TACE responser NDRG1 acts as a guardian against ferroptosis to drive tumorgenesis and metastasis in HCC

Determination of TACE Responsiveness-related Molecular Subtypes of HCC

To analyze the TACE response-related genes of HCC, we first performed a differential analysis of the HCC cohort (paracancer and HCC samples) of the GSE14520 cohort and the TACE-treated cohort (TACE responders and nonresponders) from the GSE104580 cohort. The volcano plot shows the distribution of DEG expression in these two cohorts (Fig. 1A, B), and we applied the Venn plot to intersect the HCC-related DEGs and TACE-reactivity-related DEGs. Finally, we obtained 109 shared DEGs, and we identified these 109 DEGs as genes related to the TACE response of HCC (Fig. 1C). Subsequently, we used the R package "ConsensusClusterPlus" to establish a consistent clustering of these 109 DEGs in the HCC-TACE cohort within the GSE14520 cohort, and the results showed the clustering effect when HCC samples were divided into two subtypes (Cluster A, Cluster B). This approach was optimal, with better intrasubtype consistency and stability (Fig. 1D-F). Overall survival (OS) results when analyzing the prognosis associated with these two clusters showed that Cluster A had a better survival prognosis than Cluster B (p < 0.0001) (Fig. 1G), while recurrence between Clusters A and B was not significantly different (p > 0.05) (Fig. 1H). When further examining the clinical characteristics between Clusters A and B, we found that Cluster A exhibited better survival than Cluster B (Fig. 1I) and a lower proportion of relapsed states than Cluster B (Fig. 1J). In addition, 83% of patients in Cluster A were in TNM stages I-II, while 34% of patients in Cluster B were in TNM stages III-IV (Fig. 1K). In Cluster A, 69% of patients had tumors ≤ 5 cm and 71% had AFP ≤ 300 ng/ml, while 51% of Cluster B patients had tumors > 5 cm and 71% had AFP > 300 ng/ml (Fig. 1L, M). The above results indicated that the TACE response-related molecular subtypes of HCC were closely related to the clinical characteristics of HCC patients, and the prognosis of Cluster A was significantly better than that of Cluster B.

Fig. 1figure 1

Determination of TACE responsiveness-related molecular subtypes of HCC and correlation with clinical features. A Volcano plot showing differential gene expression between paracancerous tissues and HCC tissues in the GSE14520 cohort. B Volcano plot showing differential gene expression between the HCC cohort that responded to TACE treatment and the HCC cohort that did not respond to TACE treatment within the GSE104580 cohort. Green indicates downregulated differential genes, and red indicates upregulated differential genes (|log FC|> 1 and adjusted p < 0.05). C Venn diagram showing DEGs shared between the GSE14520 and GSE104580 cohorts. D Consensus clustering matrix for k = 2. E Consistent clustered cumulative distribution function (CDF) curves for k values from 2 to 9. F CDF delta area curve. The horizontal axis represents the number of categories k, and the vertical axis represents the relative change in the area under the CDF curve. G The K-M curve shows the difference in survival between Cluster A and Cluster B. H The K-M curve shows the difference in recurrence between Cluster A and Cluster B. I Differences in survival between Cluster A and Cluster B. J Differences in recurrence status between Clusters A and B. K TNM staging between Clusters A and B. L Tumor size distribution characteristics between Clusters A and B. M AFP differences between Clusters A and B

The Predictive Value of the TRscore Scoring System Based on TACE Responsiveness-related Molecular Subtypes of HCC in the Prognosis of TACE

To further quantify individual TACE responsiveness, we constructed a TRscore scoring system using PCA to assess differences in TACE responsiveness in HCC patients. The median TRscore was used as the cutoff value to divide the patients into a high TRscore group and a low TRscore group, and the corresponding survival status, disease recurrence, and treatment differences of HCC patients with different TRscores were analyzed (Fig. 2A). When analyzing the correlation between the TRscore and the survival status and disease recurrence of HCC patients, we found that the TRscore level of the survival group was significantly lower than that of the death group (p < 0.001) (Fig. 2B), and the tumors did not recur. The group also showed a lower TRscore (p < 0.05) than the group with relapse (Fig. 2C).

Fig. 2figure 2

Predictive value of the TRscore in TACE prognosis in the GSE14520 cohort. A Corresponding clinical characteristics of patients with different TRscores. B Violin plot showing the correlation between survival status and TRscore (***p < 0.001). C Violin plot showing the correlation between relapse status and TRscore (*p < 0.05). D-F K-M curves showing the difference in survival between the high TRscore group and the low TRscore group in all HCC cohorts (D), HCC-nonTACE cohorts (E) and HCC-TACE cohorts (F). G ROC curves showing the survival prediction accuracy of the TRscore in the HCC-TACE cohort. H-J K-M curve showing the difference in recurrence rate between the high TRscore group and the low TRscore group in all HCC cohorts (H), HCC-nonTACE cohorts (I) and HCC-TACE cohorts (J). K ROC curves showing the recurrence prediction accuracy of the TR score in the HCC-TACE cohort

To verify whether the TRscore can specifically predict the TACE prognosis of HCC, we performed prognostic survival analysis in all HCC, HCC-nonTACE and HCC-TACE cohorts within the GSE14520 cohort. The results showed that there was no significant difference in survival and recurrence rates between the high- and low-TRscore groups in the HCC-nonTACE cohort (p > 0.05) (Fig. 2E, I), while the low-TRscore group of the TACE cohort showed a significant survival advantage over the high-TRscore group (p < 0.05) (Fig. 2E, F), and the recurrence rate was significantly lower than that in the corresponding high-TRscore group (p < 0.05) (Fig. 2I, J). When evaluating the prognostic prediction performance of the TRscore in the HCC-TACE cohort, the ROC results showed that the AUC for survival prediction reached 0.71, 0.77, 0.78, and 0.74 at 0.5, 1, 2, and 3 years, respectively (Fig. 2G), while the AUC for recurrence prediction was 0.67, 0.67, 0.67, and 0.66 at 0.5, 1, 2, and 3 years, respectively (Fig. 2K), suggesting that the TRscore has good predictive specificity and sensitivity. The above results indicate that the TRscore is specific for predicting the prognosis of the HCC population receiving TACE treatment. The TACE prognosis of patients with a low TRscore was significantly better than that of patients with a high TRscore, and the prediction performance was excellent.

Correlation of TRscore with Clinical Characteristics and Predictive Role of TACE Responsiveness

To elucidate the correlation between TRscore and clinical features, we analyzed the differences in the distribution of survival status, TNM stage, AFP value, tumor size, and TACE reactivity of the HCC-TACE cohort within the GSE14520 cohort under different TRscores (Fig. 3A). Further boxplot results showed that HCC patients who died and had a TNM stage III-IV, AFP > 300 ng/ml, and tumor size > 5 cm tended to have higher TRscores (p < 0.05) (Fig. 3B-E). To evaluate the predictive performance of the TRscore for TACE response in HCC, we used the GSE104580 cohort for validation, and the results showed that the TACE responder group exhibited a tendency toward lower TRscores (p < 0.05) than the TACE nonresponder group (P < 0.05) (Fig. 3F). The corresponding predicted AUC value was 0.817 (Fig. 3G, H). We also performed DCA to assess the value of the TRscore in clinical decision-making. The DCA curve results suggested that the TRscore had good predictive performance (F ig. 3I). The above results suggest that the TRscore is closely related to clinical disease characteristics such as survival status, TNM stage, AFP value, tumor size, and TACE reactivity. Patients with high TRscores have more severe disease characteristics and worse TACE prognoses than patients with low TRscores.

Fig. 3figure 3

Correlation of TRscore with clinical characteristics in the HCC-TACE cohort of the GSE14520 cohort. A Donut-shaped pie chart showing differences in clinical characteristics between the high and low TRscore groups. B-E Boxplots showing that survival status (B), TNM stage (C), AFP value (D), and tumor size (E) correlated with TRscore (*p < 0.05, **p < 0.01, ** *p < 0.001). F Boxplots showing the difference in the distribution of TR scores between TACE responders and TACE nonresponders in the HCC-TACE cohort of the GSE104580 cohort (*p < 0.05, ***p < 0.001). G Concordance curves showing concordance of correlation predictions. H AUC values of ROC curves assess the reliability of correlation prediction. I DCA curves showing clinical benefit predicted by correlation

Correlation of TRscore with Differential Expression of the Tumor Immune Microenvironment and Immune Checkpoints

Previous studies have shown that TACE treatment affects the formation and alteration of the tumor immune microenvironment [20, 21] and has potential synergistic effects with immunotherapy [22]. Therefore, we analyzed the association between the TRscore and the tumor immune microenvironment through an independent immunotherapy cohort, IMvigor210. We first explored whether the TRscore is related to the degree of immune cell infiltration and immune function in the tumor microenvironment. The results showed that there were significant differences in the infiltration levels of most immune cells in the different TRscore groups. Among them, the activated CD8 + T cells, type I CD4 + helper T cells (Th1 cells) and γδ T cells in the low TRscore group interacted with tumor-killing immune cells. The infiltration level of cells was significantly higher than that in the high-TRscore group (p < 0.0001) (Fig. 4A). In addition, compared with the high TRscore group, the low TRscore group showed stronger antigen-presenting cell (APC) costimulation, type II IFN response and immune cytolytic activity (CYT), which can induce or enhance antitumor immunity and functional levels. The low-TRscore group expressed higher levels of immune checkpoints than the high-TRscore group (p < 0.0001) (Fig. 4B). Tumor-related metabolic pathways that play a promoting role in tumor development, such as the cell cycle, viral oncogenicity, and epithelial-mesenchymal transition (EMT), were also confirmed to be positively correlated with TRscore levels (p < 0.05) (Fig. 4C). The above results suggest that the TRscore is closely related to immune cell infiltration and immune function in the tumor immune microenvironment. The low TRscore group showed stronger antitumor immunity than the high TRscore group, and the effect of immunotherapy may be better, while the high TRscore group showed stronger antitumor immunity than the high TRscore group. Groups tended to have higher expression levels of tumor-promoting signaling pathways, which were favorable for tumor development.

Fig. 4figure 4

Correlation between TRscore and the tumor immune microenvironment in the IMvigor210 cohort. A Differences in immune cell infiltration between the high- and low-TRscore groups (****p < 0.0001). B Differences in immune function between the high- and low-TRscore groups (**p < 0.01, ****p < 0.0001). C Correlations between TRscore levels, immune cells and tumor-related regulatory pathways. D Correlation between TRscore and the expression of common immune checkpoints. E-I Differences in PDL1 (E), PDL2 (F), CTLA4 (G), IDO1 (H) and GEM (I) expression between the high and low TRscore groups (***p < 0.001)

Considering that immune checkpoint inhibition is the preferred strategy for immunotherapy, we assessed the correlation between TRscore levels and common immune checkpoints (Fig. 4D). The results showed that compared with the high-TRscore group, the low-TRscore group exhibited higher expression levels of PDL1, PDL2, CTLA4, IDO1 and GEM (Fig. 4E-I), suggesting that the low-TRscore group was a more suitable candidate for immunotherapy and that more significant therapeutic effects could be obtained from immune checkpoint-targeted therapy.

Independent Prognostic Ability Assessment of the TRscore and Construction of the Nomogram Model

To assess the ability of the TRscore to independently predict TACE prognosis compared with traditional clinical features, including sex, age, ALT, tumor size, tumor number, TNM stage, and AFP metrics, we used the GSE14520 cohort within the HCC-TACE cohort with univariate and multivariate Cox regression analyses performed on these variables for 99 samples with complete clinical information. The data confirmed that tumor size (HR = 1.297), TNM stage (HR = 2.766) and TR score (HR = 1.069) were independent predictors of prognosis after TACE treatment (Fig. 5A). Based on several independent predictors of tumor size, TNM stage, and TRscore, we constructed a predictive nomogram to quantitatively predict individual TACE prognosis (Fig. 5B). The C-index results showed that among the independent prognostic factors, the TRscore had the best predictive performance (Fig. 5C). Calibration curves for nomograms showed good agreement between predicted 1-, 3-, and 5-year OS and actual observations (Fig. 5D-F). Subsequently, we performed ROC curve analysis to further verify the predictive accuracy of the nomogram. The AUCs of the nomogram for OS at 1, 2, 3, and 5 years were 0.773, 0.820, 0.780, and 0.711, respectively, which were significantly better than the predictive performance of a single independent predictor (Fig. 5G-J). To further assess the guiding value of the nomogram in clinical decision-making, we performed a DCA. We found that nomograms yielded the best net gains at 1, 2, 3, and 5 years compared with a single independent predictor (Fig. 5K-N). The above results suggest that the nomogram has high clinical applicability and excellent predictive ability for predicting the survival probability of HCC patients after TACE.

Fig. 5figure 5

Independent predictive power assessment of the TRscore for the HCC-TACE cohort in the GSE14520 cohort and establishment of the nomogram. A Univariate Cox and multivariate Cox regression analyses identify independent predictors of TACE prognosis. B Nomogram constructed based on the independent prognostic predictors tumor size, TNM stage and TRscore. C The C index shows the agreement between the nomogram-predicted OS and the actual value. D-F Calibration curves showing the accuracy of the nomogram in predicting OS at 1 year (D), 3 years (E), and 5 years (F). G-J ROC curves showing the predicted reliability of the nomogram at 1 year (G), 2 years (H), 3 years (I) and 5 years (J). K-N DCA curves showing the clinical benefit of the nomogram in the prediction of OS at 1 year (K), 2 years (L), 3 years (M) and 5 years (N)

Correlation Between Molecular Targeted Drug Sensitivity and TRscore

Molecular targeted drug therapy has also been one of the main treatment options for inoperable HCC patients in recent years. It has been proven to effectively inhibit tumor progression and prolong patient survival. The combination of molecular targeted drugs and TACE has shown good results and development prospects in recent years. Accordingly, we evaluated the sensitivity of the GSE14520 and HCC-TACE cohorts to common molecularly targeted drugs on the Genomics of Cancer Drug Sensitivity (GDSC) website using the half inhibitory concentration (IC50) as a reference standard. We found that the high-TRscore group was more sensitive (p < 0.05) to common molecularly targeted chemotherapeutics such as erlotinib, lapatinib, and temsirolimus than the low-TRscore group (Figure S2A-L). The above results suggest that the high-TRscore group is a potentially suitable group for molecular targeted drug therapy, and the combination of molecular targeted drug therapy and TACE in the high TRscore group can achieve more significant therapeutic effects.

Determination of Key Molecules Related to TACE Response in HCC and Evaluation of Prognostic Predictive Ability

To further identify the major contributors to the TRscore, the core regulators associated with TACE responsiveness of HCC, we performed random forest analysis on 109 DEGs associated with TACE responsiveness of HCC in the HCC-TACE cohort within the GSE14520 cohort. The results showed that CYP3A4, DKK1, AASS, NDRG1, CD5 L, ADH1B, SULT2A1, DCXR, ANXA10, CES2, KLKB1 and ADH1A were the key influencing factors of TACE response in HCC, of which NDRG1 and DKK1 were risk factors for TACE response (Fig. 6A, B). In previous research reports, DKK1 has been confirmed as a prognostic biomarker for various malignancies [25, 26], and it can well predict the efficacy and prognosis of TACE treatment in HCC patients [27]; however, NDRG1 was not confirmed in previous studies. There is an association between the findings and TACE responsiveness, and its role in the prognosis of HCC patients is not yet clear. Thus, we next focused on exploring the predictive role of NDRG1 in the survival prognosis of HCC.

Fig. 6figure 6

Identification of key molecules associated with TACE response in HCC and their role in prognosis. A-B Random forest analysis showing key molecules associated with TACE response in the HCC-TACE cohort of the GSE14520 cohort. C-E Boxplots showing the difference in NDRG1 expression between tumors and adjacent normal tissues in the GSE14520 cohort (C), ICGC-HCC cohort (D), and TCGA-HCC cohort (E). F Immunohistochemical results showing that the expression of NDRG1 was different in HCC tumor and normal tissues. Ruler: 20 mm. G-I Survival curve analysis showing differences in survival prognosis between the high and low NDRG1 groups in the GSE14520 cohort (G), ICGC-HCC cohort (H) and TCGA-HCC cohort (I). J-L ROC analysis showing the predictive accuracy of NDRG1 expression in the GSE14520 cohort (J), ICGC-HCC cohort (K) and TCGA-HCC cohort (L)

By analyzing the expression characteristics of NDRG1 in the GSE14520 and HCC cohorts of the ICGC and TCGA databases, we found that NDRG1 exhibited higher expression levels in tumor tissues than in normal tissues (p < 0.05) (Fig. 6C-E). Consistently, the results of immunohistochemical analysis also showed that the expression of NDRG1 was significantly increased in tumor tissues compared with adjacent normal tissues (Fig. 6F). The above results suggest that NDRG1 may be a carcinogenic factor in HCC. We then assessed the predictive value of NDRG1 in HCC prognosis. The median NDRG1 expression level was used to divide the high NDRG1 level group and the low NDRG1 level group as the cutoff value. The results of survival prediction analysis showed that the low NDRG1 level group in the GSE14520 dataset showed a more obvious survival advantage than the high NDRG1 level group (p = 0.005) (Fig. 6G). The corresponding AUC values reached 0.7, 0.62, 0.66, and 0.68 at 0.5, 1, 2, and 3 years, respectively (Fig. 6J), and the consistent results were also validated in the ICGC-HCC cohort and TCGA-HCC cohort (p < 0.05) (Fig. 6H, I), where the AUC values of the ICGC-HCC cohort were 0.75, 0.79, 0.74, and 0.74 at 0.5, 1, 2, and 3 years, respectively (Fig. 6K), and the AUC values of the TCGA-HCC cohort were 0.69, 0.69, 0.61, and 0.57 at 0.5, 1, 2, and 3 years, respectively (Fig. 6L). We also explored the clinical value of NDRG1 in predicting TACE response in HCC patients in GSE104580. The expression level of NDRG1 in TACE no-response tissues was much higher than that in TACE response tissues (Figure S3A). ROC curve, consistency analysis and DCA analysis indicated the superior clinical value of NDRG1 in predicting TACE response in HCC patients (Figure S3B-D). The above results suggest that NDRG1 can effectively predict the survival and prognosis of HCC with excellent predictive ability, as well as the TACE response in HCC patients.

The Effect of NDRG1 Knockdown on the Proliferation and Migration of HCC Cells

To further clarify the role of NDRG1 in the oncogenic progression of HCC, we explored the effect of NDRG1 expression on HCC cell proliferation and migration using in vitro cell experiments. First, we constructed NDRG1 knockdown shRNAs with two targets. Western blot results showed that NDRG1 shRNA transfection treatment significantly knocked down NDRG1 expression levels in SK-HEP1 and HCCLM3 cells (Fig. 7A, B). We then examined the effect of NDRG1 knockdown on the proliferation of SK-HEP1 and HCCLM3 cells by the CCK8 assay. The results of CCK8 experiments showed that after NDRG1 knockdown, the proliferation of SK-HEP1 and HCCLM3 cells was significantly inhibited (p < 0.001) (Fig. 7C, D). We evaluated the effect of NDRG1 knockdown on HCC cell viability by live/dead cell staining with calcein-AM/EthD-1 double staining. Calcein-AM, which fluoresces green, was used to label live cells, and EthD-1, which fluoresces red, was used to label dead cells. The results showed that the proportion of dead SK-HEP1 and HCCLM3 cells showing red fluorescence significantly increased after NDRG1 knockdown, indicating that NDRG1 knockdown could significantly inhibit the survival of HCC cells (Fig. 7E, F).The EdU experiment results also showed that the EdU-positive rate of SK-HEP1 and HCCLM3 cells in the NDRG1 knockdown group was significantly lower than that in the control group, further confirming the inhibitory effect of NDRG1 knockdown on the proliferation of SK-HEP1 and HCCLM3 cells (p < 0.01) (Fig. 7G). The above results suggest that the expression of NDRG1 is closely related to the proliferation of HCC cells and that knockdown of NDRG1 can effectively inhibit the proliferation of HCC cells.

Fig. 7figure 7

Effects of NDRG1 knockdown on HCC cell proliferation and migration. A-B Western blot results showing the knockdown efficiency of NDRG1 shRNA in SK-HEP1 (A) and HCCLM3 (B) cells. C-D CCK8 experiments demonstrate the effect of NDRG1 knockdown on the proliferation of SK-HEP1 (C) and HCCLM3 (D) cells. E–F Live and dead cell staining demonstrated the effect of NDRG1 knockdown on SK-HEP1 (E) and HCCLM3 (F) cell survival. Ruler: 100 m. G EdU experiments showing the effect of NDRG1 knockdown on the proliferation of SK-HEP1 and HCCLM3 cells (**p < 0.01, ***p < 0.001). Ruler: 100 m. H Transwell cell migration assay showing the effect of NDRG1 knockdown on the migration of SK-HEP1 and HCCLM3 cells (***p < 0.001). Ruler: 100 m. I-J Immunofluorescence experiments showing altered PCNA expression (PCNA red) in SK-HEP1 (I) and HCCLM3 cells (J) after NDRG1 knockdown. K-L Immunofluorescence experiments showing altered expression of Vimentin (Vimentin red) in SK-HEP1 (K) and HCCLM3 cells (L) after NDRG1 knockdown. Ruler: 200 μm

To explore the effect of NDRG1 knockdown on the migration of SK-HEP1 and HCCLM3 cells, we performed transwell cell migration experiments. We found that the migration ability of SK-HEP1 and HCCLM3 cells was significantly inhibited in the NDRG1 knockdown group compared with the control group (p < 0.01) (Fig. 7H). To further verify the effect of NDRG1 knockdown on the proliferation and migration of SK-HEP1 and HCCLM3 cells, we also tested the effect of NDRG1 knockdown on the expression of PCNA, a marker protein related to cell proliferation, and Vimentin, a marker protein related to cell migration, by immunofluorescence experiments. The results showed that NDRG1 knockdown effectively suppressed the expression levels of PCNA and Vimentin in SK-HEP1 and HCCLM3 cells (Fig. 7I-L). The above data indicate that NDRG1 expression is closely related to the proliferation, survival and migration of HCC cells and that knockdown of NDRG1 can effectively inhibit the proliferation, survival and migration of HCC cells.

NDRG1 Inhibition Induced HCC Cells Ferroptosis and Contribute to RLS3-Induced Ferroptosis

To explore the mechanism by which NDRG1 promotes the proliferation and metastasis of hepatoma cells, we detected ROS expression levels by flow cytometry. The results showed that the ROS expression levels of SK-HEP1 and LM3 cells were significantly up-regulated after knockdown of NDRG1 (Fig. 8A-B). We next detected ferroptosis by determining the amount of lipid peroxides in cellular membranes using BODIPY-C11 probe, and the results showed that the expression level of oxidative C11 was up-regulated after knockdown of NDRG1 (Fig. 8C-F). Therefore, we detected the expression levels of ferroptosis-related indicators (including MDA and iron ions), and the experimental results showed that after knocking down NDRG1, the expression level of MDA (Fig. 8G-H) and iron ions (Fig. 8J-K) in SK-HEP1 and LM3 cells were significantly increased. Previous studies have reported that ferroptosis is primarily characterized by cytological changes, including reduction or disappearance of mitochondrial cristae, rupture of the mitochondrial outer membrane, and mitochondrial membrane condensation. Our experimental results showed under electron microscope that after knockdown of NDRG1, mitochondrial morphology was significantly pyknotic in SK-HEP1 and LM3 cells (Fig. 8I). These above results suggest that NDRG1 expression is closely related to ferroptosis, and NDRG1 knockdown can induce ferroptosis in HCC cells. Ultimately, we treated SK-HEP1 and LM3 cells with different concentrations of RSL3-an inhibitor of glutathione peroxidase 4 (GPX4), and the results showed that knockdown of NDRG1 could induce ferroptosis in the RSL3 pathway (Fig. 8L-M). These data suggest that NDRG1 knockdown can induce ferroptosis in HCC cells and contribute to RLS3-induced ferroptosis.

Fig. 8figure 8

NDRG1 inhibition induced HCC cells ferroptosis and contribute to RLS3-induced ferroptosis. A-B FCM and quantification analysis showing increased ROS expression after NDRG1 knockdown in SK-HEP1 and LM3 cells. C-F C11-BODIPY (a marker of lipid peroxidation) probe and quantification analysis showing increasing oxidized ROS expression in SK-HEP1 (CE) and LM3 (DF) cells after knockdown of NDRG1. G-H Expression levels of MDA after NDRG1 knockdown in SK-HEP1 (G) and LM3 (H) cells. I Representative images of mitochondria within SK-HEP1 and LM3 cells under TEM after NDRG1 knockdown. J-K Expression levels of iron ions after knockdown of NDRG1 in SK-HEP1 (G) and LM3 (H) cells. L-M Effect of different concentrations of RSL3 treatment on cell viability of SK-HEP1 and LM3 cells after knockdown of NDRG1

The Regulatory Role of NDRG1 in the Growth and Tumor Metastasis of HCC

To further explore the regulatory role of NDRG1 in the growth and tumor metastasis of HCC, we constructed a xenograft tumor model of LM3 cells and a tail vein lung metastasis model. The results showed that after NDRG1 knockdown, the volume of tumors in LM3 cells was significantly reduced, the weight of the tumors was significantly reduced, and the growth rate of the tumors was significantly slowed (Fig. 9A-D). Immunohistochemical assays showed that knockdown of NDRG1 evidently inhibited the expression levels of Ki67, Vimentin and GPX4, which are proteins related to cell proliferation and migration in HCC tissues. Simultaneously, the expression levels of DHE staining was upregulated in HCC cells, which were closely related to cell ferroptosis (Fig. 9E). The above results showed that knockdown of NDRG1 obviously suppressed the tumorigenic ability of HCC cells in nude mice and induced the ferroptosis of HCC cells. The statistical results also revealed that knockdown of NDRG1 evidently inhibited Ki67, Vimentin and GPX4 expression and increased the level of DHE staining in HCC tissue (Fig. 9F-I). Next, we further explored the effect of NDRG1 expression on the lung metastasis ability of HCC cells through the tail vein lung metastasis model. The results demonstrated that compared with the control group, the tumor size and number of HCC lung metastases were markedly decreased after NDRG1 knockdown (Fig. 9J-M). This result also indicates that the expression of NDRG1 can promote the lung metastasis of HCC cells. These results suggest that NDRG1 functioned as a guardian against ferroptosis to drives tumorgenesis and metastasis in HCC.

Fig. 9figure 9

The regulatory role of NDRG1 in the growth and tumor metastasis of HCC. A-D A xenograft tumor model of LM3 cells showed that the volume and weight of the tumor were significantly reduced, and the growth rate of the tumor was significantly slowed after NDRG1 knockdown. E Immunohistochemical assays revealed that the expression levels of Ki67,Vimentin and GPX4 were inhibited and the expression levels of DHE was upregulated in HCC cells after NDRG1 knockdown. F-I The statistical results showed that knockdown of NDRG1 restricted the growth of HCC tumors. J-M We constructed a tail vein lung metastasis model, and the results of in vivo imaging (J), photography (K) and HE staining (L) demonstrated that the tumor size and number of HCC lung metastases were markedly decreased after NDRG1 knockdown

留言 (0)

沒有登入
gif