Exosomal CD44 Transmits Lymph Node Metastatic Capacity Between Gastric Cancer Cells via YAP-CPT1A-Mediated FAO Reprogramming.

Background: Lymph node metastasis (LNM) commonly occurs in gastric cancer (GC) and is tightly associated with poor prognosis. Exosome-mediated lymphangiogenesis has been considered an important driver of LNM. Whether exosomes directly transmit the LNM phenotype between GC cells and its mechanisms remain elusive.
Methods: A highly lymphatic metastatic GC cell line (HGC-27-L) was established by serial passage of parental HGC-27 cells in BALB/c nude mice. The capacities of migration, invasion and LNM; fatty acid oxidation (FAO) levels; and the role of exosome-transferred LNM phenotype were compared among HGC-27-L, HGC-27 and primary GC cell line AGS. Exosomes derived from GC cells and sera were separately isolated using ultracentrifugation and ExoQuick exosome precipitation solution, and were characterized by transmission electron microscopy, Nanosight and western blotting. Transwell assay and LNM models were conducted to evaluate the capacities of migration, invasion and LNM of GC cells in vitro and in vivo. β-oxidation rate and CPT1 activity were measured to assess FAO. CPT1A inhibitor etomoxir was used to determine the role of FAO. Label-free LC-MS/MS proteome analysis screened the differential protein profiling between HGC-27-exosomes and AGS-exosomes. Small interference RNAs and YAP inhibitor verteporfin were used to elucidate the role and mechanism of exosomal CD44. TCGA data analysis, immunochemistry staining and ELISA were performed to analyze the expression correlation and clinical significance of CD44/YAP/CPT1A.
Results: FAO was increased in lymphatic metastatic GC cells and indispensable for sustaining LNM capacity. Lymphatic metastatic GC cell-exosomes conferred LNM capacity on primary GC cells in an FAO-dependent way. Mechanistically, CD44 was identified to be enriched in HGC-27-exosomes and was a critical cargo protein regulating exosome-mediated transmission, possibly by modulating the RhoA/YAP/Prox1/CPT1A signaling axis. Abnormal expression of CD44/YAP/CPT1A in GC tissues was correlated with each other and associated with LNM status, stages, invasion and poor survival. Serum exosomal CD44 concentration was positively correlated with tumor burden in lymph nodes. Conclusions: We uncovered a novel mechanism: exosomal CD44 transmits LNM capacity between GC cells via YAP-CPT1A-mediated FAO reprogramming from the perspective of exosomes-transferred LNM phenotype. This provides potential therapeutic targets and a non-invasive biomarker for GC patients with LNM.

Introduction

Lymph node metastasis (LNM) frequently occurs whenever gastric cancer (GC) patients are diagnosed at an early or advanced stage; it is recognized as an important indicator of poor prognosis (1). Lymphadenectomy is commonly used to prevent further LNM, but it has been reported that extended lymphadenectomy might result in recurrence and cancer-related death. There is still controversy over the extent of lymph node (LN) dissection (1). Although adjuvant treatments reduce the recurrence and metastasis of GC to a certain extent, their side effects are inevitable. During the past decade, the concept of precision therapy has largely promoted progression in molecular target research. However, effective therapeutic molecular targets for LNM are still lacking. Elucidating the underlying molecular mechanism of LNM during GC progression is particularly important.

VEGF-C/VEGFR is identified as a classical pair of cytokine-receptor involved in cancer-associated lymphangiogenesis (2). Recently, a new VEGF-C-independent mechanism of LNM was revealed (3). By taking up bladder cancer cell derived exosomal long non-coding RNA (LNMAT2), lymphatic endothelial cells acquired enhanced capacities of tubule formation and migration leading to LNM in bladder cancer (3). As early as 2011, melanoma exosomes have been shown to facilitate LNM by remodeling LNs (4). These findings suggest that nano-sized exosomes released by tumor cells play an important role in LNM. Several studies have indicated that the molecular types and contents sorted into cancer cell-derived exosomes determine the LNM potential of cancer cells by crosstalk with surrounding stromal cells (5–8). In addition to communication with the tumor microenvironment, it is notable that exosomes from highly metastatic cancer cells promoted metastasis of low metastatic cancer cells (9–11). However, the exosome-mediated transmission of the LNM phenotype between cancer cells with different malignant potentials is still not well understood. Liu et al. (12) reported that CD97 enriched in highly lymphatic metastatic GC cells-derived exosomes regulated LNM capacity of GC, which suggests that highly lymphatic metastatic GC cells-exosomes are more likely involved in malignant phenotype transmission.

Metabolism reprogramming has been acknowledged as a hallmark of cancer (13). To form metastasis in LNs, cancer cells must undergo metabolic reprogramming to successfully colonize and survive in an unfavorable microenvironment. Due to the lipid-rich property of LNs, the metabolic adaption of lymphatic metastatic cancer cells is more likely associated with fatty acid metabolism (14). It has been demonstrated that melanoma cells formed LNM depending on fatty acid oxidation (FAO) (15). Moreover, CPT1A (carnitine palmitoyltransferase 1a), a rate-limiting enzyme of FAO, was shown to regulate cancer-associated lymphangiogenesis and to be associated with the status of LNM (16, 17). The studies above suggest that FAO emerges as a novel contributor to LNM, but its role and underlying mechanism in eliciting LNM of GC remain unclear.

To figure out whether lymphatic metastatic GC cell-exosomes transmit a metastatic phenotype to primary GC cells depending on FAO, we established a highly lymphatic metastatic GC cell line. We evaluated the role of exosome-mediated LNM phenotype transmission from lymphatic metastatic GC cells to primary GC cells through FAO, identified key proteins mediating exosome regulation and elucidated its downstream signaling pathway. We hope to uncover a novel mechanism of LNM from the perspective of exosome-mediated transmission and provide a potential molecular target for the detection and treatment of GC patients with LNM.

Materials and Methods

Cell Lines

Gastric cancer cell lines AGS and HGC-27 were purchased from Procell Life Science&Technology Co., Ltd (Wuhan, China). AGS was cultured in DMEM/F12 medium (Bioind, Israel) supplemented with 10% FBS (SERANA, Brandenburg, Germany), while HGC-27 was maintained in RPMI-1640 medium (Bioind) containing 10% FBS. HGC-27-L was established by injecting 1× 106 of HGC-27 cells into the left foot pads of BALB/c nude mice (Model Animal Research Center of Nanjing University, China). Three weeks later, the mice were killed according to institutional guidelines. The drained popliteal LNs were collected, cut into 1 mm3 tissue pieces and digested with 0.25% trypsin. All single cells were then harvested, suspended with RPMI-1640 medium containing 10% FBS and plated in a 3.5-cm dish. Cell culture medium was refreshed every three days until the cell number reached 1 × 106. These isolated cells were used to repeat the above process two times. The final isolated cells from LNs were named HGC-27-L. All cells were cultured in a humidified cell incubator at 37°C with 5%CO2.

Clinical Samples

Serum samples from 37 GC patients with LNM and 11 GC patients without LNM were collected from the Affiliated Tumor Hospital of Nantong University, aliquoted and stored at −80°C. Paraffin-embedded paired tissue samples of primary GC tissues and lymph node metastases from five GC patients with LNM and paraffin-embedded primary GC tissues from five patients without LNM were collected from the Affiliated Hospital of Jiangsu University.

Transwell Assay

The migration and invasion capacity of GC cells were detected using transwell as previously described (8). Briefly, fresh culture media were added into the bottom chamber. GC cells (8 × 104) suspended in serum-free media were added into the untreated upper chamber and incubated for 10 h for migration assay. The same number of GC cells were placed into the upper chamber pretreated with matrigel (BD Bioscience, San Jose, CA, USA) and incubated for 24 h for invasion assay. The migrated and invasive cells were viewed and counted in at least six random fields after extensive washing, fixation and crystal staining.

CPT1 Activity and β-Oxidation Rate Detection

Cell mitochondria of GC cells were isolated using a Cell Mitochondrial Isolation Kit (Beyotime Biotechnology, Shanghai, China). β-oxidation rate was measured according to the instructions of the Fatty Acid β-oxidation Rate Colorimetric Assay Kit (Genmed Scientifics Inc., USA). CPT1 activity was measured using a CPT1 Spectrophotometric Detection Kit (Zikerbio, Guangzhou, China). Absorbance was measured with a 722-type spectrophotometer (Shanghai Spectrum Instruments co., Ltd., China).

Exosome Isolation, Characterization and Treatment

Cellular exosomes and serum exosomes were isolated, characterized and quantified as previously described (8). Briefly, GC cell culture supernatants were processed using differential centrifugation to remove cells, cell debris and large particles. Ultracentrifugation of the processed supernatants at 110,000 g were performed to precipitate cell exosomes. Serum exosomes were isolated using ExoQuick exosome precipitation solution (EXOQ20A-1, SBI System Biosciences, Palo Alto, CA, USA). Exosomes fixed by glutaraldehyde were used for transmission electron microscopy (TEM) detection. Exosome size distribution was detected using nanoparticle tracking analysis (NTA), which was carried out using a Nanosight NS300 system (Malvern Panalytical Ltd., Malvern, UK). The exosomal protein was quantified using a BCA Protein Assay Kit (CoWin Biosciences, Shanghai, China) and detected by western blotting. AGS cells were seeded in six-well plates at the density of 8 × 104 per well and attached overnight. AGS cells were treated with fresh cell culture media containing 50μg/ml of different GC cell-exosomes for 48 h, then washed and collected for in vivo and in vitro experiments.

Label-Free Quantitation of Exosomal Protein Profiling

Exosomes derived from HGC-27 and AGS were dissolved in 50μl of PBS at a concentration of 1 μg/μl and sent to the Shanghai Applied Protein Technology company (Shanghai, China) for LC-MS/MS-based label-free quantitation detection. Each group set contained three biological replicates. Proteins were identified by searching the human UniProt database (Version 2018_02_26 with 161629 entries) using MaxQuant software. LFQ (Label Free Quantitation) algorithm was used for quantitative analysis. Differential proteins were screened based on the fold changes ≥ 2 and P value < 0.05.

Oligonucleotide Transfection

All oligonucleotides were designed and synthesized by Genepharma (Shanghai, China), and their sequences are presented in . Three pairs of small interference RNAs against CD44 (si-CD44) were transfected into GC cells with lipofectamine 2000 (Invitrogen, Thermo Fisher Scientific, Inc., MA, USA) at a final concentration of 50 nM. Negative control oligonucleotides (NC) at the same concentration were used as controls. After transfection for 48 h, cells were collected for protein purification and further analysis.

Small Molecular Chemistry Inhibitor Utilization

Etomoxir and Verteporfin (MedChem Express, NJ, USA) were dissolved in DMSO and treated GC cells at concentrations of 40 μM and 5 μM for 24 h, respectively. The cells were then refreshed with culture medium or treated with exosomes for 48 h for further analysis. AGS cells were treated with Cycloheximide (CHX) (Beyotime Biotechnology) at the concentration of 50μg/ml for 4 h followed by HGC-27-exosomes treatment for 48 h. The total protein of the cells was extracted for further analysis.

ELISA Assay

Human CD44/Heparan Sulfate Proteoglycan ELISA kits (RayBiotech Life, Inc., GA, USA) were used to detect CD44 content in serum exosomes. Briefly, exosomes isolated from 200 μl serum were suspended in 50 μl of PBS and were then lysed by RIPA buffer and subjected to detection according to the specified protocols. The final concentration of exosomal CD44 was calculated by the formula obtained by standard curve analysis. All samples and the standard were measured in triplicate.

Western Blotting

Proteins extracted from cells and exosomes were prepared and detected as previously described (18). Briefly, proteins were separated on 12% SDS−PAGE gels, transferred to polyvinylidene difluoride membranes and blocked followed by incubation with different primary antibodies against CD81 (Cat. No. A3044, dilution: 1:1000, WUHAN SANYING, Wuhan, China), Calnexin (Cat. No. 10427-2-AP, dilution: 1:20000, WUHAN SANYING), CD44 (Cat. No. A19020, dilution: 1:1200, ABclonal Technology Co., Ltd., Wuhan, China), ras homolog family member A (RhoA) (Cat. No. A13947, dilution: 1:800, ABclonal Technology), Prospero Homeobox 1 (Prox1) (Cat. No. A9047, dilution: 1:1000, ABclonal Technology), phosphor-YAP-S127 (Cat. No. AP0489, dilution: 1:1000, ABclonal Technology), YAP (Cat. No. A19134, dilution: 1:800, ABclonal Technology), CTGF (Cat. No. A11067, dilution: 1:1000, ABclonal Technology), CYR61 (Cat. No. A1111, dilution: 1:1000, ABclonal Technology), β-actin (Cat. No. AC028, dilution: 1:5000, ABclonal Technology), CPT1A (Cat. No. ab128568, dilution: 1:1000, Abcam, Shanghai, China) and TSG101 (Cat. No. ab30871, dilution: 1:1000, Abcam). The secondary antibodies HRP Goat Anti-Mouse IgG (H+L) (Cat. No. AS003, ABclonal Technology) and HRP Goat Anti-Rabbit IgG (H+L) (Cat. No. AS014, ABclonal Technology) were used at a dilution of 1:5000. Protein signals were detected as previously described (18), and gray values were analyzed using ImageJ software.

Immunohistochemistry (IHC) Staining

Primary antibodies against CD44, YAP (ABclonal Technology), CPT1A and pan cytokeratin (AE1/AE3) (Abcam) were used to detect the corresponding proteins in tissue slices using an instantw SABC-POD Kit (Boster Biological Technology) as previously stated (18). Briefly, the tissue sections were deparaffinized, rehydrated and subjected to antigen retrieval. After being blocked with BSA, the tissue sections were sequentially incubated with different primary antibodies and the corresponding biotinylated secondary antibodies. Target proteins were viewed and examined using 3,3’‐diaminobenzidine and hematoxylin counterstaining. Images were captured and scanned by the Automatic digital slice scanning system (Version: Pannoramic MIDI, 3DHIESTECH, Hungary) and assessed with Caseviewer software (version CV 2.3, 3DHISTECH). The staining scores of each protein were calculated through the intensity value (negative, 0; +, 1; ++, 2; +++, 3) multiplied by positive rate value (negative, 0; 1–25%, 1; 26–50%, 2; 51–75%, 3; 76–100%, 4).

Animal Models

Male and six-week aged BALB/c nude mice were purchased from the Changzhou Cavens Laboratory Animal company (Changzhou, China). All mice were randomly assigned to different groups, had 5 × 106 of AGS injected into the foot pads and were scarified about eight weeks later. Popliteal LNs were harvested for volume and weight measurement, paraffin tissue preparation and section staining.

TCGA Analysis

RNA sequence profiles and clinical information downloaded from the Stomach adenocarcinoma (STAD) project (306 patients) of The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/) were used to analyze the association between the expression levels of CD44/YAP/CPT1A and clinicopathological features by deleting GC patients with uncertain clinical data ( ). Gene Expression Profile Interaction Analysis (GEPIA) (http://gepia.cancer-pku.cn/) was used to explore the expression correlation among the three genes and reveal their differential expression between GC tissues and normal tissues by selecting STAD dataset and matched TCGA normal and GTEx data using Box-plot analysis. Kaplan Meier plotter (https://kmplot.com/analysis/) was used to analyze the association of the three genes (CD44:1557905_s_at; YAP:217836_s_at; CPT1A:210687_at) with five-year GC survival.

Statistics

GraphPad Prism 8 and SPSS 20.0 were used for statistical analysis. All experiments were conducted at least in triplicate. The experimental data were described as mean ± SD. Independent Student’s t-test was used to compare two groups while one-way ANOVA followed by post-hoc test (Tukey’s) was conducted to compare the three groups. The data from clinical samples were plotted and expressed as means. Clinical significance was analyzed by Chi-square test and correlation of protein staining scores were assayed by Pearson Correlation test. P < 0.05 indicates a statistically significant difference.

Results

Increased FAO Is Required for Lymphatic Metastatic GC Cells Sustaining LNM Capacity

The metastatic LN tissue-derived GC cell line HGC-27 and primary GC tissue-derived GC cell line AGS were selected as representatives of lymphatic metastatic and primary GC cells, respectively. We further established a highly lymphatic metastatic GC cell line (HGC-27-L) by serial passage of HGC-27 in vivo as indicated in . The capacities of LNM, migration and invasion of HGC-27-L were then compared with those of AGS and the parental cell HGC-27 in vivo and in vitro. The volume, weight and positive pancytokeratin-AE1/AE3 expression area of popliteal LNs obtained from the AGS, HGC-27 and HGC-27-L groups were significantly gradually increased ( ). The number of migrated and invasive cells in the three groups showed the same trend ( ). These data confirmed that a highly lymphatic metastatic GC cell line HGC-27-L was successfully established. To determine whether FAO is enhanced in lymphatic metastatic GC cells, the β-oxidation rate and CPT1 activity were assessed to evaluate FAO. As shown, β-oxidation rate and CPT1 activity were remarkably gradually increased starting from the AGS group, the HGC-27 group and then to the HGC-27-L group ( ), which suggests that FAO is highly activated in lymphatic metastatic GC cells. To analyze whether increased FAO is indispensable for GC cells sustaining LNM capacity, the CPT1A inhibitor etomoxir was used to block FAO in the two lymphatic metastatic GC cells. In vitro, the number of migrated and invasive HGC-27 and HGC-27-L cells was obviously reduced after etomoxir treatment compared to those treated with DMSO ( ). These data suggest that FAO is increased in lymphatic metastatic GC cells and is indispensable for sustaining lymphatic metastatic capacity.

Figure 1

Establishment of a highly lymphatic metastatic GC cell line HGC-27-L and lymphatic metastatic GC cells sustained LNM capacity depending on FAO (A) A flow chart illustrates the establishment of HGC-27-L by serial transplantation of parental cell HGC-27 in vivo. Images of cell morphology are presented. (magnification, 400×; scale bars, 50 μm); (B–D) In vivo comparison of LNM capacity among AGS, HGC-27 and HGC-27-L by popliteal LNs analysis. (B) Pictures of LNs. Ruler unit, mm; (C) Weight of LNs; (D) Pancytokeratin AE1/AE3 staining (magnification, 100×; scale bars, 200 μm; magnification, 400×; scale bars, 50 μm); (E, F) In vitro comparison of migration and invasion capacity among the three cell lines. (E) Morphology of migrated and invaded cells (magnification, 200×; scale bars, 100 μm); (F) Count of migrated and invaded cells; (G) β-oxidation rate detection; (H) CPT1 activity analysis; (I, J) Effect of etomoxir treatment on migration and invasion capacity of HGC-27 and HGC-27-L cells. (I) Morphology of migrated and invaded cells (magnification, 200×; scale bars, 100 μm); (J) Number of migrated and invaded cells. **P < 0.01; ***P < 0.001.

Lymphatic Metastatic GC Cell Exosomes Promote Migration, Invasion and LNM of Primary GC Cells

To investigate whether exosomes transmit the LNM phenotype from lymphatic metastatic GC cells to primary ones, exosomes from HGC-27-L, HGC-27 and AGS cells were separately isolated by ultracentrifugation and characterized by TEM, Nanosight and western blotting. The three GC cell line-exosomes exhibited the typical morphology, size distribution and specific protein markers (CD81, CD9, and TSG101) of exosomes, but they were all negative for calnexin ( and ). The primary AGS cells were incubated with the above isolated exosomes. In vitro, the number of migrated and invaded AGS cells was significantly increased by the two lymphatic metastatic GC cell-exosomes in contrast to AGS-exosomes ( ). In vivo, increased volume, weight and positive pancytokeratin-AE1/AE3 expression area of popliteal LNs formed by AGS cells were observed in the two lymphatic metastatic GC cell exosome treated groups compared to the AGS exosome treated group ( ). More notably, the effect of HGC-27-L-exosomes increasing AGS metastasis was greater than that of the HGC-27-exosomes. These data suggest that lymphatic metastatic GC cell exosomes confer LNM capacity on primary GC cells.

Figure 2

Lymphatic metastatic GC cell-exosomes transmitted LNM phenotype to primary GC cells (A–C) Characterization of exosomes from lymphatic metastatic GC cells (HGC-27 and HGC-27-L) and primary GC cells (AGS). (A) TEM analysis (magnification, 20000×; scale bars, 500 nm); (B) Nanosight detection; (C) Western blotting analysis; (D–H) Comparison of LNM capacity of AGS after treatment with GC cell-exosomes. (D, E) In vitro transwell analysis of migration and invasion capacity. (D) Images of migrated and invaded cells (magnification, 200×; scale bars, 100 μm); (E) Cell count; (F–H) LNM capacity evaluation in vivo. (F) Pictures of popliteal LNs. Ruler unit, mm; (G) Weight of LNs; (H) Pancytokeratin AE1/AE3 staining (magnification, 100×; scale bars, 200 μm; magnification, 400×; scale bars, 50 μm). ex, exosomes. *P < 0.05; ***P < 0.001.

Lymphatic Metastatic GC Cell Exosomes Increase LNM Capacity of Primary GC Cells in an FAO-Dependent Way

To determine whether FAO is increased in AGS cells treated with lymphatic metastatic GC cell-exosomes, β-oxidation rate and CPT1 activity detection showed that both HGC-27-L-exosomes and HGC-27-exosomes increased FAO in AGS cells compared to AGS-exosomes ( ). More importantly, the effect of HGC-27-L-exosomes increasing FAO was greater than that of HGC-27-exosomes. To determine whether FAO is necessary for lymphatic metastatic GC cell-exosome enhancing AGS LNM capacity, AGS cells were pretreated with etomoxir followed by incubation with lymphatic metastatic GC cell-exosomes. In vitro, even in the presence of lymphatic metastatic GC cell-exosomes, reduced numbers of migrated and invaded AGS cells were shown in etomoxir treatment groups compared to the DMSO groups ( ). Similarly, etomoxir pretreatment resulted in decreased volume, weight and positive pancytokeratin-AE1/AE3 expression area of popliteal LNs formed by AGS cells in vivo ( ). FAO blocking significantly attenuated LNM capacity of AGS cells enhanced by lymphatic metastatic GC cell-exosomes, suggesting that lymphatic metastatic GC cells-exosomes transmit a metastatic phenotype to primary GC cells depending on FAO.

Figure 3

Lymphatic metastatic GC cell-exosomes conferred LNM capacity on primary GC cells depending on FAO (A, B) FAO detection in AGS after treatment with different GC cell exosomes. (A) β-oxidation rate measurement; (B) CPT1 activity analysis; (C–G) Effect of etomoxir pretreatment on lymphatic metastatic GC cell-exosome increasing AGS LNM capacity. (C, D) In vitro migration and invasion assay. (C) Representative images of migrated and invaded cells (magnification, 200×; scale bars, 100 μm); (D) Number of migrated and invaded cells; (E–G) In vivo LNM capacity detection. (E) Images of popliteal LNs. Ruler unit, mm; (F) Weight of LNs; (G) Pancytokeratin AE1/AE3 staining in LNs (magnification, 100×; scale bars, 200 μm; magnification, 400×; scale bars, 50 μm); ex, exosomes. **P < 0.01; ***P < 0.001.

Exosomal CD44 Mediates Malignant Phenotype Transmission From Lymphatic Metastatic GC Cells to Primary GC Cells

To determine what kind of protein plays a critical role in such a malignant phenotype transmission, protein profiling in HGC-27-exosomes and AGS-exosomes was determined and compared using LC-MS/MS-based label-free quantification. One hundred and thirty-five and 175 proteins were separately identified in HGC-27-exosomes and AGS-exosomes ( ). Thirty-nine detectable proteins in both showed different contents between the two GC cell-exosomes. More importantly, 34 proteins were only detected in HGC-27-exosomes whereas 81 proteins were just detected in AGS-exosomes ( ). We focused on the detectable proteins in HGC-27-exosomes and found that CD44 was included in them. Western blotting analysis showed that CD44 protein was highly enriched in HGC-27-exosomes and HGC-27-L exosomes compared with AGS-exosomes. A higher level of CD44 protein was detected in HGC-27-L-exosomes relative to HGC-27-exosomes ( and ). The differential CD44 protein levels in AGS, HGC-27 and HGC-27-L cells were similar to those in their derived exosomes ( and ). Furthermore, CD44 protein levels were increased gradually in AGS cells after treatment with AGS-exosomes, HGC-27-exosomes and HGC-27-L-exosomes ( and ). CHX pretreatment failed to block an increase of CD44 protein by HGC-27-exosomes ( and ), which suggests that CD44 might be directly delivered into AGS cells by HGC-27-exosomes.

Figure 4

Identification of CD44 was a critical protein mediating exosome- transmission of LNM phenotype (A) A flow chart shows the identification of differential protein profile between HGC-27-exosomes and AGS-exosomes by label-free quantitation; (B) CD44 protein detection in different GC cell-exosomes; (C) Comparison of CD44 protein in GC cells; (D) Change of CD44 protein levels in AGS after treatment with exosomes; (E) Effect of CHX pretreatment on CD44 protein in AGS incubation with HGC-27-exosomes; (F) Screening for the most efficient si-CD44 in HGC-27 and HGC-27-L; (G–J) Effect of CD44 knockdown on the migration and invasion capacity (G, H) and FAO (I, J) of HGC-27 and HGC-27-L; (K) CD44 protein detection in exosomes derived from HGC-27 and HGC-27-L with CD44 knockdown; (L–O) Effect of CD44-less exosomes on migration and invasion capacity (L, M) and FAO (N, O) of AGS; (P–R) Effect of CD44-less exosomes on LNM capacity of AGS in vivo. Ruler unit, mm. CHX, cycloheximide; ex, exosomes. *P < 0.05; **P < 0.01; ***P < 0.001.

To investigate whether CD44 is critical for lymphatic metastatic GC cell exosome increasing AGS FAO and LNM capacity, three pairs of si-CD44 were transfected into HGC-27 and HGC-27-L to screen for the most effective siRNA. NC was used as a control. siRNA#2 inhibited CD44 protein levels most significantly and was renamed and used as si-CD44 for the following studies ( and ). CD44 knockdown by si-CD44 transfection reduced capacities of migration and invasion of HGC-27 and HGC-27-L cells, accompanied by a notably decreased β-oxidation rate and CPT1 activity ( ). CD44-less exosomes were separately obtained from HGC-27 and HGC-27-L with CD44 knockdown ( and ). AGS treated with CD44-less exosomes exhibited reduced capacities of migration, invasion and FAO in vitro ( ). In vivo, the volume, weight and positive pancytokeratin-AE1/AE3 expression area of popliteal LNs were significantly reduced in AGS cells treated with CD44-less exosome groups relative to the control groups ( ). These results suggest that CD44 mediates lymphatic metastatic GC cell-exosomes regulating the LNM metastatic capacity of primary GC cells.

Exosomal CD44 Promotes FAO Possibly by Modulating the RhoA/YAP/Prox1/CPT1A Signaling Axis

CPT1A is a rate-limiting enzyme of FAO. Protein analysis showed that CPT1A was increased in lymphatic GC cells and their exosome-treated AGS, but was suppressed by si-CD44 transfection ( and ). It has been proven that FAO metabolism reprogramming of melanoma cells is YAP activation-dependent (15). CD44 has been reported to regulate YAP expression and activation via RhoA (19). YAP activation has been found to maintain the expression of the transcription factor Prox1 (20), which has been shown to transcriptionally upregulate CPT1A expression (21). Based on these observations, we hypothesized that the CD44/RhoA/YAP/Prox1/CPT1A signaling axis was possibly involved in lymphatic metastatic GC cell exosomes regulating FAO in recipient cells. To confirm this hypothesis, we determined the other protein levels of this axis in GC cells, exosome-treated AGS cells and si-CD44 transfected-HGC-27 cells. As shown, the protein levels of RhoA, YAP and Prox1 were notably increased, and the phosphorylated YAP (p-YAP) levels were reduced in the two lymphatic metastatic GC cells and their exosome-treated AGS cells compared to the corresponding controls ( and ). In contrast to NC-transfected HGC-27 cells, HGC-27 with CD44 knockdown displayed reduced protein levels of RhoA, YAP and Prox1 and increased levels of p-YAP ( and ).

Figure 5

Exosomal CD44 increased FAO possibly via regulating the RhoA/YAP/Prox1/CPT1A signaling axis (A–C) Western blotting of CPT1A, RhoA, YAP, phosphorylated-YAP (p-YAP) and Prox1 in GC cells (A), exosome-treated AGS (B) and si-CD44-transfected HGC-27(C); (D) Protein analysis in Verteporfin (VP) treated HGC-27; (E–H) Effect of VP treatment on FAO (E, F), migration and invasion capacity (G, H) of HGC-27 and HGC-27-L cells; (I–L) Effect of VP pretreatment on lymphatic GC cell-exosomes increasing FAO (I, J), migration and invasion capacity (K, L) of AGS; (M–O) Effect of VP pretreatment on HGC-27-exosome enhancing LNM capacity of AGS in vivo. Ruler unit, mm. ex, exosomes. *P < 0.05; **P < 0.01; ***P < 0.001.

Among this axis, YAP might be an important modulator. To elucidate the critical role of YAP in exosomal CD44 regulation, a small molecule inhibitor of YAP, Verteporfin (VP), was used to separately treat HGC-27 and HGC-27-L. As shown in HGC-27 cells, VP treatment successfully suppressed YAP expression and signaling, which were viewed by reduced connective tissue growth factor (CTGF) and cysteine-rich 61 (CYR61) and increased p-YAP. Meanwhile, the downstream Prox1 and CPT1A in this axis were consistently decreased ( and ). VP treatment suppressed FAO, migration and invasion capacity of the two lymphatic metastatic GC cells ( ). AGS cells were then pretreated with VP before incubation with HGC-27 exosomes or HGC-27-L exosomes. In vitro, FAO, migration and invasion capacity of AGS cells enhanced by the two lymphatic metastatic SGC cell exosomes were eliminated by VP treatment ( ). In vivo, even though incubation with HGC-27-exosome, AGS cells formed reduced volume, weight and positive pancytokeratin-AE1/AE3 expression area of popliteal LNs in VP treatment groups relative to the control groups ( ). These results suggest that exosomal CD44 might regulate the RhoA/YAP/Prox1/CPT1A signaling axis to promote FAO in recipient cells and confer LNM capacity.

Clinical Significance of CD44/YAP/CPT1A in GC Tissues and Serum Exosomal CD44 Associates With LNM of GC

CD44, YAP and CPT1A are the critical molecules of the regulatory axis. To confirm their regulatory relationship and explore their clinical significance in GC, we used GEPIA to compare their expression levels between GC tissues and normal tissues and analyze their correlation. As shown, CD44, YAP and CPT1A RNA levels were all significantly increased in cancer tissues and were pairwise positively correlated ( ). Meanwhile, we downloaded clinical data and expression data of TCGA-STAD and analyzed their clinical significance. CD44 expression was associated with vital status, node metastasis status, stages and primary diagnosis. YAP expression was only correlated with invasion depth. CPT1A expression was associated with node metastasis status, stages and invasion depth ( ). Moreover, the Kaplan Meier plotter was used to evaluate the effect of the three molecules on the five-year survival of GC patients. With respect to overall survival (OS), first-progression survival (FP) or post-progression survival (PPS), GC patients with higher expression of the three molecules had shorter survival durations and lower survival rates compared to those with lower expression levels ( ). These data suggest that CD44, YAP and CPT1A RNA levels were correlated with each other, associated with metastatic pathological features and poor survival of GC patients.

Figure 6

Clinical significance of CD44/YAP/CPT1A in GC tissues and the association of serum exosomal CD44 with LNM (A, B) GEPIA analysis of CD44, YAP and CPT1A expression between 408 GC tissues and 211 normal tissues (A) and their expression correlation (B); (C) Kaplan Meier plotter analysis of CD44, YAP and CPT1A association with a five-year survival of GC patients, including overall survival (OS), first-progression survival (FP) and post-progression survival (PPS); (D–F) IHC analysis of CD44, YAP and CPT1A protein levels in primary GC tissues from GC patients with or without LNM and corresponding metastatic LNs. (D) Representative images; (E) Comparison of CD44, YAP and CPT1A staining scores; (F) Correlation of staining scores among CD44, YAP and CPT1A; (G) Comparison of serum exosomal CD44 concentration between GC patients with or without LNM; (H) Serum exosomal CD44 concentration detection in different groups of tumor-bearing animal models. ex, exosomes. *P < 0.05; **P < 0.01; ***P < 0.001.

Furthermore, we detected CD44, YAP and CPT1A by IHC staining of primary GC tissues from five patients with or without LNM and the corresponding metastatic LN tissues ( ). The overall mean expression trends of CD44 and CPT1A were gradually increased in primary sites from patients without LNM, primary sites from patients with LNM and the metastatic LNs. YAP staining scores in metastatic LNs were stronger than those in primary sites, but it did not show a difference between primary sites from patients with LNM or not ( ). Furthermore, CD44, YAP and CPT1A staining scores were positively correlated with each other ( ). Moreover, we collected sera from GC patients and divided them into two groups: with LNM or without LNM. Exosomal CD44 concentration was remarkably increased in patients with LNM ( ). Meanwhile, serum exosomal CD44 was detected and compared in the prior different groups of tumor-bearing mice. Compared to the corresponding groups, serum exosomal CD44 concentration was increased in the lymphatic GC cell group and their exosome treated groups while it was reduced in the si-CD44 group and VP treated group ( ). Collectively, CD44 and CPT1A expression levels in primary GC tissues were associated with the LNM status of GC. Serum exosomal CD44 content might be explored as a potential biomarker for LNM detection and evaluation.

Discussion

Due to the difficulties of determining an early diagnosis of GC, most GC patients are commonly diagnosed with LNs invaded by cancer cells, thereby further facilitating systemic dissemination. To elucidate the underlying mechanism of LNM is very beneficial in developing a suitable therapeutic intervention to prevent the malignant progression of GC. Currently, exosomes as cellular communication mediators have been shown to be involved in modulating sequential metastatic processes (22). Most studies have shown the importance of lymphangiogenesis contributing to LNM and exosome-mediated lymphangiogenesis (23, 24) but the role of lymphangiogenesis in LNM is still controversial (25). Therefore, from a different perspective, we revealed a novel mechanism whereby lymphatic metastatic GC cell-exosomes directly conferred the LNM phenotype on primary GC cells depending on FAO reprogramming. CD44 was identified as a critical exosomal cargo protein mediating this process possibly by regulating the RhoA/YAP/Prox1/CPT1A signaling axis ( ).

Figure 7

Schematic overview showing lymphatic metastatic GC cell-exosomal CD44 increases FAO to confer LNM phenotype on primary GC cells via the RhoA/YAP/CPT1A signaling axis. Metastatic GC cells in LNs secrete CD44-rich exosomes, which are taken up by GC cells in primary GC tissues and upregulate its intracellular CD44 protein. CD44 then increases RhoA expression, stimulates YAP expression, de-phosphorylation and nuclear translocation. YAP and TAZ function as transcriptional co-factors to promote Prox1 transcription and protein generation. Prox1 further translocates into nuclei and increases CPT1A transcription. CPT1A protein upregulation promotes FAO reprogramming and increases cell migration and invasion, leading to acquiring LNM capacity by primary GC cells. CD44, YAP and CPT1A might be potential therapeutic targets to suppress exosome-mediated LNM phenotype transfer from lymphatic metastatic GC cells to primary GC cells. LNs, lymph nodes.

A different degree of malignancy is the main manifestation of tumor cell heterogeneity, which determines the notion that exosomes secreted by tumor cells are heterogeneous and have different impacts on tumor progression (26). Extensive studies have demonstrated that highly malignant tumor cell-exosomes are better able to create favorable niches by educating stromal cells in primary tumor and secondary metastatic sites (27, 28). More importantly, exosomes from highly metastatic cancer cells even directly transfer metastatic capacity to poorly metastatic ones (9, 10). Although there are few relevant studies, these studies at least suggest the importance of exosomes transmitting malignant phenotypes between tumor cells with different metastatic potentials. To fully reveal whether exosomes transfer the LNM phenotype from lymphatic metastatic GC cells to primary GC cells, we established the lymphatic highly metastatic GC cell line HGC-27-L and compared the role of exosome-mediated LNM phenotype transmission among HGC-27-L cells, parental HGC-27 cells and primary GC cells. We observed that the capacity of exosomes promoting primary GC cell migration, invasion and LNM was highly correlated with the LNM capacity of GC cells, which further validates that exosomes secreted from GC cells with different lymphatic metastatic capacities are heterogeneous. Our findings also suggest that once LNM occurs, sequential events will be elicited by exosomes secreted from lymphatic GC cells, especially directly enhancing the metastatic ability of GC cells by exosome transfer. This may explain why LNM associates with a poor prognosis to some extent.

LNs are lipid-rich tissues and cancer cells need to utilize lipids to colonize and survive in LNs. FAO is an important way for lipid utilization and is now gradually recognized as a critical driver of LNM (14). Previous studies have reported that increased FAO is involved in enhancing GC cell stemness, chemotherapy-resistance and anoikis resistance for omental metastasis (29–31). However, the role of FAO in LNM of GC is still unknown. We revealed that FAO was increased and positively correlated with the LNM capacity of GC cells. FAO reprogramming is indispensable for lymphatic metastatic GC cells sustaining their LNM capacity and exosome-mediated transmission of LNM phenotype between GC cells. CPT1A is a rate-limiting enzyme of FAO and is deregulated in many human malignancies (32). We observed that CPT1A was abnormally increased in lymphatic metastatic GC cells. Etomoxir treatments remarkably suppressed lymphatic metastatic GC cell FAO, migration and invasion. A previous study showed a consistent finding that CPT1A protein level was very low in AGS compared to the metastatic GC cells. The research team performed CPT1A overexpression study on AGS and found that CPT1A promoted AGS migration, invasion and FAO. Etomoxir treatment attenuated the role of CPT1A in AGS (17). These findings further explain why the capacity of migration, invasion and LNM was highly correlated with CPT1A protein levels in lymphatic metastatic GC cells and their exosome-treated AGS cells. Moreover, abnormal upregulation of CPT1A in GC tissues was highly associated with LNM status and stages (17). Through TCGA data analysis, we obtained similar findings and also found that CPT1A was positively associated with invasion depth of GC. From the perspective of clinical significance analysis, CPT1A is confirmed as a promoter of migration and invasion, which explains why FAO plays a crucial role in LNM of GC.

Exosomes from GC cells with different LNM potentials have different regulatory effects, which rely on specific cargo assembled in exosomes. Exploring the differences in exosome cargo between cancer cells with different metastatic propensities is probably an important way to clarify the underlying mechanism of metastasis. Initial important studies have revealed that pro-metastasis proteins were specifically sorted into metastatic cancer cell-exosomes, played crucial roles in pre-metastatic niche formation for organ-specific metastasis and were expected to be biomarkers for cancer diagnosis and prognosis prediction (33, 34). Therefore, it is particularly exciting to investigate the difference in exosomal protein cargo between cancer cells with different malignant phenotypes. Several studies have used high throughput proteomics to identify the different protein cargo loaded within metastatic cancer cell-exosomes and demonstrated that proteins involved in any steps contributing to metastasis tend to be enriched in metastatic cancer cell-exosomes (35–37). We believe that exosomal protein cargo must vary between GC cells with different LNM capacities. Label-free quantitative analysis confirmed the differential protein profile between exosomes derived from lymphatic metastatic and primary GC cells, and specifically showed 34 proteins that were only detectable in lymphatic GC cell-exosomes. This group of proteins might provide us with important candidate proteins for elucidating the underlying mechanism of exosome-transferred LNM phenotype. Among the list of proteins, standard CD44 was preferentially chosen because CD44 has been recognized as a critical regulator of cancer metastasis (38). More recently, it has been shown that the transfer of CD44 by tumor-derived extracellular vesicles is a possible gateway for cancer metastasis (39). In agreement with this observation, the following protein detection and loss-functional analysis proved that CD44 is a critical protein mediating the GC cell exosome-transferring LNM phenotype in an FAO-dependent manner.

Although metastatic cancer cell exosomal CD44 has been shown to promote migration and invasion of poorly metastatic cancer cells, the underlying regulatory mechanism is still elusive (40). CD44 could activate multiple intercellular signaling pathways to induce nuclear translocation of crucial transcriptional factors to promote downstream target expression, in turn accelerating oncogenic progression (38). Moreover, CPT1A expression and activation were highly positively correlated with CD44 expression, which indicates that CPT1A was probably the key node molecule of CD44 regulating FAO. To establish the regulatory link between CD44 and CPT1A, YAP was preferably considered because its activation promotes LNM of melanoma through induction of FAO (15) and it can be induced by CD44 to promote cancer metastasis (41, 42). As expected, in GC cells, CD44, YAP and CPT1A protein levels were positively correlated with each other. p-YAP expression was negatively correlated with the three proteins. CD44 knockdown and YAP inactivation in GC cells further suggest the regulatory link of CD44/YAP/CPT1A with LNM of GC, which was also validated by following tissue levels and clinical significance analysis.

CD44 can be directly delivered into primary GC cells, leading to increased intracellular protein levels. To clarify the possible regulatory mechanism of CD44-YAP-CPT1A, we focused on the regulatory molecule between CD44 and YAP and between YAP and CPT1A. RhoA was chosen based on the previous finding that intercellular CD44 induced YAP expression and nuclear translocation via RhoA (19, 43). Transcriptional factor Prox1 was focused on because YAP/TAZ was proven to regulate and maintain its expression (20), and it transcriptionally upregulates CPT1A via binding to two sites separately located at the promoter and in the intergenic regions of the CPT1A gene (21). Interestingly, RhoA expression was positively correlated with CD44 and YAP expression and activation. Prox1 expression showed a similar trend to that of RhoA and was positively associated with CPT1A expression. Consistent protein expression suggests that the CD44/RhoA/YAP/Prox1/CPT1A axis might be a novel regulatory signaling pathway contributing to the LNM of GC. However, in the present study, we only validated the possible regulatory mechanism of CD44 on YAP and CPT1A through relevant molecule detection. Due to the different cellular contexts and the controversial findings regarding the regulatory role of YAP on Prox1 (44), detailed regulatory modes of this axis requires further investigation in future studies.

CD44, YAP and CPT1A intervention suggests that they are the three critical molecules of the signaling axis and are potential therapeutic targets for preventing LNM continuous progression. TCGA database analysis confirmed their regulatory relationships and suggests that they might regulate different aspects contributing to GC metastasis. Although IHC detection showed that the expression of the three molecules in primary GC tissues and metastatic LNs were heterogenous, the correlation of the staining scores among the three molecules was further validated. An increasing trend in CD44 and CPT1A protein was observed in primary sites from patients with LNM relative to those from patients without LNM, but the YAP protein was not different, which may be caused by the transcriptional co-factor property of YAP and abnormal distribution and nuclear translocation of YAP in the invasive front as reported by Lee et al. (15). In future studies, additional primary GC tissues from patients should be assessed to comprehensively evaluate the protein expression pattern of the three molecules to exclude the possibility of sampling errors. Although the expression of these three proteins varies in tissues, the content of serum exosomal CD44 is upregulated in patients with LNM. In vivo, animal models show consistent results. Collectively, these findings suggest that serum exosomal CD44 may be a promising biomarker for GC patients with LNM detection.

Conclusions

In this study, we found that FAO is indispensable for lymphatic metastatic GC cells sustaining their LNM capacity. Lymphatic metastatic GC cell-exosomes transferred the LNM phenotype to primary GC cells in an FAO-dependent way. Furthermore, CD44 was identified as a critical exosomal cargo protein mediating exosome transmission possibly by regulating the RhoA/YAP/Prox1/CPT1A signaling axis. Our study reveals a novel mechanism underlying further malignant progression of LNM from the perspective of direct exosome transmission and provides potential therapeutic targets and non-invasive biomarkers for GC patients with LNM.

Data Availability Statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ .

Ethics Statement

The studies involving human participants were reviewed and approved by the Ethics Committee of Jiangsu University and Affiliated Tumor Hospital of Nantong University. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by the Committee on Use and Care of Animals of Jiangsu University.

Author Contributions

MW and FH conceived the idea of the study. MW, WY, and XC performed the experiment. HG, JH, CW, and LW analyzed data. XS, BS, and TW interpreted the results. YY and WZ contributed to refining the ideas and helped perform the analysis with constructive discussions. All authors contributed to the article and approved the submitted version.

Funding

This work was supported by the National Natural Science Foundation of China (Grant No: 81772641, 81902510, 81772262, 81972313, 81972822) and the Suzhou Health Youth Backbone Talent of National Mentor System (Grant no: Qngg2021043).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.