Background

Colorectal cancer (CRC) is the third most prevalent malignant tumor and one of the leading causes of cancer-related death worldwide [1]. Despite progress in surgery, chemotherapy and radiotherapy, the prognosis of CRC is still poor due to its postoperative recurrence and metastasis. Therefore, it is urgent to explore the molecular mechanism of CRC and to develop new therapeutic targets.

Long noncoding RNAs (lncRNAs) are a class of transcribed RNAs with lengths greater than 200 nucleotides, and their protein coding ability is lost or restricted [2]. Recent studies have shown that lncRNAs play vital roles in regulating various biological processes, such as proliferation, differentiation, apoptosis, and chemoresistance [3]. Growing evidence has shown that the abnormal regulation or expression of lncRNAs is implicated in the tumorigenesis and progression of CRC [4, 5]. We have reported that some lncRNAs regulate CRC growth, metastasis and chemoresistance and may be potential prognostic biomarkers or therapeutic targets [6,7,8,9,10,11,12]. Recently, other groups also reported the regulatory roles of some lncRNAs, including ENO1-IT1 [13], LINC00460 [14], RAMS11 [15] and FLANC [16], in CRC development and progression. All these studies indicate the key regulatory roles of lncRNAs in CRC.

Small nucleolar RNAs (snoRNAs) are predominantly distributed in the nucleolus and play a role in guiding the sequence-specific chemical modification or processing of pre-ribosomal RNA [17]. As the host genes of snoRNAs, lncRNA small nucleolar RNA host genes (SNHGs) have been shown to be abnormally expressed in multiple cancers and regulate cell proliferation, metastasis, and chemoresistance [18, Full size image

To identify which region of SNHG17 binds to PES1, we constructed a series of SNHG17 deletion mutants based on the predicted secondary structure of SNHG17 in the AnnoLnc2 database (http://annolnc.gao-lab.org/index.php). RNA fragments were transcribed in vitro from these deletion mutant constructs and used for RNA pull-down assays. Western blotting analyses of PES1 in protein samples pulled down by these different SNHG17 constructs showed that RNA fragments with the 1–500 deletion nearly completely lost their ability to bind PES1, suggesting that the N-terminus of SNHG17 is essential for the binding of SNHG17 to PES1 (Fig. 3c). To further investigate which domain of PES1 accounts for its interaction with SNHG17, we performed RIP assays using a series of HA-tagged PES1 deletion mutants. As shown in Fig. 3d, the BRCT domain of PES1 is essential for the binding of PES1 to SNHG17 (Fig. 3d).

SNHG17 blocks PES1 ubiquitination and degradation

Given that SNHG17 interacts with the PES1 in CRC cells, we next investigated the molecular consequence of this association on PES1 expression. The results showed that the protein levels of PES1 were significantly increased in SNHG17-overexpressing HCT116 cells and decreased in SNHG17-silenced LoVo cells (Fig. 3e), whereas no obvious effect was observed on the mRNA levels of PES1 (Supplementary Fig. 5). These results suggest that SNHG17 can increase PES1 protein expression at the posttranscriptional level. To further validate this observation, we used the protein synthesis inhibitor cycloheximide (CHX) to evaluate the effect of SNHG17 on the protein stability of PES1. SNHG17 markedly increased the half-life of PES1 in HCT116 cells, whereas silencing SNHG17 expression reduced the half-life of PES1 degradation in LoVo cells (Fig. 3f). Moreover, treatment with the proteasome inhibitor MG132 attenuated the accumulation of endogenous PES1 in SNHG17-overexpressing cells and the reduction in PES1 expression in SNHG17 knockdown CRC cells (Fig. 3 g). These data suggest that SNHG17 interferes with PES1 degradation through the ubiquitin-proteasome system. Furthermore, the ubiquitination levels of PES1 significantly decreased in SNHG17-overexpressing cells (Fig. 3 h).

A recent study revealed that Trim23 is an E3 ligase for PES1 ubiquitination and degradation [26]. To investigate whether SNHG17 inhibits the effect of Trim23 on the degradation of PES1, we co-transfected SNHG17 and Trim23 expression vectors into CRC cells, and showed that Trim23 overexpression could decrease PES protein level, while SNHG17 overexpression could restore the decreased PES1 protein level caused by Trim23 overexpression (Fig. 3i). Next, we examined the effect of SNHG17 on the interaction between Trim23 and PES1 in CRC cells using IP assays. Indeed, SNHG17 knockdown notably increased this association in CRC cells, suggesting that SNHG17 increases the protein stability of PES1 by inhibiting Trim23-mediated ubiquitination of PES1 (Fig. 3j). We have confirmed that the BRCT domain of PES1 is crucial for the binding of PES1 to SNHG17 (Fig. 3d), which suggests that SNHG17 can compete with Trim23 to bind the BRCT domain of PES1, thereby inhibits Trim23-mediated ubiquitination and degradation of PES1. Therefore, we performed IP assays using a series of HA-tagged PES1 deletion mutants. As shown in Fig. 3 k, the BRCT domain of PES1 is also essential for the binding of PES1 to Trim23. Together, these results demonstrate that SNHG17 increases the protein levels of PES1 by binding PES1 and then inhibiting Trim23-mediated ubiquitination and degradation of PES1.

SNHG17 binds with miR-339-5p

Numerous studies have demonstrated that cytoplasmic lncRNAs post-transcriptionally regulate downstream genes by binding with microRNAs (miRNAs). We detected the subcellular localization of SNHG17 using RT-qPCR and FISH assays and demonstrated that SNHG17 was located in both the cytoplasm and nucleus of HCT116 and LoVo cells (Fig. 4a and Supplementary Fig. 6). We predicted miRNAs that could regulate SNHG17 using StarBase and RegRNA 2.0 tools and identified six candidate miRNAs (miR-339-5p, − 1913, − 3619-5p, − 3180-3p, − 3909, and − 5581-3p) (Fig. 4b). Subsequently, we confirmed that miR-339-5p and miR-3909 could bind to SNHG17 through luciferase assays, and miR-339-5p had the strongest inhibitory effect on SNHG17 (Fig. 4c). We also showed that miR-339-5p was significantly downregulated in CRC tissues from the TCGA CRC and GEO GSE30454 datasets (Supplementary Fig. 7). Therefore, miR-339-5p was chosen for further studies. We constructed luciferase reporter vectors of SNHG17 segments containing WT or MUT miR-339-5p binding sites. Luciferase reporter assays revealed that miR-339-5p significantly decreased the reporter activity of the SNHG17-WT plasmid, whereas no obvious changes were observed in the mutant group, indicating that SNHG17 could specifically interact with miR-339-5p (Fig. 4d). Furthermore, the RIP assay showed that both SNHG17 and miR-339-5p were enriched in Ago2-containing miRNA ribonucleoprotein complexes (miRNPs) (Fig. 4e).

Fig. 4
figure 4

SNHG17 upregulates FOSL2 by binding with miR-339-5p. a The subcellular localization of SNHG17 was detected by RT-qPCR and FISH in HCT116 and LoVo cell lines. b StarBase and RegRNA 2.0 were used to predict SNHG17-associated miRNAs. c The luciferase activity of SNHG17 after cotransfection with miRNAs was determined by dual luciferase assays. d The relative luciferase activity of cells cotransfected with SNHG17-WT or SNHG17-MUT and miR-339-5p mimic was determined in 293 T and HCT116 cells by luciferase reporter assays. e Cellular lysates from HCT116 and LoVo cells were used for RIP with Ago2 antibody and IgG antibody. The levels of SNHG17 and miR-339-5p were detected by RT-qPCR. f The StarBase and TargetScan databases were used to predict the targets of the SNHG17/miR-339-5p axis. g Luciferase reporter assays were used to determine the relative luciferase activity of FOSL2–3′ UTR-WT or FOSL2–3′ UTR-MUT after cotransfection with miR-339-5p mimics. h MiR-339-5p and pluc-FOSL2–3′ UTR-WT were cotransfected with SNHG17 to verify whether SNHG17 can function as a ceRNA of miR-339-5p. i The mRNA and protein levels of FOSL2 were determined in SNHG17-overexpressing or SNHG17-depleted CRC cells by RT-qPCR and western blotting. j The mRNA and protein levels of FOSL2 were determined in miR-339-5p-depleted or miR-339-5p-overexpressing CRC cells by RT-qPCR and western blotting. k-m CCK-8 (k), colony formation (l) and Transwell (m) assays were applied to detect the cell proliferation and migration abilities of HCT116 cells transfected with miR-339-5p and FOSL2 expression vector. n The mRNA expression levels of FOSL2 target genes were determined in HCT116 cells transfected with FOSL2 siRNA and/or SNHG17 expression plasmid by RT-qPCR.

FOSL2 is a functional target of SNHG17/miR-339-5p axis

Next, to identify the downstream target of the SNHG17/miR-339-5p axis, we analyzed the mRNA targets of the SNHG17-ceRNA network predicted by StarBase and the target genes of miR-339-5p predicted by TargetScan and identified that FOSL2 is a potential downstream target of the SNHG17/miR-339-5p axis in CRC (Fig. 4f). To further verify our speculation, we performed luciferase reporter assays in 293 T and HCT116 cells and showed that miR-339-5p significantly inhibited the reporter activity of the FOSL2 WT 3′ untranslated region (UTR) group but not that of the MUT group (Fig. 4 g). To further evaluate the relationships among SNHG17, miR-339-5p and FOSL2, pluc-FOSL2–3′ UTR-WT plasmids were cotransfected with miR-339-5p and SNHG17 into CRC cells for a luciferase assay, and the results indicated that ectopic SNHG17 expression could block the downregulation of luciferase activity induced by miR-339-5p (Fig. 4 h). Moreover, both the RNA and protein expression levels of FOSL2 were significantly increased in SNHG17-overexpressing HCT116 cells and decreased in SNHG17-depleted LoVo cells compared with their corresponding control cells (Fig. 4i). In addition, we revealed that the miR-339-5p inhibitor could increase the expression of FOSL2 compared with that in the control group, whereas ectopic miR-339-5p expression in CRC cells significantly decreased the expression of FOSL2 (Fig. 4j). Additional rescue assays confirmed that ectopic FOSL2 expression partly restored the cell proliferation and migration activities, which were decreased by miR-339-5p in CRC cells (Fig. 4 k-m). Moreover, SNHG17 overexpression partly restored the expression of the FOSL2 target genes (MYC, SNAI2 and SOX4) impaired by FOSL2 knockdown in HCT116 cells (Fig. 4n). All these results suggest that FOSL2 is a direct functional target of the SNHG17/miR-339-5p axis in CRC cells.

SNHG17 overexpression is driven by FOSL2

The mechanism mediating SNHG17 overexpression in CRC remains unclear. We scanned the binding sites of transcription factors on the regulatory element of SNHG17 by employing the JASPAR database. Interestingly, we observed several FOSL2 binding sites at the promoter of SNHG17, suggesting direct regulation of SNHG17 by FOSL2 (Fig. 5a). To validate our hypothesis, we knocked down the expression of FOSL2 in CRC cells with siRNAs and found that inhibiting FOSL2 expression significantly decreased the expression of SNHG17 (Fig. 5b and c). We then performed a luciferase assay to investigate whether FOSL2 directly regulates the transcription of SNHG17. We searched for the canonical binding site of FOSL2 predicted by JASPAR in the promoter region of SNHG17, and the three sites with the highest scores were selected for experimental verification. After mutating all three potential binding sites, luciferase assays suggested that the second binding site is the key site mediating the transcriptional regulation of SNHG17 by FOSL2 (Fig. 5d). To further substantiate this, we carried out additional luciferase reporter assays in FOSL2-depleted CRC cells. The results showed that FOSL2 knockdown reduced the activity of the WT SNHG17 promoter, which was abolished when site 2 was mutated (Fig. 5e). Moreover, ChIP assays confirmed the interaction between FOSL2 and putative binding site 2 in the SNHG17 promoter (Fig. 5f). We also examined the mRNA levels of FOSL2 in CRC tissues and revealed the upregulation of FOSL2 in CRC (Fig. 5 g). Moreover, Pearson correlation analysis showed that the expression of FOSL2 was positively correlated with that of SNHG17 in CRC tissues (Fig. 5 h, r = 0.4024, P = 0.0005). Together, these data demonstrate that FOSL2 directly drives SNHG17 overexpression in CRC by transcriptional activation.

Fig. 5
figure 5

FOSL2 binds to the promoter of SNHG17 and promotes its expression. a The JASPAR database was used to search the FOSL2 binding motif within the SNHG17 promoter. b RT-qPCR was used to verify the expression of SNHG17 and FOSL2 when FOSL2 was silenced. c The knockdown efficiency of FOSL2 was detected by western blotting. d The luciferase activity of the SNHG17 promoter (WT, #1-MUT, #2-MUT and #3-MUT) was tested via luciferase reporter assays when FOSL2 was overexpressed. Schematic depictions show the different luciferase reporter plasmids; blue represents the predicted binding sites between FOSL2 and the SNHG17 promoter; and red represents the mutant sites. e The luciferase activity of the SNHG17 promoter (WT and #2-MUT) was tested via luciferase reporter assays when FOSL2 was silenced. f ChIP assays were conducted to prove that FOSL2 could bind to the SNHG17 promoter in HCT116 cells. g The relative mRNA expression of FOSL2 was verified in cohort 1 CRC tissues and NCTs by RT-qPCR. h Bivariate correlation analysis between SNHG17 RNA expression and FOSL2 mRNA expression in cohort 1 CRC tissues.

PES1 and FOSL2 are upregulated in CRC and positively correlated with SNHG17 expression

To further confirm the relationships among SNHG17 and PES1 and FOSL2, the protein expression of PES1 and FOSL2 in CRC tissues was assessed using IHC. The results indicated that tumor tissues showed increased PES1 and FOSL2 expression compared with NCTs (Fig. 6a). Survival analyses revealed that high protein expression of PES1 or FOSL2 was closely related to poor survival (log rank = 5.446, P = 0.0196, Fig. 6b) (log rank = 5.657, P = 0.0174, Fig. 6c). Moreover, the protein expression of both PES1 and FOSL2 in human CRC tissues was positively correlated with the expression of SNHG17 (Fig. 6d and e). Meanwhile, the positive relationship between SNHG17 and PES1 (or FOSL2) levels was also observed in CRC xenografts in nude mice (Supplementary Fig. 8). The prognosis of CRC patients with high expression of SNHG17 and PES1 (FOSL2) was significantly worse than that of other CRC patients (log rank = 9.410, P = 0.0022, Fig. 6f) (log rank = 6.933, P = 0.0085, Fig. 6 g). We further showed that the protein levels of PES1 were positively correlated with those of FOSL2 in CRC tissues (r = 0.3256, P = 0.0002, Fig. 6 h), and high expression of both PES1 and FOSL2 in CRC tissues indicated a poor prognosis in CRC patients (log rank = 10.09, P = 0.0015, Fig. 6i). In summary, these data suggest that PES1 and FOSL2 are candidate downstream targets of SNHG17.

Fig. 6
figure 6

The protein expression of PES1 and FOSL2 is upregulated in CRC tissues. a Immunohistochemistry staining of PES1 and FOSL2 in CRC tissues and NCTs. b, c Kaplan-Meier survival analysis of CRC patients based on the protein expression scores of PES1 (b) or FOSL2 (c). d, e The relative expression of SNHG17 in two groups of CRC tissues with high and low protein levels of PES1 (d) or FOSL2 (e). f, g Combined influence of SNHG17 and PES1 (f) or FOSL2 (g) on the risk of CRC-related death using Kaplan-Meier survival analysis. h Bivariate correlation analysis between FOSL2 and PES1 protein expression in CRC tissues. i Combined influence of FOSL2 and PES1 on the risk of CRC-related death according to Kaplan-Meier survival analysis.

SNHG17 plays a carcinogenic role in CRC through PES1 and FOSL2

To investigate whether SNHG17 regulates CRC development and progression through PES1 and FOSL2, we designed rescue assays. It was confirmed that both PES1 and FOSL2 could inhibit cell proliferation in CRC cells, whereas the impact of SNHG17 on CRC cell proliferation was abolished by PES1 or FOSL2 knockdown (Fig. 7a-d). PES1 or FOSL2 knockdown also partly blocked the effects of SNHG17 overexpression on the migration of CRC cells (Fig. 7e and f). Collectively, we found that SNHG17 exerts tumor-promoting functions through PES1 and FOSL2 (Fig. 7 g).

Fig. 7
figure 7

SNHG17 exerts its carcinogenic role in CRC through PES1 and FOSL2. a, b CCK-8 assays were used to detect the rescuing effects of SNHG17 overexpression on cell proliferation ability in PES1 (a)- or FOSL2 (b)-silenced HCT116 cells. c, d Colony formation assays were used to detect the rescuing effects of SNHG17 overexpression on colony formation ability in PES1 (c)- or FOSL2 (d)-silenced HCT116 cells. e, f Transwell assays were performed to evaluate the rescuing effects of SNHG17 overexpression on cell migration in PES1 (e)- or FOSL2 (f)-silenced HCT116 cells. g Integrated model depicting the oncogenic effects and mechanism of SNHG17 in terms of CRC growth and metastasis.