Long noncoding RNA BMPR1B-DT promotes anoikis and reduces proliferation in ovarian cancer.

We downloaded the transcriptome data and clinical data regarding the TCGA-OV ovarian cancer samples, organised and merged them, and then downloaded the GTEx ovarian normal sample data for merging, forming a gene expression file comprising data on 379 ovarian cancer samples and 88 ovarian normal tissue samples. We obtained anoikis-related genes from GeneCards and intersected the genes in the TCGA gene expression file with the anoikis-related genes downloaded from GeneCards to generate the anoikis gene expression file. We subsequently distinguished lncRNAs and mRNAs in the TCGA gene expression database to obtain the lncRNA expression dataset. We performed Pearson correlation analysis between the lncRNA expression matrix and the anoikis gene expression matrix, using a correlation coefficient absolute value > 0.3 and P   1 and P  < 0.05, resulting in 698 differentially expressed anoikis-related lncRNAs (DEGs) between tumour and normal samples, including 451 lncRNAs with upregulated expression and 247 lncRNAs with downregulated expression in tumour samples (Fig.  1 a). Fig. 1 Identification of anoikis‑related lncRNAs and construction of a prognostic model. (a) Volcano plot of differentially expressed anoikis-related lncRNAs between tumour tissues and normal tissues. Red points indicate lncRNAs with upregulated expression, green points indicate lncRNAs with downregulated expression, and black points represent lncRNAs with no significant difference in expression. The thresholds were |log2FC| >1 and P < 0.05. (b) Forest plot showing the HRs (95% CIs) of lncRNAs significantly associated with prognosis after univariate Cox regression analysis. (c) LASSO coefficient profiles of anoikis-related lncRNAs. The x-axis represents the log of the regularisation parameter (lambda), and the y-axis represents the partial likelihood deviance. The vertical dashed line indicates the optimal lambda value. (d) LASSO coefficient profiles of anoikis-related lncRNAs. The x-axis represents the log of the regularisation parameter (lambda), and the y-axis represents the coefficients of the lncRNAs. (e) Forest plot showing the HRs (95% CIs) of lncRNAs significantly associated with prognosis after multivariate Cox regression analysis Identification of anoikis‑related lncRNAs and construction of a prognostic model. (a) Volcano plot of differentially expressed anoikis-related lncRNAs between tumour tissues and normal tissues. Red points indicate lncRNAs with upregulated expression, green points indicate lncRNAs with downregulated expression, and black points represent lncRNAs with no significant difference in expression. The thresholds were |log2FC| >1 and P < 0.05. (b) Forest plot showing the HRs (95% CIs) of lncRNAs significantly associated with prognosis after univariate Cox regression analysis. (c) LASSO coefficient profiles of anoikis-related lncRNAs. The x-axis represents the log of the regularisation parameter (lambda), and the y-axis represents the partial likelihood deviance. The vertical dashed line indicates the optimal lambda value. (d) LASSO coefficient profiles of anoikis-related lncRNAs. The x-axis represents the log of the regularisation parameter (lambda), and the y-axis represents the coefficients of the lncRNAs. (e) Forest plot showing the HRs (95% CIs) of lncRNAs significantly associated with prognosis after multivariate Cox regression analysis To further investigate the prognostic potential of these lncRNAs, we conducted a univariate Cox regression analysis (Fig.  1 b). The forest plot in Fig.  1 b shows the hazard ratios (HRs) and 95% confidence intervals (CIs) of the lncRNAs significantly associated with prognosis. The dots represent the HR point estimates, and the horizontal lines represent the 95% CIs. The P values are shown on the right. A total of 32 lncRNAs were found to have significant prognostic potential in ovarian cancer patients. Next, we applied LASSO regression analysis to identify the most robust prognostic lncRNAs (Fig.  1 c and d). The LASSO coefficient profiles in Fig.  1 c and d show the shrinkage of coefficients as the regularisation parameter (lambda) increases. The vertical dashed line in Fig.  1 c indicates the optimal lambda value. The coefficients of the lncRNAs are plotted against the log of lambda in Fig.  1 d, demonstrating how the coefficients decrease to zero as lambda increases. This process helps to minimise overfitting and select the most significant lncRNAs. Finally, we performed multivariate Cox regression analysis to confirm the independent prognostic indicators (Fig.  1 e). The forest plot in Fig.  1 e shows the HRs and 95% CIs of the lncRNAs significantly associated with prognosis after adjusting for other variables. Nine lncRNAs were identified as independent prognostic indicators and were used to create a signature related to anoikis-associated lncRNAs. The risk score calculation is as follows: risk score = (6.838182772 × expression value of AC125611.3 ) + (−0.771330461 × expression value of AL022068.1 ) + (−0.393065532 × expression value of AL023583.1 ) + (−0.491162395 × expression value of USP30-AS1) + (−0.303570064 × expression value of AC007389.5 ) + (0.356454228 × expression value of AP005233.2 ) + (0.303840459 × expression value of AL713998.1 ) + (−0.702621563 × expression value of AC138904.1 ) + (−0.159825789 × expression value of BMPR1B-DT) (Table  1 ). Table 1 Multivariate Cox results of anoikis-related LncRNAs id Coef HR HR.95 L HR.95 H pvalue AC125611.3 6.838182772 2.792497348 2.624684008 3.839631113 0.000102209 AL022068.1 −0.771330461 0.462397457 0.29413546 0.726914764 0.000832359 AL023583.1 −0.393065532 0.674984513 0.482155795 0.94493128 0.022023581 USP30-AS1 −0.491162395 0.611914694 0.458625043 0.816439482 0.000842625 AC007389.5 −0.303570064 0.738178168 0.554746412 0.982263239 0.03727446 AP005233.2 0.356454228 1.428256155 1.066240745 1.913184854 0.016847684 AL713998.1 0.303840459 1.355052853 1.006221256 1.824815589 0.045413478 AC138904.1 −0.702621563 0.495285179 0.308826056 0.794322252 0.003552144 BMPR1B-DT −0.159825789 0.852292255 0.728321749 0.997364267 0.046275421 Coef the coefficient of lncRNAs correlated with survival, HR hazard ratio, HR.95 L low 95% CI of HR, HR.95 H high 95% CI of HR Multivariate Cox results of anoikis-related LncRNAs Coef the coefficient of lncRNAs correlated with survival, HR hazard ratio, HR.95 L low 95% CI of HR, HR.95 H high 95% CI of HR The risk scores of ovarian cancer patients were calculated on the basis of the expression levels of the nine lncRNAs identified from the TCGA cohort, and the patients were randomly divided into a training cohort and a validation cohort. Within each cohort, patients were further categorised into high-risk and low-risk groups on the basis of the median risk score derived from the training cohort and validation cohort. As shown in Fig.  2 a, risk scores were clearly stratified between high-risk (orange dots) and low-risk (blue dots) ovarian cancer patients in the training cohort, indicating distinct prognostic differences. Kaplan-Meier survival analysis (Fig.  2 b) revealed that high-risk patients had a significantly lower five-year overall survival (OS) rate than low-risk patients did ( P  = 6.66e-08), highlighting the robust predictive ability of the lncRNA signature. Figure  2 c shows that more deaths occurred in the high-risk group, while the low-risk group had more survivors, with consistently higher mortality in the high-risk group across the follow-up period. Figure  2 d presents the expression distributions of the nine lncRNAs (BMPR1B-DT, AC007389.5 , etc.) between the high-risk and low-risk groups. In the high-risk group, BMPR1B-DT expression was decreased. This finding is in line with its role as a protective factor against ovarian cancer and its ability to promote anoikis and reduce the proliferation of ovarian cancer cells. The receiver operating characteristic (ROC) curve analysis (Fig.  2 e) demonstrated strong predictive performance, with 1-year, 3-year, and 5-year area under the curve (AUC) values of 0.745, 0.737, and 0.793, respectively, confirming the model’s reliability for both short-term and long-term survival prediction. Fig. 2 Correlation between the predictive signature score and the prognosis of ovarian cancer patients in the training cohort. (a) Risk score distribution: The plot shows the risk scores of ovarian cancer patients in the training cohort. The orange dots represent high-risk patients, and the blue dots represent low-risk patients. The horizontal dashed line indicates the median risk score used as the cut-off for stratification. (b) Kaplan‒Meier survival analysis: survival curves comparing the high-risk (red) and low-risk (blue) groups. The p value indicates significant differences in overall survival between the groups. (c) Survival status distribution: the number of deceased (orange) and living (blue) patients at different time points. The tables on the right show the counts of deceased and living patients in each risk group over time. (d) Expression profiles of the lncRNAs: Heatmap showing the expression levels of the nine lncRNAs between the high-risk and low-risk groups. The colours range from blue (low expression) to orange (high expression). (e) ROC curve analysis: Receiver operating characteristic (ROC) curves for predicting 1-year, 3-year, and 5-year survival. The AUC values (0.745, 0.737, and 0.793, respectively) indicate the model’s predictive performance. OS, overall survival; ROC, receiver operating characteristic; AUC, area under the curve Correlation between the predictive signature score and the prognosis of ovarian cancer patients in the training cohort. (a) Risk score distribution: The plot shows the risk scores of ovarian cancer patients in the training cohort. The orange dots represent high-risk patients, and the blue dots represent low-risk patients. The horizontal dashed line indicates the median risk score used as the cut-off for stratification. (b) Kaplan‒Meier survival analysis: survival curves comparing the high-risk (red) and low-risk (blue) groups. The p value indicates significant differences in overall survival between the groups. (c) Survival status distribution: the number of deceased (orange) and living (blue) patients at different time points. The tables on the right show the counts of deceased and living patients in each risk group over time. (d) Expression profiles of the lncRNAs: Heatmap showing the expression levels of the nine lncRNAs between the high-risk and low-risk groups. The colours range from blue (low expression) to orange (high expression). (e) ROC curve analysis: Receiver operating characteristic (ROC) curves for predicting 1-year, 3-year, and 5-year survival. The AUC values (0.745, 0.737, and 0.793, respectively) indicate the model’s predictive performance. OS, overall survival; ROC, receiver operating characteristic; AUC, area under the curve In the validation cohort, the risk score (Fig.  3 a) also clearly distinguished high-risk patients from low-risk patients. Kaplan-Meier analysis (Fig.  3 b) confirmed worse OS in high-risk patients than in low-risk patients, providing external validation of the model’s predictive efficacy. The survival status distribution pattern (Fig.  3 c) was consistent with that of the training cohort, further supporting the generalisability of the model. BMPR1B-DT expression was decreased in the high-risk group (Fig.  3 d), reinforcing its potential as a protective marker. ROC curve analysis (Fig.  3 e) revealed acceptable predictive accuracy, with AUC values indicating the model’s ability to predict survival outcomes in an independent dataset. Fig. 3 Correlation between the predictive signature score and the prognoses of ovarian cancer patients in the validation cohort. a. Risk score distribution: The plot shows the risk scores of ovarian cancer patients in the validation cohort. The orange dots represent high-risk patients, and the blue dots represent low-risk patients. b. Kaplan‒Meier survival analysis: survival curves comparing the high-risk (red) and low-risk (blue) groups. The p value indicates significant differences in overall survival between the groups. c . Survival status distribution: the number of deceased (orange) and living (blue) patients at different time points. The tables on the right show the counts of deceased and living patients in each risk group over time. (d) Expression profiles of the lncRNAs: Heatmap showing the expression levels of the nine lncRNAs between the high-risk and low-risk groups. The colours range from blue (low expression) to orange (high expression). (e) ROC curve analysis: Receiver operating characteristic (ROC) curves for predicting 1-year, 3-year, and 5-year survival. The AUC values (0.656, 0.598, and 0.703, respectively) indicate the model’s predictive performance. OS, overall survival; ROC, receiver operating characteristic; AUC, area under the curve Correlation between the predictive signature score and the prognoses of ovarian cancer patients in the validation cohort. a. Risk score distribution: The plot shows the risk scores of ovarian cancer patients in the validation cohort. The orange dots represent high-risk patients, and the blue dots represent low-risk patients. b. Kaplan‒Meier survival analysis: survival curves comparing the high-risk (red) and low-risk (blue) groups. The p value indicates significant differences in overall survival between the groups. c . Survival status distribution: the number of deceased (orange) and living (blue) patients at different time points. The tables on the right show the counts of deceased and living patients in each risk group over time. (d) Expression profiles of the lncRNAs: Heatmap showing the expression levels of the nine lncRNAs between the high-risk and low-risk groups. The colours range from blue (low expression) to orange (high expression). (e) ROC curve analysis: Receiver operating characteristic (ROC) curves for predicting 1-year, 3-year, and 5-year survival. The AUC values (0.656, 0.598, and 0.703, respectively) indicate the model’s predictive performance. OS, overall survival; ROC, receiver operating characteristic; AUC, area under the curve First, a comprehensive search of public databases such as GeneCards and GEPIA revealed that only USP30-AS1 and BMPR1B-DT had prior records among the nine key anoikis-related lncRNAs. Further analysis via GEPIA revealed that USP30-AS1 was not significantly differentially expressed between ovarian cancer tissues and normal ovarian tissues (Figure S1a), whereas BMPR1B-DT expression was significantly different (Figure S1b). Moreover, when the expression levels across all cancer tissues were considered, USP30-AS1 was highly expressed in diffuse large B-cell lymphoma (DLBC), liver hepatocellular carcinoma (LIHC), and thymoma (THYM) tissues, with relatively low expression in ovarian cancer tissues (Figure S1c). In contrast, BMPR1B-DT was prominently expressed in prostate adenocarcinoma (PRAD), uterine corpus endometrial carcinoma (UCEC), and ovarian carcinoma (OV) tissues, with notably higher expression in ovarian cancer tissues than in other cancer tissues (Figure S1d). On the basis of these results, we selected BMPR1B-DT for further study. Among the nine key anoikis-related lncRNAs identified in ovarian cancer tissues, we focused on BMPR1B-DT. This particular lncRNA has been reported to exert a substantial influence on cancer progression [ 15 – 17 ]. The analysis conducted via RT‒qPCR to assess the differential expression of lncRNAs revealed that BMPR1B-DT expression was markedly elevated in in situ ovarian cancer tissues compared with benign ovarian tissues (Fig.  4 a). We measured BMPR1B-DT expression in six ovarian cancer cell lines—A2780, SKOV3, OV4, Hey, OV8, and Caov3—along with the standard ovarian cell line IOSE80 via RT‒qPCR. The findings indicated that the expression of BMPR1B-DT in the six ovarian cancer cell lines ranked as follows: A2780 had the highest expression, followed by SKOV3, OV4, Hey, OV8, and Caov3 (Fig.  4 b). Fig. 4 Expression levels of BMPR1B-DT. (a) RT‒qPCR analysis of tissues: The expression levels of BMPR1B-DT in ovarian carcinoma in situ (OCIS) and benign ovarian tissues (N) ( P  < 0.01). (b) RT‒qPCR analysis of cell lines: The expression levels of BMPR1B-DT in the normal human ovarian cell line IOSE-80 and human ovarian cancer cell lines OV4, OV8, Hey, A2780, SKOV-3, and CAOV-3. c , d. Relative expression in OV4 and Hey cells: relative expression levels of BMPR1B-DT in OV4 and Hey ovarian cancer cells in suspension culture compared with those in adherent culture. e , f. Relative expression in SKOV3 and A2780 cells: relative expression levels of BMPR1B-DT in SKOV3 and A2780 ovarian cancer cells in suspension culture compared with those in adherent culture. g , h. Overexpression efficiency: RT‒qPCR confirmation of BMPR1B-DT overexpression in OV4 and Hey ovarian cancer cells. Control: Cells that were not transfected with any vector. NC (Lv-NC): Cells were transfected with an empty vector. OE (Lv-BMPR1B-DT): Cells that were transfected with the overexpression vector. i , j Knockdown efficiency: RT‒qPCR confirmation of BMPR1B-DT expression knockdown in SKOV3 and A2780 ovarian cancer cells. Control: Cells that were not transfected with any vector. NC (Lv-sh-NC): Cells were transfected with a nontargeting shRNA vector. Sh-1 (Lv-sh-1-BMPR1B-DT), sh-2 (Lv-sh-2-BMPR1B-DT), and sh-3 (Lv-sh-3-BMPR1B-DT): Cells that were transfected with the knockdown vectors. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 Expression levels of BMPR1B-DT. (a) RT‒qPCR analysis of tissues: The expression levels of BMPR1B-DT in ovarian carcinoma in situ (OCIS) and benign ovarian tissues (N) ( P  < 0.01). (b) RT‒qPCR analysis of cell lines: The expression levels of BMPR1B-DT in the normal human ovarian cell line IOSE-80 and human ovarian cancer cell lines OV4, OV8, Hey, A2780, SKOV-3, and CAOV-3. c , d. Relative expression in OV4 and Hey cells: relative expression levels of BMPR1B-DT in OV4 and Hey ovarian cancer cells in suspension culture compared with those in adherent culture. e , f. Relative expression in SKOV3 and A2780 cells: relative expression levels of BMPR1B-DT in SKOV3 and A2780 ovarian cancer cells in suspension culture compared with those in adherent culture. g , h. Overexpression efficiency: RT‒qPCR confirmation of BMPR1B-DT overexpression in OV4 and Hey ovarian cancer cells. Control: Cells that were not transfected with any vector. NC (Lv-NC): Cells were transfected with an empty vector. OE (Lv-BMPR1B-DT): Cells that were transfected with the overexpression vector. i , j Knockdown efficiency: RT‒qPCR confirmation of BMPR1B-DT expression knockdown in SKOV3 and A2780 ovarian cancer cells. Control: Cells that were not transfected with any vector. NC (Lv-sh-NC): Cells were transfected with a nontargeting shRNA vector. Sh-1 (Lv-sh-1-BMPR1B-DT), sh-2 (Lv-sh-2-BMPR1B-DT), and sh-3 (Lv-sh-3-BMPR1B-DT): Cells that were transfected with the knockdown vectors. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 We used ultralow-attachment plates for suspension culture to mimic the effects of anoikis [ 13 , 14 ]. We cultured OV4, Hey SKOV3, and A2780 cells in suspension and adherent cultures for 48 h. We used RT‒qPCR to determine changes in the expression of BMPR1B-DT after culture. The results revealed that, compared with that in adherent cultures, BMPR1B-DT expression in suspension cultures of OV4 and Hey ovarian cancer cells decreased after 48 h. Conversely, it increased in SKOV3 and A2780 ovarian cancer cells under suspension culture conditions after 48 h compared with that in adherent cultures (Fig.  4 c, d, e, f). We subsequently used Lv-BMPR1B-DT to induce the overexpression of BMPR1B-DT in the OV4 and Hey cell lines. Control cells were not transfected with any vector. Cells in the Lv-NC group were transfected with an empty vector. The cells in the Lv-BMPR1B-DT group were transfected with the overexpression vector. The efficiency of BMPR1B-DT overexpression was validated through RT‒qPCR analysis (Fig.  4 g, h). CCK-8 assay results indicated that after 4 days of suspension culture, OV4 and Hey cells overexpressing BMPR1B-DT exhibited significantly decreased proliferation (Fig.  5 a, b). The OV4 and Hey cells were divided into 4 groups: the adh group underwent adherent culture for 48 h and the sus, Lv-BMPR1B-DT and Lv-NC groups underwent suspension culture for 48 h. Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in OV4 and Hey cells cultured in suspension for 48 h than in those subjected to adhesion culture, and the expression levels of the antiapoptotic protein bcl-2 were lower, indicating that the cells were more likely to undergo apoptosis after suspension culture than after adhesion culture (Fig.  5 c, d, f, g). After 48 h of culture, the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were elevated in the Lv-BMPR1B-DT group compared with those in the Lv-NC group. Conversely, the expression levels of the antiapoptotic protein bcl-2 were reduced (Fig.  5 c, e, f, h). These findings suggest that BMPR1B-DT overexpression promotes the apoptosis of OV4 and Hey cells. After suspension culture for 48 h, the apoptosis rates of the OV4 and Hey cells in the Lv-BMPR1B-DT and Lv-NC groups were assessed via flow cytometry. The findings demonstrated that the overexpression of BMPR1B-DT in OV4 and Hey cells significantly increased the percentage of apoptotic cells in suspension culture (Fig.  5 i-n). These findings indicate that increased expression of BMPR1B-DT promotes anoikis and decreases proliferation in ovarian cancer cells during suspension culture. Fig. 5 The overexpression of BMPR1B-DT promoted anoikis and decreased proliferation of OV4 and Hey cells during suspension culture. a , b CCK-8 assays revealed that the proliferation of OV4 and Hey cells overexpressing BMPR1B-DT decreased significantly during suspension culture. c , d , f , g Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in OV4 and Hey cells cultured in suspension for 48 h than in those cultured in adherent media, and the expression levels of the antiapoptotic protein bcl-2 were lower. c , e , f , h Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in the Lv-BMPR1B-DT group than in the Lv-NC group and that the expression levels of the antiapoptotic protein bcl-2 were lower. i-n Flow cytometry revealed that the overexpression of BMPR1B-DT in OV4 and Hey cells significantly increased the percentage of apoptotic cells after 48 h suspension culture. Control: Cells that were not transfected with any vector. NC (Lv-NC): Cells were transfected with an empty vector. OE (Lv-BMPR1B-DT): Cells that were transfected with the overexpression vector. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 The overexpression of BMPR1B-DT promoted anoikis and decreased proliferation of OV4 and Hey cells during suspension culture. a , b CCK-8 assays revealed that the proliferation of OV4 and Hey cells overexpressing BMPR1B-DT decreased significantly during suspension culture. c , d , f , g Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in OV4 and Hey cells cultured in suspension for 48 h than in those cultured in adherent media, and the expression levels of the antiapoptotic protein bcl-2 were lower. c , e , f , h Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in the Lv-BMPR1B-DT group than in the Lv-NC group and that the expression levels of the antiapoptotic protein bcl-2 were lower. i-n Flow cytometry revealed that the overexpression of BMPR1B-DT in OV4 and Hey cells significantly increased the percentage of apoptotic cells after 48 h suspension culture. Control: Cells that were not transfected with any vector. NC (Lv-NC): Cells were transfected with an empty vector. OE (Lv-BMPR1B-DT): Cells that were transfected with the overexpression vector. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 We used Lv-sh-BMPR1B-DT to knock down BMPR1B-DT expression in SKOV3 and A2780 cells. The cells in the control group were not transfected with any vector. NC (Lv-sh-NC): Cells in the Lv-sh-NC group were transfected with a nontargeting shRNA vector. The cells transfected with the knockdown vectors (Lv-sh-1-BMPR1B-DT, Lv-sh-2-BMPR1B-DT, and Lv-sh-3-BMPR1B-DT) were divided into the sh-1 (Lv-sh-1-BMPR1B-DT) group, the sh-2 (Lv-sh-2-BMPR1B-DT) group, and the sh-3 (Lv-sh-3-BMPR1B-DT) group. The effectiveness of BMPR1B-DT expression knockdown was validated through RT‒qPCR analysis (Fig.  4 i, j). The results revealed that Lv-sh-1-BMPR1B-DT and Lv-sh-2-BMPR1B-DT effectively knocked down the expression of BMPR1B-DT. Therefore, we selected Lv-sh-1-BMPR1B-DT and Lv-sh-2-BMPR1B-DT for subsequent experiments. CCK-8 assays revealed a marked increase in the proliferative capacity of SKOV3 and A2780 cells following the knockdown of BMPR1B-DT expression over a four-day period of suspension culture (Fig.  6 a, b). SKOV3 and A2780 cells were divided into 5 groups: the adh group underwent adherent culture for 48 h; the sus group, the Lv-sh-1-BMPR1B-DT group, the Lv-sh-2-BMPR1B-DT group and the Lv-sh-NC group underwent suspension culture for 48 h. Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were greater in SKOV3 and A2780 cells cultured in suspension for 48 h than in those in adherent culture. Conversely, the expression of the antiapoptotic protein bcl-2 was reduced (Fig.  6 c, d, f, g). After 48 h of culture, Western blotting revealed that the expression of the proapoptotic proteins cleaved caspase-3 and bax were lower in the Lv-sh-1-BMPR1B-DT group and Lv-sh-2-BMPR1B-DT group than in the Lv-sh-NC group. Conversely, the expression levels of the antiapoptotic protein bcl-2 were elevated (Fig.  6 c, e, f, h). These results indicate that knockdown of BMPR1B-DT expression leads to a reduction in apoptosis in the SKOV3 and A2780 cell lines. After suspension culture for 48 h, the rates of apoptosis in the Lv-sh-1-BMPR1B-DT and Lv-sh-NC groups of SKOV3 and A2780 cells were assessed using flow cytometry. The results revealed that the reduction in BMPR1B-DT expression in the SKOV3 and A2780 cell lines led to a notable decrease in apoptosis rates following 48 h suspension culture (Fig.  6 i-n). These results suggest that the knockdown of BMPR1B-DT expression decreases anoikis and promotes proliferation during suspension culture of ovarian cancer cells. Fig. 6 Knockdown of BMPR1B-DT expression decreases anoikis and promotes proliferation of SKOV3 and A2780 cells during suspension culture. a , b CCK-8 assays revealed that the proliferation of SKOV3 and A2780 cells with BMPR1B-DT expression knocked down increased significantly during suspension culture. c , d , f , g Western blot analysis revealed that the protein expression of cleaved caspase-3 and bax were greater in SKOV3 and A2780 cells cultured in suspension for 48 h than in those in adhesion culture, and the expression levels of the antiapoptotic protein bcl-2 were lower. c , e , f , h Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were lower in the Lv-sh-1-BMPR1B-DT group and Lv-sh-2-BMPR1B-DT group than in the Lv-sh-NC group, and the expression levels of the antiapoptotic protein bcl-2 were greater. i-n Flow cytometry revealed that the knockdown of BMPR1B-DT expression in SKOV3 and A2780 cells significantly decreased the percentage of apoptotic cells after 48 h suspension culture. Control: Cells that were not transfected with any vector. NC (Lv-sh-NC): Cells were transfected with a nontargeting shRNA vector. sh1 (Lv-sh-1-BMPR1B-DT), sh2 (Lv-sh-2-BMPR1B-DT), and sh3 (Lv-sh-3-BMPR1B-DT): Cells that were transfected with the knockdown vectors. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001 Knockdown of BMPR1B-DT expression decreases anoikis and promotes proliferation of SKOV3 and A2780 cells during suspension culture. a , b CCK-8 assays revealed that the proliferation of SKOV3 and A2780 cells with BMPR1B-DT expression knocked down increased significantly during suspension culture. c , d , f , g Western blot analysis revealed that the protein expression of cleaved caspase-3 and bax were greater in SKOV3 and A2780 cells cultured in suspension for 48 h than in those in adhesion culture, and the expression levels of the antiapoptotic protein bcl-2 were lower. c , e , f , h Western blot analysis revealed that the expression levels of the proapoptotic proteins cleaved caspase-3 and bax were lower in the Lv-sh-1-BMPR1B-DT group and Lv-sh-2-BMPR1B-DT group than in the Lv-sh-NC group, and the expression levels of the antiapoptotic protein bcl-2 were greater. i-n Flow cytometry revealed that the knockdown of BMPR1B-DT expression in SKOV3 and A2780 cells significantly decreased the percentage of apoptotic cells after 48 h suspension culture. Control: Cells that were not transfected with any vector. NC (Lv-sh-NC): Cells were transfected with a nontargeting shRNA vector. sh1 (Lv-sh-1-BMPR1B-DT), sh2 (Lv-sh-2-BMPR1B-DT), and sh3 (Lv-sh-3-BMPR1B-DT): Cells that were transfected with the knockdown vectors. * P  < 0.05, ** P  < 0.01, *** P  < 0.001, **** P  < 0.0001

RNA sequencing (RNA-seq) data and clinical information regarding ovarian cancer specimens were obtained from the TCGA database ( https://portal.gdc.cancer.gov/ ). The detailed steps for data acquisition were as follows: access the TCGA database; select the Genomic Data Commons Data Portal-Repository page; and set the following search conditions: (1) in the case project, select ovaries for the primary site; TCGA for the program; TCGA-OV for the project; and cystic, mucinous and serous neoplasms for the disease type; (2) in the files project, select transcriptome profiling for the data category; gene expression quantification for the data type; RNA-Seq for the experimental strategy; and HTSeq-Counts for the workflow type. After the restrictions are set, click to enter the data download page and download the Manifest, Cart, metadata, and clinical files. The ovarian cancer transcriptome expression matrix was obtained via R software data cleaning. Since the TCGA database lacks data regarding normal ovarian tissues, the GTEx database was chosen to obtain gene sequencing data on normal ovarian tissues. The specific steps were as follows: access the GTEx database, select to download the gene expression matrix (Datasets-›Data Download - GTEx Analysis V8 - sGene read counts), and finally filter the normal ovarian tissue transcriptome expression data. Anoikis-associated genes were obtained from the GeneCards database as follows: access the official GeneCards website at https://www.genecards.org , search using the term “anoikis,” and then export and download the relevant genes. In this study, the ovarian cancer transcriptome expression matrix and the normal ovarian tissue transcriptome expression data were merged, forming a gene expression file with data on 379 ovarian cancer samples and 88 normal ovarian tissue samples. Anoikis-related genes were obtained from GeneCards. The intersection of TCGA gene expression data and these anoikis-related genes was used to obtain the anoikis gene expression profile. The TCGA gene expression data were divided into lncRNA and mRNA expression files. Pearson correlation analysis was performed between the lncRNA and anoikis gene expression matrices, with a cut-off absolute correlation coefficient > 0.3 and P   1 and P  < 0.05. Initially, the prognostic significance of lncRNAs associated with anoikis was assessed using univariate Cox regression analysis. Second, lncRNAs associated with anoikis that exhibited a significance level of P  < 0.05 in the univariate analysis were included in the least absolute shrinkage and selection operator (LASSO) regression analysis. LASSO regression is a data mining technique that incorporates a penalty function into standard multiple linear regression. This approach consistently diminishes the coefficients to simplify the model and prevent issues such as multicollinearity and overfitting. Variables with coefficients equal to zero corresponding to the optimal lambda value are excluded, and the remaining variables are included in the multivariate Cox regression analysis to calculate a risk score. Risk scores were computed for every patient by utilising the expression levels of lncRNAs via the established formula: [ 9 ] Risk score = exp₁ * β₁ + exp₂ * β₂ + … + exp i * β i . In this formula, exp i denotes the expression level of each lncRNA, and β i signifies the regression coefficient derived from the multivariate Cox regression analysis corresponding to that specific lncRNA. In our study, to ensure the robustness and generalisability of the model, we utilised the createDataPartition function in R software. By setting p  = 0.5, we ensured an equal distribution of the data into training and validation datasets. This helps prevent model overfitting and enhances model performance with new data [ 10 , 11 ]. The training and validation cohorts were hen divided into high-risk and low-risk groups on the basis of the median risk score. The survival outcomes of these groups were analysed via Kaplan‒Meier survival analyses to assess differences. Additionally, receiver operating characteristic (ROC) curves were generated to evaluate the model’s accuracy. GEPIA (Gene Expression Profiling Interactive Analysis) ( http://gepia.cancer-pku.cn/ ), a web-based tool grounded in TCGA and GTEx data, offers vital functions such as differential expression and survival analyses, strongly facilitating data mining and scientific research [ 12 ]. We analysed 9 crucial lncRNAs using the GEPIA database and selected BMPR1B-DT for further study. Five pathologically confirmed patients with high-grade serous ovarian cancer (HGSOC) were recruited in 2023 from the Department of Gynaecology of Guangdong Provincial People’s Hospital. All ovarian cancer tissues were confirmed to be HGSOC by pathological examination to reduce the potential impact on the results of tissue heterogeneity. Four noncancerous ovarian tissues were obtained from patients who had undergone hysterectomy and bilateral oophorectomy due to benign uterine diseases, such as uterine fibroids and adenomyosis. The investigation received approval from the Ethics Committee of Guangdong Provincial People’s Hospital (Approval No. KY2023-630-01). Cells were cultured in complete DMEM composed of 90% DMEM, 10% foetal bovine serum, and 1% penicillin‒streptomycin. The cells were cultured in a thermostatic incubator at 37 °C in the presence of 5% CO2. Anoikis can be induced in vitro by transferring epithelial cells from traditional adhesive culture surfaces, which are hydrophilic and promote cell adhesion, to ultralow attachment (ULA) plates (3471, Corning, USA). These plates are coated with a hydrogel layer that strongly hinders cell adhesion [ 13 , 14 ]. The following cells were obtained from Professor Shanyang He (Department of Gynaecology, Guangdong Provincial People’s Hospital, Guangdong Academy of Medical Sciences, Southern Medical University, China): the normal human ovarian cell line IOSE-80 and various human ovarian cancer cell lines: OV4, OV8, Hey, A2780, SKOV-3, and CAOV-3. Lentiviral vectors (Table S1) (Lv-BMPR1B-DT for overexpression; Lv-sh-1/2/3-BMPR1B-DT for silencing) were transfected into cells in 6-well plates (50,000 cells/well) at 70–80% confluence using Lipofectamine 3000 (L3000001; Invitrogen, USA) and Opti-MEM (31985–070; Gibco, USA) following the manufacturer’s protocol. After a 20-min incubation at room temperature, the transfection mixture was added to the cells at 37 °C in a 5% CO₂ incubator. The medium was replaced with fresh complete medium after 6 h, and the cells were harvested 48 h posttransfection. Transfected cells were selected with puromycin (P8230-25 mg, Solarbio) at 2 µg/mL for 7 days, and the medium was changed every 2 days. The surviving cells were used for further experiments. The degree of cell apoptosis was assessed via an Annexin V-FITC/PI double-stained cell apoptosis detection kit (BL110A, Biosharp, China). In summary, the concentration of cells was modified to 1 × 10 6 cells/mL. The cells were then incubated in the dark with FITC for 10 min and propidium iodide for 5 min, and the apoptosis index was evaluated via flow cytometry. For the detachment-induced apoptosis assay, the cells were cultured on ultralow attachment (ULA) plates for 48 h prior to analysis. Total cellular protein was extracted via RIPA Complete Lysis Buffer (P0039, Beyotime, China), separated via SDS‒PAGE and subsequently transferred onto PVDF membranes (ISEQ00010‒N1, Merck Millipore, USA). The membranes were blocked with protein-free rapid blocking buffer (1×) (PS108P, EpiZyme, China) and then incubated with primary antibodies overnight at 4 °C. The primary antibodies used were as follows: Caspase-3 (66470-2-Ig, Proteintech, USA), Bcl-2 (68103-1-Ig, Proteintech, USA), Bax (50599-2-Ig, Proteintech, USA), and GAPDH (10494-1-AP, Proteintech, USA). This was followed by incubation with the relevant secondary antibodies. Protein band visualisation was accomplished using a GE chemiluminescence imaging system (LAS500). RNA extraction was conducted utilising a RNA Pure Tissue & Cell Kit (CW0584S, CWBIO, China). Subsequently, complementary DNA (cDNA) was generated from total RNA via a HiFiScript gDNA Removal MasterMix Kit (CW 2020 M, CWBIO, China). Quantitative real-time polymerase chain reaction (qPCR) was performed with a Taq Pro Universal SYBR qPCR Master Mix Kit (Q712-02, Vazyme, China). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as the internal control. GAPDH is of human origin. The PCR primers were synthesised by Sangon Biotech (China), and the specific primer sequences are detailed in Table S2. Cell proliferation was assessed utilising a Cell Counting Kit-8 (CCK-8) (BS350B, Biosharp, China). In summary, 5 × 10 4 cells were seeded into each well of a 96-well ultralow attachment plate and incubated at 37 °C in an atmosphere containing 5% CO2. At designated time intervals, 10 µl of CCK-8 solution was added to each well, followed by a 1-h incubation at 37 °C. The cell viability was subsequently evaluated using a microplate reader(Multiskan GO, Thermo Fisher Scientific, Waltham, MA, USA)at a wavelength of 450 nm. We employed Pearson correlation analysis to evaluate the associations between two distinct parameters. The Kaplan‒Meier method was used to generate survival curves, and the log-rank test was applied for comparative analysis. To assess the prognostic relevance of the anoikis-related lncRNA signature, we implemented both Cox regression and LASSO regression techniques. A t test was conducted to examine the differences between the two groups. All experiments were conducted with a minimum of three independent repetitions. The data are expressed as the means ± standard deviations (SDs). Statistical significance was established at P  < 0.05, with thresholds defined as follows: * P  < 0.05, ** P  < 0.01, *** P  < 0.001, and **** P  < 0.0001. Statistical analyses were performed using R software (version 3.6) and GraphPad Prism 10 (San Diego, CA, USA).

Anoikis, a specific type of programmed cell death that occurs when cells detach from the ECM, is essential in the progression of ovarian cancer. The ability to resist anoikis is a defining characteristic of metastatic cancer cells. It enables these cells to survive and spread in nonadherent environments, such as the peritoneal cavity in ovarian cancer patients. The complex interplay of various molecular mechanisms underlying anoikis resistance in ovarian cancer patients underscores the need for continued research in this area [ 18 – 23 ]. Therefore, by targeting anoikis, novel therapeutic strategies could be developed to improve outcomes in ovarian cancer patients. In this study, we identified an anoikis-related lncRNA signature in ovarian cancer patients and developed a prognostic model. Our findings revealed that the lncRNA BMPR1B-DT is overexpressed in ovarian cancer tissues and promotes anoikis, indicating its strong potential as a therapeutic target. The development of a prognostic prediction model for ovarian cancer patients is essential for addressing the challenges posed by late-stage detection and subsequent treatment inefficacies. Anoikis resistance is a critical feature of tumour aggressiveness [ 24 – 26 ], chemotherapy resistance [ 27 ] and metastasis [ 18 , 21 , 28 – 31 ]. Therefore, investigating anoikis-related lncRNAs may provide new therapeutic and prognostic prediction strategies for ovarian cancer patients. In our study, we successfully constructed a lncRNA model based on anoikis for ovarian cancer, which not only incorporates risk scores but also predicts the prognoses of ovarian cancer patients. This approach is innovative because it systematically screens lncRNAs associated with anoikis apoptosis using transcriptomic data from public databases, offering new insights for personalised ovarian cancer treatment. We also separately tested the model’s predictive performance in both the validation and training sets, yielding satisfactory results. Compared with traditional retrospective clinical data analysis, which considers clinical parameters, our model includes expression information on anoikis-related lncRNAs, rendering the prognosis assessment more comprehensive and accurate. The continuous evolution of our models aims not only to enhance early detection but also to inform treatment decisions, ultimately improving patient outcomes in patients with ovarian cancer. The observation that five-year survival exhibited the best AUC in the ROC analysis for the validation cohort suggests that the anoikis-related lncRNAs we identified have greater predictive power for long-term outcomes in ovarian cancer patients. This finding implies that these lncRNAs may play a significant role in the biological processes that influence late-onset relapse or long-term survival, such as resistance to anoikis, metastatic potential, or dormancy of residual cancer cells. The superior predictive performance at the five-year mark could indicate that anoikis-related lncRNAs are more strongly associated with mechanisms that contribute to long-term survival or delayed recurrence. For example, these lncRNAs might regulate pathways involved in cellular dormancy, immune evasion, or adaptation to the metastatic niche [ 32 ], which are critical for the survival of disseminated tumour cells over extended periods. This aligns with the clinical behaviour of ovarian cancer, in which late relapses are not uncommon and long-term surveillance is crucial. These findings highlight the potential utility of anoikis-related lncRNAs as biomarkers for identifying patients at greater risk of late relapse, who may benefit from prolonged or tailored follow-up strategies. Additionally, targeting these lncRNAs or their associated pathways could offer new therapeutic avenues for preventing or treating late-onset recurrence in patients with ovarian cancer. LncRNAs are increasingly recognised for their roles in gynaecological malignancies, particularly in the regulation of apoptosis [ 6 ]. For example, lncRNA-ATB and lncRNA MIAT are implicated in enhancing cellular proliferation and suppressing apoptotic processes in ovarian cancer patients [ 7 , 8 ]. However, there has been no research on the effect of the lncRNA BMPR1B-DT on anoikis in ovarian cancer patients. The lncRNA BMPR1B-DT, also known as BMPR1B-AS1 (Ensembl: ENSG00000249599, Chr4: 94743668–94757681), is located on chromosome 4 near the BMPR1B gene [ 33 ]. Despite the ambiguity surrounding its mechanism of action, it is categorised as a natural antisense transcript (NAT). BMPR1B encodes a BMP receptor, a transmembrane serine/threonine kinase involved in BMP signalling, which is crucial for endochondral bone formation and embryogenesis and functions by forming heteromeric complexes with serine/threonine kinase receptors [ 34 , 35 ]. A study utilising the TCGA and GTEx RNA-seq databases identified BMPR1B-DT among differentially expressed lncRNAs associated with OC, establishing a prognostic signature predictive of survival percentages at one, three, and five years, offering both a tool for prognosis assessment and potential therapeutic targets. Furthermore, the role of BMPR1B-DT in the lncRNA‒microRNA‒mRNA regulatory network suggests its significant impact on OC pathogenesis, warranting further investigation [ 15 ]. Some studies regarding uterine corpus endometrial carcinoma [ 16 , 17 ] related to BMPR1B-DT have constructed prognostic models that include BMPR1B-DT. In these models, BMPR1B-DT consistently acts as a protective factor (with a hazard ratio of less than 1), and patients with high expression of BMPR1B-DT have better outcomes than those with low expression. The findings from these prognostic models align with the results of our study. However, few in-depth functional and mechanistic studies regarding BMPR1B-DT exist. One study confirmed that BMPR1B-DT expression enhances the proliferation and metastatic potential of endometrial cancer cells. This occurs via sponging of miR-7-2-3p, which modulates the DCLK1/Akt/NF-κB pathway [ 36 ]. A study of endometriosis patients revealed that BMPR1B-DT enhances the process of decidualisation by engaging with the RNA-binding protein IGF2BP2. The suppression of IGF2BP2 results in decreased stability of BMPR1B-DT and impedes the activation of the SMAD1/5/9 signalling cascade, which in turn reduces the decidualisation of human endometrial stromal cells (hESCs) [ 37 ]. In contrast, few investigations have examined the involvement of the lncRNA BMPR1B-DT in various cancers. Moreover, studies focusing on its functions and underlying mechanisms in the context of ovarian cancer anoikis remain notably insufficient. This investigation highlights the substantial involvement of the lncRNA BMPR1B-DT in the process of anoikis in ovarian cancer patients, thereby addressing a notable gap in current research. Our analysis of the expression levels of BMPR1B-DT in both ovarian cancer tissues and benign ovarian tissues revealed markedly elevated expression of BMPR1B-DT within the ovarian cancer tissue samples, suggesting its close association with the occurrence and progression of the disease. Further experimental validation revealed that the expression levels of BMPR1B-DT were markedly elevated in ovarian cancer cells compared with normal ovarian cells. These findings lay the groundwork for our exploration of the function of BMPR1B-DT within the TME. We also discovered that BMPR1B-DT expression effectively promoted anoikis and significantly reduced the proliferation of ovarian cancer cells under suspension culture conditions. These findings not only reveal the potential of BMPR1B-DT as a biomarker but also present novel therapeutic targets for the development of treatment strategies for ovarian cancer. The limitations of this research include reliance on publicly available data, which may introduce variability due to differences in datasets and patient demographics. Furthermore, the mechanism of the lncRNA BMPR1B-DT in anoikis remains underexplored, necessitating further functional studies to validate its biological significance and therapeutic potential. Addressing these gaps will increase the robustness of its application in clinical scenarios for ovarian cancer management. We acknowledge a limitation of our study regarding the small sample size, which included only five ovarian cancer tissues and four noncancer tissues. This may restrict the statistical power and universality of the results. Therefore, our findings require further verification in a larger sample. All ovarian cancer tissues were confirmed to be HGSOC by pathological examination to reduce the potential impact of tissue heterogeneity on the results.

In summary, we constructed an anoikis-related lncRNA prognostic model and demonstrated that it can accurately predict ovarian cancer prognosis, with higher risk scores indicating a worse prognosis for women with ovarian cancer. Additionally, our study revealed that expression of the lncRNA BMPR1B-DT is upregulated in ovarian cancer tissues and cell lines. Moreover, our experiments revealed that BMPR1B-DT promoted anoikis and inhibited proliferation under suspension culture conditions in ovarian cancer cells. These findings emphasise the potential of BMPR1B-DT as a therapeutic target, shedding light on new avenues for the diagnosis and management of ovarian cancer.

Ovarian cancer (OC) ranks among the most common malignant tumours affecting women because of its exceptionally high rates of recurrence and death. Approximately 70% of individuals diagnosed with OC are diagnosed at later stages of the illness [ 1 ]. Furthermore, 50–70% of these patients experience relapse within two years, with only a 30% five-year survival rate posttreatment [ 2 ]. High rates of metastasis are one of the reasons for the high lethality of OC. Peritoneal metastasis is a complex process that includes detachment from the primary tumour, anti-anoikis (anchorage-dependent cell death), multicellular aggregation, peritoneal implantation, and targeted organ implantation [ 3 , 4 ]. Anoikis represents a specific type of programmed cell death, or apoptosis, that is triggered when cells lose their attachment to the extracellular matrix (ECM). It primarily serves to prevent abnormal cell growth and adhesion to inappropriate ECM. The ability to evade anoikis is a defining feature of tumour metastasis, allowing cancer cells to withstand programmed cell death and persist even after losing their ECM attachment alongside cellular junctions and to regain the ability to attach for diffusion, metastasis and invasion [ 5 ]. Long noncoding RNAs (lncRNAs) are a category of RNA molecules that are longer than 200 nucleotides and lack protein-coding capabilities. They are crucial in modulating various biological processes, including cell growth, programmed cell death, and cellular movement and metastasis. However, the specific roles of lncRNAs in the OC anoikis process are still unclear. Some studies have revealed associations between lncRNAs and apoptosis, such as the lncRNAs ATB and MIAT, which promote the occurrence and progression of OC through their influence on cell growth and programmed cell death [ 6 – 8 ]. The mechanisms by which lncRNAs affect OC anoikis still require further exploration. Consequently, this study intends to explore in greater depth the particular functions of lncRNAs in the process of anoikis associated with OC, revealing its potential mechanisms in tumour progression.

Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3