Abstract

   Hepatocellular carcinoma (HCC) is a major contributor to global
   cancer-related mortality. Serglycin (SRGN) is involved in the
   progression of various cancers, and its overexpression is related to
   poor prognosis in HCC patients. Its biological role in HCC
   aggressiveness is unknown. This study aims to elucidate the mechanism
   of SRGN in HCC. Via in vitro and vivo experiments, we identified SRGN
   as a critical regulator of HCC cells migratory capability and
   metastasis. SRGN significantly correlated with maintaining
   stemness-like characteristics, emphasizing its central role in HCC
   progression. Mechanistically, SRGN activated YAP into the tumor cells'
   nucleus. Moreover, SRGN selectively upregulated CRISPLD2, establishing
   the SRGN/YAP/CRISPLD2 axis and promoting metastatic behavior in HCC
   cells. Our results revealed that CRISPLD2 is a direct target of the
   SRGN-mediated YAP/TEAD1 complex. SRGN orchestrated stemness
   maintenance, tumorigenesis, and metastasis in an autocrine way by
   selectively reactivating the novel YAP/CRISPLD2 axis. Besides,
   sorafenib with verteporfin showed a certain therapeutic effect in
   SRGN-positive individuals. Our work clarifies the mechanism by which
   SRGN promotes the invasiveness of HCC. It provides insights into
   targeting the SRGN-triggered signaling pathway as a potential new
   direction in treating HCC.

   Keywords: Hepatocellular carcinoma, SRGN, Transcriptome, Tumor
   progression, CRISPLD2

Introduction

   Hepatocellular carcinoma (HCC) is a prevalent type of malignancy that
   contributes substantially to the global burden of cancer
   mortality[51]^1. Although surgical removal, liver transplantation, and
   ablation are recognized as curative options, many HCC patients are
   either not eligible for surgery or face recurrence due to the cancer's
   local spread or the development of metastases[52]^2.

   Proteoglycans, a class of macromolecules synthesized ubiquitously
   across cell types, are broadly classified into three categories, cell
   surface-associated proteoglycans, extracellular matrix (ECM)-secreted
   proteoglycans, and intracellular proteoglycans[53]^3. Emerging evidence
   underscores the critical role of proteoglycan dysregulation in tumor
   microenvironment remodeling and cancer pathogenesis[54]^4. To date,
   serglycin (SRGN) remains the sole characterized member of the
   intracellular proteoglycan subfamily that has garnered increasing
   attention owing to pleiotropic functions that have varying capabilities
   depending on the cell and immunological microenvironment.[55]^5^,
   [56]^6.

   SRGN is a low molecular weight proteoglycan widely distributed across
   various cell types and can be both intracellular and integrated into
   the ECM[57]^7. The core protein of SRGN comprises one hundred
   fifty-eight amino acids. It is divided into three functional regions, a
   signal peptide, N-terminal, and C-terminal. The C-terminal region
   contains multiple serine and glycine repeat sequences responsible for
   binding to glycosaminoglycans, such as heparin, heparan sulfate, or
   chondroitin sulfate[58]^5^, [59]^8. SRGN is predominantly expressed in
   hematopoietic, endothelial, tumor, and embryonic stem
   cells[60]^9^-[61]^14. It plays a vital role in storing and secretion of
   various proteases, chemokines, and cytokines[62]^6^, [63]^15. It is
   also implicated in immune system functions, particularly within mast
   cells, cytotoxic T lymphocytes, macrophages, and neutrophils[64]^6^,
   [65]^16^-[66]^22. Additionally, SRGN in cancer was first discovered as
   a biomarker for acute myeloid leukemia and contributes to the immune
   evasion of tumor cells in multiple myeloma by inhibiting complement
   activity[67]^23^, [68]^24. SRGN regulates the migration and metastasis
   of head and neck squamous, nasopharyngeal, esophageal, breast, lung,
   and colorectal cancers through autocrine or paracrine[69]^16^,
   [70]^25^-[71]^29.

   Recent clinical cohort studies have demonstrated a marked
   overexpression of SRGN in HCC tissues. This heightened expression
   correlates with severe clinical outcomes, including vascular invasion,
   extrahepatic metastasis, and early postoperative recurrence of liver
   cancer. Notably, SRGN is a vital independent factor affecting HCC
   patient survival rates, where elevated protein levels strongly indicate
   a poor clinical prognosis[72]^30^, [73]^31. Although research hinted at
   a role for SRGN in promoting HCC aggressiveness, the specific
   mechanisms driving this effect are still under investigation.

   In this study, we conduct a comprehensive analysis to explore the
   biological roles of SRGN in influencing HCC cells' aggressiveness and
   its related signaling pathway.

Methods

Processing and analysis of scRNA-seq and bulk-seq data

   A full description of processing and analysis of scRNA-seq data,
   including data collection and quality control, cell type determination,
   cell subclusters metabolic analysis, copy number variation (CNV)
   analysis from scRNA-seq, pathway analysis, collection and calculation
   of functional gene module signatures and scores, cell pseudotime
   trajectory, cellchat, as well as collection and analysis of bulk-seq
   data, can be found in the [74]Supplementary Methods.

Serum samples

   Sera samples were collected from 108 HCC patients and 17 healthy
   controls at the Sun Yat University Cancer Center between 2010 and 2016.
   The samples were collected before therapeutic procedures, such as
   chemotherapy and radiotherapy. This study was reviewed and approved by
   the Ethics Committees of the Sun Yat University Cancer Center.

Serological detection of SRGN by ELISA

   The SRGN concentration in the serum of HCC patients and the supernatant
   of serum-free cultured cells was measured for 48h using SRGN ELISA Kit
   (CUSABIO, CSB-EL022664HU) according to the manufacturer's instructions.

Cell lines and cell culture

   The human HCC cell lines LM3, MHCC-97H, SK-Hep-1, Hep3B, BEL7402,
   SMMC-7721, PLC/PRF/5, and LO-2 (purchased from the Shanghai Institutes
   for Biological Sciences, Chinese Academy of Sciences) were cultured in
   Dulbecco's Modified Eagle Medium (DMEM; Gibco; c11995500bt)
   supplemented with 10% fetal bovine serum (Invitrogen; 10099141) and
   penicillin (Bioss; bs-10687PA-1). Cells were maintained at 37°C in a
   humidified incubator with 5% CO₂. Exogenous recombinant SRGN protein
   were purchased from EIAab R1072h. Baicalin and daurisoline were
   purchased from TargetMol (USA; T3054, T2775).

Real-time quantitative PCR

   Total RNA is prepared from homogenized tissues or cells and converted
   to cDNA as previously described. Quantitative PCR was performed as
   described in [75]Supplementary Table 1. The threshold cycles of all
   samples were recorded.

Western blot

   RIPA buffer was used to extract protein. Samples were centrifuged to
   generate supernatants, after which loading buffer was added to
   aliquots, and samples boiled for 10 min. Proteins were electrophoresed
   for two hours and transferred to PVDF membranes. A primary antibody was
   incubated with the membrane overnight, and a secondary antibody was
   added the next day. Membranes were then photographed under a
   fluorescence microscope.

Cell transfection

   Cells transfection was performed using siRNA-SRGN and a negative
   control with Lipofectamine 3000 (Invitrogen, L3000015).
   pQCXIH-Myc-YAP-5SA was obtained from Addgene (Catalog 33093). mRNA was
   extracted 48 hours post-transfection, followed by protein extraction at
   72 hours.

Cell proliferation assay

   HCC cell proliferation was assessed via MTS assay. Log-phase cells were
   trypsinized, counted, and seeded (1,000 cells/well) in 96-well plates.
   Daily MTS incubation (3 hrs, 37°C) for five consecutive days was
   followed by absorbance measurement (490 nm). Growth curves were
   generated from OD values.

Wound healing assay

   A pipette tip created a gap simulating a wound when transfected cells
   in 6-well plates reached 85% confluence. Non-viable cells were washed
   away with phosphate-buffered saline (PBS), and the remaining cells were
   cultured in DMEM. Wound closure was monitored through microscopic
   imaging at 0, 24, and 48 hours.

Migration assay

   Transwell migration assays were performed with uncoated pore chambers.
   For migration assessment, 3×10⁴ transfected cells in 200 µL of
   serum-free DMEM were added to the upper chamber, while 800 µL of DMEM
   with 10% FBS was added to the lower chamber. After 24 hours of
   incubation, non-migrated cells on the upper membrane were removed with
   cotton swabs. Migrated cells were stained using a Diff-Quick kit
   following the manufacturer's protocol and counted in five random fields
   per chamber.

Animal experiments

   Nude mice were divided into three groups for tumor metastasis
   experiments. Tumor cells (1.5×10⁶) in 50 μL of PBS were injected via
   the tail vein. Mice were monitored twice weekly and euthanized after 8
   weeks to evaluate lung metastases. Lungs were excised,
   paraffin-embedded, and analyzed. For tumor xenograft experiments, cells
   (5×10⁶) cultured under standard conditions (7.5% CO₂, 37°C) were
   implanted subcutaneously in 0.1 mL PBS per mouse. Tumor growth was
   measured biweekly with calipers, and volume was calculated. Mice were
   randomized into vehicle or treatment groups once tumors reached 0.2—0.4
   cm^3, and pharmacodynamic studies began at 0.5—0.8 cm^3. Sorafenib (10
   mg/kg) was administered orally daily for 21 days, while verteporfin (50
   mg/kg) was given intraperitoneally every three days in a freshly
   prepared solution (5% PEG400, 5% Tween 80). Tumors were excised and
   fixed in 10% buffered formalin at specific time points. In the
   DEN/CCl4-induced HCC model, 15 SPF C57BL/6 male mice (23±2 g) received
   intraperitoneal DEN injections (25 mg/kg) for 14 days, followed by
   another round for 30 days to promote HCC development. HCC was
   established in 10 months.

Tumor sphere analysis

   Cells were seeded at a density of 1×10³ cells per well in six-well
   ultralow attachment plates (Corning) using MEM (Invitrogen)
   supplemented with N[2] medium (Invitrogen), human EGF (10 ng/mL,
   Peprotech), and human bFGF (10 ng/mL, Peprotech). After 14 days of
   culture, the total number of spheres formed was counted.

Flow cytometry

   Add Hoechst33342 to the MHCC-97H stable knock-out SRGN and MHCC-97H
   cell suspension to make the final concentration of 5μg/ml; shake the
   cells well and put them into a constant temperature water bath at 37°C
   for 90 minutes. At the same time, the cell suspension should be shaken
   several times during the water bath, and then a low-temperature
   centrifuge at 4°C 1000 rpm for 10 minutes. Collect the precipitate and
   balance the precipitate with HBSS containing 2μg PI at 4°C to obtain
   the stained cells. Flow cytometry detection, excitation light is 350
   nm. Acquisition wavelengths are 450 and 675, and the specimen is kept
   at 4°C before detection. The side population cells with weak
   luminescence are obtained.

Luciferase assay

   Upon achieving 80% to 95% confluence of the cultured cells, the growth
   medium was carefully removed, and the cells were gently washed with PBS
   to eliminate any detached cells and residual medium. Following removing
   PBS, a minimal 1× Luc-Lysis II Buffer was added to cover the cell
   monolayer. The plates were subsequently placed on an orbital shaker at
   room temperature, where they were gently rocked for 10 to 15 minutes to
   ensure uniform lysis of the cells. The final data were derived from the
   ratio of Firefly luciferase to Renilla luciferase.

ChIP assay

   Cells were cross-linked with 1% formaldehyde for 10 minutes at room
   temperature, followed by quenching with a solution of 10× glycine for 5
   minutes. A total of 4 × 10⁶ cells were then resuspended in 200 µL of
   ChIP buffer supplemented with 5 µL of micrococcal nuclease and
   incubated at 37°C for 20 minutes. The cross-linked chromatin underwent
   sonication to generate DNA fragments ranging from 150 to 900 base pairs
   in length. Immunoprecipitation was performed utilizing 1—2 µg of YAP1
   or Histone H3 (CST, D2B12) antibodies, with normal IgG as an
   appropriate control. ChIP-qPCR subsequently analyzed the resulting
   purified DNA fragments to identify sequences associated with CRISPLD2.

Statistical analysis

   Statistical analyses for this study were conducted utilizing GraphPad
   Prism version 8.02 and R version 4.3 software. The Wilcoxon rank-sum
   test was employed to compare continuous variables between two groups.
   The Spearman correlation method was applied to assess the correlation
   between gene expression. The log-rank test was used to evaluate
   differences in cumulative survival times. Statistical significance
   symbols were displayed as follows: non-significant (ns), p < 0.05 (*),
   p < 0.01 (**), and p < 0.001 (***).

Results

Single-cell RNA-seq reveals SRGN expression in HCC progression

   To understand the cellular composition in HCC tissues, we conducted
   scRNA-seq analysis on thirteen HCC patients' samples, which comprised
   solitary tumor conditions, tumor samples with multiple hepatoma
   conditions, or tumor vascular invasion conditions (Figure [76]1A,
   [77]Supplementary Figure 1A). Following data quality control and
   filtering ([78]Supplementary Figure 1B, C), we obtained single-cell
   transcriptomes for 68,604 cells, including 24,629 cells from solitary
   tumor status specimens (ST group), 43,975 cells from specimens with
   multiple hepatoma or tumor vascular invasion status (MTVI group).

Figure 1.

   [79]Figure 1
   [80]Open in a new tab

   Cell landscapes of scRNA-seq analysis. (A) Samples for scRNA-seq
   analysis were taken from patients with solitary HCC status and patients
   with multiple hepatoma or HCC vascular invasion status. (B) UMAP is
   colored by major cell types. (C) UMAP is colored by cells from the ST
   or MTVI group. (D) UMAP and (E) stacked plots show the cell composition
   of the two groups. (F) Heatmap visualizes the marker genes for
   different cell types. (G) UMAP shows SRGN expression in malignant cells
   from the MTVI group. (H) UMAP shows SRGN expression in LVECt from the
   MTVI group. (I) UMAP shows SRGN expression in HSCs from the MTVI group.
   (J) UMAP shows SRGN expression in malignant cells from the ST group.
   (K) UMAP shows SRGN expression in LVECt from the ST group. (L) UMAP
   shows SRGN expression in HSCs from the ST group. (M) The volcano plot
   visualizes SRGN expression differences between MTVI and ST groups in
   malignant cells. (N) The volcano plot visualizes SRGN expression
   differences in LVECt between MTVI and ST groups. (O) Volcano plot
   visualizes SRGN expression differences in HSCs between MTVI and ST
   groups.

   Seven major cell types were identified according to canonical marker
   genes, including 9,605 T/NK cells (CD3E, CD3D, CD2, and IL7R), 9,320
   Kupffer cells (CD163, CD68, C1QB, and AIF1), 2,457 B cells (CD79A and
   IGHG1), 2,321 hepatic stellate cells (HSCs)(MGP, MYL9, IGFBP7, ACTA2,
   and COL1A1), 2,888 tumor liver vascular endothelial cells (LVECt)
   (PODXL, VWA1, PLVAP, and CD34), 17,307 malignant cells (TTR, TF, KRT18,
   KRT8, and EFNA1)[81]^32^, [82]^33 and 77 mast cells (CPA3, TPSAB1, and
   TPSB2) (Figure [83]1B, F). The group-specific information of each cell
   population is illustrated in Figure [84]1C-E.

   Mounting clinical evidence implicated SRGN as a critical mediator in
   tumor progression[85]^23^, [86]^34^-[87]^37. Clinical data demonstrate
   that elevated serum SRGN levels independently correlate with diminished
   overall survival and accelerated HCC recurrence, and corresponding
   immunohistochemical analysis revealed SRGN upregulation in 56.7% of HCC
   patients' specimens, contrasting sharply with minimal expression
   observed in only 3.1% of paired non-cancerous tissues[88]^31. To
   dissect SRGN's tumor microenvironment regulatory mechanisms, we
   conducted a cell-type-resolved comparative analysis between localized
   (ST group) and advanced HCC (MTVI group). Comparative feature plots and
   volcano plots showed SRGN was upregulated in carcinoma cells, LVECt,
   and HSCs when the tumor proliferated or metastasized (Figure [89]1G-O).

Dissecting the potential role of SRGN in malignant cells

   Given the heterogeneity in malignant cell composition, we extracted
   malignant cells from the ST and MTVI groups and then re-clustered
   malignant cells into seventeen subclusters. To explore the SRGN
   variation in the identified tumor cell subclusters, we computed the
   relative SRGN expression levels of these tumor cell subclusters.
   Heatmap for visualizing, the differential expression analysis revealed
   that five malignant cell populations exhibited high upregulated SRGN
   expression, including subcluster 2 (Log[2]FC=0.77,
   adjusted-p=2.14E-53), subcluster 6 (Log[2]FC=0.72,
   adjusted-p=2.12E-96), subcluster 12 (Log[2]FC=2.06,
   adjusted-p=2.33E-70), subcluster 14 (Log[2]FC=0.76,
   adjusted-p=1.89E-11), and subcluster 16 (Log[2]FC=1.63, adjusted-p=
   5.28E-10) (Figure [90]2A, B). All p-values were adjusted using the
   Bonferroni method.

Figure 2.

   [91]Figure 2
   [92]Open in a new tab

   Characterization of heterogeneity between SRGN-high malignant cells and
   SRGN-low malignant cells. (A) Subcluster of malignant cells with SRGN
   expression. (B) UMAP is colored by malignant cells from the SRGN-high
   subcluster or SRGN-low subcluster. (C) Heatmap shows the metabolic flux
   in the SRGN-high subcluster and SRGN-low subcluster. Volcano plots
   display the metabolism difference of glycolysis and TCA cycle(D),
   BCAA(E), O-linked glycan synthesis(F), and pyrimidine synthesis(G)
   between the two subclusters. (H) Dot plot shows the hallmark pathway
   condition in the two subclusters. Violin plots compare the metastasis
   score(I), proliferation score(J), ECM modeling score(K), and collagen
   formation score(L) for two subclusters. (M) SRGN-high vs. SRGN-low
   tumor size. (N) SRGN sera level comparison among different HCC
   statuses. (O) Clinical stages between SRGN-high and SRGN-low patients.
   (P) SRGN expression-AFP correlation.

   Metabolic diversity is known to be a significant contributor to tumor
   progression. We used the scFEA method to estimate malignant cell
   subclusters' metabolic flux to identify metabolic characteristics of
   the SRGN-high tumor cell population. The heatmap revealed malignant
   cell subclusters' average metabolic flux levels in different
   metabolites (Figure [93]2C). Further metabolic perturbation analysis
   indicated that the metabolic state of glycolysis, tricarboxylic acid
   cycle (TCA) cycle, branched-chain amino acids (BCAA) metabolism, O
   linked glycan synthesis, and pyrimidine synthesis were notable
   different between SRGN-high tumor cell population and SRGN-low cell
   populations (Figure [94]2D-G). In malignant cells characterized by high
   SRGN expression, within TCA cycle framework, 3-phosphoglyceric acid
   (3PD) was effectively converted into pyruvate, providing an essential
   precursor for the biosynthesis process. Succinyl-COA was more active in
   converting to succinate acid to combat oxidative stress, increasing the
   tolerance of SRGN-high tumor cells to hypoxia. Glucose was converted to
   glucose-6-phosphate (G6P) to enhance glycolysis for rapid energy
   production, fueling their rapid proliferation needs. The BCAA metabolic
   pathway, notably the conversion of phenylalanine to tyrosine, exhibits
   heightened activity in malignant cells with high SRGN expression,
   indicating a strategic metabolic adaptation that the cells employed to
   fulfill their biosynthetic demands, modulate the expression and
   activity of metabolic enzymes, regulate epigenetic modifications, and
   acclimate to the tumor microenvironment. Such metabolic fine-tuning
   confers a competitive advantage for SRGN-high tumor cell growth and
   survival. In stark contrast to malignant cells with low SRGN
   expression, (Gal)1 (GlcNAc)1 (Man)1 (Ser/Thr)1 to (Gal)1 (GlcA)1
   (GlcNAc)1 (Man)1 (S)1 (Ser/Thr)1 in O-linked glycan synthesis was
   markedly enhanced in malignant cells exhibiting high levels of SRGN.
   The pyrimidine synthesis pathway of SRGN-high malignant cells is more
   active in converting uracil to β-alanine and deoxythymine nucleotide
   (dTMP) to succinyl-COA, which supports rapid DNA replication and cell
   division. Thus, in liver tumor cells, high SRGN expression might
   closely relate to the enhancement of metabolic reprogramming.

   Results from irGSEA showed that tumor cells with elevated SRGN levels
   exhibit upregulation of multiple hallmark pathways, encompassing the
   Hedgehog signaling, unfolded protein response, early estrogen response,
   apoptosis, EMT, the p53 pathway, TGFβ signaling, increased ultraviolet
   response, hypoxia, cholesterol homeostasis, as well as the TNFA
   signaling pathway NFKB and myogenesis (Figure [95]2H). Furthermore,
   tumor cells presented SRGN-overexpressing displayed a higher
   proliferation and metastasis score, coupled with enhanced collagen
   formation and ECM modeling (Figure [96]2I-L).

   Thus, we proceeded to assess the impact of SRGN levels on the
   invasiveness of HCC by detecting clinical serum samples. Compared to
   HCC patients with low SRGN levels, HCC patients with high SRGN
   expression had larger neoplasm diameters, accompanied by significant
   tumor metastasis, and were predominantly in poorer Ⅲ-Ⅳ clinical stages
   (Figure [97]2M-O). Simultaneously, we observed positive association
   between the serum concentrations of SRGN and Alpha-Fetoprotein (AFP)
   (Figure [98]2P). In survival analyses, HCC patients with high SRGN
   serum levels had lower overall survival rates than those with lower
   ([99]Supplementary Figure 2A). Multivariate analysis confirmed that
   SRGN is an independent influencing factor correlated with poor outcomes
   in HCC patients ([100]Supplementary Figure 2B). These observations
   align with the analysis conducted at the single-cell level, providing a
   coherent scenic of SRGN's role in facilitating HCC progression.

Cellular communication of SRGN-high HCC cells with neighboring cells

   To delineate the functional influence on neighboring cells of SRGN
   secreted by SRGN-high HCC cell subpopulation within the tumor
   microenvironment, we performed CellChat analysis focusing on
   intercellular communication. Through genetic markers, NK/T cell subsets
   were classed into NK (GZMA, CCL4, CCL5, and NKG7) and T (TIGIT, CD3E,
   and CD3D) cell subsets (Figure [101]3A, B). Intercellular interactions
   analysis showed that HCC cells with high expression of SRGN acted on
   neighboring cells through the paracrine effect of SRGN (Figure
   [102]3C). Then, we analyzed the receptors of neighboring cells affected
   by secreted SRGN from SRGN-high HCC cells in the cell-cell
   communication network (Figure [103]3D). At the same time, we conducted
   a differential analysis of the expression of receptors in the cell
   communication network and screened out the receptors exhibiting
   significant responsiveness (FDR<0.05, |log2FC|>1) in recipient cells
   (Figure [104]3E). Integrated results demonstrated that SRGN secreted by
   SRGN-high HCC subpopulation exerted compartment-specific regulatory
   effects within the tumor microenvironment. Notably, NK cells exhibited
   enhanced expression of chemotaxis-associated receptors (CXCR4, ALOX5AP)
   and the adhesion mediator CD44, suggesting amplified migratory
   capacity. In T lymphocytes, concurrent upregulation of the signaling
   attenuator RGS1 and multifunctional CD44 implied dual modulation of
   activation thresholds and extracellular matrix interactions. Kupffer
   cells displayed a polarized response characterized by coordinated
   induction of myeloid activation markers (MNDA, CCL3, CTSS, NCF2,
   FCER1G, LAPTM5) alongside CXCR4 downregulation, potentially reflecting
   differentiation state shifts. B cell-specific RGS1 elevation pointed to
   altered chemokine sensing, while LVECt upregulation of ITGA5 indicated
   extracellular matrix adhesion remodeling. Mast cells manifested
   multifaceted activation through concurrent increases in inflammatory
   mediators (ALOX5AP, FCER1G), signaling regulators (RGS1), and
   migration-associated CD44, collectively suggesting SRGN-driven
   pro-inflammatory priming.

Figure 3.

   [105]Figure 3
   [106]Open in a new tab

   Cellular communication of SRGN-high HCC cells with neighboring cells.
   (A) Landscape of cell subpopulations in UMAP. (B) Genetic marker
   distinguishing NK and T cell populations. (C) SRGN-high HCC cells
   influence neighbor cells via the paracrine effect in the cell-cell chat
   network. (D) Analyzing receptors of neighboring cells impacted by
   secreted SRGN from SRGN-high HCC cells within the cell-cell
   communication network. (E) Identification of significantly differential
   expressed receptors in a cell communication network. (F) GSEA pathway
   analysis of SRGN-driven paracrine signaling effects in neighboring
   cells.

   Leveraging differential receptor expression profiles across neighboring
   cell populations, we performed systematic pathway interrogation using
   the GSEA method to explore the functional consequences of SRGN-mediated
   paracrine signaling (Figure [107]3F). NK cells exhibited dual cell
   cycle restriction through downregulating E2F targets and the G2/M
   checkpoint, paradoxically coexisting with inflammatory activation via
   upregulated TNF-α/NF-κB signaling and inflammatory responses. T
   lymphocytes underwent metabolic-stemness reprogramming marked by
   WNT/β-catenin activation alongside IL2-STAT5 signaling elevation.
   Concurrent adipogenesis pathway upregulation and KRAS hyperactivation
   implied lipid metabolic rewiring might drive T cell exhaustion. Kupffer
   cells displayed metabolic polarization featuring enhanced oxidative
   phosphorylation alongside compromised xenobiotic metabolism. The
   coordinated interferon-α response activation and estrogen signaling
   inhibition suggested SRGN-driven functional redirection toward
   pro-inflammatory phenotypes over detoxification roles. B cell profiling
   revealed metabolic adaptation for antibody production, with elevated
   glycolytic flux and cholesterol homeostasis supporting plasma cell
   differentiation. This metabolic shift occurred alongside TGF-β
   signaling suppression, potentially unleashing inflammatory responses
   through IL-6 overproduction. LVECt remodeling involved endothelial
   plasticity programs where WNT/β-catenin activation synergized with EMT
   to enhance lymphatic invasive capacity. Intriguingly, KRAS signaling
   inhibition coexisted with TNF-α/NF-κB pathway activation, implying
   compensatory inflammatory mechanisms sustaining lymphangiogenesis. Mast
   cells demonstrated proliferative dysregulation via G2/M checkpoint
   upregulation paired with immune anergy, evidenced by complement system
   suppression and impaired interferon-γ responses. These pleiotropic
   effects underscore SRGN as a regulator of immune-metabolic crosstalk in
   HCC progression.

SRGN promotes HCC cell migration and metastasis via autocrine

   SRGN emerged as a non-redundant regulator of HCC progression, grounded
   in a hierarchical evidence chain through a systematic integration of
   single-cell multi-omics, functional pathway interrogation, and clinical
   validation. scRNA-seq identified SRGN as a pan-cellular marker of
   aggressiveness, with tumor cell subclusters, LVECt, and HSCs showing
   stage-dependent upregulation during metastasis, directly aligning with
   advanced clinical phenotypes. Beyond mere expression patterns, SRGN's
   functional centrality was unmasked by its orchestration of
   metabolically fueled plasticity—scFEA-based flux analysis revealed its
   role in rewiring glycolysis, TCA cycle dynamics, and pyrimidine
   synthesis to sustain proliferation under oxidative stress, while BCAA
   metabolic shifts (phenylalanine to tyrosine) indicated adaptation to
   nutrient-scarce microenvironments. Crucially, SRGN transcended
   cell-autonomous effects to dominate immune-stromal crosstalk: paracrine
   SRGN signaling suppressed NK cell cycling (E2F/G2-M checkpoint
   downregulation), polarized Kupffer cells toward pro-tumorigenic states
   (MNDA/CCL3↑, CXCR4↓), and primed LVECt for invasive remodeling (ITGA5↑,
   WNT/β-catenin activation), collectively fostering an immuno-suppressive
   niche permissive for metastasis. The clinical translation of these
   findings—serum SRGN's independent prognostic value and correlation with
   AFP—cemented its biological and translational relevance. By converging
   cell-intrinsic metabolic adaptability, microenvironmental
   reprogramming, and dismal patient outcomes, SRGN exemplifies a
   therapeutically tractable hub where molecular mechanism meets clinical
   urgency, rationalizing its prioritization for targeted intervention in
   HCC.

   To confirm the role of SRGN in promoting tumor cell migration and
   metastasis, we quantified its mRNA and protein levels in liver cancer
   cell lines using qPCR and western blot. SRGN mRNA and protein
   expression varied across different HCC cell lines (Figure [108]4A).
   SK-Hep-1 cells were transfected with either SRGN-specific shRNA or a
   negative control. Additionally, the Hep 3B cell that stably
   overexpressed SRGN was generated. The protein expression levels of SRGN
   in these modified cell lines were subsequently confirmed by q-PCR and
   immunoblotting assays (Figure [109]4B-D). We observed that SRGN
   knockdown reduced HCC proliferation ability, and SRGN overexpression
   promoted HCC proliferation ([110]Supplementary Figure 3). Using
   conditioned medium form SK-Hep-1 cells KD2# group with different
   exogenous recombinant SRGN protein concentration via wound healing
   assays and Transwell test. A dose-dependent in exogenous recombinant
   SRGN protein concentrations was positively correlated with enhanced
   migratory and metastatic capacities of HCC cells, directly
   demonstrating SRGN's autocrine-driven role in potentiating tumor cell
   invasiveness and metastatic progression ([111]Supplementary Figure 4).
   In wound healing assays, SK-Hep-1 cells without knockdown of SRGN
   showed significantly higher wound closure width than cells treated with
   shRNA SRGN at the same time after 24 and 48 hours, indicating that
   knockdown of SRGN significantly inhibited the migration ability of
   cells (Figure [112]4E, F). Correspondingly, hep 3B cells overexpressing
   SRHN showed accelerated wound healing, and the wound closure width
   reached 63% after 70 hours (Figure [113]4E, G). In the Transwell
   experiment, SK-Hep-1 cells without SRGN-knockdown showed strong
   migration ability within 24 hours. In contrast, SK-Hep-1 cells with
   SRGN knockdown significantly reduced the number of migrations,
   indicating that the expression of SRGN was necessary for the migration
   ability of cells (Figure [114]4H, I). Similarly, Hep 3B cells
   overexpressing SRGN showed enhanced migration ability (Figure [115]4H,
   J). In vitro analysis confirmed the expression of crucial EMT markers
   in HCC cell lines, including E-cadherin, N-cadherin, vimentin, and MMP2
   (Figure [116]4K). These results indicate that SRGN promotes malignant
   cell migration by triggering EMT in HCC cell lines.

Figure 4.

   [117]Figure 4
   [118]Open in a new tab

   SRGN promotes HCC cell migration and metastasis. (A) SRGN expression
   profiling (qPCR/WB) in HCC cells. (B) SRGN knockdown efficiency
   validated by quantitative PCR. (C) SRGN overexpression confirmed by
   quantitative PCR. (D) SRGN knockdown and overexpression confirmation.
   (E) Wound-healing assay shows SRGN affects cell migration. (F)
   Quantification of wound closure rates showing significant inhibition in
   SRGN knockdown groups compared to controls. (G) Accelerated wound
   healing observed in SRGN-overexpressing cells relative to vector
   counterparts. (H) Transwell assay demonstrated SRGN's impact on HCC
   cell migration. (I) Statistical analysis of migrated cells showing
   reduction in SRGN knockdown conditions. (J) Quantification of Transwell
   assay results revealed increase in cell invasion upon SRGN
   overexpression. (K) Immunoblotting of E-cadherin, N-cadherin, Vimentin,
   and MMP2 at 72 hours post-transfection with SRGN suppression or
   overexpression. (L) SRGN inhibition reduced SK-Hep-1 cell metastasis in
   vivo. (M) Improved physiological status in SRGN knockdown groups versus
   controls. (N) Histopathological evaluation of lung tissues revealed
   decreased metastatic burden in SRGN-inhibited cohorts compared to
   controls.

   For in vivo analysis, cells with SRGN knockdown, along with their
   respective negative control cells, were administered via injection into
   the tail veins of nude mice. Following 8 weeks, the SRGN-knockdown
   group demonstrated significantly greater body weights than the control
   group (Figure [119]4L, [120]Supplementary Figure 5). Subsequently, the
   mice were humanely euthanized, and their lungs were excised for
   comprehensive examination. Strikingly, the lung tissues from mice
   injected with SRGN-knockdown cells displayed a marked decrease in the
   size and quantity of metastatic nodules compared to those from the
   negative control (Figure [121]4M, N). In aggregate, these results
   provide evidence that SRGN significantly enhanced the metastatic
   capacity of malignant cells in HCC within living organisms.

SRGN drives malignant cell stemness in HCC

   We estimated the proportion of malignant cells in different cell cycle
   phases. As the tumor progressed, the proportion of malignant cells with
   high expression of SRGN gradually increased in the G1 and G2M phases
   (Figure [122]5A), validated in an external cohort ([123]Supplementary
   Figure 6), which might be related to SRGN's involvement in regulating
   the energy and substances of these two stages of the EMT process to
   support DNA replication and cell division.

Figure 5.

   [124]Figure 5
   [125]Open in a new tab

   SRGN is significantly associated with stemness-like properties in HCC
   cells. (A) UMAP and stacked plot display the proportion of malignant
   cells in the G1, S, and G2M phases. (B) The heat map shows large-scale
   CNVs for malignant cells, with immune cells as a reference. Red
   represents gains, and blue represents losses. (C) The violin plot shows
   different CNV levels among SRGN-high malignant cells, SRGN-low
   malignant cells, and immune cells. (D) Pseudo-time trajectory analysis
   of malignant cells. Circled number one represents the trajectory root
   (2_SRGN-high cluster). (E) Dynamic variation of stemness score during
   the pseudo-time trajectory. (F) The expression of cancer stem cells
   related marker gene in subclusters. (G) Potency score comparison
   between SRGN-high and SRGN-low malignant cells.

   CNV between malignant cells was compared, using the immune cells as a
   reference point. Copy number variation inference of malignant cells
   expressing opposite levels of SRGN revealed intratumor genomic
   instability (Figure [126]5B). Tumor cells with high SRGN expression
   exhibited a higher rate of CNV changes, indicating significant genomic
   plasticity and potential activation of stemness in their cell
   populations (Figure [127]5C). Utilizing Monocle3, we delved into the
   intricate dynamics of cellular transitions within tumor states. Our
   pseudotime trajectory analysis and stemness score[128]^38 uncovered a
   lineage where SRGN-low malignant cells appeared to originate from
   SRGN-high malignant cells. Further, SRGN-high malignant cells
   maintained a comparatively higher stemness score throughout the
   differentiation trajectory (Figure [129]5D, E). Cancer stem
   cell-related markers, CD44, ABCG2, BMI1, and KRT19, expression in
   subclusters from trajectory, indicated enhanced self-renewal capacity
   in SRGN-high malignant cells (Figure [130]5F). A significant
   statistical variance in potency scores from Cytotrace analysis was
   observed between groups, with tumor cells expressing SRGN showing the
   highest developmental potential (Figure [131]5G), supporting their
   proliferation and differentiation capabilities, aligning with the
   patterns observed in pseudotime analyses.

   To investigate the role of SRGN in maintaining stemness
   characteristics, we knocked down SRGN in MHCC-97H cells (Figure
   [132]6A, B). SRGN knockdown significantly reduced the number of spheres
   formed (Figure [133]6C, D), highlighting its critical role in the
   self-renewal ability of HCC cells. Side population cells, identified by
   their ability to efflux the fluorescent dye Hoechst 33342, exhibit
   specific cancer stem cell characteristics[134]^39 and serve as cancer
   stem cell markers in HCC. Notably, SRGN knockdown in MHCC-97H cells
   significantly reduced the side population fraction (Figure [135]6E, F).
   Furthermore, the expression of cancer stem cell-related markers—ABCG2,
   Bmi-1, and Nanog—was validated, confirming that SRGN enhances the
   self-renewal and stemness of malignant HCC cells (Figure [136]6G,
   [137]Supplementary Figure 7). Side population sorting assays have
   demonstrated that SRGN was up-regulated in cells of the side
   population, revealing its association with enhanced properties
   characteristic of cancer stem cells in HCC (Figure [138]6H).
   Collectively, these findings demonstrate that SRGN plays a pivotal role
   in regulating cancer stem cell-like characteristics in HCC cells.

Figure 6.

   [139]Figure 6
   [140]Open in a new tab

   Validation of SRGN expression is related to the HCC cell's stemness
   characteristic. (A, B) Knockdown of SRGN in MHCC-97H cells verified by
   qPCR and immunoblotting. (C) The number of suspended spheres was
   reduced after SRGN knockdown in MHCC-97H cells. (D) Quantification of
   tumor sphere size and number. (E, F) Side population cell assay and
   abundance comparison. (G) Immunoblotting of cancer stem cell-related
   markers following SRGN knockdown in MHCC-97H cells (72 h
   post-transfection). (H) Relative SRGN mRNA levels in the main and side
   populations of MHCC-97H cells were determined by qPCR (normalized to
   GAPDH).

SRGN modulates effector YAP nuclear localization

   Metascape enrichment analysis demonstrated significant enrichment of
   Hippo pathways and kinase activity in malignant cells with high SRGN
   expression, implying SRGN might affect tumor cell proliferation or
   apoptosis by regulating the Hippo signaling pathway (Figure [141]7A).
   Our prior research[142]^40 demonstrated that SRGN promotes the
   expression of CD44, leading to an enhanced capacity for self-renewal
   through the activation of the MAPK pathway. CD44 is a receptor for the
   extracellular matrix ligand SRGN. In this study, we further observed a
   positive correlation between SRGN and CD44 expression in malignant
   cells at both single-cell and bulk transcriptome levels
   ([143]Supplementary Figure 8A, B) and in HCC cell lines (Figure
   [144]7B). It was presumed that the ability of SRGN to promote malignant
   cell invasion might be related to the YAP, a downstream signaling
   factor activated by CD44. Further experimental observations revealed a
   positive correlation between SRGN and YAP expression in HCC cell lines,
   supporting this hypothesis (Figure [145]7C). This correlation motivated
   us to investigate the role of SRGN in the YAP pathway further.
   Immunoblotting analysis revealed that SRGN overexpression reduced YAP
   phosphorylation (Figure [146]7D), suggesting that SRGN participated in
   the YAP pathway by inhibiting YAP phosphorylation (Figure [147]7E). A
   confocal immunofluorescence assay examining nuclear and
   cytosolic/membrane fractions showed reduced nuclear translocation of
   YAP in SK-Hep-1 cells with SRGN knockdown (Figure [148]7F). Conversely,
   SRGN overexpression in Hep 3B cells significantly increased YAP
   translocation to the nucleus (Figure [149]7G), highlighting SRGN's
   substantial impact on nuclear YAP expression. To further investigate
   the relationship between SRGN and YAP in liver cancer development, the
   HCC mice induced by DEN/CCl4 were used. Confocal immunofluorescence
   analysis demonstrated a synchronous increase in SRGN and YAP expression
   during HCC progression, with YAP predominantly localized in the nucleus
   (Figure [150]7H). Transwell migration assays were performed to evaluate
   the functional role of SRGN and its potential interaction with YAP
   signaling ([151]Supplementary Figure 9A). Quantitative results revealed
   significant differences among experimental groups ([152]Supplementary
   Figure 9B). Compared to the NC (non-knockdown control) group
   (653.7±63.9 cells), SRGN-depleted cells exhibited marked reductions in
   migration. KD#1 (336.7±48.1 cells, p=0.003 vs. NC) and KD#2 (291.0±32.8
   cells, p<0.001 vs. NC) showed decreases of 48.5% and 55.5%,
   respectively. The more potent suppression in KD#2 versus KD#1 (p=0.041)
   correlated with differential SRGN knockdown efficiency. YAP5SA
   overexpression restored migration capacity to 517.3±5.5 cells (KD#1R,
   56.9% recovery relative to KD#1; p=0.003) and 506.3±15.7 cells (KD#2R,
   58.9% recovery relative to KD#2; p<0.001). Despite this rescue, both
   KD#1R (p=0.021 vs. NC) and KD#2R (p=0.012 vs. NC) failed to reach
   normal migration levels, indicating incomplete pathway restoration.
   KD#2R achieved greater absolute rescue than KD#1R despite more
   substantial initial suppression, suggesting YAP activation
   preferentially mitigates severe migration defects. Western blot also
   showed that SRGN mediated the restoration of migration ability by
   regulating the YAP pathway ([153]Supplementary Figure 9C). Based on the
   comprehensive analysis and experimental evidence, we concluded that
   SRGN regulated YAP expression and nuclear localization, impacting tumor
   cell growth in HCC.

Figure 7.

   [154]Figure 7
   [155]Open in a new tab

   SRGN acts as a CD44 ligand to modulate YAP nuclear translocation via
   Hippo pathway deregulation. (A) Pathways enrichment analysis of
   SRGN-high malignant cells. (B) Correlation between SRGN and CD44
   expression in HCC cell lines. (C) SRGN expression induced YAP
   accumulation in HCC cells as quantified by immunoblot. (D)
   Immunoblotting of MST1, P-LATS1, LATS1, P-YAP, and YAP in SRGN
   knockdown or overexpression conditions (72 h post-transfection). (E)
   Immunoblotting of SRGN knockdown or overexpression and YAP localization
   in SK-Hep-1 and Hep 3B cells. (F, G) Confocal immunofluorescence
   analysis of YAP localization in SRGN knockdown and overexpression
   conditions. (H) Confocal immunofluorescence showing SRGN and YAP
   localization in the DEN/CCl4-induced mouse HCC model.

CRISPLD2 serve as a downstream effector gene of the SRGN/YAP axis

   In order to investigate the downstream target of SRGN that facilitates
   the metastasis of HCC, RNA sequencing was conducted on the SK-Hep-1 and
   Hep 3B cell lines to obtain expression profiling data. Recognizing
   SRGN's significant role in promoting metastasis, we used microarray
   analysis to identify SRGN-regulated genes. From the Venn plot, CRISPLD2
   emerged as a common factor in SRGN knockdown and overexpression groups
   (Figure [156]8A). At both bulk and single-cell transcriptome levels,
   SRGN and CRISPLD2 expression were synchronized (Figure [157]8B,
   [158]Supplementary Figure 10A). qPCR and immunoblotting further
   validated the upregulation of CRISPLD2 by SRGN overexpression (Figure
   [159]8C, D; [160]Supplementary Figure 10B, C). Utilizing small
   interfering RNAs to inhibit CRISPLD2 expression, we observed a
   significant decrease in vimentin levels, which parallels the findings
   obtained from SRGN knockdown experiments (Figure [161]8E). Cell growth
   assays revealed that CRISPLD2 knockdown significantly inhibited
   proliferation ([162]Supplementary Figure 10D). Similarly, the Transwell
   assay demonstrated reduced migration capacity in cells with CRISPLD2
   knockdown (Figure [163]8F, G), supporting the role of SRGN in promoting
   metastasis.

Figure 8.

   [164]Figure 8
   [165]Open in a new tab

   CRISPLD2 functions as a downstream effector of the SRGN/YAP axis. (A)
   SRGN and CRISPLD2 expression were overlapped among three transcriptomes
   groups. (B) Positive correlation between SRGN and CRISPLD2 expression
   in HCC cohorts at the bulk-seq level. (C) SRGN dose-dependently
   upregulated CRISPLD2 mRNA levels. (D) Immunoblot quantification
   confirms SRGN-induced CRISPLD2 protein elevation. (E) Immunoblotting
   shows CRISPLD2 expression in HCC cells. (F) CRISPLD2 knockdown reduced
   Transwell migration in SRGN-activated cells. (G) Migratory cell
   quantification validated CRISPLD2-dependent motility. (H) Peptide 17
   suppressed CRISPLD2 expression in SRGN-overexpressing cells. (I)
   Immunoblot validation of CRISPLD2 protein downregulation by peptide 17
   in SRGN-overexpressing cells. (J) Mechanistic schema of peptide 17
   targeting SRGN/YAP/CRISPLD2 signaling node. (K) Luciferase reporter
   assay demonstrated CRISPLD2 promoter activity. (L) CRISPLD2
   overexpression reversed SRGN knockdown-induced migration defect. (M)
   Quantitative migration analysis of SRGN-knockdown and
   CRISPLD2-overexpression. (N) SRGN silencing reduced CRISPLD2 transcript
   levels. (O) VP attenuated SRGN-mediated migration in dose-dependent
   manner. (P) Suppression of CRISPLD2 reduced SRGN-driven migration.

   The protein interaction between the YAP and TEAD1 is essential for
   facilitating the oncogenic activity of YAP. Peptide 17, an inhibitor of
   YAP-TEAD1 interaction, suppressed CRISPLD2 expression in the
   SRGN-overexpressing group but had no significant effect in the vector
   control group (Figure [166]8H). Immunoblotting confirmed a significant
   reduction in CRISPLD2 protein levels following the administration of
   peptide 17 in the group with SRGN overexpression (Figure [167]8I). By
   comparing Figure [168]8D and Figure [169]8I in Hep3B cells, we observed
   that CRISPLD2 in SRGN-overexpressed Hep 3B cells was significantly
   reduced following peptide 17 intervention. To ascertain whether SRGN
   stimulated the expression of CRISPLD2, a luciferase reporter assay
   utilizing the CRISPLD2 promoter was conducted. The results indicated a
   significant increase in activity in response to the overexpression of
   SRGN. In contrast, peptide 17 intervention suppressed SRGN's regulation
   of CRISPLD2 activity (Figure [170]8J, K). Transwell assay revealed that
   reduced SRGN levels decreased HCC cell migration, while CRISPLD2
   overexpression partially restored migration in SRGN-suppressed cells
   (Figure [171]8L, M, N; [172]Supplementary Figure 10E).

   To explore whether SRGN enhanced migration through the YAP pathway,
   cells were treated with verteporfin (VP) to disrupt YAP-TEAD1 complex
   formation. VP treatment significantly reduced preneoplastic foci and
   oval cell proliferation. SRGN-overexpressing cells treated with varying
   doses of VP exhibited reduced migration, and CRISPLD2 knockdown further
   inhibited cell motility (Figure [173]8O, P). These results uncovered
   that SRGN promotes HCC metastasis by upregulating CRISPLD2, a
   downstream effector of the SRGN/YAP pathway, which enhances HCC cell
   aggressiveness.

CRISPLD2 is a novel YAP-TEAD1 target gene regulated by SRGN

   To confirm that SRGN modulated the downstream target CRISPLD2 within
   the YAP signaling pathway, we conducted transient transfection of
   MHCC-97H cells with the YAP. Consistent with our expectations, the
   luciferase reporter assay revealed a significant increase in the
   activity of the CRISPLD2 promoter following YAP overexpression (Figure
   [174]9A). Conversely, VP treatment suppressed CRISPLD2 promoter
   activity (Figure [175]9B), suggesting that the YAP-activated signaling
   pathway directly regulated the CRISPLD2 promoter.

Figure 9.

   [176]Figure 9
   [177]Open in a new tab

   CRISPLD2 is a novel YAP-TEAD1 target gene regulated by SRGN. (A)
   Luciferase activity of the CRISPLD2 promoter in YAP-overexpressing
   cells. (B) Luciferase activity of the CRISPLD2 promoter with VP
   intervention in YAP-overexpressing cells. (C) YAP-TEAD1 binding site 1
   motif in the CRISPLD2 promoter. (D) Schematic of CRISPLD2 promoter
   binding sites within the TEAD1 motif. (E) ChIP analysis of YAP DNA in
   cells stably transfected with vector or SRGN, quantified by agarose gel
   electrophoresis and qPCR. (F) Five primer pairs were used to evaluate
   decreased YAP DNA levels after SRGN suppression. (G) Fold changes in
   promoter DNA copy numbers. (H) Conserved CRISPLD2 motif sequences
   across species.

   Via the JASPAR database[178]^41, we identified two potential YAP-TEAD1
   binding sites in the CRISPLD2 promoter region. Luciferase assays
   demonstrated that the elimination of binding site one resulted in a
   significant reduction in luciferase activity, while the deletion of
   binding site two did not produce any notable effect (Figure [179]9C).
   This study proved that the YAP-TEAD1 co-transcription factor
   facilitated the transcription of CRISPLD2 by binding to a specific
   sequence, referred to as site one (TGGATTCCTGGG). Chromatin
   immunoprecipitation analysis was performed for the subsequent
   validation. We designed five pairs of PCR primers targeting the
   CRISPLD2 promoter region (Figure [180]9D). The qPCR results, normalized
   to input levels, substantiated that YAP-TEAD1 directly bonded with the
   CRISPLD2 promoter, particularly at site 1, thereby regulating CRISPLD2
   transcription (Figure [181]9E-G). Significantly, various known
   YAP/TEAD1 targeted genes, such as BIRC5, AREG, CCND1, CTGF, and MYC,
   were not activated through SRGN-mediated YAP signaling in liver cancer
   cell lines ([182]Supplementary Figure 11). This observation indicated
   that CRISPLD2 is a novel and selective target gene of the YAP-TEAD1
   complex within the SRGN-mediated HIPPO/YAP pathway. Additionally, the
   CRISPLD2 motif exhibited a conserved sequence across multiple species
   (Figure [183]9H).

Combinatorial therapy targeting SRGN/YAP signaling

   We explored whether SRGN was directly affected by sorafenib using flow
   cytometry. After two days of sorafenib intervention, the percentage of
   apoptosis cells was measured. Comparing the apoptosis rate in the drug
   step-up concentration among KD#1, KD#2, and the control group, the
   apoptosis rate of malignant cells from the knockdown SRGN groups was
   much lower than that of the control group (Figure [184]10A, B). In the
   meantime, demonstrates a dose-dependent capacity to induce apoptosis in
   HCC cells that overexpress SRGN, compared to HCC cells with vector
   control (Figure [185]10C, D).

Figure 10.

   [186]Figure 10
   [187]Open in a new tab

   Combinatorial targeting of SRGN/YAP signaling synergizes with sorafenib
   to suppress HCC aggressiveness. (A) Flow cytometry in sorafenib-treated
   SK-Hep-1 cells. (B) Protein-level analysis of sorafenib treatment in
   SRGN-knockdown groups. (C) Flow cytometry in sorafenib-treated Hep 3B
   cells. (D) Hep 3B and Hep 3B SRGN-overexpression cells were treated
   with sorafenib, and analyzed by immunoblotting. (E) Quantification of
   apoptosis percentage in SK-Hep-1 cell groups. (F) Quantification of
   apoptosis percentage in Hep-3B cell groups. (G) VP with sorafenib
   reduced Hep 3B cell viability. (H) VP with sorafenib reduced SRGN-high
   MHCC-97H cells viability. (I) VP with sorafenib reduced SRGN-high
   MHCC-97H cells colony formation. (J) Quantification for colony
   formation assay. (K) Localization of YAP in main and side populations
   detected by confocal immunofluorescence. (L) Side population detection
   stained with Hoechst 33342 plus or minus verapamil. (M) The main
   population and side population cell number and percentage. (N) Tumor
   volume comparisons in SRGN-high SK-Hep-1 xenografts under different
   treatment schemes. (O) Tumor volume calculation. (P) IHC staining of
   YAP and cleaved caspase-3 in xenograft tumors under different treatment
   schemes.

   To examine the pathway alterations triggered by SRGN expression under
   sorafenib intervention, phosphorylated ERK (P-ERK), phosphorylated YAP
   (P-YAP), ERK, and YAP levels were analyzed. In Hep 3B cells
   overexpressing SRGN, sorafenib treatment dose-dependently induced
   caspase-3 cleavage, a hallmark of apoptosis (Figure [188]10E, F).
   Sorafenib was observed to disrupt the stimulatory effects of SRGN on
   P-YAP, P-ERK, YAP, and ERK, indicating targeting SRGN enhanced
   apoptosis via the YAP pathway. Furthermore, HCC cells that
   overexpressed SRGN demonstrated elevated levels of cleaved caspase-3
   and decreased levels of total caspase-3 compared to the vector control
   group in a dose-dependent form. Additionally, cell viability assays
   revealed VP significantly reduced sorafenib resistance in malignant
   cells with high SRGN expression (Figure [189]10G, H). Similarly, colony
   formation assays showed that combining VP with sorafenib significantly
   suppressed the colony-forming ability of SRGN-overexpressing cells
   (Figure [190]10I, J). The confocal immunofluorescence analysis
   indicated YAP predominantly localized within the nuclei of cells in
   side population, demonstrating a significant nuclear translocation
   compared to the main population (Figure [191]10K). The side population
   assay demonstrated that VP significantly reduced SRGN-overexpressing
   malignant cells in the side population from 14.67% to 0.29%.
   Additionally, sorafenib reduced the main population from 10×10⁶ to
   6.87×10⁶ cells (Figure [192]10L, M). Combining sorafenib with VP could
   significantly reduce the main and side populations of SRGN
   high-expressed malignant cells.

   Moreover, utilizing the xenograft model based on SK-Hep-1 cells in nude
   mice, in vivo effects were assessed. The results demonstrated that the
   HCC volumes were significantly smaller in the groups receiving
   combination therapy compared to those treated with individual drugs
   (sorafenib or VP alone) (Figure [193]10N, O). Immunohistochemistry
   staining further confirmed the expression of cleaved caspase-3 and YAP
   in tumor samples from the in vivo model (Figure [194]10P).
   Comprehensively, sorafenib combined with VP could effectively eliminate
   SRGN high-expressed HCC cells by targeting YAP-activated tumor cell
   progression. We also found that HCC patients characterized by elevated
   SRGN expression exhibit a significantly higher potential for immune
   escape that is unlikely to respond favorably to anti-PD1/CTLA1 therapy
   ([195]Supplementary Figure 12). Consequently, targeting the SRGN
   pathway in this patient population is imperative to realize meaningful
   therapeutic outcomes.

Screening of potential drugs targeting SRGN protein

   Given the results above, SRGN protein represents a practical and
   promising target for developing therapies to combat HCC progression. We
   performed a drug screen for SRGN (Figure [196]11A). We extracted the
   latest protein structure of the SRGN using AlphaFold2, which UniProtKB
   reviewed manually([197]https://www.uniprot.org/uniprotkb/P10124/entry)
   (Figure [198]11B). Next, the predicted SRGN protein structure was
   subjected to cavity detection using CB-Dock2's CurPocket tool[199]^42,
   a network cavity detection method based on protein surface curvature.
   We screened the compounds using DrugRep[200]^43, an online virtual
   screening server based on AutoDock Vina. Using CB-Dock2's CurPocket
   tool, we found two pockets of SRGN protein and prepared them for
   docking configurations (Figure [201]11C, D). We screened 10,640
   compounds and found 74 compounds with good binding ability to SRGN
   protein (the smaller the value of the score, the better the
   effectiveness of the docking, threshold: Vina score <-8.0), including
   22 FDA-approved drugs, 23 experimental drugs, and 19 traditional
   Chinese medicine monomers ([202]Supplementary Table 2). In addition,
   among these 74 compounds, some approved multi-targeted antitumor drugs
   had a high affinity for SRGN protein. The Vina binding score of
   sorafenib complexed with SRGN protein was -8.4 (Figure [203]11E).
   Vina's score for the regorafenib-SRGN combination was -8.1, which
   showed good potential as an anti-SRGN drug potential (Figure [204]11F).
   Daurisoline, an isoquinoline alkaloid isolated from the rhizome of
   Cortex Eucommiae, exhibited anti-inflammatory and antitumor
   effects[205]^44, and Vina score of daurisoline-SRGN complex was -9.1
   (Figure [206]11G). Baicalin, a natural flavonoid in various medicinal
   plants, presented beneficial antitumor and hepatoprotective
   bioactivities[207]^45, with a vina score of -8.8 for the baicalin-SRGN
   complex (Figure [208]11H). They indicated that small medicinal plant
   molecules might also be potential candidates for therapeutic agents for
   anti-HCC progression.

Figure 11.

   [209]Figure 11
   [210]Open in a new tab

   Potential medication screening for targeting SRGN protein. (A) Workflow
   of screening. (B) SRGN protein structure. (C) Docking pocket 1 of SRGN
   protein. (D) Docking pocket 2 of SRGN protein. The docking pockets and
   poses, and binding residues of sorafenib (E), daurisoline (F),
   regorafenib (G), and baicalin (H) with SRGN protein.

   Molecular dynamics simulations spanning 100 ns were performed to assess
   the binding stability of four SRGN complexes: sorafenib (Figure
   [211]12A), regorafenib (Figure [212]12B), baicalin (Figure [213]12C),
   and daurisoline (Figure [214]12D). The daurisoline-SRGN complex
   demonstrated superior stability across multiple structural and
   energetic parameters. The system achieved rapid equilibration, with
   root mean square deviation (RMSD) values stabilizing within the
   shortest simulation time (<5 ns) and maintaining minimal fluctuations
   (<0.25 Å), indicative of robust conformational stability. Root mean
   square fluctuation (RMSF) analysis revealed an optimal balance of
   rigidity and flexibility at the binding interface, suggesting adaptable
   yet stable ligand-protein interactions. The radius of gyration (Rg)
   remained consistently below 2 nm, confirming compact molecular packing
   and sustained binding site occupancy. Daurisoline constantly stably
   binds to the initial binding site of the SRGN protein. Notably, the
   daurisoline-SRGN complex exhibited the largest buried
   solvent-accessible surface area (SASA), exceeding other complexes,
   correlated with extensive interfacial contacts. Binding free energy
   calculations further highlighted its dominance. In vitro experiments,
   combined with the study of the previous dosage[215]^46^, [216]^47,
   daurisoline had a better inhibitory effect on the activity of SK-Hep-1
   cells and a lower expression of SRGN protein than sorafenib
   ([217]Supplementary Figure 13). The findings position daurisoline as a
   promising monomer for anti-SRGN-driven oncogenic pathways in HCC.

Figure 12.

   [218]Figure 12
   [219]Open in a new tab

   Molecular dynamics simulations of four medication monomer-SRGN
   complexes. (A) Sorafenib-SRGN complex. (B) Regorafenib-SRGN complex.
   (C) Baicalin-SRGN complex. (D) Daurisoline-SRGN complex.

Discussion

   HCC poses significant therapeutic challenges due to its tendency for
   both intrahepatic spread and distant metastasis, along with a high
   likelihood of recurrence. Proteoglycans, a class of complex
   macromolecules rich in ECM, have been implicated in cancer progression
   and metastatic behavior. SRGN, a chondroitin sulfate-bearing
   proteoglycan initially discovered within the secretory granules of
   hematopoietic cells, has been observed to be readily released through
   exocytosis. Previously clinical studies have proved that overexpressed
   glycoprotein SRGN is correlated with a worse prognosis in HCC
   patients[220]^30^, [221]^31. However, SRGN's role in tumorigenesis and
   tumor progression of HCC has not been investigated. In our study, an
   integrative approach further reveals that highly expressed SRGN
   promotes aggressive phenotypes of HCC tumor cells.

   At the single-cell level, our analysis revealed that the significantly
   elevated expression of SRGN in HCC cells was associated with multiple
   hepatoma conditions or when liver tumors infiltrated blood vessels, in
   comparison to solitary liver tumor cases. This indicates that the high
   expression of SRGN plays a role in the invasiveness of HCC malignant
   cells. Therefore, we further comprehensively analyzed the effect of
   high SRGN expression on HCC cells' tumor progression. Metabolic changes
   of tumor cells are one of the essential characteristics of tumors,
   which are causal to the development of tumors, making malignant cells
   have characteristic metabolic patterns[222]^48. The analysis of
   single-cell metabolic flux revealed significant heterogeneities in the
   metabolism of the glycolysis, TCA cycle, BCAA, O-linked glycan
   synthesis, and pyrimidine synthesis among liver malignant cells with
   high expression of SRGN compared to those exhibiting low expression of
   SRGN. The conversion of glucose to G6P in the first step of glycolysis
   is enhanced in HCC cells with high expression of SRGN. Previous
   research showed that SRGN promotes the activation of NF-κB signaling
   and increases glycolysis in microglia[223]^49. Compared with
   glycolysis, which mainly relies on glucose for production capacity, the
   tricarboxylic acid cycle can use a broader range of nutrient sources,
   including various amino acids and acetyl-CoA produced by lipid
   degradation, to produce production capacity under sufficient oxygen
   conditions. At the same time, many products have the function of
   resisting oxidative pressure. TCA enhancement is probably positively
   significant for tumor cell metastasis[224]^50. Evidence shows solid
   tumors' TCA rate increases during metastasis[225]^51. In malignant
   cells with high SRGN expression, 3PD is more efficiently converted to
   pyruvate, and succinyl CoA is catalyzed to succinate, which helps to
   utilize different energy sources to support their growth and
   proliferation, possibly providing them with additional metabolic
   flexibility[226]^52. BCAA metabolic pathway has been demonstrated to
   influence gene expression, protein metabolism, and the processes of
   apoptosis and hepatocyte regeneration[227]^53. In this study, the
   pathway of phenylalanine being catalyzed to tyrosine in
   SRGN-overexpressing malignant cells was active. A nested case-control
   study revealed that phenylalanine and tyrosine biosynthesis are
   significant pathways implicated in the etiology of HCC[228]^54.
   O-linked glycan synthesis is only active in HCC cells with high SRGN
   expression, which might potentially contribute to the EMT[229]^55.

   Tumor cells, especially highly proliferative cells, activate the
   pyrimidine biosynthesis pathway to increase the supply of nucleotides
   to meet their needs for RNA or DNA synthesis[230]^56. The activity of
   the pyrimidine synthesis pathway in SRGN-high malignant cells reflects
   their rapid proliferative state. The activation of hallmark pathways,
   including Hedgehog signaling, unfolded protein response, EMT, hypoxia,
   TGFβ signaling, cholesterol homeostasis, and the TNFA signaling pathway
   NFKB, mirrors the highly dynamic and adaptive of tumor cells with
   intensive SRGN expression in their tumorigenic niche.

   CellChat analysis revealed that SRGN-high HCC cells remodeled
   neighboring cell receptor landscapes via paracrine signaling, eliciting
   compartment-specific responses. In NK cells, upregulated CXCR4/CD44 and
   suppressed E2F/G2M checkpoints suggest enhanced migratory capacity but
   restricted proliferation, indicative of a terminally differentiated
   cytotoxic phenotype[231]^57^-[232]^60. T lymphocytes exhibited
   metabolic-stemness reprogramming via WNT and IL2-STAT5
   activation[233]^61. Kupffer cells displayed pro-inflammatory
   polarization through oxidative phosphorylation enhancement and CXCR4
   downregulation, potentially altering spatial distribution in the
   stroma[234]^62^, [235]^63. B cells showed glycolytic metabolic
   adaptation for plasma differentiation and TGF-β suppression-induced
   IL-6 overproduction[236]^64^, [237]^65. The distinct SRGN-receptor
   pathway activation across cell types suggests its role as a niche
   organizer coordinating immuno-suppression, metabolic adaptation, and
   stromal remodeling.

   Our study revealed that malignant cells with high expression of SRGN
   showed significantly enhanced activity in proliferation, migration, ECM
   modeling, and collagen formation compared with those with low
   expression, suggesting that high expression of SRGN is conducive to the
   invasive and progressive characteristics of HCC cells. SRGN protein, an
   ECM component, is higher in HCC patients than in healthy volunteers.
   Tumor neobiotic diameter is more extensive in HCC patients with high
   SRGN expression than in low-expressed patients. SRGN serum levels are
   higher in patients with metastatic HCC than those without metastases.
   The serum levels of SRGN and AFP were positively correlated, and
   patients with high SRGN levels had a worse survival outcome than those
   with low SRGN levels. These findings are consistent with previous
   studies[238]^30^, [239]^31.

   In vitro and in vivo analyses, we first revealed that the knockdown of
   SRGN significantly inhibited the growth and migration of HCC cells. In
   contrast, the overexpression of SRGN facilitated these processes. SRGN
   overexpression has also been found to have the same growth-promoting
   effect in colon cancer cells and head and neck squamous cell carcinoma
   tumor cells[240]^27^, [241]^28.

   In observing pseudotime trajectory from scRNA-seq, tumor cells with
   high expression of SRGN maintained a certain stemness during
   differentiation. Most Tumor cells with high SRGN expression showed
   unipotent differentiation potential compared to those with low SRGN
   expression. Known stem cell markers in HCC development, such as
   CD44[242]^66, ABCG2[243]^67, BMI1[244]^68, and EPCAM[245]^69, were
   highly expressed in SRGN-overexpressed tumor cell subgroups from
   various stages of differentiation. The transcriptional corepressor
   regulatory gene KRT19, which has recently been found to cause
   dedifferentiation of HCC[246]^70, was also highly expressed in
   differentiation starting point malignant cells with high SRGN
   expression. A series of vitro experiments revealed that SRGN expression
   upregulated ABCG2, BMI1, and Nanog, enhancing the sphere-forming
   ability of HCC cells and increasing the population of cells with cancer
   stem cell-like characteristics.

   Some studies demonstrated that CD44 is a cell surface receptor for SRGN
   protein, and the binding of SRGN protein to CD44 receptor can activate
   several downstream signaling pathways[247]^16^, [248]^49^, [249]^71^,
   [250]^72. Our study consistently demonstrated that SRGN protein
   positively correlated with CD44 in HCC. Meanwhile, the Hippo pathway
   was significantly enriched in SRGN-high HCC cells. It has been
   determined that CD44 was an upstream regulator of YAP[251]^73. From
   vitro and vivo experiments, we found that in HCC, SRGN protein binds to
   CD44 to activate the Hippo/YAP signaling pathway. The Hippo/YAP pathway
   controls tissue growth and apoptosis in response to developmental
   signals, cellular contact, and density. YAP, a co-transcriptional
   factor of the Hippo pathway, is activated in the development and
   progression of HCC, which drives tumor cell survival, proliferation,
   invasive migration, metastasis, and stemness of liver tumor
   cells[252]^74^, [253]^75. Then, through comprehensive transcriptomic
   analysis, we found a positive correlation between SRGN and CRISPLD2
   expression, indicating a potential regulatory link between them. This
   initial finding prompted us to delve deeper into the relationship
   between SRGN and CRISPLD2. The role of CRISPLD2 in HCC has yet to be
   observed. CRISPLD2, presently known to be involved in the work of the
   innate immune system, and its encoded secretory protein is rich in
   cysteine, which accumulates in HCC cells and contributes to their rapid
   proliferation and antioxidant stress[254]^76^-[255]^79. In further
   exploration, it was found for the first time that SRGN can regulate
   CRISPLD2 expression through the YAP axis and positively promote tumor
   cell proliferation. YAP/TEAD1 is a co-transcription factor of CRISPLD2.
   This finding is particularly intriguing because both CRISPLD2 and YAP
   pathways were known to be associated with embryonic
   development[256]^80^-[257]^83. In SRGN-mediated HCC cells, the
   YAP/CRISPLD2 axis was selectively reactivated, setting it apart from
   the activation of other known YAP pathway target genes. The above
   evidence demonstrates a unique invasive-promoting role for the
   SRGN/YAP/CRISPLD2 axis in HCC cells, potentially offering new insights
   into the molecular mechanisms underlying tumorigenesis and novel
   therapeutic strategies.

   Sorafenib is a frontline used medication for advanced metastatic liver
   cancer[258]^84, and combined with VP. It may surmount
   chemo-insensitivity stemming from the passive lysosomal sequestration
   of anti-cancer drugs[259]^85, so we also explored their effect on
   SRGN-mediated tumor progression. Our findings revealed that the
   synergistic administration of sorafenib and VP exerted notable
   anti-SRGN-high tumor cell effects in vivo and in vitro. Sorafenib plus
   VP dramatically reduced the tumor size and caner stem cells. Massive
   molecular docking and dynamics simulations revealed daurisoline, a
   natural isoquinoline alkaloid, as a novel potential SRGN-targeting
   inhibitor in HCC[260]^46. These findings offer promising novel insights
   for underlying individual precision treatment in SRGN-positive HCC and
   hold potential therapeutic strategies for other types of cancer.

Conclusion

   In conclusion, our work confirms that SRGN promotes the invasiveness of
   HCC. SRGN supports the proliferation and metastatic potential and
   sustains stemness characteristics of HCC cells via the autocrine
   activation of the YAP/CRISPLD2 signaling pathway. Our findings provide
   insights into targeting SRGN-triggered signaling as a promising
   strategy for reversing tumor therapeutic resistance. Therapeutic
   approaches targeting SRGN may be a new direction in treating HCC.

Supplementary Material

   Supplementary methods, figures and tables.
   [261]ijbsv21p3262s1.pdf^ (1.9MB, pdf)

Acknowledgments