Lnc408 is highly expressed in BCSCs
Our previous study revealed that epithelial–mesenchymal transition (EMT) leads to enhanced stemness characteristic of breast cancer cells21. Using lncRNA array, we identified a series of dysregulated lncRNAs in BCSCs, 21 of which were stemness-associated lncRNAs (Fig. 1A), and some of these lncRNAs were further validated by qRT-PCR (Fig. S1A). Among these lncRNAs, ENST00000422408, termed lnc408 here after, was one of the most highly expressed lncRNAs in BCSCs. To confirm the relationship between lnc408 expression and BCSCs, lnc408 levels were evaluated in CSCs derived from various breast cancer cells, and we found that higher level of lnc408 was in EMT breast cancer cell lines (MDA-MB-468, Hs578T, MDA-MB-231, and BT549) than that in non-EMT cell lines (SKBr3, MDA-MB-453, MCF-7, and T47D). In addition, lnc408 expression was positively correlated with the protein level of stemness marker (e.g., CD44 and SOX2) in both breast cancer cells and clinical samples (Fig. S1B and Fig. 1C).
CD44+/CD24− has been widely used as BCSC surface markers, thus we sorted CD44+/CD24− subpopulation (termed as BCSCs) from various breast cancer cells and clinical tumor samples, and other remaining subpopulations (CD44+/CD24+, CD44−/CD24+, CD44−/CD24−) were termed as non-BCSCs. The higher level of lnc408 was further confirmed in these CD44+/CD24− subpopulation compared with their non-BCSCs (Figs. 1D, E). Moreover, the gradually elevated lnc408 was observed in the BCSC spheres from first to third generation (Fig. 1F). Kaplan–Meier survival curves showed that patients with high expression of lnc408 had a shorter overall survival than those with low expression of lnc408 (Fig. S1C). These data demonstrate that lnc408 is highly expressed in BCSCs.
Lnc408 is required for self-renewal maintenance of BCSCs
To determine the role of lnc408 in BCSC self-renewal, we silenced lnc408 by separately transfecting two lentivirus-mediated shRNAs into breast cancer cells and primary breast cancer cells derived from clinical samples (Fig. 2A and Fig. S2A). Lnc408 knockdown dramatically reduced sphere formation of breast cancer cells and primary breast cancer cells (Fig. 2B and Fig. S2B). Moreover, the mRNA (Fig. 2C and Fig. S2C) and protein (Fig. 2D and Fig. S2D) levels of representative breast cancer stemness markers, including SOX2, Nanog, and CD44 were notably decreased in the lnc408-knocked down cells compared to their scrambled control cells (sh Ctrl). In line with this, lnc408 knockdown attenuated CD44 and c-Myc expressions in the BCSC spheres checked by IF staining (Fig. S2E). Subsequently, we explored the effects of lnc408 on tumor-initiating capacity using lnc408 wild-type and knocked down primary breast cancer cells isolated from breast cancer tumors, which were subcutaneously injected into mammary fat pad of nude mice. Lnc408 knockdown led to the significant reduce of tumor burden compared with the control primary cancer cells, as detected by a limiting dilution xenograft analysis (Fig. 2E). Furthermore, lnc408 knockdown apparently decreased tumorigenic cell frequency (Fig. 2F). Taken together, these data indicate that lnc408 is required for the maintenance of self-renewal capability of BCSCs.
Lnc408 endows non-BCSCs to acquire stemness property
To further clarify the relationship between lnc408 and BCSCs stemness, ectopic lnc408 was stably transfected into non-CSC subpopulation derived from breast cancer cell lines and primary cancer cells (Fig. 3A and Fig. S3A). Compared with non-BCSCs, lnc408 overexpression can restore the non-BCSCs with the sphere formation abilities (Fig. 3B and Fig. S3B). Correspondingly, ectopic lnc408 obviously augmented both mRNA (Fig. 3C and Fig. S3C) and protein (Fig. 3D and Fig. S3D) expressions of breast cancer stemness markers (e.g., SOX2, Nanog, and CD44). Moreover, the effect of ectopic lnc408 on tumorigenesis of non-BCSC population was assessed using limiting dilution experiments in vivo. As few as 1 × 103 of lnc408-overexpressing BT549 cells was sufficient for tumor initiation (2/10 vs 0/10), 1 × 104 of lnc408-overexpressing BT549 cells led to more tumor initiation (4/10 vs 1/10), and injection of 1 × 105 of lnc408-overexpressing BT549 cells resulted in tumorigenesis in most of mice in comparison of non-BCSC cells (8/10 vs 2/10; Fig. 3E), suggesting that lnc408 can enable non-CSCs to acquire stemness characteristic in breast cancer.
Lnc408 suppresses the transcription of CBY1 to maintain BCSCs self-renewal
Next, we asked how lnc408 involves in stemness maintenance of BCSC. It has been reported that lncRNAs can act in cis-regulatory effects to regulate neighboring gene expression25. We found that lnc408 was an antisense lncRNA composed of two exons and spanning 1.08 kilobases (kb), and locating on chromosome 22 near to chibby 1 (CBY1) gene (Fig. 4A). Furthermore, lnc408 was confirmed to be a lncRNA with no coding potentiality (Fig. S4A and Fig. S4B), and mainly distributed in nuclei of breast cancer cells checked by RNA-FISH assay (Fig. 4B).
Since lncRNAs could exert their functions through interacting with proteins to regulate gene expression in the nuclei26, an RNA pull-down assay with biotin-labeled lnc408 was performed to seek the potential lnc408-associated proteins. SP3 was identified to bind with lnc408 in BCSCs (Fig. 4C, D), and the interaction of lnc408 with SP3 was further validated by immunoprecipitation (RIP; Fig. 4E). In addition, we confirmed that lnc408 knockdown promoted CBY1 mRNA (Fig. 4F) and protein (Fig. 4G) expression in BCSCs. Similarly, efficient silence of SP3 (Fig. S4C, D) led to the enhanced mRNA (Fig. 4H) and protein (Fig. 4I) levels of CBY1, which is consistent with the finding that SP3 could serve inhibitory transcription function27,28. These data suggest that lnc408 recruits SP3 to suppress the transcription of CBY1.
To investigate the role of CBY1 in the self-renewal of BCSC, ectopic CBY1 was efficiently transfected into BCSCs (Fig. S5A, B). As shown in Fig. 5A, CBY1 overexpression markedly reduced sphere number and their average diameter. Correspondingly, ectopic CBY1 significantly decreased SOX2, Nanog, and CD44, the representative breast cancer stemness marker expressions both in mRNA (Fig. S5C) and protein (Fig. 5B) levels, implying that CBY1 has the ability to inhibit the stemness of BCSCs. To further verify the function of CBY1 in stemness maintenance of BCSCs, CBY1 was transfected into non-BCSC with ectopic lnc408, and mammosphere formation and CSC-associated markers were assessed. As shown, the ectopic lnc408-induced mammosphere formation potentials (Fig. 5C) and stemness-related protein expressions (Fig. 5D, E) were notably decreased by CBY1. Besides, the expression of CBY1 was negatively related with the levels of CD44 and SOX2, two of known stemness biomarkers, in our cohort of clinical samples (Fig. S5D), these findings were further confirmed in large samples downloaded from Metabric database (Fig. 5F). Indeed, the breast cancer patients with high level of CBY1 had a higher overall survival in comparison with these with low level of CBY1 (Fig. 5G). Overall, lnc408 may involve in maintenance of BCSCs by inhibiting CBY1.
Lnc408-mediated decrease of CBY1 activates Wnt/β-catenin signaling
Subsequently, we wondered how lnc408-mediated CBY1 involves in regulating the stemness of BCSCs. By using bioinformatics analysis, we found CBY1 could interact with 14-3-3, AKT, β-catenin, and other proteins (Fig. 6A). It has been reported that the phosphorylation of CBY1 by ATK lead to an activated CBY1, which binds with 14-3-3 and β-catenin to form a ternary complex, and causing β-catenin to be pumped out of nuclei29. Agreeing with the previous findings, we confirmed that CBY1 could interact with 14-3-3 and β-catenin to form a ternary complex, mutation of CBY1 at 20 sites from serine into alanine (mutant CBY1 S20A) caused the disintegration of the complex (Fig. 6B) in 293T cells. Indeed, the trimer of CBY1/14-3-3/β-catenin was mainly existed in non-BCSCs rather than in BCSCs, which could be abrogated when either ectopic lnc408 was transfected into non-BCSC or CBY1 was silenced in non-BCSC (Fig. 6C, left panel). However, restoration of CBY1 expression or knockdown of lnc408 in BCSC facilitated the ternary complex formation (Fig. 6C, right panel). Consistently with the ternary complex formation in non-BCSCs or disintegration in BCSCs, higher level of cytoplasmic β-catenin than nuclear β-catenin was detected in non-BCSCs, the distribution of cytoplasmic or nuclear β-catenin in non-BCSC could be reversed by overexpression of lnc408 or knockdown of CBY1 (Fig. 6D, left panel); however, the lower level of cytoplasmic β-catenin and higher level of nuclear β-catenin in BCSCs could also be reversed by ectopic CBY1 or silence of lnc408 (Fig. 6D, right panel). Correspondingly, CBY1-mediated the ternary complex formation and nuclear β-catenin accumulation were closed with the change of mammosphere formation capability in BCSC derived from BT549 cells (Fig. 6E), in which ectopic CBY1 induced decrease of nuclear β-catenin blunted mammosphere formation, SKL2001 (an agonist of Wnt signaling) treatment rescued the spheres formation ability (Fig. 6E). In addition, previous study has found that stemness-related gene c-Myc and Klf4 were at downstream of Wnt/β-catenin signaling in bovine embryonic stem cells (ESCs)30. Similarly, the same trend of stemess-related gene expression, including c-Myc and Klf4 was observed in these BCSCs (Fig. 6F, G). These data demonstrate that lnc408-mediated decrease of CBY1 stimulates Wnt/β-catenin signaling and acts a pivotal role in stemness maintenance of BCSCs.
Lnc408/CBY1 axis facilitates BCSC tumorigenesis in vivo
Lastly, the effect of lnc408/CBY1 axis on tumorigenesis and tumor growth was assessed in vivo by injection of lnc408-knocked down BCSCs and lnc408-overexpressing non-BCSCs into female nude mice. As expectedly, in comparison with BCSC with high level of endogenous lnc408, knockdown of lnc408 notably decreased tumorigenesis and tumor growth; while ectopic lnc408 in the non-BCSCs significantly promoted tumorigenesis and increased tumor growth, and the lnc408-droved tumor initiation was severely blunted by reexpression of CBY1 in non-BCSC (Fig. 7A). The changes of c-MYC protein, another known stemness biomarker, were detected in the corresponding tumor tissues (Fig. 7B). Meanwhile, nuclear β-catenin protein, the downstream target of lnc408/CBY1 axis, was significantly increased in the tumor derived from injected BCSC (BCSC/shCtrl) compared to the tumor from injected non-BCSCs (non-BCSC/Vec), and it was reduced in tumor derived from lnc408-silenced BCSCs (Fig. 7C). The mice injected with lnc408-overexpressing non-BCSC had high rate of tumor initiation (Fig. 7A), which was compatible with the increased expression of nuclear β-catenin in the tumor (Fig. 7C); the mice injected with non-BCSC transfected with both ectopic lnc408 and CBY1 (non-BCSC/lnc408/CBY1) just had few tumor initiation (Fig. 7A), due to lowest nuclear β-catenin, which accompanied with high level of phosphorylated cytoplasm β-catenin, in the tumor (Fig. 7C), suggesting an essential role of lnc408/CBY1 axis in promoting tumor initiation of BCSCs via regulation of β-catenin signaling. To expand this finding, we assessed the lnc408 expression and the corresponding nuclear β-catenin and c-Myc protein levels in clinical tumor tissues. As shown in Fig. 7D, the enhanced nuclear β-catenin (n-β-catenin) and c-MYC (the representative stemness-related marker protein) accompanied with decreased phosphorylated cytoplasm β-catenin were detected in the randomly selected lnc408 high expressed tumors in comparison with the lnc408 low expressed tumors (Fig. 7D). In conclusion, in non-BCSCs, CBY1 forms ternary complex with 14-3-3 and β-catenin to result in the complex being bumped out of nucleus and the degradation of β-catenin; in BCSCs, lnc408 recruits SP3 to cause the transcription depression of CBY1, which prevents the formation of CBY1/14-3-3/β-catenin complex, and β-catenin is accumulated in the nucleus to promote the self-renewal and stemness of BCSCs (Fig. 7E).