biochemical ink structures^[234]22.
Cell viability assessment
Following 3D bioprinting, live and dead cells were identified using
Calcein AM (C3099, Thermo Fisher, USA) and Propidium Iodide (PI)
(DN1005-010, innibio, USA). Cells were cultured in a medium containing
1 μM Calcein AM at 37 °C for 30 min, followed by three washes with PBS.
Subsequently, cells were incubated in a medium with 1 μM PI for 10 min
and washed three times with PBS. Images were captured using a confocal
microscope (Carl Zeiss AG, Germany, model 880), with each image
representing a different field of view. Experiments were performed on
three independent samples, and cell viability was calculated using
ImageJ software.
For additional cell viability assessment, cells were treated according
to the instructions of the CCK-8 kit (C0041, Beyotime, Shanghai,
China). Cell viability was assessed using the CCK-8 assay at 48 h
post-culturing. Each measurement involved the addition of 10 μL of
CCK-8 detection solution, followed by a 4 h incubation in a cell
culture incubator. Subsequently, the absorbance at 450 nm was measured
using a microplate reader to calculate cell viability^[235]22.
Hematoxylin and Eosin (H&E) staining
H&E staining was performed using the Hematoxylin and Eosin staining kit
(C0105S, Beyotime, Shanghai, China). Sections of 3D bioprinted
structures were fixed in 4% paraformaldehyde, dehydrated, cleared,
embedded in paraffin wax, sectioned into 5 μm thick slices using a
microtome, and subsequently processed through baking,
deparaffinization, hydration to water, hematoxylin staining, rinsing
with distilled water, immersion in 95% ethanol, eosin staining,
differentiation in 70% acid ethanol, dehydration, clearing in xylene,
and mounting with neutral resin. The prepared slides were observed for
morphological changes in tissue under an optical microscope^[236]22.
In vivo animal experiments
Twenty-seven SPF female BALB/c nude mice, aged 6-8 weeks and weighing
between 18 and 25 grams, were obtained from Beijing Vital River
Laboratory Animal Technology Co., Ltd. (Beijing, China) and sourced
from lot number 409. The mice were individually housed in cages in an
SPF animal facility with a 12-h light-dark cycle, maintaining a
humidity range of 60%–65% and a temperature range of 22–25 degrees
Celsius. They had ad libitum access to food and water for one week of
acclimatization before commencing the experiments. The health status of
the mice was observed prior to the initiation of the experiments. This
experimental protocol and animal procedures were approved by the
Institutional Animal Ethics Committee.
Establishment of Subcutaneous OC Transplant Nude Mouse Model: Eighteen
nude mice were randomly divided into 3 groups: SKOV3 group (injected
with 4 × 10^6 SKOV3 cells in 100 µL PBS), SKOV3+CAFs group (mixture of
SKOV3 and CAFs cells in a 1:1 ratio, injected with 4 × 10^6 cells in
100 µL PBS), and SKOV3+CAFs-sh-GLUT1 group (mixture of SKOV3 and
CAFs-sh-GLUT1 cells in a 1:1 ratio, injected with 4 × 10^6 cells in
100 µL PBS), each group consisting of 6 nude mice. 4 × 10^6 cells from
each group were injected subcutaneously into the mice’s backs to
establish subcutaneous xenograft models. Starting from day 8, the width
(W) and length (L) of the tumors in each group of mice were measured
using a caliper, and measurements were taken every 4 days to monitor
tumor growth. The tumor volume (V) was calculated using the formula V =
(W^2 × L) / 2. On day 35 post-injection, the mice were euthanized,
tumors dissected, photographed, and tumor weights recorded.
Establishment of Peritoneal Metastasis Nude Mouse Model: Nine nude mice
were randomly divided into 3 groups: SKOV3 group (injected with
4 × 10^6 SKOV3 cells in 100 µL PBS), SKOV3+CAFs group (mixture of SKOV3
and CAFs cells in a 1:1 ratio, injected with 4 × 10^6 cells in 100 µL
PBS), and SKOV3+CAFs-sh-GLUT1 group (mixture of SKOV3 and CAFs-sh-GLUT1
cells in a 1:1 ratio, injected with 4 × 10^6 cells in 100 µL PBS), each
group consisting of 3 nude mice. Before the experiment, firefly
luciferase reporter gene plasmid (D2102, Beyotime, Shanghai, China) was
transfected into the SKOV3 cells, followed by intraperitoneal injection
to construct a peritoneal metastasis model. Body weights were measured
every 4 days post-cell injection. In the fourth week post-cell
injection, mice were intraperitoneally injected with 150 mg/kg of
luciferin (ST196, Beyotime, Shanghai, China), followed by anesthesia
induction with 2% isoflurane and O[2] mixture, live imaging using the
IVIS Lumina II in-vivo imaging system (IVIS Lumina Series, PerkinElmer,
USA) to observe tumor growth and metastasis. Subsequently, euthanasia
was performed, and visible tumor nodules were excised, weighed, and
analyzed using the Living Image software (PerkinElmer, USA) for
quantitative evaluation.
TUNEL detection of apoptosis in cells
TUNEL staining was performed on tissue cells using the TUNEL assay kit
(C1088, Beyotime, Shanghai, China). In brief, tumor tissues were fixed
in 4% paraformaldehyde for 30 min, washed three times with PBS,
permeabilized with PBS containing 0.3% Triton X-100 for 3 min, and then
incubated at room temperature for 5 min followed by two additional PBS
washes. Subsequently, 50 μL of TUNEL detection solution was added, and
the cells were incubated in the dark at 37 °C for 60 min. After washing
three times with PBS, the cells were counterstained with DAPI
(10 μg/mL) for 10 min and mounted with an anti-fade mounting medium
before observation under a fluorescence microscope. The Cy3 excitation
wavelength was 550 nm, and the emission wavelength was 570 nm (red
fluorescence). The percentage of apoptotic cells was calculated using
Image-Pro Plus 6.0 software^[237]53.
Statistical and reproducibility
Data were obtained from at least three independent experiments and
presented as mean ± standard deviation (Mean ± SD). Two-sample
independent t-tests were used for comparisons between the two groups.
One-way analysis of variance (ANOVA) was employed for comparisons
involving three or more groups. Post hoc pairwise comparisons between
groups were conducted using Tukey’s Honestly Significant Difference
(HSD) test when significant differences were detected by the ANOVA. For
non-normally distributed or heteroscedastic data, the Mann-Whitney U
test or Kruskal-Wallis H test was applied. All statistical analyses
were performed using GraphPad Prism 9.5.0 (GraphPad Software, Inc.) and
R version 4.2.1 (R Foundation for Statistical Computing). All
experiments were repeated at least three times independently, and
biological replicates were defined as independently cultured cell
samples or separate animals used in each experimental condition. For
animal experiments, sample sizes were determined based on previous
studies and power calculations to ensure adequate statistical power. A
significance level of 0.05 was set, with a two-sided p-value less than
0.05 considered statistically significant and greater than 0.05
considered not statistically significant.
Supplementary information
[238]Supplementary information^ (3.6MB, pdf)
[239]42003_2025_8380_MOESM2_ESM.pdf^ (38.1KB, pdf)
Description of Additional Supplementary Files
[240]Supplementary Data 1^ (31.6KB, xlsx)
[241]nr-reporting-summary^ (2.5MB, pdf)
Acknowledgements