RETRACTED ARTICLE: Long non-coding RNA 91H regulates IGF2 expression by interacting with IGF2BP2 and promotes tumorigenesis in colorectal cancer

We, the Editors and Publisher of the journal Artificial Cells, Nanomedicine, and Biotechnology, have retracted the following article: Tianyi Gao, Xiangxiang Liu, Bangshun He, Yuqin Pan & Shukui Wang (2020) Long non-coding RNA 91H regulates IGF2 expression by interacting with IGF2BP2 and promotes tumorigenesis in colorectal cancer, Artificial Cells, Nanomedicine, and Biotechnology, 48:1, 664-671, DOI: 10.1080/21691401.2020.1727491 Since publication, the authors noticed duplication in Figure 1A and 4A. As this error directly impacts the reported results and conclusions the authors alerted the issue to the Editor and Publisher to request correction. The authors acknowledged that they had made mistakes and the Editor and Publisher found there to be fundamental flaws present within the requested corrections, as well as the rest of the paper. The authors have agreed to the retraction of this article. We have been informed in our decision-making by our policy on publishing ethics and integrity and the COPE guidelines on retractions. The retracted article will remain online to maintain the scholarly record, but it will be digitally watermarked on each page as ‘Retracted’.


Introduction
Colorectal cancer (CRC) is one of the leading causes of cancerrelated morbidity and mortality [1]. Increasing evidence indicates that long non-coding RNAs (lncRNAs) regulate diverse cellular processes, including cell growth, differentiation, apoptosis and cancer progression [2]. Therefore, a better understanding of lncRNAs function in CRC progression would be a new area for finding potential prognostic biomarkers which could further improve CRC diagnosis and therapy. lncRNA 91H, a 119.392-kb long noncoding antisense transcripts, is located at the position of the H19/IGF2 locus (accession number NC_000011.9). It has been proved to be consistently overexpressed in a number of human cancer tissues, including CRC which is significantly associated with adverse clinical characteristics and plays an important role in tumour development [3][4][5]. Unfortunately, the exact role of 91H in tumorigenesis is poorly described.
Recently, the aberrant activation of the insulin-like growth factor (IGF) system has been reported to be associated with poor prognosis and therapeutic resistance in several cancers, including CRC [6][7][8]. Particularly, IGF2, a neighbouring gene expressed from the paternal allele, encodes for a protein of 7.5 kDa involved in embryonic growth, is commonly upregulated in CRC [9,10]. Early studies had showed that lncRNA 91H could up-regulate IGF2 expression by epigenetic modifications which could further promote breast cancer development [5,11].
In light of these observations, lncRNA 91H might also play a critical role in the development of CRC through IGF2 modification. However, few studies have investigated the potential mechanism between lncRNA 91H and IGF2 expression in CRC. So, we performed this study to deeply explore the biological functions of lncRNA 91H in CRC development and its mechanism in IGF2 expression.

Clinical samples
CRC tumour tissues and matched adjacent normal tissues were included from 28 patients who underwent surgery at Nanjing First Hospital Affiliated to Nanjing Medical University, between 2011 and 2014 [3]. Particularly, adjacent normal tissues were taken 5-10 cm away from the tumour tissues. Before DNA and RNA extraction, all the specimens were frozen in liquid nitrogen after surgery and stored at -80 C. Patients received chemotherapy or radiotherapy at preoperation were excluded. Clinical information of these patients was collected including age, sex, TNM stage, tumour grade and family history of disease.

Scratch wound assay
HCT-8 cells were separated into three groups: siRNA-91H, siRNA-NC and blank control (cells treated with nothing) and cultured in six-well plates for 48 h. Then, a 20 lL pipette tip was used to create wounds in cells. Free-floating cells and debris were removed using PBS. Then, the cells were observed at 0 and 48 h. Each experiment was repeated in triplicate. The results were recorded by wound healing percentage (ImageJ, Bethesda, MD, USA) [12]. Wound healing percentage¼(original area-area after 48 h)/original area.

Migration and invasion assays
Migration and invasion assays were performed with Transwell chambers [3,12]. HCT-8 cells were separated into three groups: siRNA-91H, siRNA-NC and blank control. Then, cells (1 Â 10 5 ) were cultured in the upper chamber of transwell plates with serum-free DMEM (2 lg/lL) or the upper transwell chambers coated with Matrigel for invasion assay, in a 24well format with 6 mm diameters. Six hundred microlitres DMEM with 5% FBS were added to the bottom chamber. Culturing for 24 h, cells were fixed with methanol and stained with Wright's stain. The numbers of migrating cells were counted in five random areas under a light microscope and the average value of five fields was recorded.

Monodansylcadaverine (MDC) staining of autophagic vacuoles
Cells were treated with 1 M of TAM dissolved in Me2SO/EtOH (1:1, v/v) or incubated for four days with 20 M PDMP in the absence of TAM for four days. Then, 100 nM Myrio dissolved in methanol was added with TAM in the first group. An MDC stock solution (0.1 M in Me2SO) was applied to cells for 30 min at 37 C by diluting 1:1000 in DMEM. Washing with phosphate-buffered saline, the cell autophagy particles were finally dyed green fluorescent and examined under fluorescence microscopy (Axioplan, Zeiss, Oberkochen, Germany).

Immunofluorescence (IF)
Cells (5 Â 10 5 ) treated with or without siRNAs were seeded on sterile coverslips in fully supplemented keratinocyte media and allowed to adhere for approximately 4 h. After washing with PBS twice, cells were fixed with fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Then, fixed cells were incubated with antibodies IGF2 (1:50; R&D Systems Inc., Minneapolis, MN), FITC or Rhodamine B-conjugated anti-rabbit secondary antibody (1:100; Sigma, Oakville, Canada) and cellular nuclei were stained with DAPI [13].

5-aza-CdR treatment
The demethylating agent 5-aza-CdR (Sigma, St. Louis, MO, USA) was diluted in DMSO at 2.5 mM, 5 mM and 10 mM and kept frozen at -20 C as our previous study [13]. Twenty-four hours later, Hct-8 cells (5 Â 10 2 ) seeded in six-well plates were incubated with 5-aza-CdR and replaced with fresh medium with 5-aza-CdR every 24 h for a period of 48 h.

RNA-binding protein immunoprecipitation
RNA-binding protein immunoprecipitation (RIP) was performed with the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA). Following the protocols, CRC cells were treated with RIP lysis buffer. Next, clear supernatant and IGF2BP2 antibody (or immunoglobulin G (IgG)/SNRNP70 control) were mixed to perform the immunoprecipitation. After washing, RNAs binding to IGF2BP2 were eluted and quantified. RT-PCR and realtime PCR were performed to examine whether certain RNAs were co-immunoprecipitated with the IGF2BP2 antibody.

RNA pulldown and mass spectrometry analysis
As reported in our previous study [3,12]

Statistical analysis
Differences in 91H expression, IGF2 expression, and comparison of continuous data were analysed using an independent t-test or ANOVA test, and categorical data were analyzed by chi-square test. All analyses were performed using SPSS 18.0 for Windows (SPSS, Chicago, IL, USA) and p values of less than .05 were considered significant.

Results
The expression of lncRNA 91H in CRC By performing qRT-PCR, the expression of lncRNA 91H in cell lines HCT-8, DLD-1 and HCT-116 was significantly higher than FHC with p < .05 (Supplemental Figure 1). Then, lncRNA 91H expression of 28 tumour tissues and matched adjacent normal tissues was also compared which showed that the expression of lncRNA 91H in patients tumour tissues (1.41 ± 1.34) was significantly higher than adjacent normal tissues (0.94 ± 0.60, p < .05). Furthermore, the results were compared with patients' clinicopathological features which are shown in Table 1. Significant differences were not found in the comparison with age, sex, TNM, grade and family history. But we found that patients with high lncRNA 91H expression usually had a high risk in tumour metastasis.

Biological functions of lncRNA 91H in cells
The expression level of LncRNA 91H was detected higher in HCT-8 cells than the other cells (Supplemental Figure 1). Hence, HCT-8 cells were chosen to explore the potential biological functions of lncRNA 91H. After transfection with siRNA-91H or siRNA-NC for 48 h, lncRNA 91H was effectively silenced in HCT-8 cells (p < .05, Supplemental Figure 2). Afterward, a scratch wound assay was first performed to evaluate the effect of lncRNA 91H on cell motility. Cells treated with siRNA-91H resulted in significant decreased motility and wound recovery than cells transfected with negative control (Figure 1(A)). Then, migration and invasion assays were performed to further study the effect of lncRNA 91H on CRC cells. The results demonstrated that the migration and the invasiveness were both significantly decreased in transfected cells with siRNA-91H, compared with negative control cells (p < .05, Figure 1(B)). These findings both indicated that lncRNA 91H might be closely correlated with proliferation, migration, and invasion of CRC cell lines.
Moreover, MDC was used to detect the autophagy in CRC cells. The results showed that the autophagy was significantly inhibited in transfected cells with siRNA-91H, compared with negative control (p < .05, Figure 2(A)). Moreover, the expression of mTOR was also tested in the study which further demonstrated cells with lncRNA 91H silence usually had a higher mTOR expression than the other groups (p < .05, Figure 2(B)).

lncRNA 91H increased IGF2 expression correlated with IGF2BP2
CRC cells was first treated with 5-aza-CdR as our previous study [13]. However, lncRNA 91H and IGF2 expression did not show any significant differences in cells treated with different 5-aza-CdR concentration which showed that IGF2 ICR methylation might not play an important role in lncRNA 91H and  Figure 3). Then, we conducted lncRNA 91H interference in Hct-8 and DLD-1 cell lines. By Western blot, we found that the IGF2 expression both significantly decreased in siRNA-91H groups compared with siRNA-NC and blank groups (p < .05, Figure 3(A)). Similarly, the results of IF also showed that Hct-8 and DLD-1 cells treated with siRNA-91H both had a lower IGF2 expression than the other groups which confirmed the Western blot results before (Figure 3(B)). Then, IGF2BP2 was also silenced in Hct-8 and DLD-1 cell lines which demonstrated that though lncRNA 91H expression was not changed in all the groups, IGF2 expression was significantly lower in siRNA-IGF2BP2 groups than siRNA-NC and blank groups by Western blot (p < .05, Figure 4(A)). By the way, IF was also performed which declared that cells treated with siRNA-IGF2BP2 had a lower IGF2 expression than the others (Figure 4(B)). Furthermore, scratch wound, migration and invasion assays were also performed in cells treated with IGF2BP2 siRNA which showed that cells transfected with IGF2BP2 siRNA had a decline in cell motility, migration and invasion compared with negative controls (p < .05, Supplementary Figure 4). In addition, the siRNA IGF2 was used to guarantee the test effect. As shown in Supplementary Figure 6, we found that IGF2 expression significantly decreased after treating with siRNA IGF2 which was consistent with the results of Figures 2  and 3.

lncRNA 91H interacted with IGF2BP2 in vitro
Then, RNA pull down mass spectrometry analysis was used to screen the potential factors involved in lncRNA 91H regulation which demonstrated that IGF2BP2 protein highly combined with lncRNA 91H in CRC cells ( Figure 5(A)). Then, RIP was performed later to further determine the interaction

R E T R A C T E D
between lncRNA 91H and IGF2BP2 which showed that lncRNA 91H also highly combined with IGF2BP2 protein compared with IgG negative control and SNRNP70, a positive control ( Figure 5(B)).

Discussion
Though overexpressed lncRNA 91H had been reported in lots of tumours. Few studies have described the potential mechanism of lncRNA91H in cancer. In this study, we demonstrated for the first time, that up-regulated lncRNA 91H usually leads to a high IGF2 expression by interacting with IGF2BP2 which would promote CRC progression and cause a high risk in tumour metastasis. Nowadays, lncRNAs had become a hot topic based on their emergence as novel regulators in the cancer paradigm. Emerging investigations indicated that most lncRNAs might regulate cancer progression by completely combining with miRNA to produce ceRNA or promote protein stabilization at the post-transcript level [14]. However, recent studies reported that some lncRNAs could directly interact with functional domains of signalling proteins to suppress cancer metastasis [15]. Consistent with the view, our results also found that lncRNA 91H could participate in tumorigenesis by directly regulating IGF2 protein expression through interacting with IGF2BP2 protein which might be a new way in CRC lncRNA study. Our early study found that IGF2 ICR methylation might play an important role in IGF2 and lncRNA 91H expression in oesophageal squamous cell carcinoma [13]. However, in this study, the correlation was not found in CRC cells which might be due to their different pathological type.
The IGF2 mRNA binding protein (IGF2BP or IMP) family is comprised of three related members: IGF2BP1, IGF2BP2 and IGF2BP3 [16]. As the foetal transcripts of IGF2, IGF2BP family is normally expressed in adults' tissues but become reactive in cancer [17]. The unregulated expression of them had been demonstrated in many studies, especially IGF2BP2 which had been pointed to a tumorigenic role in individual cancer types, often with a poor prognosis [18][19][20]. In this study, we similarly found that cells with IGF2BP2 silence usually caused a low IGF2 expression which confirmed the association that IGF2BP2 played an important role in IGF2 expression. Then by RNA pull down and RIP, a strong interaction had been found between lncRNA 91H and IGF2BP2. Furthermore, RNA interference was conducted which also proved that lncRNA 91H and IGF2BP2 both participated in IGF2 protein expression.
By the way, early study had proved that IGF2 could bind to IGF1R which would further activate the PI3K/AKT and mTOR pathways so that to mediate diverse cellular functions, such as growth, proliferation, apoptosis, differentiation, cell adhesion and motility [21]. In this study, cells treated with siRNA-91H usually had a decreased proliferation, migration and invasion which might be due to the decreased regulation of lncRNA 91H on IGF2 expression that lead to the inactivation of PI3K/AKT and mTOR pathways.
In this study, we fortunately found lncRNA 91H high expression was significantly related to CRC metastasis and CRC cells' motility, migration and invasion. The epithelial-mesenchymal transition (EMT) in cancer had been reported to play an essential role in metastasis [22]. Early studies had demonstrated that EMT and autophagy signalling, an evolutionarily conserved lysosome-dependent catabolic process which degraded cells' components in order to recycle substrates to exert optimally and adapt to tough circumstances, were linked to one another [23,24]. Early studies had reported that through IGF1R, IGF2 could enhance tumour autophagy which played an important role in tumour progression [25]. As a result, we also observed a significantly decreased autophagy since lncRNA 91H was silenced in CRC cells and also found a association between lncRNA 91H and mTOR expression, a autophagy inhibiting protein, which suggested that lncRNA 91H might also affect tumour autophagy by regulating IGF2 and mTOR expression.
In early study, HOTS, an antisense transcript to H19, had been reported by Onyango and Feinberg [26]. To figure out whether HOTS affected IGF2 expression, as previous study [11], the siRNA used against 91H was out of the HOTS sequence. Fortunately, significant down-regulate IGF2 expression was observed which might indicate the observed sh91Hinduced phenotype was HOTS-independent.

Conclusions
In summary, lncRNA 91H was highly expressed in CRC cells and tumour tissues. Patients with high lncRNA 91H expression usually had a high risk in tumour metastasis which might be due to its effect on tumour proliferation, migration, invasion and autophagy. Moreover, our results suggested that lncRNA 91H promotes IGF2 expression by interacting with IGF2BP2 which would provide a new strategy in finding potential CRC diagnostic biomarkers and therapeutic targets.